diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,120419 +1,5619 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.6949017643232586,
+  "epoch": 0.016160506147052525,
   "eval_steps": 500,
-  "global_step": 172000,
+  "global_step": 8000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 4.040126536763131e-05,
-      "grad_norm": 37380620.0,
-      "learning_rate": 2e-08,
-      "loss": 1265640.2,
+      "epoch": 2.0200632683815657e-05,
+      "grad_norm": 316.38043212890625,
+      "learning_rate": 2e-09,
+      "loss": 26.2272,
       "step": 10
     },
     {
-      "epoch": 8.080253073526263e-05,
-      "grad_norm": 9634068.0,
-      "learning_rate": 4e-08,
-      "loss": 1485744.3,
+      "epoch": 4.040126536763131e-05,
+      "grad_norm": 576.4369506835938,
+      "learning_rate": 4e-09,
+      "loss": 32.9983,
       "step": 20
     },
     {
-      "epoch": 0.00012120379610289395,
-      "grad_norm": 14280229.0,
-      "learning_rate": 6e-08,
-      "loss": 1357241.2,
+      "epoch": 6.060189805144697e-05,
+      "grad_norm": 472.15478515625,
+      "learning_rate": 6e-09,
+      "loss": 23.2158,
       "step": 30
     },
     {
-      "epoch": 0.00016160506147052525,
-      "grad_norm": 47433768.0,
-      "learning_rate": 8e-08,
-      "loss": 1486355.4,
+      "epoch": 8.080253073526263e-05,
+      "grad_norm": 20.11992835998535,
+      "learning_rate": 8e-09,
+      "loss": 19.9305,
       "step": 40
     },
     {
-      "epoch": 0.00020200632683815657,
-      "grad_norm": 36763048.0,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 1302607.4,
+      "epoch": 0.00010100316341907829,
+      "grad_norm": 766.17529296875,
+      "learning_rate": 1e-08,
+      "loss": 35.6484,
       "step": 50
     },
     {
-      "epoch": 0.0002424075922057879,
-      "grad_norm": 58413324.0,
-      "learning_rate": 1.2e-07,
-      "loss": 1164601.1,
+      "epoch": 0.00012120379610289395,
+      "grad_norm": 649.330810546875,
+      "learning_rate": 1.2e-08,
+      "loss": 23.5938,
       "step": 60
     },
     {
-      "epoch": 0.0002828088575734192,
-      "grad_norm": 20137730.0,
-      "learning_rate": 1.4e-07,
-      "loss": 1635228.2,
+      "epoch": 0.0001414044287867096,
+      "grad_norm": 638.9619750976562,
+      "learning_rate": 1.4000000000000001e-08,
+      "loss": 20.7426,
       "step": 70
     },
     {
-      "epoch": 0.0003232101229410505,
-      "grad_norm": 110655184.0,
-      "learning_rate": 1.6e-07,
-      "loss": 1525543.8,
+      "epoch": 0.00016160506147052525,
+      "grad_norm": 703.6967163085938,
+      "learning_rate": 1.6e-08,
+      "loss": 33.9524,
       "step": 80
     },
     {
-      "epoch": 0.0003636113883086818,
-      "grad_norm": 14134144.0,
-      "learning_rate": 1.8e-07,
-      "loss": 1310257.0,
+      "epoch": 0.0001818056941543409,
+      "grad_norm": 785.22998046875,
+      "learning_rate": 1.8000000000000002e-08,
+      "loss": 29.1029,
       "step": 90
     },
     {
-      "epoch": 0.00040401265367631315,
-      "grad_norm": 19299640.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 851063.5,
+      "epoch": 0.00020200632683815657,
+      "grad_norm": 676.3937377929688,
+      "learning_rate": 2e-08,
+      "loss": 26.1458,
       "step": 100
     },
     {
-      "epoch": 0.00044441391904394446,
-      "grad_norm": 17030668.0,
-      "learning_rate": 2.2e-07,
-      "loss": 1405910.8,
+      "epoch": 0.00022220695952197223,
+      "grad_norm": 246.762451171875,
+      "learning_rate": 2.2000000000000002e-08,
+      "loss": 25.4453,
       "step": 110
     },
     {
-      "epoch": 0.0004848151844115758,
-      "grad_norm": 5774159.0,
-      "learning_rate": 2.4e-07,
-      "loss": 1021530.2,
+      "epoch": 0.0002424075922057879,
+      "grad_norm": 1033.1109619140625,
+      "learning_rate": 2.4e-08,
+      "loss": 38.0849,
       "step": 120
     },
     {
-      "epoch": 0.000525216449779207,
-      "grad_norm": 39249404.0,
-      "learning_rate": 2.6e-07,
-      "loss": 1118213.8,
+      "epoch": 0.0002626082248896035,
+      "grad_norm": 724.1307983398438,
+      "learning_rate": 2.6e-08,
+      "loss": 42.023,
       "step": 130
     },
     {
-      "epoch": 0.0005656177151468384,
-      "grad_norm": 3952896.25,
-      "learning_rate": 2.8e-07,
-      "loss": 802054.8,
+      "epoch": 0.0002828088575734192,
+      "grad_norm": 493.44427490234375,
+      "learning_rate": 2.8000000000000003e-08,
+      "loss": 27.2123,
       "step": 140
     },
     {
-      "epoch": 0.0006060189805144697,
-      "grad_norm": 17492202.0,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 782337.9,
+      "epoch": 0.00030300949025723485,
+      "grad_norm": 665.4328002929688,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 34.0098,
       "step": 150
     },
     {
-      "epoch": 0.000646420245882101,
-      "grad_norm": 14565722.0,
-      "learning_rate": 3.2e-07,
-      "loss": 762118.95,
+      "epoch": 0.0003232101229410505,
+      "grad_norm": 844.9791259765625,
+      "learning_rate": 3.2e-08,
+      "loss": 34.3665,
       "step": 160
     },
     {
-      "epoch": 0.0006868215112497323,
-      "grad_norm": 12661834.0,
-      "learning_rate": 3.4e-07,
-      "loss": 617497.6,
+      "epoch": 0.00034341075562486617,
+      "grad_norm": 32.71566390991211,
+      "learning_rate": 3.4e-08,
+      "loss": 33.4008,
       "step": 170
     },
     {
-      "epoch": 0.0007272227766173637,
-      "grad_norm": 4945727.5,
-      "learning_rate": 3.6e-07,
-      "loss": 360572.95,
+      "epoch": 0.0003636113883086818,
+      "grad_norm": 1014.241455078125,
+      "learning_rate": 3.6000000000000005e-08,
+      "loss": 37.7545,
       "step": 180
     },
     {
-      "epoch": 0.000767624041984995,
-      "grad_norm": 3134309.5,
-      "learning_rate": 3.8e-07,
-      "loss": 247792.15,
+      "epoch": 0.0003838120209924975,
+      "grad_norm": 883.66259765625,
+      "learning_rate": 3.8e-08,
+      "loss": 27.2009,
       "step": 190
     },
     {
-      "epoch": 0.0008080253073526263,
-      "grad_norm": 23188752.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 302759.75,
+      "epoch": 0.00040401265367631315,
+      "grad_norm": 984.9612426757812,
+      "learning_rate": 4e-08,
+      "loss": 16.6661,
       "step": 200
     },
     {
-      "epoch": 0.0008484265727202576,
-      "grad_norm": 2640093.75,
-      "learning_rate": 4.2e-07,
-      "loss": 177911.875,
+      "epoch": 0.0004242132863601288,
+      "grad_norm": 553.194091796875,
+      "learning_rate": 4.2e-08,
+      "loss": 36.3195,
       "step": 210
     },
     {
-      "epoch": 0.0008888278380878889,
-      "grad_norm": 3475973.0,
-      "learning_rate": 4.4e-07,
-      "loss": 124657.5875,
+      "epoch": 0.00044441391904394446,
+      "grad_norm": 313.7496032714844,
+      "learning_rate": 4.4000000000000004e-08,
+      "loss": 16.8871,
       "step": 220
     },
     {
-      "epoch": 0.0009292291034555202,
-      "grad_norm": 7612417.5,
-      "learning_rate": 4.6e-07,
-      "loss": 133348.3,
+      "epoch": 0.0004646145517277601,
+      "grad_norm": 525.6804809570312,
+      "learning_rate": 4.6e-08,
+      "loss": 17.3601,
       "step": 230
     },
     {
-      "epoch": 0.0009696303688231516,
-      "grad_norm": 3544448.5,
-      "learning_rate": 4.8e-07,
-      "loss": 108178.025,
+      "epoch": 0.0004848151844115758,
+      "grad_norm": 206.4374237060547,
+      "learning_rate": 4.8e-08,
+      "loss": 34.4888,
       "step": 240
     },
     {
-      "epoch": 0.0010100316341907828,
-      "grad_norm": 1792657.375,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 147530.7625,
+      "epoch": 0.0005050158170953914,
+      "grad_norm": 665.89453125,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 40.5891,
       "step": 250
     },
     {
-      "epoch": 0.001050432899558414,
-      "grad_norm": 807344.25,
-      "learning_rate": 5.2e-07,
-      "loss": 46811.725,
+      "epoch": 0.000525216449779207,
+      "grad_norm": 1490.542724609375,
+      "learning_rate": 5.2e-08,
+      "loss": 26.4835,
       "step": 260
     },
     {
-      "epoch": 0.0010908341649260454,
-      "grad_norm": 9978717.0,
-      "learning_rate": 5.4e-07,
-      "loss": 46902.9469,
+      "epoch": 0.0005454170824630227,
+      "grad_norm": 157.42401123046875,
+      "learning_rate": 5.400000000000001e-08,
+      "loss": 10.1466,
       "step": 270
     },
     {
-      "epoch": 0.0011312354302936767,
-      "grad_norm": 1081468.125,
-      "learning_rate": 5.6e-07,
-      "loss": 61798.675,
+      "epoch": 0.0005656177151468384,
+      "grad_norm": 896.1552124023438,
+      "learning_rate": 5.6000000000000005e-08,
+      "loss": 46.9781,
       "step": 280
     },
     {
-      "epoch": 0.001171636695661308,
-      "grad_norm": 836054.125,
-      "learning_rate": 5.8e-07,
-      "loss": 18199.025,
+      "epoch": 0.000585818347830654,
+      "grad_norm": 1364.9993896484375,
+      "learning_rate": 5.8e-08,
+      "loss": 40.9419,
       "step": 290
     },
     {
-      "epoch": 0.0012120379610289394,
-      "grad_norm": 713206.125,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 16179.9375,
+      "epoch": 0.0006060189805144697,
+      "grad_norm": 1216.1474609375,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 30.2287,
       "step": 300
     },
     {
-      "epoch": 0.0012524392263965707,
-      "grad_norm": 736075.8125,
-      "learning_rate": 6.2e-07,
-      "loss": 18567.8047,
+      "epoch": 0.0006262196131982854,
+      "grad_norm": 328.39678955078125,
+      "learning_rate": 6.2e-08,
+      "loss": 38.7195,
       "step": 310
     },
     {
-      "epoch": 0.001292840491764202,
-      "grad_norm": 524556.5,
-      "learning_rate": 6.4e-07,
-      "loss": 18508.9297,
+      "epoch": 0.000646420245882101,
+      "grad_norm": 433.4118957519531,
+      "learning_rate": 6.4e-08,
+      "loss": 22.1306,
       "step": 320
     },
     {
-      "epoch": 0.0013332417571318333,
-      "grad_norm": 74135.0703125,
-      "learning_rate": 6.6e-07,
-      "loss": 7819.6453,
+      "epoch": 0.0006666208785659167,
+      "grad_norm": 420.328857421875,
+      "learning_rate": 6.600000000000001e-08,
+      "loss": 27.7195,
       "step": 330
     },
     {
-      "epoch": 0.0013736430224994647,
-      "grad_norm": 94151.4921875,
-      "learning_rate": 6.8e-07,
-      "loss": 6291.5781,
+      "epoch": 0.0006868215112497323,
+      "grad_norm": 185.5102081298828,
+      "learning_rate": 6.8e-08,
+      "loss": 37.9056,
       "step": 340
     },
     {
-      "epoch": 0.001414044287867096,
-      "grad_norm": 17262.64453125,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 2254.5939,
+      "epoch": 0.000707022143933548,
+      "grad_norm": 337.02044677734375,
+      "learning_rate": 7e-08,
+      "loss": 22.7152,
       "step": 350
     },
     {
-      "epoch": 0.0014544455532347273,
-      "grad_norm": 6977.82470703125,
-      "learning_rate": 7.2e-07,
-      "loss": 806.7898,
+      "epoch": 0.0007272227766173637,
+      "grad_norm": 670.1768188476562,
+      "learning_rate": 7.200000000000001e-08,
+      "loss": 28.3785,
       "step": 360
     },
     {
-      "epoch": 0.0014948468186023586,
-      "grad_norm": 720.22705078125,
-      "learning_rate": 7.400000000000001e-07,
-      "loss": 1413.3216,
+      "epoch": 0.0007474234093011793,
+      "grad_norm": 475.4079895019531,
+      "learning_rate": 7.400000000000001e-08,
+      "loss": 22.1121,
       "step": 370
     },
     {
-      "epoch": 0.00153524808396999,
-      "grad_norm": 12234.6259765625,
-      "learning_rate": 7.6e-07,
-      "loss": 569.8974,
+      "epoch": 0.000767624041984995,
+      "grad_norm": 607.6908569335938,
+      "learning_rate": 7.6e-08,
+      "loss": 31.8589,
       "step": 380
     },
     {
-      "epoch": 0.0015756493493376213,
-      "grad_norm": 12985.419921875,
-      "learning_rate": 7.8e-07,
-      "loss": 645.2221,
+      "epoch": 0.0007878246746688106,
+      "grad_norm": 800.4915161132812,
+      "learning_rate": 7.8e-08,
+      "loss": 21.862,
       "step": 390
     },
     {
-      "epoch": 0.0016160506147052526,
-      "grad_norm": 17890.517578125,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 556.4382,
+      "epoch": 0.0008080253073526263,
+      "grad_norm": 1185.45849609375,
+      "learning_rate": 8e-08,
+      "loss": 35.8806,
       "step": 400
     },
     {
-      "epoch": 0.001656451880072884,
-      "grad_norm": 1073.23046875,
-      "learning_rate": 8.200000000000001e-07,
-      "loss": 579.9454,
+      "epoch": 0.000828225940036442,
+      "grad_norm": 452.5521240234375,
+      "learning_rate": 8.200000000000002e-08,
+      "loss": 17.0217,
       "step": 410
     },
     {
-      "epoch": 0.0016968531454405152,
-      "grad_norm": 946.015380859375,
-      "learning_rate": 8.4e-07,
-      "loss": 339.6989,
+      "epoch": 0.0008484265727202576,
+      "grad_norm": 125.98954010009766,
+      "learning_rate": 8.4e-08,
+      "loss": 29.4705,
       "step": 420
     },
     {
-      "epoch": 0.0017372544108081465,
-      "grad_norm": 919.689697265625,
-      "learning_rate": 8.6e-07,
-      "loss": 544.2018,
+      "epoch": 0.0008686272054040733,
+      "grad_norm": 301.3433532714844,
+      "learning_rate": 8.6e-08,
+      "loss": 22.8971,
       "step": 430
     },
     {
-      "epoch": 0.0017776556761757779,
-      "grad_norm": 1076.61962890625,
-      "learning_rate": 8.8e-07,
-      "loss": 477.4546,
+      "epoch": 0.0008888278380878889,
+      "grad_norm": 578.7197875976562,
+      "learning_rate": 8.800000000000001e-08,
+      "loss": 20.5219,
       "step": 440
     },
     {
-      "epoch": 0.0018180569415434092,
-      "grad_norm": 6896.8623046875,
-      "learning_rate": 9e-07,
-      "loss": 551.458,
+      "epoch": 0.0009090284707717046,
+      "grad_norm": 1337.2423095703125,
+      "learning_rate": 9e-08,
+      "loss": 38.1059,
       "step": 450
     },
     {
-      "epoch": 0.0018584582069110405,
-      "grad_norm": 1263.240478515625,
-      "learning_rate": 9.2e-07,
-      "loss": 478.671,
+      "epoch": 0.0009292291034555202,
+      "grad_norm": 971.8609619140625,
+      "learning_rate": 9.2e-08,
+      "loss": 28.7212,
       "step": 460
     },
     {
-      "epoch": 0.0018988594722786718,
-      "grad_norm": 6494.14501953125,
-      "learning_rate": 9.400000000000001e-07,
-      "loss": 551.3243,
+      "epoch": 0.0009494297361393359,
+      "grad_norm": 635.0576171875,
+      "learning_rate": 9.400000000000001e-08,
+      "loss": 28.1718,
       "step": 470
     },
     {
-      "epoch": 0.0019392607376463031,
-      "grad_norm": 1367.335205078125,
-      "learning_rate": 9.6e-07,
-      "loss": 566.3516,
+      "epoch": 0.0009696303688231516,
+      "grad_norm": 347.902587890625,
+      "learning_rate": 9.6e-08,
+      "loss": 35.6753,
       "step": 480
     },
     {
-      "epoch": 0.0019796620030139342,
-      "grad_norm": 2059.615478515625,
-      "learning_rate": 9.8e-07,
-      "loss": 440.9915,
+      "epoch": 0.0009898310015069671,
+      "grad_norm": 804.568115234375,
+      "learning_rate": 9.8e-08,
+      "loss": 52.5154,
       "step": 490
     },
     {
-      "epoch": 0.0020200632683815656,
-      "grad_norm": 1089.5028076171875,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 450.5991,
+      "epoch": 0.0010100316341907828,
+      "grad_norm": 357.3641662597656,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 23.8221,
       "step": 500
     },
     {
-      "epoch": 0.002060464533749197,
-      "grad_norm": 1123.5267333984375,
-      "learning_rate": 1.0200000000000002e-06,
-      "loss": 450.4043,
+      "epoch": 0.0010302322668745984,
+      "grad_norm": 695.1265869140625,
+      "learning_rate": 1.0200000000000001e-07,
+      "loss": 29.7152,
       "step": 510
     },
     {
-      "epoch": 0.002100865799116828,
-      "grad_norm": 1164.44677734375,
-      "learning_rate": 1.04e-06,
-      "loss": 401.3136,
+      "epoch": 0.001050432899558414,
+      "grad_norm": 466.5853271484375,
+      "learning_rate": 1.04e-07,
+      "loss": 23.3024,
       "step": 520
     },
     {
-      "epoch": 0.0021412670644844595,
-      "grad_norm": 1139.4189453125,
-      "learning_rate": 1.06e-06,
-      "loss": 510.3279,
+      "epoch": 0.0010706335322422298,
+      "grad_norm": 725.106201171875,
+      "learning_rate": 1.0600000000000001e-07,
+      "loss": 20.0807,
       "step": 530
     },
     {
-      "epoch": 0.002181668329852091,
-      "grad_norm": 1584.498779296875,
-      "learning_rate": 1.08e-06,
-      "loss": 276.7529,
+      "epoch": 0.0010908341649260454,
+      "grad_norm": 500.51708984375,
+      "learning_rate": 1.0800000000000001e-07,
+      "loss": 16.9731,
       "step": 540
     },
     {
-      "epoch": 0.002222069595219722,
-      "grad_norm": 13004.5224609375,
-      "learning_rate": 1.1e-06,
-      "loss": 517.8619,
+      "epoch": 0.001111034797609861,
+      "grad_norm": 643.19580078125,
+      "learning_rate": 1.1e-07,
+      "loss": 35.556,
       "step": 550
     },
     {
-      "epoch": 0.0022624708605873535,
-      "grad_norm": 6897.6220703125,
-      "learning_rate": 1.12e-06,
-      "loss": 371.9735,
+      "epoch": 0.0011312354302936767,
+      "grad_norm": 606.3302001953125,
+      "learning_rate": 1.1200000000000001e-07,
+      "loss": 27.9096,
       "step": 560
     },
     {
-      "epoch": 0.002302872125954985,
-      "grad_norm": 5952.30859375,
-      "learning_rate": 1.14e-06,
-      "loss": 321.4458,
+      "epoch": 0.0011514360629774924,
+      "grad_norm": 455.76116943359375,
+      "learning_rate": 1.1400000000000001e-07,
+      "loss": 27.992,
       "step": 570
     },
     {
-      "epoch": 0.002343273391322616,
-      "grad_norm": 1368.4573974609375,
-      "learning_rate": 1.16e-06,
-      "loss": 444.6263,
+      "epoch": 0.001171636695661308,
+      "grad_norm": 890.6718139648438,
+      "learning_rate": 1.16e-07,
+      "loss": 20.3737,
       "step": 580
     },
     {
-      "epoch": 0.0023836746566902474,
-      "grad_norm": 6530.33154296875,
-      "learning_rate": 1.18e-06,
-      "loss": 576.2317,
+      "epoch": 0.0011918373283451237,
+      "grad_norm": 521.8313598632812,
+      "learning_rate": 1.1800000000000001e-07,
+      "loss": 21.4314,
       "step": 590
     },
     {
-      "epoch": 0.0024240759220578788,
-      "grad_norm": 918.2885131835938,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 319.5849,
+      "epoch": 0.0012120379610289394,
+      "grad_norm": 526.7328491210938,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 35.9307,
       "step": 600
     },
     {
-      "epoch": 0.00246447718742551,
-      "grad_norm": 5709.6923828125,
-      "learning_rate": 1.2200000000000002e-06,
-      "loss": 544.9418,
+      "epoch": 0.001232238593712755,
+      "grad_norm": 491.2201843261719,
+      "learning_rate": 1.22e-07,
+      "loss": 15.4924,
       "step": 610
     },
     {
-      "epoch": 0.0025048784527931414,
-      "grad_norm": 767.517822265625,
-      "learning_rate": 1.24e-06,
-      "loss": 475.3013,
+      "epoch": 0.0012524392263965707,
+      "grad_norm": 1669.0023193359375,
+      "learning_rate": 1.24e-07,
+      "loss": 31.0749,
       "step": 620
     },
     {
-      "epoch": 0.0025452797181607727,
-      "grad_norm": 1049.8690185546875,
-      "learning_rate": 1.26e-06,
-      "loss": 505.3646,
+      "epoch": 0.0012726398590803864,
+      "grad_norm": 747.734619140625,
+      "learning_rate": 1.2600000000000002e-07,
+      "loss": 27.6358,
       "step": 630
     },
     {
-      "epoch": 0.002585680983528404,
-      "grad_norm": 6113.7568359375,
-      "learning_rate": 1.28e-06,
-      "loss": 371.3924,
+      "epoch": 0.001292840491764202,
+      "grad_norm": 415.5186767578125,
+      "learning_rate": 1.28e-07,
+      "loss": 34.9266,
       "step": 640
     },
     {
-      "epoch": 0.0026260822488960354,
-      "grad_norm": 961.7456665039062,
-      "learning_rate": 1.3e-06,
-      "loss": 434.1364,
+      "epoch": 0.0013130411244480177,
+      "grad_norm": 812.7051391601562,
+      "learning_rate": 1.3e-07,
+      "loss": 28.9554,
       "step": 650
     },
     {
-      "epoch": 0.0026664835142636667,
-      "grad_norm": 827.18310546875,
-      "learning_rate": 1.32e-06,
-      "loss": 325.7226,
+      "epoch": 0.0013332417571318333,
+      "grad_norm": 391.8579406738281,
+      "learning_rate": 1.3200000000000002e-07,
+      "loss": 16.3211,
       "step": 660
     },
     {
-      "epoch": 0.002706884779631298,
-      "grad_norm": 800.2735595703125,
-      "learning_rate": 1.34e-06,
-      "loss": 362.5384,
+      "epoch": 0.001353442389815649,
+      "grad_norm": 569.1922607421875,
+      "learning_rate": 1.34e-07,
+      "loss": 40.6097,
       "step": 670
     },
     {
-      "epoch": 0.0027472860449989293,
-      "grad_norm": 2117.720703125,
-      "learning_rate": 1.36e-06,
-      "loss": 459.5714,
+      "epoch": 0.0013736430224994647,
+      "grad_norm": 784.5740356445312,
+      "learning_rate": 1.36e-07,
+      "loss": 38.8948,
       "step": 680
     },
     {
-      "epoch": 0.0027876873103665606,
-      "grad_norm": 11787.2021484375,
-      "learning_rate": 1.3800000000000001e-06,
-      "loss": 452.4481,
+      "epoch": 0.0013938436551832803,
+      "grad_norm": 284.9936218261719,
+      "learning_rate": 1.3800000000000002e-07,
+      "loss": 28.3477,
       "step": 690
     },
     {
-      "epoch": 0.002828088575734192,
-      "grad_norm": 1027.6024169921875,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 370.987,
+      "epoch": 0.001414044287867096,
+      "grad_norm": 518.3870239257812,
+      "learning_rate": 1.4e-07,
+      "loss": 25.8091,
       "step": 700
     },
     {
-      "epoch": 0.0028684898411018233,
-      "grad_norm": 2432.72802734375,
-      "learning_rate": 1.4200000000000002e-06,
-      "loss": 396.6002,
+      "epoch": 0.0014342449205509116,
+      "grad_norm": 448.25848388671875,
+      "learning_rate": 1.4200000000000003e-07,
+      "loss": 21.6013,
       "step": 710
     },
     {
-      "epoch": 0.0029088911064694546,
-      "grad_norm": 4263.763671875,
-      "learning_rate": 1.44e-06,
-      "loss": 514.6806,
+      "epoch": 0.0014544455532347273,
+      "grad_norm": 175.99990844726562,
+      "learning_rate": 1.4400000000000002e-07,
+      "loss": 30.1821,
       "step": 720
     },
     {
-      "epoch": 0.002949292371837086,
-      "grad_norm": 1061.5069580078125,
-      "learning_rate": 1.46e-06,
-      "loss": 367.6588,
+      "epoch": 0.001474646185918543,
+      "grad_norm": 863.377685546875,
+      "learning_rate": 1.46e-07,
+      "loss": 51.8515,
       "step": 730
     },
     {
-      "epoch": 0.0029896936372047172,
-      "grad_norm": 7566.0791015625,
-      "learning_rate": 1.4800000000000002e-06,
-      "loss": 387.3527,
+      "epoch": 0.0014948468186023586,
+      "grad_norm": 175.85592651367188,
+      "learning_rate": 1.4800000000000003e-07,
+      "loss": 25.4575,
       "step": 740
     },
     {
-      "epoch": 0.0030300949025723486,
-      "grad_norm": 1594.1578369140625,
-      "learning_rate": 1.5e-06,
-      "loss": 286.798,
+      "epoch": 0.0015150474512861743,
+      "grad_norm": 640.6859741210938,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 26.2666,
       "step": 750
     },
     {
-      "epoch": 0.00307049616793998,
-      "grad_norm": 1798.3486328125,
-      "learning_rate": 1.52e-06,
-      "loss": 265.2009,
+      "epoch": 0.00153524808396999,
+      "grad_norm": 811.0674438476562,
+      "learning_rate": 1.52e-07,
+      "loss": 24.3395,
       "step": 760
     },
     {
-      "epoch": 0.003110897433307611,
-      "grad_norm": 1887.8646240234375,
-      "learning_rate": 1.54e-06,
-      "loss": 511.3708,
+      "epoch": 0.0015554487166538056,
+      "grad_norm": 628.3569946289062,
+      "learning_rate": 1.5400000000000003e-07,
+      "loss": 25.5079,
       "step": 770
     },
     {
-      "epoch": 0.0031512986986752425,
-      "grad_norm": 1046.9554443359375,
-      "learning_rate": 1.56e-06,
-      "loss": 330.1199,
+      "epoch": 0.0015756493493376213,
+      "grad_norm": 796.8837890625,
+      "learning_rate": 1.56e-07,
+      "loss": 22.1897,
       "step": 780
     },
     {
-      "epoch": 0.003191699964042874,
-      "grad_norm": 14393.11328125,
-      "learning_rate": 1.5800000000000003e-06,
-      "loss": 459.7904,
+      "epoch": 0.001595849982021437,
+      "grad_norm": 634.3350830078125,
+      "learning_rate": 1.5800000000000004e-07,
+      "loss": 13.3538,
       "step": 790
     },
     {
-      "epoch": 0.003232101229410505,
-      "grad_norm": 1148.2391357421875,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 297.0352,
+      "epoch": 0.0016160506147052526,
+      "grad_norm": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 40.9726,
       "step": 800
     },
     {
-      "epoch": 0.0032725024947781365,
-      "grad_norm": 1655.5443115234375,
-      "learning_rate": 1.62e-06,
-      "loss": 380.9321,
+      "epoch": 0.0016362512473890682,
+      "grad_norm": 891.0134887695312,
+      "learning_rate": 1.62e-07,
+      "loss": 33.7623,
       "step": 810
     },
     {
-      "epoch": 0.003312903760145768,
-      "grad_norm": 1723.20947265625,
-      "learning_rate": 1.6400000000000002e-06,
-      "loss": 356.5419,
+      "epoch": 0.001656451880072884,
+      "grad_norm": 124.5013198852539,
+      "learning_rate": 1.6400000000000004e-07,
+      "loss": 32.6563,
       "step": 820
     },
     {
-      "epoch": 0.003353305025513399,
-      "grad_norm": 962.674560546875,
-      "learning_rate": 1.6600000000000002e-06,
-      "loss": 244.5568,
+      "epoch": 0.0016766525127566996,
+      "grad_norm": 557.0490112304688,
+      "learning_rate": 1.66e-07,
+      "loss": 11.069,
       "step": 830
     },
     {
-      "epoch": 0.0033937062908810304,
-      "grad_norm": 1860.3314208984375,
-      "learning_rate": 1.68e-06,
-      "loss": 314.3494,
+      "epoch": 0.0016968531454405152,
+      "grad_norm": 370.4107971191406,
+      "learning_rate": 1.68e-07,
+      "loss": 18.7746,
       "step": 840
     },
     {
-      "epoch": 0.0034341075562486618,
-      "grad_norm": 677.5274047851562,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 389.2205,
+      "epoch": 0.0017170537781243309,
+      "grad_norm": 470.0067443847656,
+      "learning_rate": 1.7000000000000001e-07,
+      "loss": 53.0731,
       "step": 850
     },
     {
-      "epoch": 0.003474508821616293,
-      "grad_norm": 847.7564697265625,
-      "learning_rate": 1.72e-06,
-      "loss": 421.029,
+      "epoch": 0.0017372544108081465,
+      "grad_norm": 250.24252319335938,
+      "learning_rate": 1.72e-07,
+      "loss": 19.3808,
       "step": 860
     },
     {
-      "epoch": 0.0035149100869839244,
-      "grad_norm": 1153.310791015625,
-      "learning_rate": 1.7399999999999999e-06,
-      "loss": 338.7876,
+      "epoch": 0.0017574550434919622,
+      "grad_norm": 1099.9942626953125,
+      "learning_rate": 1.74e-07,
+      "loss": 50.3866,
       "step": 870
     },
     {
-      "epoch": 0.0035553113523515557,
-      "grad_norm": 702.7095947265625,
-      "learning_rate": 1.76e-06,
-      "loss": 375.8682,
+      "epoch": 0.0017776556761757779,
+      "grad_norm": 599.7510375976562,
+      "learning_rate": 1.7600000000000001e-07,
+      "loss": 23.1743,
       "step": 880
     },
     {
-      "epoch": 0.003595712617719187,
-      "grad_norm": 898.8328857421875,
-      "learning_rate": 1.7800000000000001e-06,
-      "loss": 437.3662,
+      "epoch": 0.0017978563088595935,
+      "grad_norm": 401.0577087402344,
+      "learning_rate": 1.78e-07,
+      "loss": 18.497,
       "step": 890
     },
     {
-      "epoch": 0.0036361138830868184,
-      "grad_norm": 1392.5733642578125,
-      "learning_rate": 1.8e-06,
-      "loss": 327.5625,
+      "epoch": 0.0018180569415434092,
+      "grad_norm": 1774.6864013671875,
+      "learning_rate": 1.8e-07,
+      "loss": 39.7275,
       "step": 900
     },
     {
-      "epoch": 0.0036765151484544497,
-      "grad_norm": 1176.8975830078125,
-      "learning_rate": 1.8200000000000002e-06,
-      "loss": 331.6111,
+      "epoch": 0.0018382575742272248,
+      "grad_norm": 950.1702880859375,
+      "learning_rate": 1.8200000000000002e-07,
+      "loss": 34.1336,
       "step": 910
     },
     {
-      "epoch": 0.003716916413822081,
-      "grad_norm": 844.1708374023438,
-      "learning_rate": 1.84e-06,
-      "loss": 462.9411,
+      "epoch": 0.0018584582069110405,
+      "grad_norm": 681.1715087890625,
+      "learning_rate": 1.84e-07,
+      "loss": 22.6729,
       "step": 920
     },
     {
-      "epoch": 0.0037573176791897123,
-      "grad_norm": 1110.1351318359375,
-      "learning_rate": 1.86e-06,
-      "loss": 291.5723,
+      "epoch": 0.0018786588395948562,
+      "grad_norm": 548.5640869140625,
+      "learning_rate": 1.86e-07,
+      "loss": 17.3724,
       "step": 930
     },
     {
-      "epoch": 0.0037977189445573436,
-      "grad_norm": 1204.96826171875,
-      "learning_rate": 1.8800000000000002e-06,
-      "loss": 453.21,
+      "epoch": 0.0018988594722786718,
+      "grad_norm": 570.0945434570312,
+      "learning_rate": 1.8800000000000002e-07,
+      "loss": 40.4859,
       "step": 940
     },
     {
-      "epoch": 0.003838120209924975,
-      "grad_norm": 788.923095703125,
-      "learning_rate": 1.9e-06,
-      "loss": 338.4679,
+      "epoch": 0.0019190601049624875,
+      "grad_norm": 411.9521789550781,
+      "learning_rate": 1.9e-07,
+      "loss": 19.4746,
       "step": 950
     },
     {
-      "epoch": 0.0038785214752926063,
-      "grad_norm": 828.4852294921875,
-      "learning_rate": 1.92e-06,
-      "loss": 304.8583,
+      "epoch": 0.0019392607376463031,
+      "grad_norm": 781.8963623046875,
+      "learning_rate": 1.92e-07,
+      "loss": 54.7843,
       "step": 960
     },
     {
-      "epoch": 0.003918922740660237,
-      "grad_norm": 3126.61181640625,
-      "learning_rate": 1.94e-06,
-      "loss": 368.0714,
+      "epoch": 0.0019594613703301186,
+      "grad_norm": 296.8689270019531,
+      "learning_rate": 1.9400000000000002e-07,
+      "loss": 29.0034,
       "step": 970
     },
     {
-      "epoch": 0.0039593240060278685,
-      "grad_norm": 3534.13232421875,
-      "learning_rate": 1.96e-06,
-      "loss": 321.7589,
+      "epoch": 0.0019796620030139342,
+      "grad_norm": 842.5889282226562,
+      "learning_rate": 1.96e-07,
+      "loss": 37.4423,
       "step": 980
     },
     {
-      "epoch": 0.0039997252713955,
-      "grad_norm": 676.08251953125,
-      "learning_rate": 1.98e-06,
-      "loss": 336.8801,
+      "epoch": 0.00199986263569775,
+      "grad_norm": 948.009033203125,
+      "learning_rate": 1.9800000000000003e-07,
+      "loss": 30.0773,
       "step": 990
     },
     {
-      "epoch": 0.004040126536763131,
-      "grad_norm": 1422.3154296875,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 310.1541,
+      "epoch": 0.0020200632683815656,
+      "grad_norm": 215.3368377685547,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 26.6014,
       "step": 1000
     },
     {
-      "epoch": 0.0040805278021307624,
-      "grad_norm": 754.2792358398438,
-      "learning_rate": 2.02e-06,
-      "loss": 294.2774,
+      "epoch": 0.0020402639010653812,
+      "grad_norm": 494.67578125,
+      "learning_rate": 2.02e-07,
+      "loss": 21.6492,
       "step": 1010
     },
     {
-      "epoch": 0.004120929067498394,
-      "grad_norm": 1239.5504150390625,
-      "learning_rate": 2.0400000000000004e-06,
-      "loss": 379.9203,
+      "epoch": 0.002060464533749197,
+      "grad_norm": 234.77099609375,
+      "learning_rate": 2.0400000000000003e-07,
+      "loss": 24.3645,
       "step": 1020
     },
     {
-      "epoch": 0.004161330332866025,
-      "grad_norm": 1107.9769287109375,
-      "learning_rate": 2.06e-06,
-      "loss": 394.3968,
+      "epoch": 0.0020806651664330125,
+      "grad_norm": 327.6997985839844,
+      "learning_rate": 2.0600000000000002e-07,
+      "loss": 24.5405,
       "step": 1030
     },
     {
-      "epoch": 0.004201731598233656,
-      "grad_norm": 747.5755615234375,
-      "learning_rate": 2.08e-06,
-      "loss": 357.4806,
+      "epoch": 0.002100865799116828,
+      "grad_norm": 611.9917602539062,
+      "learning_rate": 2.08e-07,
+      "loss": 24.577,
       "step": 1040
     },
     {
-      "epoch": 0.004242132863601288,
-      "grad_norm": 897.203125,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 376.0113,
+      "epoch": 0.002121066431800644,
+      "grad_norm": 133.8150177001953,
+      "learning_rate": 2.1000000000000003e-07,
+      "loss": 33.8902,
       "step": 1050
     },
     {
-      "epoch": 0.004282534128968919,
-      "grad_norm": 3561.832763671875,
-      "learning_rate": 2.12e-06,
-      "loss": 385.5077,
+      "epoch": 0.0021412670644844595,
+      "grad_norm": 163.73504638671875,
+      "learning_rate": 2.1200000000000002e-07,
+      "loss": 17.783,
       "step": 1060
     },
     {
-      "epoch": 0.00432293539433655,
-      "grad_norm": 2549.35205078125,
-      "learning_rate": 2.14e-06,
-      "loss": 448.231,
+      "epoch": 0.002161467697168275,
+      "grad_norm": 193.5026397705078,
+      "learning_rate": 2.14e-07,
+      "loss": 20.8054,
       "step": 1070
     },
     {
-      "epoch": 0.004363336659704182,
-      "grad_norm": 746.3321533203125,
-      "learning_rate": 2.16e-06,
-      "loss": 308.5054,
+      "epoch": 0.002181668329852091,
+      "grad_norm": 624.8909301757812,
+      "learning_rate": 2.1600000000000003e-07,
+      "loss": 10.6967,
       "step": 1080
     },
     {
-      "epoch": 0.004403737925071813,
-      "grad_norm": 4975.05615234375,
-      "learning_rate": 2.1800000000000003e-06,
-      "loss": 604.6913,
+      "epoch": 0.0022018689625359065,
+      "grad_norm": 430.49560546875,
+      "learning_rate": 2.1800000000000002e-07,
+      "loss": 45.0397,
       "step": 1090
     },
     {
-      "epoch": 0.004444139190439444,
-      "grad_norm": 865.9938354492188,
-      "learning_rate": 2.2e-06,
-      "loss": 239.4941,
+      "epoch": 0.002222069595219722,
+      "grad_norm": 1073.7850341796875,
+      "learning_rate": 2.2e-07,
+      "loss": 42.1204,
       "step": 1100
     },
     {
-      "epoch": 0.004484540455807076,
-      "grad_norm": 1072.074462890625,
-      "learning_rate": 2.2200000000000003e-06,
-      "loss": 401.7543,
+      "epoch": 0.002242270227903538,
+      "grad_norm": 486.751708984375,
+      "learning_rate": 2.2200000000000003e-07,
+      "loss": 23.1064,
       "step": 1110
     },
     {
-      "epoch": 0.004524941721174707,
-      "grad_norm": 1182.0032958984375,
-      "learning_rate": 2.24e-06,
-      "loss": 330.9987,
+      "epoch": 0.0022624708605873535,
+      "grad_norm": 1003.8533935546875,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 43.3564,
       "step": 1120
     },
     {
-      "epoch": 0.004565342986542338,
-      "grad_norm": 814.4903564453125,
-      "learning_rate": 2.26e-06,
-      "loss": 379.418,
+      "epoch": 0.002282671493271169,
+      "grad_norm": 572.0302734375,
+      "learning_rate": 2.26e-07,
+      "loss": 23.8053,
       "step": 1130
     },
     {
-      "epoch": 0.00460574425190997,
-      "grad_norm": 2526.64501953125,
-      "learning_rate": 2.28e-06,
-      "loss": 389.4854,
+      "epoch": 0.002302872125954985,
+      "grad_norm": 447.3976745605469,
+      "learning_rate": 2.2800000000000003e-07,
+      "loss": 18.1645,
       "step": 1140
     },
     {
-      "epoch": 0.004646145517277601,
-      "grad_norm": 1358.009033203125,
-      "learning_rate": 2.3e-06,
-      "loss": 432.4065,
+      "epoch": 0.0023230727586388005,
+      "grad_norm": 931.8524780273438,
+      "learning_rate": 2.3000000000000002e-07,
+      "loss": 30.1086,
       "step": 1150
     },
     {
-      "epoch": 0.004686546782645232,
-      "grad_norm": 1503.1463623046875,
-      "learning_rate": 2.32e-06,
-      "loss": 417.3679,
+      "epoch": 0.002343273391322616,
+      "grad_norm": 875.0267333984375,
+      "learning_rate": 2.32e-07,
+      "loss": 44.5649,
       "step": 1160
     },
     {
-      "epoch": 0.0047269480480128636,
-      "grad_norm": 2745.59619140625,
-      "learning_rate": 2.34e-06,
-      "loss": 339.9085,
+      "epoch": 0.0023634740240064318,
+      "grad_norm": 683.9933471679688,
+      "learning_rate": 2.3400000000000003e-07,
+      "loss": 28.6991,
       "step": 1170
     },
     {
-      "epoch": 0.004767349313380495,
-      "grad_norm": 636.4331665039062,
-      "learning_rate": 2.36e-06,
-      "loss": 324.7576,
+      "epoch": 0.0023836746566902474,
+      "grad_norm": 1272.7320556640625,
+      "learning_rate": 2.3600000000000002e-07,
+      "loss": 53.9565,
       "step": 1180
     },
     {
-      "epoch": 0.004807750578748126,
-      "grad_norm": 1743.3262939453125,
-      "learning_rate": 2.38e-06,
-      "loss": 337.6445,
+      "epoch": 0.002403875289374063,
+      "grad_norm": 540.0521240234375,
+      "learning_rate": 2.3800000000000004e-07,
+      "loss": 23.6139,
       "step": 1190
     },
     {
-      "epoch": 0.0048481518441157575,
-      "grad_norm": 912.2630004882812,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 342.137,
+      "epoch": 0.0024240759220578788,
+      "grad_norm": 578.9132690429688,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 22.7089,
       "step": 1200
     },
     {
-      "epoch": 0.004888553109483389,
-      "grad_norm": 1142.1710205078125,
-      "learning_rate": 2.42e-06,
-      "loss": 280.1062,
+      "epoch": 0.0024442765547416944,
+      "grad_norm": 382.25244140625,
+      "learning_rate": 2.42e-07,
+      "loss": 33.4145,
       "step": 1210
     },
     {
-      "epoch": 0.00492895437485102,
-      "grad_norm": 981.51806640625,
-      "learning_rate": 2.4400000000000004e-06,
-      "loss": 291.3894,
+      "epoch": 0.00246447718742551,
+      "grad_norm": 1396.89697265625,
+      "learning_rate": 2.44e-07,
+      "loss": 35.8553,
       "step": 1220
     },
     {
-      "epoch": 0.0049693556402186515,
-      "grad_norm": 1795.322265625,
-      "learning_rate": 2.46e-06,
-      "loss": 349.2701,
+      "epoch": 0.0024846778201093257,
+      "grad_norm": 507.25848388671875,
+      "learning_rate": 2.46e-07,
+      "loss": 39.8554,
       "step": 1230
     },
     {
-      "epoch": 0.005009756905586283,
-      "grad_norm": 1389.8359375,
-      "learning_rate": 2.48e-06,
-      "loss": 363.471,
+      "epoch": 0.0025048784527931414,
+      "grad_norm": 236.29849243164062,
+      "learning_rate": 2.48e-07,
+      "loss": 20.3849,
       "step": 1240
     },
     {
-      "epoch": 0.005050158170953914,
-      "grad_norm": 1739.7503662109375,
-      "learning_rate": 2.5e-06,
-      "loss": 414.4333,
+      "epoch": 0.002525079085476957,
+      "grad_norm": 766.2069702148438,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 35.435,
       "step": 1250
     },
     {
-      "epoch": 0.0050905594363215454,
-      "grad_norm": 1108.7142333984375,
-      "learning_rate": 2.52e-06,
-      "loss": 376.2317,
+      "epoch": 0.0025452797181607727,
+      "grad_norm": 92.45375061035156,
+      "learning_rate": 2.5200000000000003e-07,
+      "loss": 33.7958,
       "step": 1260
     },
     {
-      "epoch": 0.005130960701689177,
-      "grad_norm": 1234.8914794921875,
-      "learning_rate": 2.54e-06,
-      "loss": 432.4711,
+      "epoch": 0.0025654803508445884,
+      "grad_norm": 206.84521484375,
+      "learning_rate": 2.54e-07,
+      "loss": 19.2633,
       "step": 1270
     },
     {
-      "epoch": 0.005171361967056808,
-      "grad_norm": 1600.1077880859375,
-      "learning_rate": 2.56e-06,
-      "loss": 273.5047,
+      "epoch": 0.002585680983528404,
+      "grad_norm": 1618.84521484375,
+      "learning_rate": 2.56e-07,
+      "loss": 36.3744,
       "step": 1280
     },
     {
-      "epoch": 0.005211763232424439,
-      "grad_norm": 5998.88134765625,
-      "learning_rate": 2.5800000000000003e-06,
-      "loss": 417.7149,
+      "epoch": 0.0026058816162122197,
+      "grad_norm": 322.9664001464844,
+      "learning_rate": 2.58e-07,
+      "loss": 19.4322,
       "step": 1290
     },
     {
-      "epoch": 0.005252164497792071,
-      "grad_norm": 1022.2550659179688,
-      "learning_rate": 2.6e-06,
-      "loss": 337.5613,
+      "epoch": 0.0026260822488960354,
+      "grad_norm": 479.2985534667969,
+      "learning_rate": 2.6e-07,
+      "loss": 44.4554,
       "step": 1300
     },
     {
-      "epoch": 0.005292565763159702,
-      "grad_norm": 821.97998046875,
-      "learning_rate": 2.6200000000000003e-06,
-      "loss": 355.1617,
+      "epoch": 0.002646282881579851,
+      "grad_norm": 839.0703125,
+      "learning_rate": 2.6200000000000004e-07,
+      "loss": 25.7475,
       "step": 1310
     },
     {
-      "epoch": 0.005332967028527333,
-      "grad_norm": 0.0,
-      "learning_rate": 2.64e-06,
-      "loss": 441.1887,
+      "epoch": 0.0026664835142636667,
+      "grad_norm": 496.9524841308594,
+      "learning_rate": 2.6400000000000003e-07,
+      "loss": 14.5702,
       "step": 1320
     },
     {
-      "epoch": 0.005373368293894965,
-      "grad_norm": 1180.22607421875,
-      "learning_rate": 2.66e-06,
-      "loss": 349.4191,
+      "epoch": 0.0026866841469474823,
+      "grad_norm": 206.1656951904297,
+      "learning_rate": 2.66e-07,
+      "loss": 30.1201,
       "step": 1330
     },
     {
-      "epoch": 0.005413769559262596,
-      "grad_norm": 1288.874755859375,
-      "learning_rate": 2.68e-06,
-      "loss": 358.9353,
+      "epoch": 0.002706884779631298,
+      "grad_norm": 688.9887084960938,
+      "learning_rate": 2.68e-07,
+      "loss": 20.7835,
       "step": 1340
     },
     {
-      "epoch": 0.005454170824630227,
-      "grad_norm": 858.6513671875,
-      "learning_rate": 2.7e-06,
-      "loss": 298.1967,
+      "epoch": 0.0027270854123151137,
+      "grad_norm": 173.92941284179688,
+      "learning_rate": 2.7e-07,
+      "loss": 38.9963,
       "step": 1350
     },
     {
-      "epoch": 0.005494572089997859,
-      "grad_norm": 826.8380737304688,
-      "learning_rate": 2.72e-06,
-      "loss": 266.4493,
+      "epoch": 0.0027472860449989293,
+      "grad_norm": 416.7894287109375,
+      "learning_rate": 2.72e-07,
+      "loss": 40.8383,
       "step": 1360
     },
     {
-      "epoch": 0.00553497335536549,
-      "grad_norm": 1527.0496826171875,
-      "learning_rate": 2.74e-06,
-      "loss": 385.1624,
+      "epoch": 0.002767486677682745,
+      "grad_norm": 602.2549438476562,
+      "learning_rate": 2.7400000000000004e-07,
+      "loss": 22.1494,
       "step": 1370
     },
     {
-      "epoch": 0.005575374620733121,
-      "grad_norm": 1727.4271240234375,
-      "learning_rate": 2.7600000000000003e-06,
-      "loss": 409.4935,
+      "epoch": 0.0027876873103665606,
+      "grad_norm": 1489.0169677734375,
+      "learning_rate": 2.7600000000000004e-07,
+      "loss": 30.0528,
       "step": 1380
     },
     {
-      "epoch": 0.005615775886100753,
-      "grad_norm": 1242.2200927734375,
-      "learning_rate": 2.78e-06,
-      "loss": 423.3242,
+      "epoch": 0.0028078879430503763,
+      "grad_norm": 410.69769287109375,
+      "learning_rate": 2.7800000000000003e-07,
+      "loss": 36.4594,
       "step": 1390
     },
     {
-      "epoch": 0.005656177151468384,
-      "grad_norm": 784.5136108398438,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 378.621,
+      "epoch": 0.002828088575734192,
+      "grad_norm": 99.08312225341797,
+      "learning_rate": 2.8e-07,
+      "loss": 19.1275,
       "step": 1400
     },
     {
-      "epoch": 0.005696578416836015,
-      "grad_norm": 686.701416015625,
-      "learning_rate": 2.82e-06,
-      "loss": 366.9333,
+      "epoch": 0.0028482892084180076,
+      "grad_norm": 579.4361572265625,
+      "learning_rate": 2.82e-07,
+      "loss": 23.6028,
       "step": 1410
     },
     {
-      "epoch": 0.0057369796822036466,
-      "grad_norm": 2546.25390625,
-      "learning_rate": 2.8400000000000003e-06,
-      "loss": 372.8342,
+      "epoch": 0.0028684898411018233,
+      "grad_norm": 932.9219970703125,
+      "learning_rate": 2.8400000000000005e-07,
+      "loss": 33.0755,
       "step": 1420
     },
     {
-      "epoch": 0.005777380947571278,
-      "grad_norm": 1681.410400390625,
-      "learning_rate": 2.86e-06,
-      "loss": 383.352,
+      "epoch": 0.002888690473785639,
+      "grad_norm": 736.8807373046875,
+      "learning_rate": 2.8600000000000005e-07,
+      "loss": 33.5045,
       "step": 1430
     },
     {
-      "epoch": 0.005817782212938909,
-      "grad_norm": 1604.5748291015625,
-      "learning_rate": 2.88e-06,
-      "loss": 390.5337,
+      "epoch": 0.0029088911064694546,
+      "grad_norm": 1126.778076171875,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 33.3801,
       "step": 1440
     },
     {
-      "epoch": 0.0058581834783065405,
-      "grad_norm": 642.487060546875,
-      "learning_rate": 2.9e-06,
-      "loss": 304.9479,
+      "epoch": 0.0029290917391532703,
+      "grad_norm": 1085.3497314453125,
+      "learning_rate": 2.9000000000000003e-07,
+      "loss": 28.1494,
       "step": 1450
     },
     {
-      "epoch": 0.005898584743674172,
-      "grad_norm": 2208.190185546875,
-      "learning_rate": 2.92e-06,
-      "loss": 419.7739,
+      "epoch": 0.002949292371837086,
+      "grad_norm": 374.22149658203125,
+      "learning_rate": 2.92e-07,
+      "loss": 32.5694,
       "step": 1460
     },
     {
-      "epoch": 0.005938986009041803,
-      "grad_norm": 558.7657470703125,
-      "learning_rate": 2.9400000000000002e-06,
-      "loss": 243.1836,
+      "epoch": 0.0029694930045209016,
+      "grad_norm": 238.65066528320312,
+      "learning_rate": 2.94e-07,
+      "loss": 18.5137,
       "step": 1470
     },
     {
-      "epoch": 0.0059793872744094345,
-      "grad_norm": 866.2562866210938,
-      "learning_rate": 2.9600000000000005e-06,
-      "loss": 254.9255,
+      "epoch": 0.0029896936372047172,
+      "grad_norm": 2642.603759765625,
+      "learning_rate": 2.9600000000000006e-07,
+      "loss": 33.0043,
       "step": 1480
     },
     {
-      "epoch": 0.006019788539777066,
-      "grad_norm": 827.721923828125,
-      "learning_rate": 2.9800000000000003e-06,
-      "loss": 301.6956,
+      "epoch": 0.003009894269888533,
+      "grad_norm": 371.8793640136719,
+      "learning_rate": 2.9800000000000005e-07,
+      "loss": 22.4324,
       "step": 1490
     },
     {
-      "epoch": 0.006060189805144697,
-      "grad_norm": 1020.5379638671875,
-      "learning_rate": 3e-06,
-      "loss": 236.8173,
+      "epoch": 0.0030300949025723486,
+      "grad_norm": 463.6688232421875,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 19.1007,
       "step": 1500
     },
     {
-      "epoch": 0.0061005910705123284,
-      "grad_norm": 12217.869140625,
-      "learning_rate": 3.0200000000000003e-06,
-      "loss": 315.658,
+      "epoch": 0.0030502955352561642,
+      "grad_norm": 710.45068359375,
+      "learning_rate": 3.0200000000000003e-07,
+      "loss": 17.671,
       "step": 1510
     },
     {
-      "epoch": 0.00614099233587996,
-      "grad_norm": 1178.291259765625,
-      "learning_rate": 3.04e-06,
-      "loss": 300.3993,
+      "epoch": 0.00307049616793998,
+      "grad_norm": 231.896240234375,
+      "learning_rate": 3.04e-07,
+      "loss": 25.546,
       "step": 1520
     },
     {
-      "epoch": 0.006181393601247591,
-      "grad_norm": 2459.315185546875,
-      "learning_rate": 3.06e-06,
-      "loss": 315.2213,
+      "epoch": 0.0030906968006237955,
+      "grad_norm": 263.27044677734375,
+      "learning_rate": 3.06e-07,
+      "loss": 27.5296,
       "step": 1530
     },
     {
-      "epoch": 0.006221794866615222,
-      "grad_norm": 0.0,
-      "learning_rate": 3.08e-06,
-      "loss": 271.6832,
+      "epoch": 0.003110897433307611,
+      "grad_norm": 447.924072265625,
+      "learning_rate": 3.0800000000000006e-07,
+      "loss": 65.7333,
       "step": 1540
     },
     {
-      "epoch": 0.006262196131982854,
-      "grad_norm": 1094.0899658203125,
-      "learning_rate": 3.1e-06,
-      "loss": 273.5612,
+      "epoch": 0.003131098065991427,
+      "grad_norm": 415.6585998535156,
+      "learning_rate": 3.1000000000000005e-07,
+      "loss": 17.1337,
       "step": 1550
     },
     {
-      "epoch": 0.006302597397350485,
-      "grad_norm": 1611.595703125,
-      "learning_rate": 3.12e-06,
-      "loss": 291.6243,
+      "epoch": 0.0031512986986752425,
+      "grad_norm": 743.8123779296875,
+      "learning_rate": 3.12e-07,
+      "loss": 31.4509,
       "step": 1560
     },
     {
-      "epoch": 0.006342998662718116,
-      "grad_norm": 1336.66259765625,
-      "learning_rate": 3.14e-06,
-      "loss": 322.2612,
+      "epoch": 0.003171499331359058,
+      "grad_norm": 379.4447326660156,
+      "learning_rate": 3.14e-07,
+      "loss": 33.4806,
       "step": 1570
     },
     {
-      "epoch": 0.006383399928085748,
-      "grad_norm": 2437.919189453125,
-      "learning_rate": 3.1600000000000007e-06,
-      "loss": 281.6811,
+      "epoch": 0.003191699964042874,
+      "grad_norm": 1378.80615234375,
+      "learning_rate": 3.160000000000001e-07,
+      "loss": 41.5661,
       "step": 1580
     },
     {
-      "epoch": 0.006423801193453379,
-      "grad_norm": 945.635498046875,
-      "learning_rate": 3.1800000000000005e-06,
-      "loss": 331.3091,
+      "epoch": 0.0032119005967266895,
+      "grad_norm": 721.19384765625,
+      "learning_rate": 3.1800000000000007e-07,
+      "loss": 24.3997,
       "step": 1590
     },
     {
-      "epoch": 0.00646420245882101,
-      "grad_norm": 1270.96826171875,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 337.9903,
+      "epoch": 0.003232101229410505,
+      "grad_norm": 447.39154052734375,
+      "learning_rate": 3.2e-07,
+      "loss": 20.2326,
       "step": 1600
     },
     {
-      "epoch": 0.006504603724188642,
-      "grad_norm": 1054.9232177734375,
-      "learning_rate": 3.22e-06,
-      "loss": 371.3227,
+      "epoch": 0.003252301862094321,
+      "grad_norm": 616.3694458007812,
+      "learning_rate": 3.22e-07,
+      "loss": 28.001,
       "step": 1610
     },
     {
-      "epoch": 0.006545004989556273,
-      "grad_norm": 1814.677734375,
-      "learning_rate": 3.24e-06,
-      "loss": 390.2153,
+      "epoch": 0.0032725024947781365,
+      "grad_norm": 942.26904296875,
+      "learning_rate": 3.24e-07,
+      "loss": 34.0719,
       "step": 1620
     },
     {
-      "epoch": 0.006585406254923904,
-      "grad_norm": 648.9537963867188,
-      "learning_rate": 3.2599999999999997e-06,
-      "loss": 368.1877,
+      "epoch": 0.003292703127461952,
+      "grad_norm": 340.3528137207031,
+      "learning_rate": 3.26e-07,
+      "loss": 23.7475,
       "step": 1630
     },
     {
-      "epoch": 0.006625807520291536,
-      "grad_norm": 1473.499267578125,
-      "learning_rate": 3.2800000000000004e-06,
-      "loss": 386.7809,
+      "epoch": 0.003312903760145768,
+      "grad_norm": 616.3563842773438,
+      "learning_rate": 3.280000000000001e-07,
+      "loss": 25.5421,
       "step": 1640
     },
     {
-      "epoch": 0.006666208785659167,
-      "grad_norm": 2527.870849609375,
-      "learning_rate": 3.3e-06,
-      "loss": 414.1417,
+      "epoch": 0.0033331043928295835,
+      "grad_norm": 287.14404296875,
+      "learning_rate": 3.3e-07,
+      "loss": 21.6099,
       "step": 1650
     },
     {
-      "epoch": 0.006706610051026798,
-      "grad_norm": 1029.5302734375,
-      "learning_rate": 3.3200000000000004e-06,
-      "loss": 258.2289,
+      "epoch": 0.003353305025513399,
+      "grad_norm": 325.98907470703125,
+      "learning_rate": 3.32e-07,
+      "loss": 19.8994,
       "step": 1660
     },
     {
-      "epoch": 0.0067470113163944296,
-      "grad_norm": 748.2296142578125,
-      "learning_rate": 3.34e-06,
-      "loss": 348.1511,
+      "epoch": 0.0033735056581972148,
+      "grad_norm": 455.73468017578125,
+      "learning_rate": 3.34e-07,
+      "loss": 24.4489,
       "step": 1670
     },
     {
-      "epoch": 0.006787412581762061,
-      "grad_norm": 2257.800537109375,
-      "learning_rate": 3.36e-06,
-      "loss": 326.2408,
+      "epoch": 0.0033937062908810304,
+      "grad_norm": 970.495361328125,
+      "learning_rate": 3.36e-07,
+      "loss": 25.1155,
       "step": 1680
     },
     {
-      "epoch": 0.006827813847129692,
-      "grad_norm": 987.0545654296875,
-      "learning_rate": 3.38e-06,
-      "loss": 285.0146,
+      "epoch": 0.003413906923564846,
+      "grad_norm": 820.18359375,
+      "learning_rate": 3.38e-07,
+      "loss": 35.97,
       "step": 1690
     },
     {
-      "epoch": 0.0068682151124973235,
-      "grad_norm": 3093.51171875,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 357.8721,
+      "epoch": 0.0034341075562486618,
+      "grad_norm": 135.357177734375,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 22.5324,
       "step": 1700
     },
     {
-      "epoch": 0.006908616377864955,
-      "grad_norm": 769.6875610351562,
-      "learning_rate": 3.4200000000000003e-06,
-      "loss": 312.937,
+      "epoch": 0.0034543081889324774,
+      "grad_norm": 280.0291748046875,
+      "learning_rate": 3.42e-07,
+      "loss": 27.6606,
       "step": 1710
     },
     {
-      "epoch": 0.006949017643232586,
-      "grad_norm": 5268.0009765625,
-      "learning_rate": 3.44e-06,
-      "loss": 437.4973,
+      "epoch": 0.003474508821616293,
+      "grad_norm": 345.8559265136719,
+      "learning_rate": 3.44e-07,
+      "loss": 39.866,
       "step": 1720
     },
     {
-      "epoch": 0.0069894189086002175,
-      "grad_norm": 875.3474731445312,
-      "learning_rate": 3.46e-06,
-      "loss": 303.2312,
+      "epoch": 0.0034947094543001087,
+      "grad_norm": 705.6763305664062,
+      "learning_rate": 3.46e-07,
+      "loss": 18.8156,
       "step": 1730
     },
     {
-      "epoch": 0.007029820173967849,
-      "grad_norm": 988.2050170898438,
-      "learning_rate": 3.4799999999999997e-06,
-      "loss": 358.3378,
+      "epoch": 0.0035149100869839244,
+      "grad_norm": 416.8535461425781,
+      "learning_rate": 3.48e-07,
+      "loss": 33.5592,
       "step": 1740
     },
     {
-      "epoch": 0.00707022143933548,
-      "grad_norm": 1848.81884765625,
-      "learning_rate": 3.5000000000000004e-06,
-      "loss": 354.324,
+      "epoch": 0.00353511071966774,
+      "grad_norm": 552.7217407226562,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 12.8888,
       "step": 1750
     },
     {
-      "epoch": 0.0071106227047031114,
-      "grad_norm": 985.3126220703125,
-      "learning_rate": 3.52e-06,
-      "loss": 387.196,
+      "epoch": 0.0035553113523515557,
+      "grad_norm": 208.27780151367188,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 27.6918,
       "step": 1760
     },
     {
-      "epoch": 0.007151023970070743,
-      "grad_norm": 847.1213989257812,
-      "learning_rate": 3.5400000000000004e-06,
-      "loss": 298.6457,
+      "epoch": 0.0035755119850353714,
+      "grad_norm": 537.1244506835938,
+      "learning_rate": 3.54e-07,
+      "loss": 44.3155,
       "step": 1770
     },
     {
-      "epoch": 0.007191425235438374,
-      "grad_norm": 1782.748779296875,
-      "learning_rate": 3.5600000000000002e-06,
-      "loss": 410.3063,
+      "epoch": 0.003595712617719187,
+      "grad_norm": 284.45733642578125,
+      "learning_rate": 3.56e-07,
+      "loss": 28.734,
       "step": 1780
     },
     {
-      "epoch": 0.007231826500806005,
-      "grad_norm": 731.6522216796875,
-      "learning_rate": 3.58e-06,
-      "loss": 233.0396,
+      "epoch": 0.0036159132504030027,
+      "grad_norm": 387.14532470703125,
+      "learning_rate": 3.58e-07,
+      "loss": 18.956,
       "step": 1790
     },
     {
-      "epoch": 0.007272227766173637,
-      "grad_norm": 498.69366455078125,
-      "learning_rate": 3.6e-06,
-      "loss": 287.9018,
+      "epoch": 0.0036361138830868184,
+      "grad_norm": 370.3746337890625,
+      "learning_rate": 3.6e-07,
+      "loss": 27.5186,
       "step": 1800
     },
     {
-      "epoch": 0.007312629031541268,
-      "grad_norm": 906.2838134765625,
-      "learning_rate": 3.6200000000000005e-06,
-      "loss": 345.1769,
+      "epoch": 0.003656314515770634,
+      "grad_norm": 302.4902038574219,
+      "learning_rate": 3.6200000000000004e-07,
+      "loss": 30.0567,
       "step": 1810
     },
     {
-      "epoch": 0.007353030296908899,
-      "grad_norm": 713.8414916992188,
-      "learning_rate": 3.6400000000000003e-06,
-      "loss": 291.5412,
+      "epoch": 0.0036765151484544497,
+      "grad_norm": 381.73919677734375,
+      "learning_rate": 3.6400000000000003e-07,
+      "loss": 29.6831,
       "step": 1820
     },
     {
-      "epoch": 0.007393431562276531,
-      "grad_norm": 1315.886962890625,
-      "learning_rate": 3.66e-06,
-      "loss": 394.6411,
+      "epoch": 0.0036967157811382653,
+      "grad_norm": 463.9573974609375,
+      "learning_rate": 3.66e-07,
+      "loss": 51.3698,
       "step": 1830
     },
     {
-      "epoch": 0.007433832827644162,
-      "grad_norm": 1429.893798828125,
-      "learning_rate": 3.68e-06,
-      "loss": 247.4847,
+      "epoch": 0.003716916413822081,
+      "grad_norm": 156.73703002929688,
+      "learning_rate": 3.68e-07,
+      "loss": 27.487,
       "step": 1840
     },
     {
-      "epoch": 0.007474234093011793,
-      "grad_norm": 1034.032470703125,
-      "learning_rate": 3.7e-06,
-      "loss": 406.1072,
+      "epoch": 0.0037371170465058967,
+      "grad_norm": 124.4668197631836,
+      "learning_rate": 3.7e-07,
+      "loss": 14.7367,
       "step": 1850
     },
     {
-      "epoch": 0.007514635358379425,
-      "grad_norm": 955.8413696289062,
-      "learning_rate": 3.72e-06,
-      "loss": 392.0067,
+      "epoch": 0.0037573176791897123,
+      "grad_norm": 733.3515625,
+      "learning_rate": 3.72e-07,
+      "loss": 28.2449,
       "step": 1860
     },
     {
-      "epoch": 0.007555036623747056,
-      "grad_norm": 0.0,
-      "learning_rate": 3.7400000000000006e-06,
-      "loss": 235.0364,
+      "epoch": 0.003777518311873528,
+      "grad_norm": 337.3455810546875,
+      "learning_rate": 3.7400000000000004e-07,
+      "loss": 38.7027,
       "step": 1870
     },
     {
-      "epoch": 0.007595437889114687,
-      "grad_norm": 13884.802734375,
-      "learning_rate": 3.7600000000000004e-06,
-      "loss": 336.9673,
+      "epoch": 0.0037977189445573436,
+      "grad_norm": 987.9752197265625,
+      "learning_rate": 3.7600000000000003e-07,
+      "loss": 46.9669,
       "step": 1880
     },
     {
-      "epoch": 0.007635839154482319,
-      "grad_norm": 3753.5439453125,
-      "learning_rate": 3.7800000000000002e-06,
-      "loss": 362.8436,
+      "epoch": 0.0038179195772411593,
+      "grad_norm": 186.5377655029297,
+      "learning_rate": 3.78e-07,
+      "loss": 38.3815,
       "step": 1890
     },
     {
-      "epoch": 0.00767624041984995,
-      "grad_norm": 1046.0020751953125,
-      "learning_rate": 3.8e-06,
-      "loss": 370.7911,
+      "epoch": 0.003838120209924975,
+      "grad_norm": 49.196285247802734,
+      "learning_rate": 3.8e-07,
+      "loss": 18.3016,
       "step": 1900
     },
     {
-      "epoch": 0.007716641685217581,
-      "grad_norm": 2058.227783203125,
-      "learning_rate": 3.82e-06,
-      "loss": 356.5814,
+      "epoch": 0.0038583208426087906,
+      "grad_norm": 309.545166015625,
+      "learning_rate": 3.82e-07,
+      "loss": 17.9834,
       "step": 1910
     },
     {
-      "epoch": 0.0077570429505852126,
-      "grad_norm": 833.1560668945312,
-      "learning_rate": 3.84e-06,
-      "loss": 362.6392,
+      "epoch": 0.0038785214752926063,
+      "grad_norm": 540.118408203125,
+      "learning_rate": 3.84e-07,
+      "loss": 29.434,
       "step": 1920
     },
     {
-      "epoch": 0.007797444215952844,
-      "grad_norm": 1860.0916748046875,
-      "learning_rate": 3.86e-06,
-      "loss": 327.3118,
+      "epoch": 0.003898722107976422,
+      "grad_norm": 20.59901237487793,
+      "learning_rate": 3.8600000000000004e-07,
+      "loss": 34.1916,
       "step": 1930
     },
     {
-      "epoch": 0.007837845481320474,
-      "grad_norm": 2047.00146484375,
-      "learning_rate": 3.88e-06,
-      "loss": 431.6941,
+      "epoch": 0.003918922740660237,
+      "grad_norm": 742.4661254882812,
+      "learning_rate": 3.8800000000000003e-07,
+      "loss": 26.6047,
       "step": 1940
     },
     {
-      "epoch": 0.007878246746688106,
-      "grad_norm": 1433.5374755859375,
-      "learning_rate": 3.9e-06,
-      "loss": 283.749,
+      "epoch": 0.003939123373344053,
+      "grad_norm": 275.4309387207031,
+      "learning_rate": 3.9e-07,
+      "loss": 15.0218,
       "step": 1950
     },
     {
-      "epoch": 0.007918648012055737,
-      "grad_norm": 1340.44287109375,
-      "learning_rate": 3.92e-06,
-      "loss": 323.1868,
+      "epoch": 0.0039593240060278685,
+      "grad_norm": 547.0128173828125,
+      "learning_rate": 3.92e-07,
+      "loss": 30.1546,
       "step": 1960
     },
     {
-      "epoch": 0.007959049277423368,
-      "grad_norm": 5642.4560546875,
-      "learning_rate": 3.9399999999999995e-06,
-      "loss": 428.7537,
+      "epoch": 0.003979524638711684,
+      "grad_norm": 109.45767211914062,
+      "learning_rate": 3.94e-07,
+      "loss": 22.6386,
       "step": 1970
     },
     {
-      "epoch": 0.007999450542791,
-      "grad_norm": 519.6063232421875,
-      "learning_rate": 3.96e-06,
-      "loss": 304.1046,
+      "epoch": 0.0039997252713955,
+      "grad_norm": 498.2757263183594,
+      "learning_rate": 3.9600000000000005e-07,
+      "loss": 28.6018,
       "step": 1980
     },
     {
-      "epoch": 0.008039851808158631,
-      "grad_norm": 1227.509765625,
-      "learning_rate": 3.98e-06,
-      "loss": 295.7083,
+      "epoch": 0.0040199259040793155,
+      "grad_norm": 2670.8740234375,
+      "learning_rate": 3.9800000000000004e-07,
+      "loss": 45.6287,
       "step": 1990
     },
     {
-      "epoch": 0.008080253073526262,
-      "grad_norm": 788.1338500976562,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 383.1164,
+      "epoch": 0.004040126536763131,
+      "grad_norm": 733.6322021484375,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 22.9162,
       "step": 2000
     },
     {
-      "epoch": 0.008120654338893894,
-      "grad_norm": 1502.9964599609375,
-      "learning_rate": 4.0200000000000005e-06,
-      "loss": 312.3206,
+      "epoch": 0.004060327169446947,
+      "grad_norm": 774.7802734375,
+      "learning_rate": 4.02e-07,
+      "loss": 19.3823,
       "step": 2010
     },
     {
-      "epoch": 0.008161055604261525,
-      "grad_norm": 6165.92529296875,
-      "learning_rate": 4.04e-06,
-      "loss": 307.8958,
+      "epoch": 0.0040805278021307624,
+      "grad_norm": 323.9580078125,
+      "learning_rate": 4.04e-07,
+      "loss": 23.0773,
       "step": 2020
     },
     {
-      "epoch": 0.008201456869629156,
-      "grad_norm": 915.6934814453125,
-      "learning_rate": 4.06e-06,
-      "loss": 374.9606,
+      "epoch": 0.004100728434814578,
+      "grad_norm": 1654.9927978515625,
+      "learning_rate": 4.06e-07,
+      "loss": 37.3542,
       "step": 2030
     },
     {
-      "epoch": 0.008241858134996788,
-      "grad_norm": 730.7386474609375,
-      "learning_rate": 4.080000000000001e-06,
-      "loss": 329.4409,
+      "epoch": 0.004120929067498394,
+      "grad_norm": 831.7189331054688,
+      "learning_rate": 4.0800000000000005e-07,
+      "loss": 19.7393,
       "step": 2040
     },
     {
-      "epoch": 0.008282259400364419,
-      "grad_norm": 1635.5831298828125,
-      "learning_rate": 4.1000000000000006e-06,
-      "loss": 209.305,
+      "epoch": 0.004141129700182209,
+      "grad_norm": 921.7901611328125,
+      "learning_rate": 4.1000000000000004e-07,
+      "loss": 44.1891,
       "step": 2050
     },
     {
-      "epoch": 0.00832266066573205,
-      "grad_norm": 4476.35888671875,
-      "learning_rate": 4.12e-06,
-      "loss": 415.3398,
+      "epoch": 0.004161330332866025,
+      "grad_norm": 432.77813720703125,
+      "learning_rate": 4.1200000000000004e-07,
+      "loss": 17.2742,
       "step": 2060
     },
     {
-      "epoch": 0.008363061931099681,
-      "grad_norm": 1422.126220703125,
-      "learning_rate": 4.14e-06,
-      "loss": 360.3677,
+      "epoch": 0.004181530965549841,
+      "grad_norm": 1103.80029296875,
+      "learning_rate": 4.1400000000000003e-07,
+      "loss": 34.2473,
       "step": 2070
     },
     {
-      "epoch": 0.008403463196467313,
-      "grad_norm": 620.626220703125,
-      "learning_rate": 4.16e-06,
-      "loss": 310.8916,
+      "epoch": 0.004201731598233656,
+      "grad_norm": 534.106689453125,
+      "learning_rate": 4.16e-07,
+      "loss": 45.6846,
       "step": 2080
     },
     {
-      "epoch": 0.008443864461834944,
-      "grad_norm": 1371.109130859375,
-      "learning_rate": 4.18e-06,
-      "loss": 339.1394,
+      "epoch": 0.004221932230917472,
+      "grad_norm": 867.72412109375,
+      "learning_rate": 4.18e-07,
+      "loss": 47.1584,
       "step": 2090
     },
     {
-      "epoch": 0.008484265727202575,
-      "grad_norm": 1084.7042236328125,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 282.2104,
+      "epoch": 0.004242132863601288,
+      "grad_norm": 230.99696350097656,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 18.7068,
       "step": 2100
     },
     {
-      "epoch": 0.008524666992570207,
-      "grad_norm": 678.981689453125,
-      "learning_rate": 4.22e-06,
-      "loss": 248.9129,
+      "epoch": 0.004262333496285103,
+      "grad_norm": 531.1705322265625,
+      "learning_rate": 4.2200000000000005e-07,
+      "loss": 33.7492,
       "step": 2110
     },
     {
-      "epoch": 0.008565068257937838,
-      "grad_norm": 1121.5936279296875,
-      "learning_rate": 4.24e-06,
-      "loss": 333.5619,
+      "epoch": 0.004282534128968919,
+      "grad_norm": 897.9804077148438,
+      "learning_rate": 4.2400000000000004e-07,
+      "loss": 29.6693,
       "step": 2120
     },
     {
-      "epoch": 0.00860546952330547,
-      "grad_norm": 795.3826293945312,
-      "learning_rate": 4.26e-06,
-      "loss": 293.3469,
+      "epoch": 0.004302734761652735,
+      "grad_norm": 1721.717529296875,
+      "learning_rate": 4.2600000000000003e-07,
+      "loss": 39.9116,
       "step": 2130
     },
     {
-      "epoch": 0.0086458707886731,
-      "grad_norm": 1109.0904541015625,
-      "learning_rate": 4.28e-06,
-      "loss": 365.5112,
+      "epoch": 0.00432293539433655,
+      "grad_norm": 799.24658203125,
+      "learning_rate": 4.28e-07,
+      "loss": 36.7716,
       "step": 2140
     },
     {
-      "epoch": 0.008686272054040732,
-      "grad_norm": 778.438232421875,
-      "learning_rate": 4.2999999999999995e-06,
-      "loss": 211.6606,
+      "epoch": 0.004343136027020366,
+      "grad_norm": 243.7528076171875,
+      "learning_rate": 4.3e-07,
+      "loss": 17.6017,
       "step": 2150
     },
     {
-      "epoch": 0.008726673319408363,
-      "grad_norm": 1132.38818359375,
-      "learning_rate": 4.32e-06,
-      "loss": 386.3405,
+      "epoch": 0.004363336659704182,
+      "grad_norm": 611.0807495117188,
+      "learning_rate": 4.3200000000000006e-07,
+      "loss": 28.1989,
       "step": 2160
     },
     {
-      "epoch": 0.008767074584775995,
-      "grad_norm": 986.1419677734375,
-      "learning_rate": 4.34e-06,
-      "loss": 326.4904,
+      "epoch": 0.004383537292387997,
+      "grad_norm": 146.26116943359375,
+      "learning_rate": 4.3400000000000005e-07,
+      "loss": 35.3578,
       "step": 2170
     },
     {
-      "epoch": 0.008807475850143626,
-      "grad_norm": 880.376220703125,
-      "learning_rate": 4.360000000000001e-06,
-      "loss": 308.4621,
+      "epoch": 0.004403737925071813,
+      "grad_norm": 1156.3270263671875,
+      "learning_rate": 4.3600000000000004e-07,
+      "loss": 41.4281,
       "step": 2180
     },
     {
-      "epoch": 0.008847877115511257,
-      "grad_norm": 2747.5908203125,
-      "learning_rate": 4.38e-06,
-      "loss": 240.2334,
+      "epoch": 0.004423938557755629,
+      "grad_norm": 309.5587463378906,
+      "learning_rate": 4.3800000000000003e-07,
+      "loss": 18.0717,
       "step": 2190
     },
     {
-      "epoch": 0.008888278380878889,
-      "grad_norm": 1022.6881103515625,
-      "learning_rate": 4.4e-06,
-      "loss": 348.5623,
+      "epoch": 0.004444139190439444,
+      "grad_norm": 573.9265747070312,
+      "learning_rate": 4.4e-07,
+      "loss": 21.9999,
       "step": 2200
     },
     {
-      "epoch": 0.00892867964624652,
-      "grad_norm": 701.6885986328125,
-      "learning_rate": 4.420000000000001e-06,
-      "loss": 239.2583,
+      "epoch": 0.00446433982312326,
+      "grad_norm": 86.95663452148438,
+      "learning_rate": 4.4200000000000007e-07,
+      "loss": 33.4672,
       "step": 2210
     },
     {
-      "epoch": 0.008969080911614151,
-      "grad_norm": 812.2908325195312,
-      "learning_rate": 4.440000000000001e-06,
-      "loss": 305.8652,
+      "epoch": 0.004484540455807076,
+      "grad_norm": 818.9146118164062,
+      "learning_rate": 4.4400000000000006e-07,
+      "loss": 30.7386,
       "step": 2220
     },
     {
-      "epoch": 0.009009482176981783,
-      "grad_norm": 936.9786376953125,
-      "learning_rate": 4.4600000000000005e-06,
-      "loss": 406.4304,
+      "epoch": 0.004504741088490891,
+      "grad_norm": 523.1422729492188,
+      "learning_rate": 4.4600000000000005e-07,
+      "loss": 29.29,
       "step": 2230
     },
     {
-      "epoch": 0.009049883442349414,
-      "grad_norm": 849.5304565429688,
-      "learning_rate": 4.48e-06,
-      "loss": 365.7475,
+      "epoch": 0.004524941721174707,
+      "grad_norm": 1081.639892578125,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 24.7567,
       "step": 2240
     },
     {
-      "epoch": 0.009090284707717045,
-      "grad_norm": 1223.7852783203125,
-      "learning_rate": 4.5e-06,
-      "loss": 295.9342,
+      "epoch": 0.004545142353858523,
+      "grad_norm": 438.20458984375,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 24.615,
       "step": 2250
     },
     {
-      "epoch": 0.009130685973084677,
-      "grad_norm": 2563.434814453125,
-      "learning_rate": 4.52e-06,
-      "loss": 282.5878,
+      "epoch": 0.004565342986542338,
+      "grad_norm": 248.22909545898438,
+      "learning_rate": 4.52e-07,
+      "loss": 25.8396,
       "step": 2260
     },
     {
-      "epoch": 0.009171087238452308,
-      "grad_norm": 3302.397705078125,
-      "learning_rate": 4.540000000000001e-06,
-      "loss": 292.655,
+      "epoch": 0.004585543619226154,
+      "grad_norm": 861.9166870117188,
+      "learning_rate": 4.5400000000000007e-07,
+      "loss": 29.9438,
       "step": 2270
     },
     {
-      "epoch": 0.00921148850381994,
-      "grad_norm": 726.690185546875,
-      "learning_rate": 4.56e-06,
-      "loss": 278.9063,
+      "epoch": 0.00460574425190997,
+      "grad_norm": 823.51318359375,
+      "learning_rate": 4.5600000000000006e-07,
+      "loss": 30.2976,
       "step": 2280
     },
     {
-      "epoch": 0.00925188976918757,
-      "grad_norm": 611.144287109375,
-      "learning_rate": 4.58e-06,
-      "loss": 297.2235,
+      "epoch": 0.004625944884593785,
+      "grad_norm": 1101.48779296875,
+      "learning_rate": 4.5800000000000005e-07,
+      "loss": 32.0947,
       "step": 2290
     },
     {
-      "epoch": 0.009292291034555202,
-      "grad_norm": 962.685302734375,
-      "learning_rate": 4.6e-06,
-      "loss": 224.3765,
+      "epoch": 0.004646145517277601,
+      "grad_norm": 512.0722045898438,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 37.1648,
       "step": 2300
     },
     {
-      "epoch": 0.009332692299922833,
-      "grad_norm": 792.4270629882812,
-      "learning_rate": 4.62e-06,
-      "loss": 339.5073,
+      "epoch": 0.004666346149961417,
+      "grad_norm": 1179.549560546875,
+      "learning_rate": 4.6200000000000003e-07,
+      "loss": 41.9576,
       "step": 2310
     },
     {
-      "epoch": 0.009373093565290464,
-      "grad_norm": 993.0485229492188,
-      "learning_rate": 4.64e-06,
-      "loss": 268.8986,
+      "epoch": 0.004686546782645232,
+      "grad_norm": 472.5940246582031,
+      "learning_rate": 4.64e-07,
+      "loss": 22.1231,
       "step": 2320
     },
     {
-      "epoch": 0.009413494830658096,
-      "grad_norm": 10737.74609375,
-      "learning_rate": 4.66e-06,
-      "loss": 290.0178,
+      "epoch": 0.004706747415329048,
+      "grad_norm": 0.0,
+      "learning_rate": 4.6600000000000007e-07,
+      "loss": 22.0882,
       "step": 2330
     },
     {
-      "epoch": 0.009453896096025727,
-      "grad_norm": 1019.1024780273438,
-      "learning_rate": 4.68e-06,
-      "loss": 290.3607,
+      "epoch": 0.0047269480480128636,
+      "grad_norm": 783.9161376953125,
+      "learning_rate": 4.6800000000000006e-07,
+      "loss": 29.3172,
       "step": 2340
     },
     {
-      "epoch": 0.009494297361393358,
-      "grad_norm": 582.261962890625,
-      "learning_rate": 4.7e-06,
-      "loss": 263.2255,
+      "epoch": 0.004747148680696679,
+      "grad_norm": 1057.4423828125,
+      "learning_rate": 4.7000000000000005e-07,
+      "loss": 28.8642,
       "step": 2350
     },
     {
-      "epoch": 0.00953469862676099,
-      "grad_norm": 928.1985473632812,
-      "learning_rate": 4.72e-06,
-      "loss": 496.6325,
+      "epoch": 0.004767349313380495,
+      "grad_norm": 563.2919921875,
+      "learning_rate": 4.7200000000000004e-07,
+      "loss": 21.4855,
       "step": 2360
     },
     {
-      "epoch": 0.009575099892128621,
-      "grad_norm": 1354.3148193359375,
-      "learning_rate": 4.74e-06,
-      "loss": 317.678,
+      "epoch": 0.0047875499460643105,
+      "grad_norm": 594.2389526367188,
+      "learning_rate": 4.7400000000000004e-07,
+      "loss": 38.7323,
       "step": 2370
     },
     {
-      "epoch": 0.009615501157496252,
-      "grad_norm": 1359.7713623046875,
-      "learning_rate": 4.76e-06,
-      "loss": 263.0336,
+      "epoch": 0.004807750578748126,
+      "grad_norm": 207.3581085205078,
+      "learning_rate": 4.760000000000001e-07,
+      "loss": 28.6296,
       "step": 2380
     },
     {
-      "epoch": 0.009655902422863884,
-      "grad_norm": 759.7086791992188,
-      "learning_rate": 4.780000000000001e-06,
-      "loss": 328.0415,
+      "epoch": 0.004827951211431942,
+      "grad_norm": 874.9584350585938,
+      "learning_rate": 4.78e-07,
+      "loss": 22.6688,
       "step": 2390
     },
     {
-      "epoch": 0.009696303688231515,
-      "grad_norm": 593.6005249023438,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 368.25,
+      "epoch": 0.0048481518441157575,
+      "grad_norm": 76.04383087158203,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 21.4943,
       "step": 2400
     },
     {
-      "epoch": 0.009736704953599146,
-      "grad_norm": 732.736328125,
-      "learning_rate": 4.8200000000000004e-06,
-      "loss": 175.4389,
+      "epoch": 0.004868352476799573,
+      "grad_norm": 961.9263305664062,
+      "learning_rate": 4.82e-07,
+      "loss": 17.3612,
       "step": 2410
     },
     {
-      "epoch": 0.009777106218966778,
-      "grad_norm": 1101.7640380859375,
-      "learning_rate": 4.84e-06,
-      "loss": 293.6888,
+      "epoch": 0.004888553109483389,
+      "grad_norm": 789.2753295898438,
+      "learning_rate": 4.84e-07,
+      "loss": 17.5381,
       "step": 2420
     },
     {
-      "epoch": 0.009817507484334409,
-      "grad_norm": 4335.19873046875,
-      "learning_rate": 4.86e-06,
-      "loss": 348.8911,
+      "epoch": 0.0049087537421672045,
+      "grad_norm": 526.267578125,
+      "learning_rate": 4.86e-07,
+      "loss": 15.962,
       "step": 2430
     },
     {
-      "epoch": 0.00985790874970204,
-      "grad_norm": 1072.8892822265625,
-      "learning_rate": 4.880000000000001e-06,
-      "loss": 231.7467,
+      "epoch": 0.00492895437485102,
+      "grad_norm": 909.1647338867188,
+      "learning_rate": 4.88e-07,
+      "loss": 31.8372,
       "step": 2440
     },
     {
-      "epoch": 0.009898310015069672,
-      "grad_norm": 608.339111328125,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 241.5925,
+      "epoch": 0.004949155007534836,
+      "grad_norm": 392.2228698730469,
+      "learning_rate": 4.900000000000001e-07,
+      "loss": 28.5269,
       "step": 2450
     },
     {
-      "epoch": 0.009938711280437303,
-      "grad_norm": 1088.7662353515625,
-      "learning_rate": 4.92e-06,
-      "loss": 185.2763,
+      "epoch": 0.0049693556402186515,
+      "grad_norm": 572.967529296875,
+      "learning_rate": 4.92e-07,
+      "loss": 21.5063,
       "step": 2460
     },
     {
-      "epoch": 0.009979112545804934,
-      "grad_norm": 901.8641967773438,
-      "learning_rate": 4.94e-06,
-      "loss": 355.3011,
+      "epoch": 0.004989556272902467,
+      "grad_norm": 1103.427001953125,
+      "learning_rate": 4.940000000000001e-07,
+      "loss": 32.4583,
       "step": 2470
     },
     {
-      "epoch": 0.010019513811172566,
-      "grad_norm": 1227.393798828125,
-      "learning_rate": 4.96e-06,
-      "loss": 310.6582,
+      "epoch": 0.005009756905586283,
+      "grad_norm": 816.0195922851562,
+      "learning_rate": 4.96e-07,
+      "loss": 27.6086,
       "step": 2480
     },
     {
-      "epoch": 0.010059915076540197,
-      "grad_norm": 820.9786376953125,
-      "learning_rate": 4.98e-06,
-      "loss": 388.506,
+      "epoch": 0.0050299575382700985,
+      "grad_norm": 440.2334289550781,
+      "learning_rate": 4.98e-07,
+      "loss": 29.0266,
       "step": 2490
     },
     {
-      "epoch": 0.010100316341907828,
-      "grad_norm": 631.5418701171875,
-      "learning_rate": 5e-06,
-      "loss": 186.973,
+      "epoch": 0.005050158170953914,
+      "grad_norm": 528.32421875,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 47.6612,
       "step": 2500
     },
     {
-      "epoch": 0.01014071760727546,
-      "grad_norm": 976.1464233398438,
-      "learning_rate": 5.02e-06,
-      "loss": 260.0689,
+      "epoch": 0.00507035880363773,
+      "grad_norm": 596.6814575195312,
+      "learning_rate": 5.02e-07,
+      "loss": 26.5699,
       "step": 2510
     },
     {
-      "epoch": 0.010181118872643091,
-      "grad_norm": 537.5474853515625,
-      "learning_rate": 5.04e-06,
-      "loss": 245.8573,
+      "epoch": 0.0050905594363215454,
+      "grad_norm": 503.26715087890625,
+      "learning_rate": 5.040000000000001e-07,
+      "loss": 35.7597,
       "step": 2520
     },
     {
-      "epoch": 0.010221520138010722,
-      "grad_norm": 1130.83154296875,
-      "learning_rate": 5.06e-06,
-      "loss": 213.0081,
+      "epoch": 0.005110760069005361,
+      "grad_norm": 402.7926330566406,
+      "learning_rate": 5.06e-07,
+      "loss": 49.0305,
       "step": 2530
     },
     {
-      "epoch": 0.010261921403378354,
-      "grad_norm": 1045.5367431640625,
-      "learning_rate": 5.08e-06,
-      "loss": 273.8134,
+      "epoch": 0.005130960701689177,
+      "grad_norm": 734.1798706054688,
+      "learning_rate": 5.08e-07,
+      "loss": 33.3107,
       "step": 2540
     },
     {
-      "epoch": 0.010302322668745985,
-      "grad_norm": 934.2299194335938,
-      "learning_rate": 5.1e-06,
-      "loss": 351.3407,
+      "epoch": 0.005151161334372992,
+      "grad_norm": 430.47332763671875,
+      "learning_rate": 5.1e-07,
+      "loss": 17.1561,
       "step": 2550
     },
     {
-      "epoch": 0.010342723934113616,
-      "grad_norm": 603.625,
-      "learning_rate": 5.12e-06,
-      "loss": 189.3821,
+      "epoch": 0.005171361967056808,
+      "grad_norm": 171.5887451171875,
+      "learning_rate": 5.12e-07,
+      "loss": 24.0431,
       "step": 2560
     },
     {
-      "epoch": 0.010383125199481247,
-      "grad_norm": 2282.93115234375,
-      "learning_rate": 5.140000000000001e-06,
-      "loss": 306.1943,
+      "epoch": 0.005191562599740624,
+      "grad_norm": 303.52740478515625,
+      "learning_rate": 5.140000000000001e-07,
+      "loss": 25.7299,
       "step": 2570
     },
     {
-      "epoch": 0.010423526464848879,
-      "grad_norm": 1158.8792724609375,
-      "learning_rate": 5.1600000000000006e-06,
-      "loss": 293.8449,
+      "epoch": 0.005211763232424439,
+      "grad_norm": 103.0549087524414,
+      "learning_rate": 5.16e-07,
+      "loss": 50.7616,
       "step": 2580
     },
     {
-      "epoch": 0.01046392773021651,
-      "grad_norm": 1000.369140625,
-      "learning_rate": 5.18e-06,
-      "loss": 252.3016,
+      "epoch": 0.005231963865108255,
+      "grad_norm": 713.0372314453125,
+      "learning_rate": 5.180000000000001e-07,
+      "loss": 30.2423,
       "step": 2590
     },
     {
-      "epoch": 0.010504328995584141,
-      "grad_norm": 1298.2783203125,
-      "learning_rate": 5.2e-06,
-      "loss": 304.2064,
+      "epoch": 0.005252164497792071,
+      "grad_norm": 860.4334716796875,
+      "learning_rate": 5.2e-07,
+      "loss": 22.214,
       "step": 2600
     },
     {
-      "epoch": 0.010544730260951773,
-      "grad_norm": 2631.5205078125,
-      "learning_rate": 5.220000000000001e-06,
-      "loss": 272.8798,
+      "epoch": 0.005272365130475886,
+      "grad_norm": 2196.531494140625,
+      "learning_rate": 5.22e-07,
+      "loss": 48.7065,
       "step": 2610
     },
     {
-      "epoch": 0.010585131526319404,
-      "grad_norm": 1061.5810546875,
-      "learning_rate": 5.240000000000001e-06,
-      "loss": 403.6962,
+      "epoch": 0.005292565763159702,
+      "grad_norm": 193.64439392089844,
+      "learning_rate": 5.240000000000001e-07,
+      "loss": 20.2737,
       "step": 2620
     },
     {
-      "epoch": 0.010625532791687035,
-      "grad_norm": 835.53369140625,
-      "learning_rate": 5.2600000000000005e-06,
-      "loss": 367.6394,
+      "epoch": 0.005312766395843518,
+      "grad_norm": 609.69140625,
+      "learning_rate": 5.26e-07,
+      "loss": 47.1675,
       "step": 2630
     },
     {
-      "epoch": 0.010665934057054667,
-      "grad_norm": 867.9429931640625,
-      "learning_rate": 5.28e-06,
-      "loss": 321.1476,
+      "epoch": 0.005332967028527333,
+      "grad_norm": 1125.103759765625,
+      "learning_rate": 5.280000000000001e-07,
+      "loss": 44.6597,
       "step": 2640
     },
     {
-      "epoch": 0.010706335322422298,
-      "grad_norm": 3146.14013671875,
-      "learning_rate": 5.3e-06,
-      "loss": 266.9892,
+      "epoch": 0.005353167661211149,
+      "grad_norm": 624.8217163085938,
+      "learning_rate": 5.3e-07,
+      "loss": 29.0659,
       "step": 2650
     },
     {
-      "epoch": 0.01074673658778993,
-      "grad_norm": 640.1483764648438,
-      "learning_rate": 5.32e-06,
-      "loss": 206.9268,
+      "epoch": 0.005373368293894965,
+      "grad_norm": 366.8795471191406,
+      "learning_rate": 5.32e-07,
+      "loss": 28.1106,
       "step": 2660
     },
     {
-      "epoch": 0.01078713785315756,
-      "grad_norm": 842.3226928710938,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 325.724,
+      "epoch": 0.00539356892657878,
+      "grad_norm": 585.3496704101562,
+      "learning_rate": 5.340000000000001e-07,
+      "loss": 22.2636,
       "step": 2670
     },
     {
-      "epoch": 0.010827539118525192,
-      "grad_norm": 717.4343872070312,
-      "learning_rate": 5.36e-06,
-      "loss": 314.9767,
+      "epoch": 0.005413769559262596,
+      "grad_norm": 785.0697631835938,
+      "learning_rate": 5.36e-07,
+      "loss": 54.8097,
       "step": 2680
     },
     {
-      "epoch": 0.010867940383892823,
-      "grad_norm": 5089.775390625,
-      "learning_rate": 5.38e-06,
-      "loss": 329.5498,
+      "epoch": 0.005433970191946412,
+      "grad_norm": 433.1402893066406,
+      "learning_rate": 5.380000000000001e-07,
+      "loss": 14.2988,
       "step": 2690
     },
     {
-      "epoch": 0.010908341649260455,
-      "grad_norm": 1157.4693603515625,
-      "learning_rate": 5.4e-06,
-      "loss": 339.251,
+      "epoch": 0.005454170824630227,
+      "grad_norm": 614.6635131835938,
+      "learning_rate": 5.4e-07,
+      "loss": 36.9219,
       "step": 2700
     },
     {
-      "epoch": 0.010948742914628086,
-      "grad_norm": 1268.937744140625,
-      "learning_rate": 5.42e-06,
-      "loss": 361.0937,
+      "epoch": 0.005474371457314043,
+      "grad_norm": 1097.97314453125,
+      "learning_rate": 5.420000000000001e-07,
+      "loss": 32.1943,
       "step": 2710
     },
     {
-      "epoch": 0.010989144179995717,
-      "grad_norm": 855.2471313476562,
-      "learning_rate": 5.44e-06,
-      "loss": 285.6735,
+      "epoch": 0.005494572089997859,
+      "grad_norm": 221.27752685546875,
+      "learning_rate": 5.44e-07,
+      "loss": 17.7383,
       "step": 2720
     },
     {
-      "epoch": 0.011029545445363349,
-      "grad_norm": 777.139892578125,
-      "learning_rate": 5.46e-06,
-      "loss": 275.3022,
+      "epoch": 0.005514772722681674,
+      "grad_norm": 846.3823852539062,
+      "learning_rate": 5.46e-07,
+      "loss": 32.384,
       "step": 2730
     },
     {
-      "epoch": 0.01106994671073098,
-      "grad_norm": 2281.784423828125,
-      "learning_rate": 5.48e-06,
-      "loss": 264.5153,
+      "epoch": 0.00553497335536549,
+      "grad_norm": 1222.7154541015625,
+      "learning_rate": 5.480000000000001e-07,
+      "loss": 40.9341,
       "step": 2740
     },
     {
-      "epoch": 0.011110347976098611,
-      "grad_norm": 737.9111328125,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 200.1999,
+      "epoch": 0.005555173988049306,
+      "grad_norm": 1121.3111572265625,
+      "learning_rate": 5.5e-07,
+      "loss": 23.7586,
       "step": 2750
     },
     {
-      "epoch": 0.011150749241466243,
-      "grad_norm": 7024.08251953125,
-      "learning_rate": 5.5200000000000005e-06,
-      "loss": 316.7326,
+      "epoch": 0.005575374620733121,
+      "grad_norm": 1418.0941162109375,
+      "learning_rate": 5.520000000000001e-07,
+      "loss": 44.7959,
       "step": 2760
     },
     {
-      "epoch": 0.011191150506833874,
-      "grad_norm": 1147.371826171875,
-      "learning_rate": 5.54e-06,
-      "loss": 238.9098,
+      "epoch": 0.005595575253416937,
+      "grad_norm": 352.4377746582031,
+      "learning_rate": 5.54e-07,
+      "loss": 18.5532,
       "step": 2770
     },
     {
-      "epoch": 0.011231551772201505,
-      "grad_norm": 873.0858154296875,
-      "learning_rate": 5.56e-06,
-      "loss": 249.4478,
+      "epoch": 0.005615775886100753,
+      "grad_norm": 682.9746704101562,
+      "learning_rate": 5.560000000000001e-07,
+      "loss": 62.9547,
       "step": 2780
     },
     {
-      "epoch": 0.011271953037569137,
-      "grad_norm": 2017.5811767578125,
-      "learning_rate": 5.580000000000001e-06,
-      "loss": 280.6727,
+      "epoch": 0.005635976518784568,
+      "grad_norm": 69.55480194091797,
+      "learning_rate": 5.580000000000001e-07,
+      "loss": 44.0819,
       "step": 2790
     },
     {
-      "epoch": 0.011312354302936768,
-      "grad_norm": 689.9805908203125,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 238.7173,
+      "epoch": 0.005656177151468384,
+      "grad_norm": 178.6747589111328,
+      "learning_rate": 5.6e-07,
+      "loss": 28.2389,
       "step": 2800
     },
     {
-      "epoch": 0.0113527555683044,
-      "grad_norm": 1248.52490234375,
-      "learning_rate": 5.62e-06,
-      "loss": 243.3881,
+      "epoch": 0.0056763777841522,
+      "grad_norm": 662.1491088867188,
+      "learning_rate": 5.620000000000001e-07,
+      "loss": 19.7388,
       "step": 2810
     },
     {
-      "epoch": 0.01139315683367203,
-      "grad_norm": 756.8421020507812,
-      "learning_rate": 5.64e-06,
-      "loss": 192.3346,
+      "epoch": 0.005696578416836015,
+      "grad_norm": 310.38323974609375,
+      "learning_rate": 5.64e-07,
+      "loss": 39.9212,
       "step": 2820
     },
     {
-      "epoch": 0.011433558099039662,
-      "grad_norm": 1675.668701171875,
-      "learning_rate": 5.66e-06,
-      "loss": 323.5421,
+      "epoch": 0.005716779049519831,
+      "grad_norm": 256.0929870605469,
+      "learning_rate": 5.660000000000001e-07,
+      "loss": 22.1921,
       "step": 2830
     },
     {
-      "epoch": 0.011473959364407293,
-      "grad_norm": 730.6026000976562,
-      "learning_rate": 5.680000000000001e-06,
-      "loss": 235.7116,
+      "epoch": 0.0057369796822036466,
+      "grad_norm": 934.5228881835938,
+      "learning_rate": 5.680000000000001e-07,
+      "loss": 32.7585,
       "step": 2840
     },
     {
-      "epoch": 0.011514360629774924,
-      "grad_norm": 962.4594116210938,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 284.3505,
+      "epoch": 0.005757180314887462,
+      "grad_norm": 703.0299072265625,
+      "learning_rate": 5.7e-07,
+      "loss": 37.5443,
       "step": 2850
     },
     {
-      "epoch": 0.011554761895142556,
-      "grad_norm": 737.1028442382812,
-      "learning_rate": 5.72e-06,
-      "loss": 279.0431,
+      "epoch": 0.005777380947571278,
+      "grad_norm": 687.6595458984375,
+      "learning_rate": 5.720000000000001e-07,
+      "loss": 24.1511,
       "step": 2860
     },
     {
-      "epoch": 0.011595163160510187,
-      "grad_norm": 1873.3394775390625,
-      "learning_rate": 5.74e-06,
-      "loss": 293.4966,
+      "epoch": 0.0057975815802550935,
+      "grad_norm": 400.9228820800781,
+      "learning_rate": 5.74e-07,
+      "loss": 23.191,
       "step": 2870
     },
     {
-      "epoch": 0.011635564425877818,
-      "grad_norm": 699.8680419921875,
-      "learning_rate": 5.76e-06,
-      "loss": 201.6691,
+      "epoch": 0.005817782212938909,
+      "grad_norm": 810.8984375,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 28.8146,
       "step": 2880
     },
     {
-      "epoch": 0.01167596569124545,
-      "grad_norm": 1195.6563720703125,
-      "learning_rate": 5.78e-06,
-      "loss": 339.2745,
+      "epoch": 0.005837982845622725,
+      "grad_norm": 528.00634765625,
+      "learning_rate": 5.78e-07,
+      "loss": 27.1717,
       "step": 2890
     },
     {
-      "epoch": 0.011716366956613081,
-      "grad_norm": 1244.9783935546875,
-      "learning_rate": 5.8e-06,
-      "loss": 295.2553,
+      "epoch": 0.0058581834783065405,
+      "grad_norm": 282.5981140136719,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 19.146,
       "step": 2900
     },
     {
-      "epoch": 0.011756768221980712,
-      "grad_norm": 1545.2462158203125,
-      "learning_rate": 5.82e-06,
-      "loss": 380.4514,
+      "epoch": 0.005878384110990356,
+      "grad_norm": 767.5343627929688,
+      "learning_rate": 5.820000000000001e-07,
+      "loss": 29.7168,
       "step": 2910
     },
     {
-      "epoch": 0.011797169487348344,
-      "grad_norm": 580.0228271484375,
-      "learning_rate": 5.84e-06,
-      "loss": 219.7412,
+      "epoch": 0.005898584743674172,
+      "grad_norm": 1321.3330078125,
+      "learning_rate": 5.84e-07,
+      "loss": 51.2614,
       "step": 2920
     },
     {
-      "epoch": 0.011837570752715975,
-      "grad_norm": 741.6790771484375,
-      "learning_rate": 5.86e-06,
-      "loss": 223.957,
+      "epoch": 0.0059187853763579875,
+      "grad_norm": 422.82183837890625,
+      "learning_rate": 5.860000000000001e-07,
+      "loss": 25.8771,
       "step": 2930
     },
     {
-      "epoch": 0.011877972018083606,
-      "grad_norm": 633.643798828125,
-      "learning_rate": 5.8800000000000005e-06,
-      "loss": 292.0679,
+      "epoch": 0.005938986009041803,
+      "grad_norm": 170.95977783203125,
+      "learning_rate": 5.88e-07,
+      "loss": 16.7884,
       "step": 2940
     },
     {
-      "epoch": 0.011918373283451238,
-      "grad_norm": 11992.681640625,
-      "learning_rate": 5.9e-06,
-      "loss": 365.1848,
+      "epoch": 0.005959186641725619,
+      "grad_norm": 752.912841796875,
+      "learning_rate": 5.900000000000001e-07,
+      "loss": 24.5547,
       "step": 2950
     },
     {
-      "epoch": 0.011958774548818869,
-      "grad_norm": 2407.924072265625,
-      "learning_rate": 5.920000000000001e-06,
-      "loss": 284.2829,
+      "epoch": 0.0059793872744094345,
+      "grad_norm": 47.68494415283203,
+      "learning_rate": 5.920000000000001e-07,
+      "loss": 22.6978,
       "step": 2960
     },
     {
-      "epoch": 0.0119991758141865,
-      "grad_norm": 9947.2861328125,
-      "learning_rate": 5.940000000000001e-06,
-      "loss": 336.0969,
+      "epoch": 0.00599958790709325,
+      "grad_norm": 537.8870849609375,
+      "learning_rate": 5.94e-07,
+      "loss": 49.1119,
       "step": 2970
     },
     {
-      "epoch": 0.012039577079554132,
-      "grad_norm": 1071.607177734375,
-      "learning_rate": 5.9600000000000005e-06,
-      "loss": 288.0666,
+      "epoch": 0.006019788539777066,
+      "grad_norm": 487.58251953125,
+      "learning_rate": 5.960000000000001e-07,
+      "loss": 33.8852,
       "step": 2980
     },
     {
-      "epoch": 0.012079978344921763,
-      "grad_norm": 1607.6715087890625,
-      "learning_rate": 5.98e-06,
-      "loss": 334.0282,
+      "epoch": 0.0060399891724608815,
+      "grad_norm": 578.7280883789062,
+      "learning_rate": 5.98e-07,
+      "loss": 19.5557,
       "step": 2990
     },
     {
-      "epoch": 0.012120379610289394,
-      "grad_norm": 616.33642578125,
-      "learning_rate": 6e-06,
-      "loss": 176.2524,
+      "epoch": 0.006060189805144697,
+      "grad_norm": 357.2806701660156,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 13.2956,
       "step": 3000
     },
     {
-      "epoch": 0.012160780875657026,
-      "grad_norm": 1787.1187744140625,
-      "learning_rate": 6.02e-06,
-      "loss": 262.5472,
+      "epoch": 0.006080390437828513,
+      "grad_norm": 49.976646423339844,
+      "learning_rate": 6.02e-07,
+      "loss": 23.7729,
       "step": 3010
     },
     {
-      "epoch": 0.012201182141024657,
-      "grad_norm": 680.6809692382812,
-      "learning_rate": 6.040000000000001e-06,
-      "loss": 195.7003,
+      "epoch": 0.0061005910705123284,
+      "grad_norm": 1392.4190673828125,
+      "learning_rate": 6.040000000000001e-07,
+      "loss": 27.7141,
       "step": 3020
     },
     {
-      "epoch": 0.012241583406392288,
-      "grad_norm": 1115.2088623046875,
-      "learning_rate": 6.0600000000000004e-06,
-      "loss": 291.6268,
+      "epoch": 0.006120791703196144,
+      "grad_norm": 547.042236328125,
+      "learning_rate": 6.060000000000001e-07,
+      "loss": 22.2183,
       "step": 3030
     },
     {
-      "epoch": 0.01228198467175992,
-      "grad_norm": 1072.4854736328125,
-      "learning_rate": 6.08e-06,
-      "loss": 399.983,
+      "epoch": 0.00614099233587996,
+      "grad_norm": 363.829833984375,
+      "learning_rate": 6.08e-07,
+      "loss": 25.6521,
       "step": 3040
     },
     {
-      "epoch": 0.01232238593712755,
-      "grad_norm": 1406.6270751953125,
-      "learning_rate": 6.1e-06,
-      "loss": 332.8581,
+      "epoch": 0.006161192968563775,
+      "grad_norm": 1138.9224853515625,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 28.1272,
       "step": 3050
     },
     {
-      "epoch": 0.012362787202495182,
-      "grad_norm": 915.8823852539062,
-      "learning_rate": 6.12e-06,
-      "loss": 242.4036,
+      "epoch": 0.006181393601247591,
+      "grad_norm": 388.612548828125,
+      "learning_rate": 6.12e-07,
+      "loss": 25.8512,
       "step": 3060
     },
     {
-      "epoch": 0.012403188467862813,
-      "grad_norm": 1265.71826171875,
-      "learning_rate": 6.1400000000000005e-06,
-      "loss": 240.312,
+      "epoch": 0.006201594233931407,
+      "grad_norm": 633.0219116210938,
+      "learning_rate": 6.140000000000001e-07,
+      "loss": 19.1016,
       "step": 3070
     },
     {
-      "epoch": 0.012443589733230445,
-      "grad_norm": 1575.8834228515625,
-      "learning_rate": 6.16e-06,
-      "loss": 207.945,
+      "epoch": 0.006221794866615222,
+      "grad_norm": 0.0,
+      "learning_rate": 6.160000000000001e-07,
+      "loss": 38.8512,
       "step": 3080
     },
     {
-      "epoch": 0.012483990998598076,
-      "grad_norm": 1585.046142578125,
-      "learning_rate": 6.18e-06,
-      "loss": 328.1433,
+      "epoch": 0.006241995499299038,
+      "grad_norm": 579.1844482421875,
+      "learning_rate": 6.180000000000001e-07,
+      "loss": 23.9865,
       "step": 3090
     },
     {
-      "epoch": 0.012524392263965707,
-      "grad_norm": 3717.249755859375,
-      "learning_rate": 6.2e-06,
-      "loss": 368.012,
+      "epoch": 0.006262196131982854,
+      "grad_norm": 860.4633178710938,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 20.4271,
       "step": 3100
     },
     {
-      "epoch": 0.012564793529333339,
-      "grad_norm": 784.789306640625,
-      "learning_rate": 6.22e-06,
-      "loss": 339.9229,
+      "epoch": 0.006282396764666669,
+      "grad_norm": 406.67791748046875,
+      "learning_rate": 6.22e-07,
+      "loss": 32.6324,
       "step": 3110
     },
     {
-      "epoch": 0.01260519479470097,
-      "grad_norm": 1485.326171875,
-      "learning_rate": 6.24e-06,
-      "loss": 345.8715,
+      "epoch": 0.006302597397350485,
+      "grad_norm": 110.9633560180664,
+      "learning_rate": 6.24e-07,
+      "loss": 14.6998,
       "step": 3120
     },
     {
-      "epoch": 0.012645596060068601,
-      "grad_norm": 883.3239135742188,
-      "learning_rate": 6.26e-06,
-      "loss": 215.2944,
+      "epoch": 0.006322798030034301,
+      "grad_norm": 445.50244140625,
+      "learning_rate": 6.260000000000001e-07,
+      "loss": 16.6768,
       "step": 3130
     },
     {
-      "epoch": 0.012685997325436233,
-      "grad_norm": 665.870849609375,
-      "learning_rate": 6.28e-06,
-      "loss": 247.7601,
+      "epoch": 0.006342998662718116,
+      "grad_norm": 1032.125244140625,
+      "learning_rate": 6.28e-07,
+      "loss": 37.2087,
       "step": 3140
     },
     {
-      "epoch": 0.012726398590803864,
-      "grad_norm": 4960.85546875,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 272.3599,
+      "epoch": 0.006363199295401932,
+      "grad_norm": 284.1348876953125,
+      "learning_rate": 6.3e-07,
+      "loss": 25.7215,
       "step": 3150
     },
     {
-      "epoch": 0.012766799856171495,
-      "grad_norm": 1986.215576171875,
-      "learning_rate": 6.320000000000001e-06,
-      "loss": 261.7445,
+      "epoch": 0.006383399928085748,
+      "grad_norm": 1127.0638427734375,
+      "learning_rate": 6.320000000000002e-07,
+      "loss": 43.177,
       "step": 3160
     },
     {
-      "epoch": 0.012807201121539127,
-      "grad_norm": 689.2445678710938,
-      "learning_rate": 6.34e-06,
-      "loss": 186.6741,
+      "epoch": 0.006403600560769563,
+      "grad_norm": 858.960205078125,
+      "learning_rate": 6.34e-07,
+      "loss": 29.6178,
       "step": 3170
     },
     {
-      "epoch": 0.012847602386906758,
-      "grad_norm": 1272.3314208984375,
-      "learning_rate": 6.360000000000001e-06,
-      "loss": 362.3708,
+      "epoch": 0.006423801193453379,
+      "grad_norm": 632.6044921875,
+      "learning_rate": 6.360000000000001e-07,
+      "loss": 24.2948,
       "step": 3180
     },
     {
-      "epoch": 0.01288800365227439,
-      "grad_norm": 1243.4554443359375,
-      "learning_rate": 6.38e-06,
-      "loss": 223.126,
+      "epoch": 0.006444001826137195,
+      "grad_norm": 385.60772705078125,
+      "learning_rate": 6.38e-07,
+      "loss": 29.6091,
       "step": 3190
     },
     {
-      "epoch": 0.01292840491764202,
-      "grad_norm": 0.0,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 220.924,
+      "epoch": 0.00646420245882101,
+      "grad_norm": 735.461181640625,
+      "learning_rate": 6.4e-07,
+      "loss": 36.3189,
       "step": 3200
     },
     {
-      "epoch": 0.012968806183009652,
-      "grad_norm": 1140.0751953125,
-      "learning_rate": 6.4199999999999995e-06,
-      "loss": 304.4165,
+      "epoch": 0.006484403091504826,
+      "grad_norm": 806.14208984375,
+      "learning_rate": 6.42e-07,
+      "loss": 40.9133,
       "step": 3210
     },
     {
-      "epoch": 0.013009207448377283,
-      "grad_norm": 865.4572143554688,
-      "learning_rate": 6.44e-06,
-      "loss": 225.4173,
+      "epoch": 0.006504603724188642,
+      "grad_norm": 197.8910675048828,
+      "learning_rate": 6.44e-07,
+      "loss": 22.794,
       "step": 3220
     },
     {
-      "epoch": 0.013049608713744915,
-      "grad_norm": 1003.0341796875,
-      "learning_rate": 6.460000000000001e-06,
-      "loss": 318.8464,
+      "epoch": 0.006524804356872457,
+      "grad_norm": 464.4063720703125,
+      "learning_rate": 6.460000000000001e-07,
+      "loss": 25.8888,
       "step": 3230
     },
     {
-      "epoch": 0.013090009979112546,
-      "grad_norm": 3230.90966796875,
-      "learning_rate": 6.48e-06,
-      "loss": 266.8362,
+      "epoch": 0.006545004989556273,
+      "grad_norm": 897.4555053710938,
+      "learning_rate": 6.48e-07,
+      "loss": 45.3982,
       "step": 3240
     },
     {
-      "epoch": 0.013130411244480177,
-      "grad_norm": 5314.744140625,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 356.5823,
+      "epoch": 0.006565205622240089,
+      "grad_norm": 1184.5333251953125,
+      "learning_rate": 6.5e-07,
+      "loss": 45.8197,
       "step": 3250
     },
     {
-      "epoch": 0.013170812509847809,
-      "grad_norm": 1545.9931640625,
-      "learning_rate": 6.519999999999999e-06,
-      "loss": 267.9298,
+      "epoch": 0.006585406254923904,
+      "grad_norm": 245.31578063964844,
+      "learning_rate": 6.52e-07,
+      "loss": 22.9282,
       "step": 3260
     },
     {
-      "epoch": 0.01321121377521544,
-      "grad_norm": 1124.9794921875,
-      "learning_rate": 6.54e-06,
-      "loss": 350.8778,
+      "epoch": 0.00660560688760772,
+      "grad_norm": 396.6833190917969,
+      "learning_rate": 6.54e-07,
+      "loss": 36.6589,
       "step": 3270
     },
     {
-      "epoch": 0.013251615040583071,
-      "grad_norm": 1584.3470458984375,
-      "learning_rate": 6.560000000000001e-06,
-      "loss": 268.6283,
+      "epoch": 0.006625807520291536,
+      "grad_norm": 827.7760620117188,
+      "learning_rate": 6.560000000000002e-07,
+      "loss": 27.5275,
       "step": 3280
     },
     {
-      "epoch": 0.013292016305950703,
-      "grad_norm": 1276.875244140625,
-      "learning_rate": 6.58e-06,
-      "loss": 401.9857,
+      "epoch": 0.006646008152975351,
+      "grad_norm": 698.2625122070312,
+      "learning_rate": 6.58e-07,
+      "loss": 43.7663,
       "step": 3290
     },
     {
-      "epoch": 0.013332417571318334,
-      "grad_norm": 1092.377197265625,
-      "learning_rate": 6.6e-06,
-      "loss": 331.368,
+      "epoch": 0.006666208785659167,
+      "grad_norm": 804.8634033203125,
+      "learning_rate": 6.6e-07,
+      "loss": 42.3979,
       "step": 3300
     },
     {
-      "epoch": 0.013372818836685965,
-      "grad_norm": 1179.2205810546875,
-      "learning_rate": 6.62e-06,
-      "loss": 177.4306,
+      "epoch": 0.006686409418342983,
+      "grad_norm": 381.66253662109375,
+      "learning_rate": 6.62e-07,
+      "loss": 28.6512,
       "step": 3310
     },
     {
-      "epoch": 0.013413220102053596,
-      "grad_norm": 1153.752685546875,
-      "learning_rate": 6.640000000000001e-06,
-      "loss": 229.9581,
+      "epoch": 0.006706610051026798,
+      "grad_norm": 274.6836242675781,
+      "learning_rate": 6.64e-07,
+      "loss": 13.3558,
       "step": 3320
     },
     {
-      "epoch": 0.013453621367421228,
-      "grad_norm": 998.432373046875,
-      "learning_rate": 6.660000000000001e-06,
-      "loss": 266.7427,
+      "epoch": 0.006726810683710614,
+      "grad_norm": 640.2205200195312,
+      "learning_rate": 6.660000000000002e-07,
+      "loss": 31.3027,
       "step": 3330
     },
     {
-      "epoch": 0.013494022632788859,
-      "grad_norm": 1151.6092529296875,
-      "learning_rate": 6.68e-06,
-      "loss": 331.803,
+      "epoch": 0.0067470113163944296,
+      "grad_norm": 545.2610473632812,
+      "learning_rate": 6.68e-07,
+      "loss": 19.2501,
       "step": 3340
     },
     {
-      "epoch": 0.01353442389815649,
-      "grad_norm": 2472.0546875,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 236.3656,
+      "epoch": 0.006767211949078245,
+      "grad_norm": 652.1705322265625,
+      "learning_rate": 6.7e-07,
+      "loss": 37.1378,
       "step": 3350
     },
     {
-      "epoch": 0.013574825163524122,
-      "grad_norm": 1243.1534423828125,
-      "learning_rate": 6.72e-06,
-      "loss": 180.7836,
+      "epoch": 0.006787412581762061,
+      "grad_norm": 1061.9658203125,
+      "learning_rate": 6.72e-07,
+      "loss": 21.7822,
       "step": 3360
     },
     {
-      "epoch": 0.013615226428891753,
-      "grad_norm": 1034.3489990234375,
-      "learning_rate": 6.740000000000001e-06,
-      "loss": 343.4018,
+      "epoch": 0.0068076132144458765,
+      "grad_norm": 556.8729248046875,
+      "learning_rate": 6.74e-07,
+      "loss": 24.4347,
       "step": 3370
     },
     {
-      "epoch": 0.013655627694259384,
-      "grad_norm": 905.5031127929688,
-      "learning_rate": 6.76e-06,
-      "loss": 140.6687,
+      "epoch": 0.006827813847129692,
+      "grad_norm": 522.956787109375,
+      "learning_rate": 6.76e-07,
+      "loss": 27.4032,
       "step": 3380
     },
     {
-      "epoch": 0.013696028959627016,
-      "grad_norm": 1340.9151611328125,
-      "learning_rate": 6.78e-06,
-      "loss": 289.7602,
+      "epoch": 0.006848014479813508,
+      "grad_norm": 531.7928466796875,
+      "learning_rate": 6.78e-07,
+      "loss": 19.0332,
       "step": 3390
     },
     {
-      "epoch": 0.013736430224994647,
-      "grad_norm": 960.6358642578125,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 226.2994,
+      "epoch": 0.0068682151124973235,
+      "grad_norm": 804.1568603515625,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 34.7435,
       "step": 3400
     },
     {
-      "epoch": 0.013776831490362278,
-      "grad_norm": 1423.4761962890625,
-      "learning_rate": 6.82e-06,
-      "loss": 430.8338,
+      "epoch": 0.006888415745181139,
+      "grad_norm": 1052.230712890625,
+      "learning_rate": 6.82e-07,
+      "loss": 25.6694,
       "step": 3410
     },
     {
-      "epoch": 0.01381723275572991,
-      "grad_norm": 1304.45263671875,
-      "learning_rate": 6.840000000000001e-06,
-      "loss": 328.4004,
+      "epoch": 0.006908616377864955,
+      "grad_norm": 526.1654663085938,
+      "learning_rate": 6.84e-07,
+      "loss": 22.8071,
       "step": 3420
     },
     {
-      "epoch": 0.013857634021097541,
-      "grad_norm": 9199.7373046875,
-      "learning_rate": 6.8599999999999995e-06,
-      "loss": 270.6068,
+      "epoch": 0.0069288170105487705,
+      "grad_norm": 1492.260986328125,
+      "learning_rate": 6.86e-07,
+      "loss": 20.8931,
       "step": 3430
     },
     {
-      "epoch": 0.013898035286465172,
-      "grad_norm": 776.5551147460938,
-      "learning_rate": 6.88e-06,
-      "loss": 176.4451,
+      "epoch": 0.006949017643232586,
+      "grad_norm": 2416.557373046875,
+      "learning_rate": 6.88e-07,
+      "loss": 45.9088,
       "step": 3440
     },
     {
-      "epoch": 0.013938436551832804,
-      "grad_norm": 774.5919799804688,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 258.9075,
+      "epoch": 0.006969218275916402,
+      "grad_norm": 138.57374572753906,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 20.0898,
       "step": 3450
     },
     {
-      "epoch": 0.013978837817200435,
-      "grad_norm": 1491.8228759765625,
-      "learning_rate": 6.92e-06,
-      "loss": 295.3933,
+      "epoch": 0.0069894189086002175,
+      "grad_norm": 517.4945678710938,
+      "learning_rate": 6.92e-07,
+      "loss": 26.9102,
       "step": 3460
     },
     {
-      "epoch": 0.014019239082568066,
-      "grad_norm": 1491.8526611328125,
-      "learning_rate": 6.9400000000000005e-06,
-      "loss": 239.9352,
+      "epoch": 0.007009619541284033,
+      "grad_norm": 426.5979919433594,
+      "learning_rate": 6.94e-07,
+      "loss": 39.5767,
       "step": 3470
     },
     {
-      "epoch": 0.014059640347935698,
-      "grad_norm": 1934.498779296875,
-      "learning_rate": 6.9599999999999994e-06,
-      "loss": 275.0346,
+      "epoch": 0.007029820173967849,
+      "grad_norm": 208.76681518554688,
+      "learning_rate": 6.96e-07,
+      "loss": 42.304,
       "step": 3480
     },
     {
-      "epoch": 0.014100041613303329,
-      "grad_norm": 1181.4918212890625,
-      "learning_rate": 6.98e-06,
-      "loss": 238.241,
+      "epoch": 0.0070500208066516645,
+      "grad_norm": 478.20050048828125,
+      "learning_rate": 6.98e-07,
+      "loss": 23.9574,
       "step": 3490
     },
     {
-      "epoch": 0.01414044287867096,
-      "grad_norm": 1785.65478515625,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 163.911,
+      "epoch": 0.00707022143933548,
+      "grad_norm": 1073.7861328125,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 40.8396,
       "step": 3500
     },
     {
-      "epoch": 0.014180844144038592,
-      "grad_norm": 601.7042846679688,
-      "learning_rate": 7.0200000000000006e-06,
-      "loss": 200.9242,
+      "epoch": 0.007090422072019296,
+      "grad_norm": 164.99581909179688,
+      "learning_rate": 7.02e-07,
+      "loss": 54.3387,
       "step": 3510
     },
     {
-      "epoch": 0.014221245409406223,
-      "grad_norm": 1390.438720703125,
-      "learning_rate": 7.04e-06,
-      "loss": 225.1192,
+      "epoch": 0.0071106227047031114,
+      "grad_norm": 559.994873046875,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 23.7497,
       "step": 3520
     },
     {
-      "epoch": 0.014261646674773854,
-      "grad_norm": 1239.8173828125,
-      "learning_rate": 7.06e-06,
-      "loss": 207.541,
+      "epoch": 0.007130823337386927,
+      "grad_norm": 248.2763671875,
+      "learning_rate": 7.06e-07,
+      "loss": 14.4129,
       "step": 3530
     },
     {
-      "epoch": 0.014302047940141486,
-      "grad_norm": 5047.65478515625,
-      "learning_rate": 7.080000000000001e-06,
-      "loss": 200.282,
+      "epoch": 0.007151023970070743,
+      "grad_norm": 115.21528625488281,
+      "learning_rate": 7.08e-07,
+      "loss": 20.9172,
       "step": 3540
     },
     {
-      "epoch": 0.014342449205509117,
-      "grad_norm": 684.9512939453125,
-      "learning_rate": 7.1e-06,
-      "loss": 200.1313,
+      "epoch": 0.007171224602754558,
+      "grad_norm": 529.1926879882812,
+      "learning_rate": 7.1e-07,
+      "loss": 29.7155,
       "step": 3550
     },
     {
-      "epoch": 0.014382850470876748,
-      "grad_norm": 538.3458862304688,
-      "learning_rate": 7.1200000000000004e-06,
-      "loss": 271.5458,
+      "epoch": 0.007191425235438374,
+      "grad_norm": 931.08056640625,
+      "learning_rate": 7.12e-07,
+      "loss": 48.6689,
       "step": 3560
     },
     {
-      "epoch": 0.01442325173624438,
-      "grad_norm": 1814.7567138671875,
-      "learning_rate": 7.140000000000001e-06,
-      "loss": 227.9855,
+      "epoch": 0.00721162586812219,
+      "grad_norm": 325.41485595703125,
+      "learning_rate": 7.140000000000001e-07,
+      "loss": 19.5409,
       "step": 3570
     },
     {
-      "epoch": 0.01446365300161201,
-      "grad_norm": 617.8508911132812,
-      "learning_rate": 7.16e-06,
-      "loss": 152.5538,
+      "epoch": 0.007231826500806005,
+      "grad_norm": 255.26974487304688,
+      "learning_rate": 7.16e-07,
+      "loss": 13.3323,
       "step": 3580
     },
     {
-      "epoch": 0.014504054266979642,
-      "grad_norm": 1195.5587158203125,
-      "learning_rate": 7.180000000000001e-06,
-      "loss": 220.7645,
+      "epoch": 0.007252027133489821,
+      "grad_norm": 174.78330993652344,
+      "learning_rate": 7.18e-07,
+      "loss": 27.6813,
       "step": 3590
     },
     {
-      "epoch": 0.014544455532347273,
-      "grad_norm": 963.2885131835938,
-      "learning_rate": 7.2e-06,
-      "loss": 237.8198,
+      "epoch": 0.007272227766173637,
+      "grad_norm": 259.9703369140625,
+      "learning_rate": 7.2e-07,
+      "loss": 23.6667,
       "step": 3600
     },
     {
-      "epoch": 0.014584856797714905,
-      "grad_norm": 1143.399658203125,
-      "learning_rate": 7.22e-06,
-      "loss": 246.5691,
+      "epoch": 0.007292428398857452,
+      "grad_norm": 510.34515380859375,
+      "learning_rate": 7.22e-07,
+      "loss": 43.1148,
       "step": 3610
     },
     {
-      "epoch": 0.014625258063082536,
-      "grad_norm": 936.9075927734375,
-      "learning_rate": 7.240000000000001e-06,
-      "loss": 225.2431,
+      "epoch": 0.007312629031541268,
+      "grad_norm": 369.45806884765625,
+      "learning_rate": 7.240000000000001e-07,
+      "loss": 18.297,
       "step": 3620
     },
     {
-      "epoch": 0.014665659328450167,
-      "grad_norm": 505.3612365722656,
-      "learning_rate": 7.26e-06,
-      "loss": 209.9184,
+      "epoch": 0.007332829664225084,
+      "grad_norm": 492.9736022949219,
+      "learning_rate": 7.26e-07,
+      "loss": 34.2994,
       "step": 3630
     },
     {
-      "epoch": 0.014706060593817799,
-      "grad_norm": 4472.5458984375,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 251.8955,
+      "epoch": 0.007353030296908899,
+      "grad_norm": 611.5191650390625,
+      "learning_rate": 7.280000000000001e-07,
+      "loss": 40.5437,
       "step": 3640
     },
     {
-      "epoch": 0.01474646185918543,
-      "grad_norm": 1152.7425537109375,
-      "learning_rate": 7.2999999999999996e-06,
-      "loss": 267.1611,
+      "epoch": 0.007373230929592715,
+      "grad_norm": 478.23126220703125,
+      "learning_rate": 7.3e-07,
+      "loss": 42.8498,
       "step": 3650
     },
     {
-      "epoch": 0.014786863124553061,
-      "grad_norm": 1535.3359375,
-      "learning_rate": 7.32e-06,
-      "loss": 258.2474,
+      "epoch": 0.007393431562276531,
+      "grad_norm": 822.2637329101562,
+      "learning_rate": 7.32e-07,
+      "loss": 42.8805,
       "step": 3660
     },
     {
-      "epoch": 0.014827264389920693,
-      "grad_norm": 788.0407104492188,
-      "learning_rate": 7.340000000000001e-06,
-      "loss": 308.7459,
+      "epoch": 0.007413632194960346,
+      "grad_norm": 572.3778686523438,
+      "learning_rate": 7.340000000000001e-07,
+      "loss": 21.5523,
       "step": 3670
     },
     {
-      "epoch": 0.014867665655288324,
-      "grad_norm": 3632.613037109375,
-      "learning_rate": 7.36e-06,
-      "loss": 270.6488,
+      "epoch": 0.007433832827644162,
+      "grad_norm": 610.9832153320312,
+      "learning_rate": 7.36e-07,
+      "loss": 24.501,
       "step": 3680
     },
     {
-      "epoch": 0.014908066920655955,
-      "grad_norm": 1338.3414306640625,
-      "learning_rate": 7.3800000000000005e-06,
-      "loss": 402.1751,
+      "epoch": 0.007454033460327978,
+      "grad_norm": 444.3849792480469,
+      "learning_rate": 7.380000000000001e-07,
+      "loss": 29.0815,
       "step": 3690
     },
     {
-      "epoch": 0.014948468186023587,
-      "grad_norm": 2017.5732421875,
-      "learning_rate": 7.4e-06,
-      "loss": 335.9066,
+      "epoch": 0.007474234093011793,
+      "grad_norm": 334.1302490234375,
+      "learning_rate": 7.4e-07,
+      "loss": 29.4803,
       "step": 3700
     },
     {
-      "epoch": 0.014988869451391218,
-      "grad_norm": 866.8572998046875,
-      "learning_rate": 7.420000000000001e-06,
-      "loss": 228.7708,
+      "epoch": 0.007494434725695609,
+      "grad_norm": 484.7538757324219,
+      "learning_rate": 7.420000000000001e-07,
+      "loss": 34.7338,
       "step": 3710
     },
     {
-      "epoch": 0.01502927071675885,
-      "grad_norm": 2627.818359375,
-      "learning_rate": 7.44e-06,
-      "loss": 237.4983,
+      "epoch": 0.007514635358379425,
+      "grad_norm": 557.7028198242188,
+      "learning_rate": 7.44e-07,
+      "loss": 45.3084,
       "step": 3720
     },
     {
-      "epoch": 0.01506967198212648,
-      "grad_norm": 1014.5845947265625,
-      "learning_rate": 7.4600000000000006e-06,
-      "loss": 289.3517,
+      "epoch": 0.00753483599106324,
+      "grad_norm": 419.6079406738281,
+      "learning_rate": 7.46e-07,
+      "loss": 38.1011,
       "step": 3730
     },
     {
-      "epoch": 0.015110073247494112,
-      "grad_norm": 1206.6943359375,
-      "learning_rate": 7.480000000000001e-06,
-      "loss": 253.7986,
+      "epoch": 0.007555036623747056,
+      "grad_norm": 0.0,
+      "learning_rate": 7.480000000000001e-07,
+      "loss": 34.1749,
       "step": 3740
     },
     {
-      "epoch": 0.015150474512861743,
-      "grad_norm": 1855.4755859375,
-      "learning_rate": 7.5e-06,
-      "loss": 384.5513,
+      "epoch": 0.007575237256430872,
+      "grad_norm": 608.0559692382812,
+      "learning_rate": 7.5e-07,
+      "loss": 12.6068,
       "step": 3750
     },
     {
-      "epoch": 0.015190875778229375,
-      "grad_norm": 745.6865844726562,
-      "learning_rate": 7.520000000000001e-06,
-      "loss": 246.6602,
+      "epoch": 0.007595437889114687,
+      "grad_norm": 1825.6767578125,
+      "learning_rate": 7.520000000000001e-07,
+      "loss": 36.8878,
       "step": 3760
     },
     {
-      "epoch": 0.015231277043597006,
-      "grad_norm": 1637.106201171875,
-      "learning_rate": 7.54e-06,
-      "loss": 291.9528,
+      "epoch": 0.007615638521798503,
+      "grad_norm": 347.3094787597656,
+      "learning_rate": 7.54e-07,
+      "loss": 26.2681,
       "step": 3770
     },
     {
-      "epoch": 0.015271678308964637,
-      "grad_norm": 2112.7109375,
-      "learning_rate": 7.5600000000000005e-06,
-      "loss": 228.7163,
+      "epoch": 0.007635839154482319,
+      "grad_norm": 1999.683837890625,
+      "learning_rate": 7.56e-07,
+      "loss": 40.3283,
       "step": 3780
     },
     {
-      "epoch": 0.015312079574332269,
-      "grad_norm": 1741.5140380859375,
-      "learning_rate": 7.580000000000001e-06,
-      "loss": 238.9058,
+      "epoch": 0.007656039787166134,
+      "grad_norm": 520.2977905273438,
+      "learning_rate": 7.580000000000001e-07,
+      "loss": 34.4978,
       "step": 3790
     },
     {
-      "epoch": 0.0153524808396999,
-      "grad_norm": 991.5599365234375,
-      "learning_rate": 7.6e-06,
-      "loss": 172.588,
+      "epoch": 0.00767624041984995,
+      "grad_norm": 457.4718322753906,
+      "learning_rate": 7.6e-07,
+      "loss": 37.1305,
       "step": 3800
     },
     {
-      "epoch": 0.015392882105067531,
-      "grad_norm": 1450.9693603515625,
-      "learning_rate": 7.620000000000001e-06,
-      "loss": 298.7596,
+      "epoch": 0.007696441052533766,
+      "grad_norm": 241.23533630371094,
+      "learning_rate": 7.620000000000001e-07,
+      "loss": 34.1395,
       "step": 3810
     },
     {
-      "epoch": 0.015433283370435162,
-      "grad_norm": 602.7521362304688,
-      "learning_rate": 7.64e-06,
-      "loss": 212.8531,
+      "epoch": 0.007716641685217581,
+      "grad_norm": 1090.16943359375,
+      "learning_rate": 7.64e-07,
+      "loss": 32.3401,
       "step": 3820
     },
     {
-      "epoch": 0.015473684635802794,
-      "grad_norm": 1867.60546875,
-      "learning_rate": 7.660000000000001e-06,
-      "loss": 314.0657,
+      "epoch": 0.007736842317901397,
+      "grad_norm": 799.810546875,
+      "learning_rate": 7.660000000000001e-07,
+      "loss": 18.844,
       "step": 3830
     },
     {
-      "epoch": 0.015514085901170425,
-      "grad_norm": 6575.3671875,
-      "learning_rate": 7.68e-06,
-      "loss": 181.4101,
+      "epoch": 0.0077570429505852126,
+      "grad_norm": 513.3098754882812,
+      "learning_rate": 7.68e-07,
+      "loss": 56.5756,
       "step": 3840
     },
     {
-      "epoch": 0.015554487166538056,
-      "grad_norm": 811.929443359375,
-      "learning_rate": 7.7e-06,
-      "loss": 203.4754,
+      "epoch": 0.007777243583269028,
+      "grad_norm": 412.8589172363281,
+      "learning_rate": 7.7e-07,
+      "loss": 28.4811,
       "step": 3850
     },
     {
-      "epoch": 0.015594888431905688,
-      "grad_norm": 1337.7786865234375,
-      "learning_rate": 7.72e-06,
-      "loss": 251.0148,
+      "epoch": 0.007797444215952844,
+      "grad_norm": 818.9476318359375,
+      "learning_rate": 7.720000000000001e-07,
+      "loss": 23.9097,
       "step": 3860
     },
     {
-      "epoch": 0.01563528969727332,
-      "grad_norm": 1872.110595703125,
-      "learning_rate": 7.74e-06,
-      "loss": 220.2635,
+      "epoch": 0.00781764484863666,
+      "grad_norm": 660.37158203125,
+      "learning_rate": 7.74e-07,
+      "loss": 33.1159,
       "step": 3870
     },
     {
-      "epoch": 0.01567569096264095,
-      "grad_norm": 421.63671875,
-      "learning_rate": 7.76e-06,
-      "loss": 219.4246,
+      "epoch": 0.007837845481320474,
+      "grad_norm": 554.0394287109375,
+      "learning_rate": 7.760000000000001e-07,
+      "loss": 41.6907,
       "step": 3880
     },
     {
-      "epoch": 0.01571609222800858,
-      "grad_norm": 901.9820556640625,
-      "learning_rate": 7.78e-06,
-      "loss": 253.3245,
+      "epoch": 0.00785804611400429,
+      "grad_norm": 609.9306640625,
+      "learning_rate": 7.78e-07,
+      "loss": 23.2773,
       "step": 3890
     },
     {
-      "epoch": 0.01575649349337621,
-      "grad_norm": 1378.5067138671875,
-      "learning_rate": 7.8e-06,
-      "loss": 272.5337,
+      "epoch": 0.007878246746688106,
+      "grad_norm": 509.7363586425781,
+      "learning_rate": 7.8e-07,
+      "loss": 27.8946,
       "step": 3900
     },
     {
-      "epoch": 0.015796894758743844,
-      "grad_norm": 899.2632446289062,
-      "learning_rate": 7.820000000000001e-06,
-      "loss": 221.5219,
+      "epoch": 0.007898447379371922,
+      "grad_norm": 617.0547485351562,
+      "learning_rate": 7.820000000000001e-07,
+      "loss": 40.5662,
       "step": 3910
     },
     {
-      "epoch": 0.015837296024111474,
-      "grad_norm": 2511.511474609375,
-      "learning_rate": 7.84e-06,
-      "loss": 183.7072,
+      "epoch": 0.007918648012055737,
+      "grad_norm": 45.31270217895508,
+      "learning_rate": 7.84e-07,
+      "loss": 19.5,
       "step": 3920
     },
     {
-      "epoch": 0.015877697289479107,
-      "grad_norm": 1480.0848388671875,
-      "learning_rate": 7.860000000000001e-06,
-      "loss": 226.5134,
+      "epoch": 0.007938848644739553,
+      "grad_norm": 936.0264282226562,
+      "learning_rate": 7.860000000000001e-07,
+      "loss": 31.398,
       "step": 3930
     },
     {
-      "epoch": 0.015918098554846737,
-      "grad_norm": 1251.0631103515625,
-      "learning_rate": 7.879999999999999e-06,
-      "loss": 220.8395,
+      "epoch": 0.007959049277423368,
+      "grad_norm": 4046.48974609375,
+      "learning_rate": 7.88e-07,
+      "loss": 61.3853,
       "step": 3940
     },
     {
-      "epoch": 0.01595849982021437,
-      "grad_norm": 1017.7593994140625,
-      "learning_rate": 7.9e-06,
-      "loss": 299.8441,
+      "epoch": 0.007979249910107185,
+      "grad_norm": 678.829345703125,
+      "learning_rate": 7.900000000000001e-07,
+      "loss": 52.0202,
       "step": 3950
     },
     {
-      "epoch": 0.015998901085582,
-      "grad_norm": 750.101318359375,
-      "learning_rate": 7.92e-06,
-      "loss": 256.8377,
+      "epoch": 0.007999450542791,
+      "grad_norm": 174.16665649414062,
+      "learning_rate": 7.920000000000001e-07,
+      "loss": 18.2646,
       "step": 3960
     },
     {
-      "epoch": 0.016039302350949632,
-      "grad_norm": 850.2648315429688,
-      "learning_rate": 7.94e-06,
-      "loss": 300.5525,
+      "epoch": 0.008019651175474816,
+      "grad_norm": 631.7503051757812,
+      "learning_rate": 7.94e-07,
+      "loss": 22.1172,
       "step": 3970
     },
     {
-      "epoch": 0.016079703616317262,
-      "grad_norm": 688.6962280273438,
-      "learning_rate": 7.96e-06,
-      "loss": 238.5739,
+      "epoch": 0.008039851808158631,
+      "grad_norm": 863.083984375,
+      "learning_rate": 7.960000000000001e-07,
+      "loss": 31.1579,
       "step": 3980
     },
     {
-      "epoch": 0.016120104881684895,
-      "grad_norm": 1160.3623046875,
-      "learning_rate": 7.98e-06,
-      "loss": 230.5895,
+      "epoch": 0.008060052440842447,
+      "grad_norm": 395.8502197265625,
+      "learning_rate": 7.98e-07,
+      "loss": 36.5626,
       "step": 3990
     },
     {
-      "epoch": 0.016160506147052525,
-      "grad_norm": 768.995849609375,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 273.8474,
+      "epoch": 0.008080253073526262,
+      "grad_norm": 551.8558349609375,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 27.3743,
       "step": 4000
     },
     {
-      "epoch": 0.016200907412420158,
-      "grad_norm": 733.1216430664062,
-      "learning_rate": 8.02e-06,
-      "loss": 213.2722,
+      "epoch": 0.008100453706210079,
+      "grad_norm": 350.1429138183594,
+      "learning_rate": 8.02e-07,
+      "loss": 27.0405,
       "step": 4010
     },
     {
-      "epoch": 0.016241308677787787,
-      "grad_norm": 0.0,
-      "learning_rate": 8.040000000000001e-06,
-      "loss": 222.9751,
+      "epoch": 0.008120654338893894,
+      "grad_norm": 1742.940185546875,
+      "learning_rate": 8.04e-07,
+      "loss": 44.1298,
       "step": 4020
     },
     {
-      "epoch": 0.01628170994315542,
-      "grad_norm": 800.3453979492188,
-      "learning_rate": 8.06e-06,
-      "loss": 268.3867,
+      "epoch": 0.00814085497157771,
+      "grad_norm": 773.5975341796875,
+      "learning_rate": 8.060000000000001e-07,
+      "loss": 22.7084,
       "step": 4030
     },
     {
-      "epoch": 0.01632211120852305,
-      "grad_norm": 1095.6239013671875,
-      "learning_rate": 8.08e-06,
-      "loss": 321.233,
+      "epoch": 0.008161055604261525,
+      "grad_norm": 996.5304565429688,
+      "learning_rate": 8.08e-07,
+      "loss": 40.6753,
       "step": 4040
     },
     {
-      "epoch": 0.016362512473890683,
-      "grad_norm": 2067.449462890625,
-      "learning_rate": 8.1e-06,
-      "loss": 218.0929,
+      "epoch": 0.008181256236945341,
+      "grad_norm": 671.2755126953125,
+      "learning_rate": 8.100000000000001e-07,
+      "loss": 39.0983,
       "step": 4050
     },
     {
-      "epoch": 0.016402913739258312,
-      "grad_norm": 1188.5040283203125,
-      "learning_rate": 8.12e-06,
-      "loss": 326.4176,
+      "epoch": 0.008201456869629156,
+      "grad_norm": 350.6563415527344,
+      "learning_rate": 8.12e-07,
+      "loss": 30.7569,
       "step": 4060
     },
     {
-      "epoch": 0.016443315004625945,
-      "grad_norm": 1143.6793212890625,
-      "learning_rate": 8.14e-06,
-      "loss": 244.4551,
+      "epoch": 0.008221657502312973,
+      "grad_norm": 1149.395263671875,
+      "learning_rate": 8.140000000000001e-07,
+      "loss": 28.9356,
       "step": 4070
     },
     {
-      "epoch": 0.016483716269993575,
-      "grad_norm": 878.0779418945312,
-      "learning_rate": 8.160000000000001e-06,
-      "loss": 235.9173,
+      "epoch": 0.008241858134996788,
+      "grad_norm": 990.6906127929688,
+      "learning_rate": 8.160000000000001e-07,
+      "loss": 39.5627,
       "step": 4080
     },
     {
-      "epoch": 0.016524117535361208,
-      "grad_norm": 737.4267578125,
-      "learning_rate": 8.18e-06,
-      "loss": 331.0821,
+      "epoch": 0.008262058767680604,
+      "grad_norm": 364.4676208496094,
+      "learning_rate": 8.18e-07,
+      "loss": 14.7625,
       "step": 4090
     },
     {
-      "epoch": 0.016564518800728838,
-      "grad_norm": 1495.6761474609375,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 248.3052,
+      "epoch": 0.008282259400364419,
+      "grad_norm": 811.5744018554688,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 22.5022,
       "step": 4100
     },
     {
-      "epoch": 0.01660492006609647,
-      "grad_norm": 814.0003051757812,
-      "learning_rate": 8.22e-06,
-      "loss": 199.0726,
+      "epoch": 0.008302460033048235,
+      "grad_norm": 582.2328491210938,
+      "learning_rate": 8.22e-07,
+      "loss": 39.7949,
       "step": 4110
     },
     {
-      "epoch": 0.0166453213314641,
-      "grad_norm": 634.8568725585938,
-      "learning_rate": 8.24e-06,
-      "loss": 313.8576,
+      "epoch": 0.00832266066573205,
+      "grad_norm": 1127.8995361328125,
+      "learning_rate": 8.240000000000001e-07,
+      "loss": 37.47,
       "step": 4120
     },
     {
-      "epoch": 0.016685722596831733,
-      "grad_norm": 714.36328125,
-      "learning_rate": 8.26e-06,
-      "loss": 215.2515,
+      "epoch": 0.008342861298415867,
+      "grad_norm": 502.9598693847656,
+      "learning_rate": 8.260000000000001e-07,
+      "loss": 50.2378,
       "step": 4130
     },
     {
-      "epoch": 0.016726123862199363,
-      "grad_norm": 1302.689697265625,
-      "learning_rate": 8.28e-06,
-      "loss": 262.0066,
+      "epoch": 0.008363061931099681,
+      "grad_norm": 630.9242553710938,
+      "learning_rate": 8.280000000000001e-07,
+      "loss": 25.1314,
       "step": 4140
     },
     {
-      "epoch": 0.016766525127566996,
-      "grad_norm": 1403.1646728515625,
-      "learning_rate": 8.3e-06,
-      "loss": 314.0872,
+      "epoch": 0.008383262563783498,
+      "grad_norm": 632.32958984375,
+      "learning_rate": 8.300000000000001e-07,
+      "loss": 42.2978,
       "step": 4150
     },
     {
-      "epoch": 0.016806926392934626,
-      "grad_norm": 944.2760009765625,
-      "learning_rate": 8.32e-06,
-      "loss": 276.365,
+      "epoch": 0.008403463196467313,
+      "grad_norm": 513.2865600585938,
+      "learning_rate": 8.32e-07,
+      "loss": 22.9661,
       "step": 4160
     },
     {
-      "epoch": 0.01684732765830226,
-      "grad_norm": 2125.658447265625,
-      "learning_rate": 8.34e-06,
-      "loss": 337.9024,
+      "epoch": 0.00842366382915113,
+      "grad_norm": 437.45941162109375,
+      "learning_rate": 8.340000000000001e-07,
+      "loss": 24.9955,
       "step": 4170
     },
     {
-      "epoch": 0.016887728923669888,
-      "grad_norm": 2500.66455078125,
-      "learning_rate": 8.36e-06,
-      "loss": 241.9205,
+      "epoch": 0.008443864461834944,
+      "grad_norm": 732.6044311523438,
+      "learning_rate": 8.36e-07,
+      "loss": 37.1115,
       "step": 4180
     },
     {
-      "epoch": 0.01692813018903752,
-      "grad_norm": 637.8106689453125,
-      "learning_rate": 8.380000000000001e-06,
-      "loss": 239.1331,
+      "epoch": 0.00846406509451876,
+      "grad_norm": 270.413330078125,
+      "learning_rate": 8.380000000000001e-07,
+      "loss": 20.6703,
       "step": 4190
     },
     {
-      "epoch": 0.01696853145440515,
-      "grad_norm": 1099.7269287109375,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 170.0579,
+      "epoch": 0.008484265727202575,
+      "grad_norm": 8.429944038391113,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 25.673,
       "step": 4200
     },
     {
-      "epoch": 0.017008932719772784,
-      "grad_norm": 777.8555297851562,
-      "learning_rate": 8.42e-06,
-      "loss": 269.5526,
+      "epoch": 0.008504466359886392,
+      "grad_norm": 555.270751953125,
+      "learning_rate": 8.42e-07,
+      "loss": 22.8865,
       "step": 4210
     },
     {
-      "epoch": 0.017049333985140414,
-      "grad_norm": 624.0740356445312,
-      "learning_rate": 8.44e-06,
-      "loss": 180.0498,
+      "epoch": 0.008524666992570207,
+      "grad_norm": 638.0531005859375,
+      "learning_rate": 8.440000000000001e-07,
+      "loss": 21.5478,
       "step": 4220
     },
     {
-      "epoch": 0.017089735250508047,
-      "grad_norm": 1288.8555908203125,
-      "learning_rate": 8.46e-06,
-      "loss": 287.3699,
+      "epoch": 0.008544867625254023,
+      "grad_norm": 1256.038818359375,
+      "learning_rate": 8.46e-07,
+      "loss": 41.5287,
       "step": 4230
     },
     {
-      "epoch": 0.017130136515875676,
-      "grad_norm": 689.3584594726562,
-      "learning_rate": 8.48e-06,
-      "loss": 308.6741,
+      "epoch": 0.008565068257937838,
+      "grad_norm": 414.123046875,
+      "learning_rate": 8.480000000000001e-07,
+      "loss": 19.9983,
       "step": 4240
     },
     {
-      "epoch": 0.01717053778124331,
-      "grad_norm": 1398.036376953125,
-      "learning_rate": 8.500000000000002e-06,
-      "loss": 266.2167,
+      "epoch": 0.008585268890621655,
+      "grad_norm": 594.0344848632812,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 33.3682,
       "step": 4250
     },
     {
-      "epoch": 0.01721093904661094,
-      "grad_norm": 789.7576293945312,
-      "learning_rate": 8.52e-06,
-      "loss": 255.249,
+      "epoch": 0.00860546952330547,
+      "grad_norm": 580.2907104492188,
+      "learning_rate": 8.520000000000001e-07,
+      "loss": 26.1064,
       "step": 4260
     },
     {
-      "epoch": 0.017251340311978572,
-      "grad_norm": 815.0007934570312,
-      "learning_rate": 8.540000000000001e-06,
-      "loss": 194.3588,
+      "epoch": 0.008625670155989286,
+      "grad_norm": 910.2868041992188,
+      "learning_rate": 8.540000000000001e-07,
+      "loss": 25.9703,
       "step": 4270
     },
     {
-      "epoch": 0.0172917415773462,
-      "grad_norm": 1327.666259765625,
-      "learning_rate": 8.56e-06,
-      "loss": 329.2384,
+      "epoch": 0.0086458707886731,
+      "grad_norm": 401.0901184082031,
+      "learning_rate": 8.56e-07,
+      "loss": 31.2655,
       "step": 4280
     },
     {
-      "epoch": 0.017332142842713835,
-      "grad_norm": 626.3024291992188,
-      "learning_rate": 8.580000000000001e-06,
-      "loss": 137.7933,
+      "epoch": 0.008666071421356917,
+      "grad_norm": 586.83447265625,
+      "learning_rate": 8.580000000000001e-07,
+      "loss": 22.6068,
       "step": 4290
     },
     {
-      "epoch": 0.017372544108081464,
-      "grad_norm": 1358.869384765625,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 234.6338,
+      "epoch": 0.008686272054040732,
+      "grad_norm": 384.7112731933594,
+      "learning_rate": 8.6e-07,
+      "loss": 20.9198,
       "step": 4300
     },
     {
-      "epoch": 0.017412945373449097,
-      "grad_norm": 935.4063110351562,
-      "learning_rate": 8.62e-06,
-      "loss": 326.7994,
+      "epoch": 0.008706472686724549,
+      "grad_norm": 861.9733276367188,
+      "learning_rate": 8.620000000000001e-07,
+      "loss": 30.633,
       "step": 4310
     },
     {
-      "epoch": 0.017453346638816727,
-      "grad_norm": 1433.051025390625,
-      "learning_rate": 8.64e-06,
-      "loss": 242.4297,
+      "epoch": 0.008726673319408363,
+      "grad_norm": 508.044189453125,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 37.4199,
       "step": 4320
     },
     {
-      "epoch": 0.01749374790418436,
-      "grad_norm": 1025.8055419921875,
-      "learning_rate": 8.66e-06,
-      "loss": 168.3089,
+      "epoch": 0.00874687395209218,
+      "grad_norm": 757.0943603515625,
+      "learning_rate": 8.66e-07,
+      "loss": 31.3223,
       "step": 4330
     },
     {
-      "epoch": 0.01753414916955199,
-      "grad_norm": 2163.64794921875,
-      "learning_rate": 8.68e-06,
-      "loss": 246.4337,
+      "epoch": 0.008767074584775995,
+      "grad_norm": 481.5451965332031,
+      "learning_rate": 8.680000000000001e-07,
+      "loss": 30.192,
       "step": 4340
     },
     {
-      "epoch": 0.017574550434919622,
-      "grad_norm": 1108.7593994140625,
-      "learning_rate": 8.7e-06,
-      "loss": 211.9252,
+      "epoch": 0.008787275217459811,
+      "grad_norm": 1465.626220703125,
+      "learning_rate": 8.7e-07,
+      "loss": 36.7876,
       "step": 4350
     },
     {
-      "epoch": 0.017614951700287252,
-      "grad_norm": 1170.3670654296875,
-      "learning_rate": 8.720000000000001e-06,
-      "loss": 196.4572,
+      "epoch": 0.008807475850143626,
+      "grad_norm": 537.8186645507812,
+      "learning_rate": 8.720000000000001e-07,
+      "loss": 18.696,
       "step": 4360
     },
     {
-      "epoch": 0.017655352965654885,
-      "grad_norm": 822.1085815429688,
-      "learning_rate": 8.740000000000001e-06,
-      "loss": 236.9191,
+      "epoch": 0.008827676482827443,
+      "grad_norm": 421.6712951660156,
+      "learning_rate": 8.740000000000001e-07,
+      "loss": 16.2779,
       "step": 4370
     },
     {
-      "epoch": 0.017695754231022515,
-      "grad_norm": 551.1714477539062,
-      "learning_rate": 8.76e-06,
-      "loss": 262.1803,
+      "epoch": 0.008847877115511257,
+      "grad_norm": 625.7432861328125,
+      "learning_rate": 8.760000000000001e-07,
+      "loss": 23.2634,
       "step": 4380
     },
     {
-      "epoch": 0.017736155496390148,
-      "grad_norm": 941.1051635742188,
-      "learning_rate": 8.78e-06,
-      "loss": 197.1894,
+      "epoch": 0.008868077748195074,
+      "grad_norm": 630.1290893554688,
+      "learning_rate": 8.780000000000001e-07,
+      "loss": 22.0466,
       "step": 4390
     },
     {
-      "epoch": 0.017776556761757777,
-      "grad_norm": 812.2485961914062,
-      "learning_rate": 8.8e-06,
-      "loss": 304.7698,
+      "epoch": 0.008888278380878889,
+      "grad_norm": 19.57767105102539,
+      "learning_rate": 8.8e-07,
+      "loss": 32.943,
       "step": 4400
     },
     {
-      "epoch": 0.01781695802712541,
-      "grad_norm": 740.1641845703125,
-      "learning_rate": 8.82e-06,
-      "loss": 207.9336,
+      "epoch": 0.008908479013562705,
+      "grad_norm": 563.815185546875,
+      "learning_rate": 8.820000000000001e-07,
+      "loss": 20.2435,
       "step": 4410
     },
     {
-      "epoch": 0.01785735929249304,
-      "grad_norm": 1118.2490234375,
-      "learning_rate": 8.840000000000002e-06,
-      "loss": 201.0788,
+      "epoch": 0.00892867964624652,
+      "grad_norm": 565.490234375,
+      "learning_rate": 8.840000000000001e-07,
+      "loss": 30.1036,
       "step": 4420
     },
     {
-      "epoch": 0.017897760557860673,
-      "grad_norm": 2099.23583984375,
-      "learning_rate": 8.86e-06,
-      "loss": 222.6882,
+      "epoch": 0.008948880278930336,
+      "grad_norm": 553.708984375,
+      "learning_rate": 8.860000000000001e-07,
+      "loss": 24.4606,
       "step": 4430
     },
     {
-      "epoch": 0.017938161823228303,
-      "grad_norm": 1029.5682373046875,
-      "learning_rate": 8.880000000000001e-06,
-      "loss": 194.8024,
+      "epoch": 0.008969080911614151,
+      "grad_norm": 433.3461608886719,
+      "learning_rate": 8.880000000000001e-07,
+      "loss": 24.9225,
       "step": 4440
     },
     {
-      "epoch": 0.017978563088595936,
-      "grad_norm": 1188.818603515625,
-      "learning_rate": 8.9e-06,
-      "loss": 206.5788,
+      "epoch": 0.008989281544297968,
+      "grad_norm": 939.7255859375,
+      "learning_rate": 8.900000000000001e-07,
+      "loss": 32.4033,
       "step": 4450
     },
     {
-      "epoch": 0.018018964353963565,
-      "grad_norm": 820.3424682617188,
-      "learning_rate": 8.920000000000001e-06,
-      "loss": 222.0988,
+      "epoch": 0.009009482176981783,
+      "grad_norm": 207.4413604736328,
+      "learning_rate": 8.920000000000001e-07,
+      "loss": 38.8472,
       "step": 4460
     },
     {
-      "epoch": 0.0180593656193312,
-      "grad_norm": 1431.5162353515625,
-      "learning_rate": 8.939999999999999e-06,
-      "loss": 210.3697,
+      "epoch": 0.0090296828096656,
+      "grad_norm": 1215.0419921875,
+      "learning_rate": 8.94e-07,
+      "loss": 41.1395,
       "step": 4470
     },
     {
-      "epoch": 0.018099766884698828,
-      "grad_norm": 1079.22314453125,
-      "learning_rate": 8.96e-06,
-      "loss": 236.7236,
+      "epoch": 0.009049883442349414,
+      "grad_norm": 761.4100341796875,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 31.6368,
       "step": 4480
     },
     {
-      "epoch": 0.01814016815006646,
-      "grad_norm": 1381.0302734375,
-      "learning_rate": 8.98e-06,
-      "loss": 200.5311,
+      "epoch": 0.00907008407503323,
+      "grad_norm": 815.1241455078125,
+      "learning_rate": 8.980000000000001e-07,
+      "loss": 25.794,
       "step": 4490
     },
     {
-      "epoch": 0.01818056941543409,
-      "grad_norm": 799.3221435546875,
-      "learning_rate": 9e-06,
-      "loss": 229.802,
+      "epoch": 0.009090284707717045,
+      "grad_norm": 664.5227661132812,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 26.1395,
       "step": 4500
     },
     {
-      "epoch": 0.018220970680801724,
-      "grad_norm": 1207.725830078125,
-      "learning_rate": 9.02e-06,
-      "loss": 214.4318,
+      "epoch": 0.009110485340400862,
+      "grad_norm": 621.558349609375,
+      "learning_rate": 9.020000000000001e-07,
+      "loss": 20.145,
       "step": 4510
     },
     {
-      "epoch": 0.018261371946169353,
-      "grad_norm": 725.3284301757812,
-      "learning_rate": 9.04e-06,
-      "loss": 193.9023,
+      "epoch": 0.009130685973084677,
+      "grad_norm": 1977.96240234375,
+      "learning_rate": 9.04e-07,
+      "loss": 31.9802,
       "step": 4520
     },
     {
-      "epoch": 0.018301773211536986,
-      "grad_norm": 1104.83642578125,
-      "learning_rate": 9.06e-06,
-      "loss": 220.9035,
+      "epoch": 0.009150886605768493,
+      "grad_norm": 638.37109375,
+      "learning_rate": 9.060000000000001e-07,
+      "loss": 20.8006,
       "step": 4530
     },
     {
-      "epoch": 0.018342174476904616,
-      "grad_norm": 1509.99462890625,
-      "learning_rate": 9.080000000000001e-06,
-      "loss": 188.4938,
+      "epoch": 0.009171087238452308,
+      "grad_norm": 1067.8172607421875,
+      "learning_rate": 9.080000000000001e-07,
+      "loss": 55.018,
       "step": 4540
     },
     {
-      "epoch": 0.01838257574227225,
-      "grad_norm": 1129.591064453125,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 308.8094,
+      "epoch": 0.009191287871136124,
+      "grad_norm": 1054.7120361328125,
+      "learning_rate": 9.100000000000001e-07,
+      "loss": 28.5558,
       "step": 4550
     },
     {
-      "epoch": 0.01842297700763988,
-      "grad_norm": 2700.010986328125,
-      "learning_rate": 9.12e-06,
-      "loss": 310.8748,
+      "epoch": 0.00921148850381994,
+      "grad_norm": 297.5315856933594,
+      "learning_rate": 9.120000000000001e-07,
+      "loss": 29.3597,
       "step": 4560
     },
     {
-      "epoch": 0.01846337827300751,
-      "grad_norm": 1198.5032958984375,
-      "learning_rate": 9.14e-06,
-      "loss": 219.3821,
+      "epoch": 0.009231689136503756,
+      "grad_norm": 683.9236450195312,
+      "learning_rate": 9.140000000000001e-07,
+      "loss": 20.7864,
       "step": 4570
     },
     {
-      "epoch": 0.01850377953837514,
-      "grad_norm": 907.4744873046875,
-      "learning_rate": 9.16e-06,
-      "loss": 272.8794,
+      "epoch": 0.00925188976918757,
+      "grad_norm": 822.9036865234375,
+      "learning_rate": 9.160000000000001e-07,
+      "loss": 26.9667,
       "step": 4580
     },
     {
-      "epoch": 0.018544180803742774,
-      "grad_norm": 480.4779968261719,
-      "learning_rate": 9.180000000000002e-06,
-      "loss": 303.4508,
+      "epoch": 0.009272090401871387,
+      "grad_norm": 630.2400512695312,
+      "learning_rate": 9.180000000000001e-07,
+      "loss": 12.3359,
       "step": 4590
     },
     {
-      "epoch": 0.018584582069110404,
-      "grad_norm": 5315.6416015625,
-      "learning_rate": 9.2e-06,
-      "loss": 298.2705,
+      "epoch": 0.009292291034555202,
+      "grad_norm": 608.8886108398438,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 23.5769,
       "step": 4600
     },
     {
-      "epoch": 0.018624983334478037,
-      "grad_norm": 872.8356323242188,
-      "learning_rate": 9.220000000000002e-06,
-      "loss": 203.8982,
+      "epoch": 0.009312491667239018,
+      "grad_norm": 827.4570922851562,
+      "learning_rate": 9.220000000000001e-07,
+      "loss": 18.0923,
       "step": 4610
     },
     {
-      "epoch": 0.018665384599845666,
-      "grad_norm": 593.9172973632812,
-      "learning_rate": 9.24e-06,
-      "loss": 175.9667,
+      "epoch": 0.009332692299922833,
+      "grad_norm": 412.2288818359375,
+      "learning_rate": 9.240000000000001e-07,
+      "loss": 38.5852,
       "step": 4620
     },
     {
-      "epoch": 0.0187057858652133,
-      "grad_norm": 869.2783203125,
-      "learning_rate": 9.260000000000001e-06,
-      "loss": 180.2353,
+      "epoch": 0.00935289293260665,
+      "grad_norm": 950.310302734375,
+      "learning_rate": 9.260000000000001e-07,
+      "loss": 28.0527,
       "step": 4630
     },
     {
-      "epoch": 0.01874618713058093,
-      "grad_norm": 1448.7855224609375,
-      "learning_rate": 9.28e-06,
-      "loss": 180.4345,
+      "epoch": 0.009373093565290464,
+      "grad_norm": 442.42523193359375,
+      "learning_rate": 9.28e-07,
+      "loss": 20.4062,
       "step": 4640
     },
     {
-      "epoch": 0.018786588395948562,
-      "grad_norm": 2569.0322265625,
-      "learning_rate": 9.3e-06,
-      "loss": 263.0333,
+      "epoch": 0.009393294197974281,
+      "grad_norm": 524.7567749023438,
+      "learning_rate": 9.300000000000001e-07,
+      "loss": 27.0559,
       "step": 4650
     },
     {
-      "epoch": 0.01882698966131619,
-      "grad_norm": 1611.3876953125,
-      "learning_rate": 9.32e-06,
-      "loss": 286.5421,
+      "epoch": 0.009413494830658096,
+      "grad_norm": 602.6281127929688,
+      "learning_rate": 9.320000000000001e-07,
+      "loss": 26.432,
       "step": 4660
     },
     {
-      "epoch": 0.018867390926683825,
-      "grad_norm": 762.2740478515625,
-      "learning_rate": 9.34e-06,
-      "loss": 272.6539,
+      "epoch": 0.009433695463341912,
+      "grad_norm": 1005.23974609375,
+      "learning_rate": 9.340000000000001e-07,
+      "loss": 30.3101,
       "step": 4670
     },
     {
-      "epoch": 0.018907792192051454,
-      "grad_norm": 1819.1978759765625,
-      "learning_rate": 9.36e-06,
-      "loss": 197.3641,
+      "epoch": 0.009453896096025727,
+      "grad_norm": 511.9738464355469,
+      "learning_rate": 9.360000000000001e-07,
+      "loss": 20.6791,
       "step": 4680
     },
     {
-      "epoch": 0.018948193457419087,
-      "grad_norm": 1281.1929931640625,
-      "learning_rate": 9.38e-06,
-      "loss": 199.3052,
+      "epoch": 0.009474096728709544,
+      "grad_norm": 503.60693359375,
+      "learning_rate": 9.380000000000001e-07,
+      "loss": 26.5743,
       "step": 4690
     },
     {
-      "epoch": 0.018988594722786717,
-      "grad_norm": 4967.193359375,
-      "learning_rate": 9.4e-06,
-      "loss": 230.312,
+      "epoch": 0.009494297361393358,
+      "grad_norm": 674.8211059570312,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 31.4228,
       "step": 4700
     },
     {
-      "epoch": 0.01902899598815435,
-      "grad_norm": 1749.0260009765625,
-      "learning_rate": 9.420000000000001e-06,
-      "loss": 245.5051,
+      "epoch": 0.009514497994077175,
+      "grad_norm": 473.37396240234375,
+      "learning_rate": 9.420000000000002e-07,
+      "loss": 44.0672,
       "step": 4710
     },
     {
-      "epoch": 0.01906939725352198,
-      "grad_norm": 0.0,
-      "learning_rate": 9.44e-06,
-      "loss": 190.8588,
+      "epoch": 0.00953469862676099,
+      "grad_norm": 831.3383178710938,
+      "learning_rate": 9.440000000000001e-07,
+      "loss": 47.9142,
       "step": 4720
     },
     {
-      "epoch": 0.019109798518889613,
-      "grad_norm": 1861.452880859375,
-      "learning_rate": 9.460000000000001e-06,
-      "loss": 190.6377,
+      "epoch": 0.009554899259444806,
+      "grad_norm": 2031.18798828125,
+      "learning_rate": 9.460000000000001e-07,
+      "loss": 43.8193,
       "step": 4730
     },
     {
-      "epoch": 0.019150199784257242,
-      "grad_norm": 2230.98876953125,
-      "learning_rate": 9.48e-06,
-      "loss": 252.1078,
+      "epoch": 0.009575099892128621,
+      "grad_norm": 1036.27978515625,
+      "learning_rate": 9.480000000000001e-07,
+      "loss": 19.8682,
       "step": 4740
     },
     {
-      "epoch": 0.019190601049624875,
-      "grad_norm": 803.5484008789062,
-      "learning_rate": 9.5e-06,
-      "loss": 254.6172,
+      "epoch": 0.009595300524812438,
+      "grad_norm": 367.123046875,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 19.759,
       "step": 4750
     },
     {
-      "epoch": 0.019231002314992505,
-      "grad_norm": 765.6219482421875,
-      "learning_rate": 9.52e-06,
-      "loss": 249.983,
+      "epoch": 0.009615501157496252,
+      "grad_norm": 1182.7408447265625,
+      "learning_rate": 9.520000000000002e-07,
+      "loss": 29.7404,
       "step": 4760
     },
     {
-      "epoch": 0.019271403580360138,
-      "grad_norm": 932.5681762695312,
-      "learning_rate": 9.54e-06,
-      "loss": 235.5316,
+      "epoch": 0.009635701790180069,
+      "grad_norm": 326.5006408691406,
+      "learning_rate": 9.54e-07,
+      "loss": 62.1775,
       "step": 4770
     },
     {
-      "epoch": 0.019311804845727767,
-      "grad_norm": 981.9686889648438,
-      "learning_rate": 9.560000000000002e-06,
-      "loss": 229.3628,
+      "epoch": 0.009655902422863884,
+      "grad_norm": 206.4904022216797,
+      "learning_rate": 9.56e-07,
+      "loss": 20.3732,
       "step": 4780
     },
     {
-      "epoch": 0.0193522061110954,
-      "grad_norm": 436.2134704589844,
-      "learning_rate": 9.58e-06,
-      "loss": 158.716,
+      "epoch": 0.0096761030555477,
+      "grad_norm": 148.8917999267578,
+      "learning_rate": 9.58e-07,
+      "loss": 32.3255,
       "step": 4790
     },
     {
-      "epoch": 0.01939260737646303,
-      "grad_norm": 792.7994995117188,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 170.3208,
+      "epoch": 0.009696303688231515,
+      "grad_norm": 392.4678649902344,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 36.1733,
       "step": 4800
     },
     {
-      "epoch": 0.019433008641830663,
-      "grad_norm": 844.0726318359375,
-      "learning_rate": 9.62e-06,
-      "loss": 224.4318,
+      "epoch": 0.009716504320915332,
+      "grad_norm": 1750.965087890625,
+      "learning_rate": 9.62e-07,
+      "loss": 28.3648,
       "step": 4810
     },
     {
-      "epoch": 0.019473409907198293,
-      "grad_norm": 2132.043701171875,
-      "learning_rate": 9.640000000000001e-06,
-      "loss": 252.1749,
+      "epoch": 0.009736704953599146,
+      "grad_norm": 149.6742706298828,
+      "learning_rate": 9.64e-07,
+      "loss": 16.249,
       "step": 4820
     },
     {
-      "epoch": 0.019513811172565926,
-      "grad_norm": 2546.78271484375,
-      "learning_rate": 9.66e-06,
-      "loss": 207.1322,
+      "epoch": 0.009756905586282963,
+      "grad_norm": 268.1315002441406,
+      "learning_rate": 9.660000000000002e-07,
+      "loss": 24.6534,
       "step": 4830
     },
     {
-      "epoch": 0.019554212437933555,
-      "grad_norm": 1447.4708251953125,
-      "learning_rate": 9.68e-06,
-      "loss": 163.8928,
+      "epoch": 0.009777106218966778,
+      "grad_norm": 312.3633728027344,
+      "learning_rate": 9.68e-07,
+      "loss": 19.7529,
       "step": 4840
     },
     {
-      "epoch": 0.01959461370330119,
-      "grad_norm": 1765.4603271484375,
-      "learning_rate": 9.7e-06,
-      "loss": 285.0885,
+      "epoch": 0.009797306851650594,
+      "grad_norm": 335.4103698730469,
+      "learning_rate": 9.7e-07,
+      "loss": 25.3802,
       "step": 4850
     },
     {
-      "epoch": 0.019635014968668818,
-      "grad_norm": 385.1683349609375,
-      "learning_rate": 9.72e-06,
-      "loss": 159.1181,
+      "epoch": 0.009817507484334409,
+      "grad_norm": 588.1651000976562,
+      "learning_rate": 9.72e-07,
+      "loss": 37.7548,
       "step": 4860
     },
     {
-      "epoch": 0.01967541623403645,
-      "grad_norm": 3586.742919921875,
-      "learning_rate": 9.74e-06,
-      "loss": 281.9318,
+      "epoch": 0.009837708117018226,
+      "grad_norm": 1084.7984619140625,
+      "learning_rate": 9.740000000000001e-07,
+      "loss": 28.1253,
       "step": 4870
     },
     {
-      "epoch": 0.01971581749940408,
-      "grad_norm": 1368.7694091796875,
-      "learning_rate": 9.760000000000001e-06,
-      "loss": 191.7896,
+      "epoch": 0.00985790874970204,
+      "grad_norm": 841.3825073242188,
+      "learning_rate": 9.76e-07,
+      "loss": 17.4928,
       "step": 4880
     },
     {
-      "epoch": 0.019756218764771714,
-      "grad_norm": 826.7830810546875,
-      "learning_rate": 9.78e-06,
-      "loss": 174.5084,
+      "epoch": 0.009878109382385857,
+      "grad_norm": 258.460205078125,
+      "learning_rate": 9.78e-07,
+      "loss": 26.3879,
       "step": 4890
     },
     {
-      "epoch": 0.019796620030139343,
-      "grad_norm": 890.3446044921875,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 290.2659,
+      "epoch": 0.009898310015069672,
+      "grad_norm": 299.3013916015625,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 13.678,
       "step": 4900
     },
     {
-      "epoch": 0.019837021295506976,
-      "grad_norm": 1318.1915283203125,
-      "learning_rate": 9.820000000000001e-06,
-      "loss": 272.5934,
+      "epoch": 0.009918510647753488,
+      "grad_norm": 191.6232452392578,
+      "learning_rate": 9.82e-07,
+      "loss": 21.8842,
       "step": 4910
     },
     {
-      "epoch": 0.019877422560874606,
-      "grad_norm": 772.6763916015625,
-      "learning_rate": 9.84e-06,
-      "loss": 151.6077,
+      "epoch": 0.009938711280437303,
+      "grad_norm": 565.6702270507812,
+      "learning_rate": 9.84e-07,
+      "loss": 11.0409,
       "step": 4920
     },
     {
-      "epoch": 0.01991782382624224,
-      "grad_norm": 1587.762451171875,
-      "learning_rate": 9.86e-06,
-      "loss": 248.4316,
+      "epoch": 0.00995891191312112,
+      "grad_norm": 291.95526123046875,
+      "learning_rate": 9.86e-07,
+      "loss": 36.9195,
       "step": 4930
     },
     {
-      "epoch": 0.01995822509160987,
-      "grad_norm": 1252.715576171875,
-      "learning_rate": 9.88e-06,
-      "loss": 258.2403,
+      "epoch": 0.009979112545804934,
+      "grad_norm": 428.879150390625,
+      "learning_rate": 9.880000000000001e-07,
+      "loss": 26.7375,
       "step": 4940
     },
     {
-      "epoch": 0.0199986263569775,
-      "grad_norm": 1054.4132080078125,
-      "learning_rate": 9.900000000000002e-06,
-      "loss": 212.0352,
+      "epoch": 0.00999931317848875,
+      "grad_norm": 177.6942138671875,
+      "learning_rate": 9.9e-07,
+      "loss": 26.29,
       "step": 4950
     },
     {
-      "epoch": 0.02003902762234513,
-      "grad_norm": 1351.8643798828125,
-      "learning_rate": 9.92e-06,
-      "loss": 135.1369,
+      "epoch": 0.010019513811172566,
+      "grad_norm": 662.8630981445312,
+      "learning_rate": 9.92e-07,
+      "loss": 25.1648,
       "step": 4960
     },
     {
-      "epoch": 0.020079428887712764,
-      "grad_norm": 0.0,
-      "learning_rate": 9.940000000000001e-06,
-      "loss": 259.982,
+      "epoch": 0.010039714443856382,
+      "grad_norm": 489.55889892578125,
+      "learning_rate": 9.940000000000001e-07,
+      "loss": 41.3737,
       "step": 4970
     },
     {
-      "epoch": 0.020119830153080394,
-      "grad_norm": 1335.3504638671875,
-      "learning_rate": 9.96e-06,
-      "loss": 232.5153,
+      "epoch": 0.010059915076540197,
+      "grad_norm": 350.4941101074219,
+      "learning_rate": 9.96e-07,
+      "loss": 42.1294,
       "step": 4980
     },
     {
-      "epoch": 0.020160231418448027,
-      "grad_norm": 677.4559326171875,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 240.9019,
+      "epoch": 0.010080115709224013,
+      "grad_norm": 320.653076171875,
+      "learning_rate": 9.98e-07,
+      "loss": 26.9559,
       "step": 4990
     },
     {
-      "epoch": 0.020200632683815656,
-      "grad_norm": 1134.4188232421875,
-      "learning_rate": 1e-05,
-      "loss": 172.6664,
+      "epoch": 0.010100316341907828,
+      "grad_norm": 231.248046875,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 6.5142,
       "step": 5000
     },
     {
-      "epoch": 0.02024103394918329,
-      "grad_norm": 1606.3121337890625,
-      "learning_rate": 1.002e-05,
-      "loss": 218.7504,
+      "epoch": 0.010120516974591645,
+      "grad_norm": 521.7716674804688,
+      "learning_rate": 1.002e-06,
+      "loss": 31.8698,
       "step": 5010
     },
     {
-      "epoch": 0.02028143521455092,
-      "grad_norm": 1248.30615234375,
-      "learning_rate": 1.004e-05,
-      "loss": 143.733,
+      "epoch": 0.01014071760727546,
+      "grad_norm": 800.10888671875,
+      "learning_rate": 1.004e-06,
+      "loss": 28.1427,
       "step": 5020
     },
     {
-      "epoch": 0.020321836479918552,
-      "grad_norm": 1033.4849853515625,
-      "learning_rate": 1.006e-05,
-      "loss": 252.8728,
+      "epoch": 0.010160918239959276,
+      "grad_norm": 420.4501037597656,
+      "learning_rate": 1.006e-06,
+      "loss": 28.2438,
       "step": 5030
     },
     {
-      "epoch": 0.020362237745286182,
-      "grad_norm": 1231.0445556640625,
-      "learning_rate": 1.008e-05,
-      "loss": 243.1658,
+      "epoch": 0.010181118872643091,
+      "grad_norm": 480.6717224121094,
+      "learning_rate": 1.0080000000000001e-06,
+      "loss": 20.2761,
       "step": 5040
     },
     {
-      "epoch": 0.020402639010653815,
-      "grad_norm": 1193.17236328125,
-      "learning_rate": 1.0100000000000002e-05,
-      "loss": 181.7209,
+      "epoch": 0.010201319505326907,
+      "grad_norm": 1004.8648681640625,
+      "learning_rate": 1.01e-06,
+      "loss": 22.479,
       "step": 5050
     },
     {
-      "epoch": 0.020443040276021444,
-      "grad_norm": 0.0,
-      "learning_rate": 1.012e-05,
-      "loss": 276.8536,
+      "epoch": 0.010221520138010722,
+      "grad_norm": 591.0374755859375,
+      "learning_rate": 1.012e-06,
+      "loss": 12.5654,
       "step": 5060
     },
     {
-      "epoch": 0.020483441541389077,
-      "grad_norm": 1306.9161376953125,
-      "learning_rate": 1.0140000000000001e-05,
-      "loss": 205.8765,
+      "epoch": 0.010241720770694539,
+      "grad_norm": 921.0972290039062,
+      "learning_rate": 1.0140000000000002e-06,
+      "loss": 28.9,
       "step": 5070
     },
     {
-      "epoch": 0.020523842806756707,
-      "grad_norm": 3135.527099609375,
-      "learning_rate": 1.016e-05,
-      "loss": 227.9693,
+      "epoch": 0.010261921403378354,
+      "grad_norm": 537.0875244140625,
+      "learning_rate": 1.016e-06,
+      "loss": 14.163,
       "step": 5080
     },
     {
-      "epoch": 0.02056424407212434,
-      "grad_norm": 2155.43994140625,
-      "learning_rate": 1.018e-05,
-      "loss": 284.9991,
+      "epoch": 0.01028212203606217,
+      "grad_norm": 672.0023193359375,
+      "learning_rate": 1.018e-06,
+      "loss": 40.8729,
       "step": 5090
     },
     {
-      "epoch": 0.02060464533749197,
-      "grad_norm": 528.388427734375,
-      "learning_rate": 1.02e-05,
-      "loss": 214.0478,
+      "epoch": 0.010302322668745985,
+      "grad_norm": 761.9652709960938,
+      "learning_rate": 1.02e-06,
+      "loss": 29.6747,
       "step": 5100
     },
     {
-      "epoch": 0.020645046602859603,
-      "grad_norm": 924.0242309570312,
-      "learning_rate": 1.022e-05,
-      "loss": 253.1813,
+      "epoch": 0.010322523301429801,
+      "grad_norm": 934.7273559570312,
+      "learning_rate": 1.0220000000000001e-06,
+      "loss": 23.6142,
       "step": 5110
     },
     {
-      "epoch": 0.020685447868227232,
-      "grad_norm": 1019.5430297851562,
-      "learning_rate": 1.024e-05,
-      "loss": 193.0214,
+      "epoch": 0.010342723934113616,
+      "grad_norm": 364.7134704589844,
+      "learning_rate": 1.024e-06,
+      "loss": 29.4059,
       "step": 5120
     },
     {
-      "epoch": 0.020725849133594865,
-      "grad_norm": 3559.9951171875,
-      "learning_rate": 1.026e-05,
-      "loss": 180.7234,
+      "epoch": 0.010362924566797433,
+      "grad_norm": 662.9189453125,
+      "learning_rate": 1.026e-06,
+      "loss": 25.676,
       "step": 5130
     },
     {
-      "epoch": 0.020766250398962495,
-      "grad_norm": 686.9107055664062,
-      "learning_rate": 1.0280000000000002e-05,
-      "loss": 177.0602,
+      "epoch": 0.010383125199481247,
+      "grad_norm": 1470.47802734375,
+      "learning_rate": 1.0280000000000002e-06,
+      "loss": 58.6466,
       "step": 5140
     },
     {
-      "epoch": 0.020806651664330128,
-      "grad_norm": 616.1886596679688,
-      "learning_rate": 1.03e-05,
-      "loss": 150.3324,
+      "epoch": 0.010403325832165064,
+      "grad_norm": 1297.0006103515625,
+      "learning_rate": 1.03e-06,
+      "loss": 34.2537,
       "step": 5150
     },
     {
-      "epoch": 0.020847052929697758,
-      "grad_norm": 768.7440185546875,
-      "learning_rate": 1.0320000000000001e-05,
-      "loss": 218.8612,
+      "epoch": 0.010423526464848879,
+      "grad_norm": 546.3383178710938,
+      "learning_rate": 1.032e-06,
+      "loss": 18.1663,
       "step": 5160
     },
     {
-      "epoch": 0.02088745419506539,
-      "grad_norm": 927.5042724609375,
-      "learning_rate": 1.0340000000000001e-05,
-      "loss": 247.3814,
+      "epoch": 0.010443727097532695,
+      "grad_norm": 1221.919189453125,
+      "learning_rate": 1.0340000000000002e-06,
+      "loss": 27.4059,
       "step": 5170
     },
     {
-      "epoch": 0.02092785546043302,
-      "grad_norm": 1545.854736328125,
-      "learning_rate": 1.036e-05,
-      "loss": 202.6357,
+      "epoch": 0.01046392773021651,
+      "grad_norm": 460.9737243652344,
+      "learning_rate": 1.0360000000000001e-06,
+      "loss": 16.2116,
       "step": 5180
     },
     {
-      "epoch": 0.020968256725800653,
-      "grad_norm": 950.5018310546875,
-      "learning_rate": 1.038e-05,
-      "loss": 245.9979,
+      "epoch": 0.010484128362900327,
+      "grad_norm": 413.9505920410156,
+      "learning_rate": 1.038e-06,
+      "loss": 21.2082,
       "step": 5190
     },
     {
-      "epoch": 0.021008657991168283,
-      "grad_norm": 829.0062866210938,
-      "learning_rate": 1.04e-05,
-      "loss": 244.4165,
+      "epoch": 0.010504328995584141,
+      "grad_norm": 227.77557373046875,
+      "learning_rate": 1.04e-06,
+      "loss": 27.9558,
       "step": 5200
     },
     {
-      "epoch": 0.021049059256535916,
-      "grad_norm": 1268.8380126953125,
-      "learning_rate": 1.042e-05,
-      "loss": 216.124,
+      "epoch": 0.010524529628267958,
+      "grad_norm": 606.6741943359375,
+      "learning_rate": 1.0420000000000001e-06,
+      "loss": 20.9653,
       "step": 5210
     },
     {
-      "epoch": 0.021089460521903546,
-      "grad_norm": 1645.070068359375,
-      "learning_rate": 1.0440000000000002e-05,
-      "loss": 224.5982,
+      "epoch": 0.010544730260951773,
+      "grad_norm": 541.234619140625,
+      "learning_rate": 1.044e-06,
+      "loss": 36.7112,
       "step": 5220
     },
     {
-      "epoch": 0.02112986178727118,
-      "grad_norm": 1425.95947265625,
-      "learning_rate": 1.046e-05,
-      "loss": 241.4969,
+      "epoch": 0.01056493089363559,
+      "grad_norm": 145.1033477783203,
+      "learning_rate": 1.046e-06,
+      "loss": 30.0167,
       "step": 5230
     },
     {
-      "epoch": 0.021170263052638808,
-      "grad_norm": 1239.9669189453125,
-      "learning_rate": 1.0480000000000001e-05,
-      "loss": 296.2492,
+      "epoch": 0.010585131526319404,
+      "grad_norm": 373.5177917480469,
+      "learning_rate": 1.0480000000000002e-06,
+      "loss": 31.7274,
       "step": 5240
     },
     {
-      "epoch": 0.02121066431800644,
-      "grad_norm": 3276.483154296875,
-      "learning_rate": 1.05e-05,
-      "loss": 257.1036,
+      "epoch": 0.01060533215900322,
+      "grad_norm": 563.7691650390625,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 30.9538,
       "step": 5250
     },
     {
-      "epoch": 0.02125106558337407,
-      "grad_norm": 1606.2689208984375,
-      "learning_rate": 1.0520000000000001e-05,
-      "loss": 273.2055,
+      "epoch": 0.010625532791687035,
+      "grad_norm": 170.43959045410156,
+      "learning_rate": 1.052e-06,
+      "loss": 44.6468,
       "step": 5260
     },
     {
-      "epoch": 0.021291466848741704,
-      "grad_norm": 2068.524658203125,
-      "learning_rate": 1.0539999999999999e-05,
-      "loss": 209.8239,
+      "epoch": 0.010645733424370852,
+      "grad_norm": 695.9778442382812,
+      "learning_rate": 1.054e-06,
+      "loss": 42.9755,
       "step": 5270
     },
     {
-      "epoch": 0.021331868114109333,
-      "grad_norm": 1133.3856201171875,
-      "learning_rate": 1.056e-05,
-      "loss": 248.1236,
+      "epoch": 0.010665934057054667,
+      "grad_norm": 390.760009765625,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 28.6079,
       "step": 5280
     },
     {
-      "epoch": 0.021372269379476967,
-      "grad_norm": 1275.0018310546875,
-      "learning_rate": 1.058e-05,
-      "loss": 238.0178,
+      "epoch": 0.010686134689738483,
+      "grad_norm": 483.7234191894531,
+      "learning_rate": 1.058e-06,
+      "loss": 22.693,
       "step": 5290
     },
     {
-      "epoch": 0.021412670644844596,
-      "grad_norm": 759.3450927734375,
-      "learning_rate": 1.06e-05,
-      "loss": 269.6895,
+      "epoch": 0.010706335322422298,
+      "grad_norm": 694.895263671875,
+      "learning_rate": 1.06e-06,
+      "loss": 31.7093,
       "step": 5300
     },
     {
-      "epoch": 0.02145307191021223,
-      "grad_norm": 4057.62744140625,
-      "learning_rate": 1.062e-05,
-      "loss": 268.4212,
+      "epoch": 0.010726535955106115,
+      "grad_norm": 0.03282935544848442,
+      "learning_rate": 1.0620000000000002e-06,
+      "loss": 29.5697,
       "step": 5310
     },
     {
-      "epoch": 0.02149347317557986,
-      "grad_norm": 933.5989379882812,
-      "learning_rate": 1.064e-05,
-      "loss": 200.5304,
+      "epoch": 0.01074673658778993,
+      "grad_norm": 197.685546875,
+      "learning_rate": 1.064e-06,
+      "loss": 25.7651,
       "step": 5320
     },
     {
-      "epoch": 0.021533874440947492,
-      "grad_norm": 709.9401245117188,
-      "learning_rate": 1.0660000000000001e-05,
-      "loss": 227.5934,
+      "epoch": 0.010766937220473746,
+      "grad_norm": 692.8652954101562,
+      "learning_rate": 1.066e-06,
+      "loss": 28.2902,
       "step": 5330
     },
     {
-      "epoch": 0.02157427570631512,
-      "grad_norm": 1363.613525390625,
-      "learning_rate": 1.0680000000000001e-05,
-      "loss": 195.4376,
+      "epoch": 0.01078713785315756,
+      "grad_norm": 295.2193603515625,
+      "learning_rate": 1.0680000000000002e-06,
+      "loss": 35.8585,
       "step": 5340
     },
     {
-      "epoch": 0.021614676971682754,
-      "grad_norm": 860.6060180664062,
-      "learning_rate": 1.0700000000000001e-05,
-      "loss": 220.5302,
+      "epoch": 0.010807338485841377,
+      "grad_norm": 179.0888214111328,
+      "learning_rate": 1.0700000000000001e-06,
+      "loss": 13.3468,
       "step": 5350
     },
     {
-      "epoch": 0.021655078237050384,
-      "grad_norm": 809.3516845703125,
-      "learning_rate": 1.072e-05,
-      "loss": 289.8672,
+      "epoch": 0.010827539118525192,
+      "grad_norm": 369.4256896972656,
+      "learning_rate": 1.072e-06,
+      "loss": 42.9108,
       "step": 5360
     },
     {
-      "epoch": 0.021695479502418017,
-      "grad_norm": 688.4315185546875,
-      "learning_rate": 1.074e-05,
-      "loss": 231.9644,
+      "epoch": 0.010847739751209009,
+      "grad_norm": 380.3387756347656,
+      "learning_rate": 1.074e-06,
+      "loss": 35.2701,
       "step": 5370
     },
     {
-      "epoch": 0.021735880767785647,
-      "grad_norm": 2667.476318359375,
-      "learning_rate": 1.076e-05,
-      "loss": 270.5944,
+      "epoch": 0.010867940383892823,
+      "grad_norm": 735.1011962890625,
+      "learning_rate": 1.0760000000000002e-06,
+      "loss": 37.2414,
       "step": 5380
     },
     {
-      "epoch": 0.02177628203315328,
-      "grad_norm": 1891.450927734375,
-      "learning_rate": 1.0780000000000002e-05,
-      "loss": 249.0208,
+      "epoch": 0.01088814101657664,
+      "grad_norm": 459.38720703125,
+      "learning_rate": 1.078e-06,
+      "loss": 36.9988,
       "step": 5390
     },
     {
-      "epoch": 0.02181668329852091,
-      "grad_norm": 3481.829345703125,
-      "learning_rate": 1.08e-05,
-      "loss": 301.6664,
+      "epoch": 0.010908341649260455,
+      "grad_norm": 1027.1058349609375,
+      "learning_rate": 1.08e-06,
+      "loss": 22.0428,
       "step": 5400
     },
     {
-      "epoch": 0.021857084563888542,
-      "grad_norm": 956.6407470703125,
-      "learning_rate": 1.0820000000000001e-05,
-      "loss": 245.2214,
+      "epoch": 0.010928542281944271,
+      "grad_norm": 2424.2099609375,
+      "learning_rate": 1.0820000000000002e-06,
+      "loss": 31.5498,
       "step": 5410
     },
     {
-      "epoch": 0.021897485829256172,
-      "grad_norm": 4194.12841796875,
-      "learning_rate": 1.084e-05,
-      "loss": 260.2813,
+      "epoch": 0.010948742914628086,
+      "grad_norm": 194.22132873535156,
+      "learning_rate": 1.0840000000000001e-06,
+      "loss": 73.3618,
       "step": 5420
     },
     {
-      "epoch": 0.021937887094623805,
-      "grad_norm": 3680.693603515625,
-      "learning_rate": 1.0860000000000001e-05,
-      "loss": 249.7451,
+      "epoch": 0.010968943547311902,
+      "grad_norm": 756.6248168945312,
+      "learning_rate": 1.086e-06,
+      "loss": 74.6274,
       "step": 5430
     },
     {
-      "epoch": 0.021978288359991435,
-      "grad_norm": 1049.5306396484375,
-      "learning_rate": 1.088e-05,
-      "loss": 227.2931,
+      "epoch": 0.010989144179995717,
+      "grad_norm": 472.0841369628906,
+      "learning_rate": 1.088e-06,
+      "loss": 22.9419,
       "step": 5440
     },
     {
-      "epoch": 0.022018689625359068,
-      "grad_norm": 10489.931640625,
-      "learning_rate": 1.09e-05,
-      "loss": 334.8502,
+      "epoch": 0.011009344812679534,
+      "grad_norm": 600.0186157226562,
+      "learning_rate": 1.0900000000000002e-06,
+      "loss": 25.0131,
       "step": 5450
     },
     {
-      "epoch": 0.022059090890726697,
-      "grad_norm": 1121.204833984375,
-      "learning_rate": 1.092e-05,
-      "loss": 220.4895,
+      "epoch": 0.011029545445363349,
+      "grad_norm": 305.5976257324219,
+      "learning_rate": 1.092e-06,
+      "loss": 31.3711,
       "step": 5460
     },
     {
-      "epoch": 0.02209949215609433,
-      "grad_norm": 1375.3389892578125,
-      "learning_rate": 1.094e-05,
-      "loss": 212.8332,
+      "epoch": 0.011049746078047165,
+      "grad_norm": 108.77481842041016,
+      "learning_rate": 1.094e-06,
+      "loss": 17.3308,
       "step": 5470
     },
     {
-      "epoch": 0.02213989342146196,
-      "grad_norm": 2596.01318359375,
-      "learning_rate": 1.096e-05,
-      "loss": 242.3982,
+      "epoch": 0.01106994671073098,
+      "grad_norm": 1116.296142578125,
+      "learning_rate": 1.0960000000000002e-06,
+      "loss": 37.9978,
       "step": 5480
     },
     {
-      "epoch": 0.022180294686829593,
-      "grad_norm": 1790.5255126953125,
-      "learning_rate": 1.098e-05,
-      "loss": 228.799,
+      "epoch": 0.011090147343414796,
+      "grad_norm": 390.4914245605469,
+      "learning_rate": 1.0980000000000001e-06,
+      "loss": 14.3431,
       "step": 5490
     },
     {
-      "epoch": 0.022220695952197222,
-      "grad_norm": 1782.9239501953125,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 247.0614,
+      "epoch": 0.011110347976098611,
+      "grad_norm": 598.1354370117188,
+      "learning_rate": 1.1e-06,
+      "loss": 30.984,
       "step": 5500
     },
     {
-      "epoch": 0.022261097217564856,
-      "grad_norm": 3071.753173828125,
-      "learning_rate": 1.1020000000000001e-05,
-      "loss": 190.9815,
+      "epoch": 0.011130548608782428,
+      "grad_norm": 711.5731201171875,
+      "learning_rate": 1.1020000000000002e-06,
+      "loss": 30.1973,
       "step": 5510
     },
     {
-      "epoch": 0.022301498482932485,
-      "grad_norm": 778.1385498046875,
-      "learning_rate": 1.1040000000000001e-05,
-      "loss": 208.1198,
+      "epoch": 0.011150749241466243,
+      "grad_norm": 371.409423828125,
+      "learning_rate": 1.1040000000000001e-06,
+      "loss": 24.4026,
       "step": 5520
     },
     {
-      "epoch": 0.022341899748300118,
-      "grad_norm": 4282.1689453125,
-      "learning_rate": 1.106e-05,
-      "loss": 187.0153,
+      "epoch": 0.011170949874150059,
+      "grad_norm": 665.7667846679688,
+      "learning_rate": 1.106e-06,
+      "loss": 19.5368,
       "step": 5530
     },
     {
-      "epoch": 0.022382301013667748,
-      "grad_norm": 3060.544921875,
-      "learning_rate": 1.108e-05,
-      "loss": 243.2019,
+      "epoch": 0.011191150506833874,
+      "grad_norm": 502.86566162109375,
+      "learning_rate": 1.108e-06,
+      "loss": 18.791,
       "step": 5540
     },
     {
-      "epoch": 0.02242270227903538,
-      "grad_norm": 1174.474853515625,
-      "learning_rate": 1.11e-05,
-      "loss": 240.466,
+      "epoch": 0.01121135113951769,
+      "grad_norm": 404.4401550292969,
+      "learning_rate": 1.1100000000000002e-06,
+      "loss": 20.3444,
       "step": 5550
     },
     {
-      "epoch": 0.02246310354440301,
-      "grad_norm": 835.2548828125,
-      "learning_rate": 1.112e-05,
-      "loss": 195.4574,
+      "epoch": 0.011231551772201505,
+      "grad_norm": 323.6523132324219,
+      "learning_rate": 1.1120000000000001e-06,
+      "loss": 36.3596,
       "step": 5560
     },
     {
-      "epoch": 0.022503504809770643,
-      "grad_norm": 679.6596069335938,
-      "learning_rate": 1.114e-05,
-      "loss": 116.5087,
+      "epoch": 0.011251752404885322,
+      "grad_norm": 743.5354614257812,
+      "learning_rate": 1.114e-06,
+      "loss": 29.4033,
       "step": 5570
     },
     {
-      "epoch": 0.022543906075138273,
-      "grad_norm": 4872.7197265625,
-      "learning_rate": 1.1160000000000002e-05,
-      "loss": 208.2604,
+      "epoch": 0.011271953037569137,
+      "grad_norm": 281.8879699707031,
+      "learning_rate": 1.1160000000000002e-06,
+      "loss": 23.6649,
       "step": 5580
     },
     {
-      "epoch": 0.022584307340505906,
-      "grad_norm": 507.1872253417969,
-      "learning_rate": 1.118e-05,
-      "loss": 240.8744,
+      "epoch": 0.011292153670252953,
+      "grad_norm": 812.1799926757812,
+      "learning_rate": 1.1180000000000001e-06,
+      "loss": 34.1818,
       "step": 5590
     },
     {
-      "epoch": 0.022624708605873536,
-      "grad_norm": 1005.294189453125,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 184.084,
+      "epoch": 0.011312354302936768,
+      "grad_norm": 148.88833618164062,
+      "learning_rate": 1.12e-06,
+      "loss": 28.1046,
       "step": 5600
     },
     {
-      "epoch": 0.02266510987124117,
-      "grad_norm": 7254.05029296875,
-      "learning_rate": 1.122e-05,
-      "loss": 216.7616,
+      "epoch": 0.011332554935620584,
+      "grad_norm": 471.3416748046875,
+      "learning_rate": 1.122e-06,
+      "loss": 24.278,
       "step": 5610
     },
     {
-      "epoch": 0.0227055111366088,
-      "grad_norm": 1527.6485595703125,
-      "learning_rate": 1.124e-05,
-      "loss": 197.5231,
+      "epoch": 0.0113527555683044,
+      "grad_norm": 353.2952575683594,
+      "learning_rate": 1.1240000000000002e-06,
+      "loss": 30.7999,
       "step": 5620
     },
     {
-      "epoch": 0.02274591240197643,
-      "grad_norm": 1460.7916259765625,
-      "learning_rate": 1.126e-05,
-      "loss": 280.5447,
+      "epoch": 0.011372956200988216,
+      "grad_norm": 585.4158325195312,
+      "learning_rate": 1.126e-06,
+      "loss": 16.3788,
       "step": 5630
     },
     {
-      "epoch": 0.02278631366734406,
-      "grad_norm": 1024.780517578125,
-      "learning_rate": 1.128e-05,
-      "loss": 248.3504,
+      "epoch": 0.01139315683367203,
+      "grad_norm": 629.95703125,
+      "learning_rate": 1.128e-06,
+      "loss": 22.8701,
       "step": 5640
     },
     {
-      "epoch": 0.022826714932711694,
-      "grad_norm": 1925.320068359375,
-      "learning_rate": 1.13e-05,
-      "loss": 226.1158,
+      "epoch": 0.011413357466355847,
+      "grad_norm": 1085.2030029296875,
+      "learning_rate": 1.1300000000000002e-06,
+      "loss": 31.6779,
       "step": 5650
     },
     {
-      "epoch": 0.022867116198079324,
-      "grad_norm": 1352.9869384765625,
-      "learning_rate": 1.132e-05,
-      "loss": 243.9569,
+      "epoch": 0.011433558099039662,
+      "grad_norm": 1200.4349365234375,
+      "learning_rate": 1.1320000000000001e-06,
+      "loss": 26.5144,
       "step": 5660
     },
     {
-      "epoch": 0.022907517463446957,
-      "grad_norm": 1091.9189453125,
-      "learning_rate": 1.134e-05,
-      "loss": 186.5363,
+      "epoch": 0.011453758731723478,
+      "grad_norm": 869.242919921875,
+      "learning_rate": 1.134e-06,
+      "loss": 26.5842,
       "step": 5670
     },
     {
-      "epoch": 0.022947918728814586,
-      "grad_norm": 544.18994140625,
-      "learning_rate": 1.1360000000000001e-05,
-      "loss": 178.7458,
+      "epoch": 0.011473959364407293,
+      "grad_norm": 107.99213409423828,
+      "learning_rate": 1.1360000000000002e-06,
+      "loss": 27.156,
       "step": 5680
     },
     {
-      "epoch": 0.02298831999418222,
-      "grad_norm": 0.0,
-      "learning_rate": 1.1380000000000001e-05,
-      "loss": 237.9985,
+      "epoch": 0.01149415999709111,
+      "grad_norm": 908.802734375,
+      "learning_rate": 1.1380000000000002e-06,
+      "loss": 34.6566,
       "step": 5690
     },
     {
-      "epoch": 0.02302872125954985,
-      "grad_norm": 1084.4674072265625,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 201.7748,
+      "epoch": 0.011514360629774924,
+      "grad_norm": 497.6568298339844,
+      "learning_rate": 1.14e-06,
+      "loss": 21.8019,
       "step": 5700
     },
     {
-      "epoch": 0.023069122524917482,
-      "grad_norm": 1035.4659423828125,
-      "learning_rate": 1.142e-05,
-      "loss": 207.2124,
+      "epoch": 0.011534561262458741,
+      "grad_norm": 293.9489440917969,
+      "learning_rate": 1.142e-06,
+      "loss": 22.4049,
       "step": 5710
     },
     {
-      "epoch": 0.02310952379028511,
-      "grad_norm": 733.1090087890625,
-      "learning_rate": 1.144e-05,
-      "loss": 178.1889,
+      "epoch": 0.011554761895142556,
+      "grad_norm": 295.17230224609375,
+      "learning_rate": 1.1440000000000002e-06,
+      "loss": 37.8035,
       "step": 5720
     },
     {
-      "epoch": 0.023149925055652745,
-      "grad_norm": 876.4588012695312,
-      "learning_rate": 1.146e-05,
-      "loss": 158.9155,
+      "epoch": 0.011574962527826372,
+      "grad_norm": 475.5271301269531,
+      "learning_rate": 1.1460000000000001e-06,
+      "loss": 24.764,
       "step": 5730
     },
     {
-      "epoch": 0.023190326321020374,
-      "grad_norm": 657.6229248046875,
-      "learning_rate": 1.148e-05,
-      "loss": 194.4797,
+      "epoch": 0.011595163160510187,
+      "grad_norm": 1417.2218017578125,
+      "learning_rate": 1.148e-06,
+      "loss": 31.2254,
       "step": 5740
     },
     {
-      "epoch": 0.023230727586388007,
-      "grad_norm": 1495.9139404296875,
-      "learning_rate": 1.1500000000000002e-05,
-      "loss": 222.9885,
+      "epoch": 0.011615363793194004,
+      "grad_norm": 271.8209533691406,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 14.2182,
       "step": 5750
     },
     {
-      "epoch": 0.023271128851755637,
-      "grad_norm": 2601.6259765625,
-      "learning_rate": 1.152e-05,
-      "loss": 304.1291,
+      "epoch": 0.011635564425877818,
+      "grad_norm": 612.4166259765625,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 18.7975,
       "step": 5760
     },
     {
-      "epoch": 0.02331153011712327,
-      "grad_norm": 1693.1273193359375,
-      "learning_rate": 1.1540000000000001e-05,
-      "loss": 199.6243,
+      "epoch": 0.011655765058561635,
+      "grad_norm": 314.46990966796875,
+      "learning_rate": 1.154e-06,
+      "loss": 19.6362,
       "step": 5770
     },
     {
-      "epoch": 0.0233519313824909,
-      "grad_norm": 945.2630615234375,
-      "learning_rate": 1.156e-05,
-      "loss": 126.1058,
+      "epoch": 0.01167596569124545,
+      "grad_norm": 627.936767578125,
+      "learning_rate": 1.156e-06,
+      "loss": 38.5957,
       "step": 5780
     },
     {
-      "epoch": 0.023392332647858533,
-      "grad_norm": 616.036865234375,
-      "learning_rate": 1.1580000000000001e-05,
-      "loss": 265.3511,
+      "epoch": 0.011696166323929266,
+      "grad_norm": 1000.9813842773438,
+      "learning_rate": 1.1580000000000002e-06,
+      "loss": 34.4684,
       "step": 5790
     },
     {
-      "epoch": 0.023432733913226162,
-      "grad_norm": 908.6356811523438,
-      "learning_rate": 1.16e-05,
-      "loss": 215.007,
+      "epoch": 0.011716366956613081,
+      "grad_norm": 637.470703125,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 34.59,
       "step": 5800
     },
     {
-      "epoch": 0.023473135178593795,
-      "grad_norm": 1166.1875,
-      "learning_rate": 1.162e-05,
-      "loss": 171.8394,
+      "epoch": 0.011736567589296898,
+      "grad_norm": 750.3966674804688,
+      "learning_rate": 1.162e-06,
+      "loss": 44.1128,
       "step": 5810
     },
     {
-      "epoch": 0.023513536443961425,
-      "grad_norm": 936.7083740234375,
-      "learning_rate": 1.164e-05,
-      "loss": 209.5779,
+      "epoch": 0.011756768221980712,
+      "grad_norm": 1139.5133056640625,
+      "learning_rate": 1.1640000000000002e-06,
+      "loss": 23.7449,
       "step": 5820
     },
     {
-      "epoch": 0.023553937709329058,
-      "grad_norm": 689.0086669921875,
-      "learning_rate": 1.166e-05,
-      "loss": 165.4548,
+      "epoch": 0.011776968854664529,
+      "grad_norm": 370.25567626953125,
+      "learning_rate": 1.1660000000000001e-06,
+      "loss": 17.9951,
       "step": 5830
     },
     {
-      "epoch": 0.023594338974696687,
-      "grad_norm": 1105.93212890625,
-      "learning_rate": 1.168e-05,
-      "loss": 275.4574,
+      "epoch": 0.011797169487348344,
+      "grad_norm": 194.2151336669922,
+      "learning_rate": 1.168e-06,
+      "loss": 23.3352,
       "step": 5840
     },
     {
-      "epoch": 0.02363474024006432,
-      "grad_norm": 2241.417236328125,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 359.449,
+      "epoch": 0.01181737012003216,
+      "grad_norm": 812.2650756835938,
+      "learning_rate": 1.1700000000000002e-06,
+      "loss": 21.2374,
       "step": 5850
     },
     {
-      "epoch": 0.02367514150543195,
-      "grad_norm": 1302.2642822265625,
-      "learning_rate": 1.172e-05,
-      "loss": 259.1196,
+      "epoch": 0.011837570752715975,
+      "grad_norm": 422.30401611328125,
+      "learning_rate": 1.1720000000000002e-06,
+      "loss": 19.3065,
       "step": 5860
     },
     {
-      "epoch": 0.023715542770799583,
-      "grad_norm": 1307.6990966796875,
-      "learning_rate": 1.1740000000000001e-05,
-      "loss": 215.3923,
+      "epoch": 0.011857771385399792,
+      "grad_norm": 1042.8720703125,
+      "learning_rate": 1.1740000000000001e-06,
+      "loss": 38.1709,
       "step": 5870
     },
     {
-      "epoch": 0.023755944036167213,
-      "grad_norm": 523.6953735351562,
-      "learning_rate": 1.1760000000000001e-05,
-      "loss": 189.7961,
+      "epoch": 0.011877972018083606,
+      "grad_norm": 463.48712158203125,
+      "learning_rate": 1.176e-06,
+      "loss": 20.1112,
       "step": 5880
     },
     {
-      "epoch": 0.023796345301534846,
-      "grad_norm": 1645.471923828125,
-      "learning_rate": 1.178e-05,
-      "loss": 197.7059,
+      "epoch": 0.011898172650767423,
+      "grad_norm": 251.95362854003906,
+      "learning_rate": 1.1780000000000002e-06,
+      "loss": 19.2058,
       "step": 5890
     },
     {
-      "epoch": 0.023836746566902475,
-      "grad_norm": 4609.2265625,
-      "learning_rate": 1.18e-05,
-      "loss": 312.7925,
+      "epoch": 0.011918373283451238,
+      "grad_norm": 644.2528076171875,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 55.8893,
       "step": 5900
     },
     {
-      "epoch": 0.02387714783227011,
-      "grad_norm": 886.41015625,
-      "learning_rate": 1.182e-05,
-      "loss": 227.9965,
+      "epoch": 0.011938573916135054,
+      "grad_norm": 546.8013305664062,
+      "learning_rate": 1.182e-06,
+      "loss": 65.9524,
       "step": 5910
     },
     {
-      "epoch": 0.023917549097637738,
-      "grad_norm": 1345.154052734375,
-      "learning_rate": 1.1840000000000002e-05,
-      "loss": 143.6501,
+      "epoch": 0.011958774548818869,
+      "grad_norm": 649.1343994140625,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 27.965,
       "step": 5920
     },
     {
-      "epoch": 0.02395795036300537,
-      "grad_norm": 2758.603515625,
-      "learning_rate": 1.186e-05,
-      "loss": 189.3002,
+      "epoch": 0.011978975181502685,
+      "grad_norm": 526.4266357421875,
+      "learning_rate": 1.1860000000000002e-06,
+      "loss": 34.2897,
       "step": 5930
     },
     {
-      "epoch": 0.023998351628373,
-      "grad_norm": 933.2481079101562,
-      "learning_rate": 1.1880000000000001e-05,
-      "loss": 170.3514,
+      "epoch": 0.0119991758141865,
+      "grad_norm": 1568.1280517578125,
+      "learning_rate": 1.188e-06,
+      "loss": 35.1816,
       "step": 5940
     },
     {
-      "epoch": 0.024038752893740634,
-      "grad_norm": 1366.0477294921875,
-      "learning_rate": 1.19e-05,
-      "loss": 204.9711,
+      "epoch": 0.012019376446870317,
+      "grad_norm": 318.4168395996094,
+      "learning_rate": 1.19e-06,
+      "loss": 22.1523,
       "step": 5950
     },
     {
-      "epoch": 0.024079154159108263,
-      "grad_norm": 870.7377319335938,
-      "learning_rate": 1.1920000000000001e-05,
-      "loss": 182.9471,
+      "epoch": 0.012039577079554132,
+      "grad_norm": 402.7890930175781,
+      "learning_rate": 1.1920000000000002e-06,
+      "loss": 33.6881,
       "step": 5960
     },
     {
-      "epoch": 0.024119555424475896,
-      "grad_norm": 1715.831298828125,
-      "learning_rate": 1.1940000000000001e-05,
-      "loss": 206.0364,
+      "epoch": 0.012059777712237948,
+      "grad_norm": 497.604736328125,
+      "learning_rate": 1.1940000000000001e-06,
+      "loss": 32.794,
       "step": 5970
     },
     {
-      "epoch": 0.024159956689843526,
-      "grad_norm": 1668.3841552734375,
-      "learning_rate": 1.196e-05,
-      "loss": 226.4282,
+      "epoch": 0.012079978344921763,
+      "grad_norm": 444.6767883300781,
+      "learning_rate": 1.196e-06,
+      "loss": 37.6437,
       "step": 5980
     },
     {
-      "epoch": 0.02420035795521116,
-      "grad_norm": 1048.4296875,
-      "learning_rate": 1.198e-05,
-      "loss": 109.3125,
+      "epoch": 0.01210017897760558,
+      "grad_norm": 863.6239624023438,
+      "learning_rate": 1.1980000000000002e-06,
+      "loss": 26.2152,
       "step": 5990
     },
     {
-      "epoch": 0.02424075922057879,
-      "grad_norm": 879.1497192382812,
-      "learning_rate": 1.2e-05,
-      "loss": 201.5203,
+      "epoch": 0.012120379610289394,
+      "grad_norm": 592.9383544921875,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 14.3755,
       "step": 6000
     },
     {
-      "epoch": 0.02428116048594642,
-      "grad_norm": 1213.1204833984375,
-      "learning_rate": 1.202e-05,
-      "loss": 163.2346,
+      "epoch": 0.01214058024297321,
+      "grad_norm": 1376.8350830078125,
+      "learning_rate": 1.202e-06,
+      "loss": 36.9957,
       "step": 6010
     },
     {
-      "epoch": 0.02432156175131405,
-      "grad_norm": 1427.281494140625,
-      "learning_rate": 1.204e-05,
-      "loss": 174.4642,
+      "epoch": 0.012160780875657026,
+      "grad_norm": 894.4027709960938,
+      "learning_rate": 1.204e-06,
+      "loss": 25.3654,
       "step": 6020
     },
     {
-      "epoch": 0.024361963016681684,
-      "grad_norm": 1276.7529296875,
-      "learning_rate": 1.206e-05,
-      "loss": 204.9076,
+      "epoch": 0.012180981508340842,
+      "grad_norm": 587.0653686523438,
+      "learning_rate": 1.2060000000000002e-06,
+      "loss": 29.8984,
       "step": 6030
     },
     {
-      "epoch": 0.024402364282049314,
-      "grad_norm": 2190.304443359375,
-      "learning_rate": 1.2080000000000001e-05,
-      "loss": 159.9595,
+      "epoch": 0.012201182141024657,
+      "grad_norm": 121.9262466430664,
+      "learning_rate": 1.2080000000000001e-06,
+      "loss": 20.7306,
       "step": 6040
     },
     {
-      "epoch": 0.024442765547416947,
-      "grad_norm": 785.987060546875,
-      "learning_rate": 1.2100000000000001e-05,
-      "loss": 209.5406,
+      "epoch": 0.012221382773708473,
+      "grad_norm": 488.919677734375,
+      "learning_rate": 1.21e-06,
+      "loss": 27.6776,
       "step": 6050
     },
     {
-      "epoch": 0.024483166812784576,
-      "grad_norm": 611.1929931640625,
-      "learning_rate": 1.2120000000000001e-05,
-      "loss": 221.088,
+      "epoch": 0.012241583406392288,
+      "grad_norm": 461.9215393066406,
+      "learning_rate": 1.2120000000000002e-06,
+      "loss": 29.0396,
       "step": 6060
     },
     {
-      "epoch": 0.02452356807815221,
-      "grad_norm": 1893.1925048828125,
-      "learning_rate": 1.214e-05,
-      "loss": 215.4067,
+      "epoch": 0.012261784039076105,
+      "grad_norm": 399.4093322753906,
+      "learning_rate": 1.214e-06,
+      "loss": 60.5062,
       "step": 6070
     },
     {
-      "epoch": 0.02456396934351984,
-      "grad_norm": 4432.23095703125,
-      "learning_rate": 1.216e-05,
-      "loss": 221.237,
+      "epoch": 0.01228198467175992,
+      "grad_norm": 167.42552185058594,
+      "learning_rate": 1.216e-06,
+      "loss": 29.8078,
       "step": 6080
     },
     {
-      "epoch": 0.02460437060888747,
-      "grad_norm": 1176.1290283203125,
-      "learning_rate": 1.2180000000000002e-05,
-      "loss": 220.3091,
+      "epoch": 0.012302185304443734,
+      "grad_norm": 412.9476623535156,
+      "learning_rate": 1.2180000000000002e-06,
+      "loss": 25.1903,
       "step": 6090
     },
     {
-      "epoch": 0.0246447718742551,
-      "grad_norm": 1318.6954345703125,
-      "learning_rate": 1.22e-05,
-      "loss": 250.0935,
+      "epoch": 0.01232238593712755,
+      "grad_norm": 402.8238220214844,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 39.7322,
       "step": 6100
     },
     {
-      "epoch": 0.02468517313962273,
-      "grad_norm": 1313.9632568359375,
-      "learning_rate": 1.2220000000000002e-05,
-      "loss": 226.1311,
+      "epoch": 0.012342586569811366,
+      "grad_norm": 503.3050537109375,
+      "learning_rate": 1.2220000000000001e-06,
+      "loss": 19.9504,
       "step": 6110
     },
     {
-      "epoch": 0.024725574404990364,
-      "grad_norm": 1210.4168701171875,
-      "learning_rate": 1.224e-05,
-      "loss": 348.8165,
+      "epoch": 0.012362787202495182,
+      "grad_norm": 251.07025146484375,
+      "learning_rate": 1.224e-06,
+      "loss": 29.8188,
       "step": 6120
     },
     {
-      "epoch": 0.024765975670357994,
-      "grad_norm": 1387.2384033203125,
-      "learning_rate": 1.2260000000000001e-05,
-      "loss": 208.124,
+      "epoch": 0.012382987835178997,
+      "grad_norm": 1289.874755859375,
+      "learning_rate": 1.2260000000000002e-06,
+      "loss": 26.9208,
       "step": 6130
     },
     {
-      "epoch": 0.024806376935725627,
-      "grad_norm": 1911.8060302734375,
-      "learning_rate": 1.2280000000000001e-05,
-      "loss": 208.8157,
+      "epoch": 0.012403188467862813,
+      "grad_norm": 839.939453125,
+      "learning_rate": 1.2280000000000001e-06,
+      "loss": 29.809,
       "step": 6140
     },
     {
-      "epoch": 0.024846778201093257,
-      "grad_norm": 770.5685424804688,
-      "learning_rate": 1.23e-05,
-      "loss": 205.734,
+      "epoch": 0.012423389100546628,
+      "grad_norm": 0.0,
+      "learning_rate": 1.23e-06,
+      "loss": 10.9996,
       "step": 6150
     },
     {
-      "epoch": 0.02488717946646089,
-      "grad_norm": 718.35498046875,
-      "learning_rate": 1.232e-05,
-      "loss": 200.1725,
+      "epoch": 0.012443589733230445,
+      "grad_norm": 687.5811157226562,
+      "learning_rate": 1.2320000000000002e-06,
+      "loss": 30.5085,
       "step": 6160
     },
     {
-      "epoch": 0.02492758073182852,
-      "grad_norm": 1146.25244140625,
-      "learning_rate": 1.234e-05,
-      "loss": 197.4777,
+      "epoch": 0.01246379036591426,
+      "grad_norm": 127.76237487792969,
+      "learning_rate": 1.234e-06,
+      "loss": 29.9668,
       "step": 6170
     },
     {
-      "epoch": 0.024967981997196152,
-      "grad_norm": 773.2630004882812,
-      "learning_rate": 1.236e-05,
-      "loss": 217.971,
+      "epoch": 0.012483990998598076,
+      "grad_norm": 1359.7757568359375,
+      "learning_rate": 1.2360000000000001e-06,
+      "loss": 30.7715,
       "step": 6180
     },
     {
-      "epoch": 0.025008383262563782,
-      "grad_norm": 888.3221435546875,
-      "learning_rate": 1.238e-05,
-      "loss": 198.487,
+      "epoch": 0.012504191631281891,
+      "grad_norm": 436.6647644042969,
+      "learning_rate": 1.238e-06,
+      "loss": 33.9171,
       "step": 6190
     },
     {
-      "epoch": 0.025048784527931415,
-      "grad_norm": 971.0499267578125,
-      "learning_rate": 1.24e-05,
-      "loss": 239.6828,
+      "epoch": 0.012524392263965707,
+      "grad_norm": 281.2616882324219,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 41.9189,
       "step": 6200
     },
     {
-      "epoch": 0.025089185793299044,
-      "grad_norm": 10692.5146484375,
-      "learning_rate": 1.2420000000000001e-05,
-      "loss": 258.4363,
+      "epoch": 0.012544592896649522,
+      "grad_norm": 758.2962036132812,
+      "learning_rate": 1.2420000000000001e-06,
+      "loss": 33.9747,
       "step": 6210
     },
     {
-      "epoch": 0.025129587058666678,
-      "grad_norm": 791.8817138671875,
-      "learning_rate": 1.244e-05,
-      "loss": 213.3801,
+      "epoch": 0.012564793529333339,
+      "grad_norm": 684.6299438476562,
+      "learning_rate": 1.244e-06,
+      "loss": 35.7783,
       "step": 6220
     },
     {
-      "epoch": 0.025169988324034307,
-      "grad_norm": 1000.6807861328125,
-      "learning_rate": 1.2460000000000001e-05,
-      "loss": 181.9501,
+      "epoch": 0.012584994162017154,
+      "grad_norm": 1113.127197265625,
+      "learning_rate": 1.2460000000000002e-06,
+      "loss": 48.0597,
       "step": 6230
     },
     {
-      "epoch": 0.02521038958940194,
-      "grad_norm": 892.7822875976562,
-      "learning_rate": 1.248e-05,
-      "loss": 167.6201,
+      "epoch": 0.01260519479470097,
+      "grad_norm": 840.9987182617188,
+      "learning_rate": 1.248e-06,
+      "loss": 22.5444,
       "step": 6240
     },
     {
-      "epoch": 0.02525079085476957,
-      "grad_norm": 537.8553466796875,
-      "learning_rate": 1.25e-05,
-      "loss": 242.3135,
+      "epoch": 0.012625395427384785,
+      "grad_norm": 520.2071533203125,
+      "learning_rate": 1.25e-06,
+      "loss": 15.9313,
       "step": 6250
     },
     {
-      "epoch": 0.025291192120137203,
-      "grad_norm": 1086.2513427734375,
-      "learning_rate": 1.252e-05,
-      "loss": 222.7271,
+      "epoch": 0.012645596060068601,
+      "grad_norm": 556.5119018554688,
+      "learning_rate": 1.2520000000000003e-06,
+      "loss": 32.9128,
       "step": 6260
     },
     {
-      "epoch": 0.025331593385504832,
-      "grad_norm": 4348.87353515625,
-      "learning_rate": 1.2540000000000002e-05,
-      "loss": 316.0617,
+      "epoch": 0.012665796692752416,
+      "grad_norm": 228.95762634277344,
+      "learning_rate": 1.2540000000000002e-06,
+      "loss": 16.7764,
       "step": 6270
     },
     {
-      "epoch": 0.025371994650872465,
-      "grad_norm": 1642.019287109375,
-      "learning_rate": 1.256e-05,
-      "loss": 202.9976,
+      "epoch": 0.012685997325436233,
+      "grad_norm": 316.0266418457031,
+      "learning_rate": 1.256e-06,
+      "loss": 25.5342,
       "step": 6280
     },
     {
-      "epoch": 0.025412395916240095,
-      "grad_norm": 685.2844848632812,
-      "learning_rate": 1.258e-05,
-      "loss": 147.7885,
+      "epoch": 0.012706197958120048,
+      "grad_norm": 18.748266220092773,
+      "learning_rate": 1.258e-06,
+      "loss": 42.3436,
       "step": 6290
     },
     {
-      "epoch": 0.025452797181607728,
-      "grad_norm": 2664.384765625,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 178.4338,
+      "epoch": 0.012726398590803864,
+      "grad_norm": 1323.001953125,
+      "learning_rate": 1.26e-06,
+      "loss": 32.5139,
       "step": 6300
     },
     {
-      "epoch": 0.025493198446975358,
-      "grad_norm": 727.255126953125,
-      "learning_rate": 1.2620000000000001e-05,
-      "loss": 142.211,
+      "epoch": 0.012746599223487679,
+      "grad_norm": 942.7252197265625,
+      "learning_rate": 1.2620000000000002e-06,
+      "loss": 29.162,
       "step": 6310
     },
     {
-      "epoch": 0.02553359971234299,
-      "grad_norm": 2616.48486328125,
-      "learning_rate": 1.2640000000000003e-05,
-      "loss": 215.0363,
+      "epoch": 0.012766799856171495,
+      "grad_norm": 1211.36962890625,
+      "learning_rate": 1.2640000000000003e-06,
+      "loss": 29.0307,
       "step": 6320
     },
     {
-      "epoch": 0.02557400097771062,
-      "grad_norm": 1724.4073486328125,
-      "learning_rate": 1.2659999999999999e-05,
-      "loss": 268.9592,
+      "epoch": 0.01278700048885531,
+      "grad_norm": 802.3575439453125,
+      "learning_rate": 1.266e-06,
+      "loss": 21.0279,
       "step": 6330
     },
     {
-      "epoch": 0.025614402243078253,
-      "grad_norm": 757.4139404296875,
-      "learning_rate": 1.268e-05,
-      "loss": 245.0375,
+      "epoch": 0.012807201121539127,
+      "grad_norm": 740.271728515625,
+      "learning_rate": 1.268e-06,
+      "loss": 20.676,
       "step": 6340
     },
     {
-      "epoch": 0.025654803508445883,
-      "grad_norm": 804.900634765625,
-      "learning_rate": 1.27e-05,
-      "loss": 258.7294,
+      "epoch": 0.012827401754222941,
+      "grad_norm": 419.27020263671875,
+      "learning_rate": 1.2700000000000001e-06,
+      "loss": 51.1324,
       "step": 6350
     },
     {
-      "epoch": 0.025695204773813516,
-      "grad_norm": 1432.1217041015625,
-      "learning_rate": 1.2720000000000002e-05,
-      "loss": 226.9255,
+      "epoch": 0.012847602386906758,
+      "grad_norm": 692.2057495117188,
+      "learning_rate": 1.2720000000000003e-06,
+      "loss": 36.6126,
       "step": 6360
     },
     {
-      "epoch": 0.025735606039181146,
-      "grad_norm": 1246.9285888671875,
-      "learning_rate": 1.2740000000000002e-05,
-      "loss": 190.6593,
+      "epoch": 0.012867803019590573,
+      "grad_norm": 132.77894592285156,
+      "learning_rate": 1.2740000000000002e-06,
+      "loss": 21.4999,
       "step": 6370
     },
     {
-      "epoch": 0.02577600730454878,
-      "grad_norm": 1320.610107421875,
-      "learning_rate": 1.276e-05,
-      "loss": 254.1634,
+      "epoch": 0.01288800365227439,
+      "grad_norm": 595.9766845703125,
+      "learning_rate": 1.276e-06,
+      "loss": 28.293,
       "step": 6380
     },
     {
-      "epoch": 0.025816408569916408,
-      "grad_norm": 5738.62939453125,
-      "learning_rate": 1.278e-05,
-      "loss": 214.6841,
+      "epoch": 0.012908204284958204,
+      "grad_norm": 1459.1014404296875,
+      "learning_rate": 1.278e-06,
+      "loss": 21.5185,
       "step": 6390
     },
     {
-      "epoch": 0.02585680983528404,
-      "grad_norm": 649.080810546875,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 170.7529,
+      "epoch": 0.01292840491764202,
+      "grad_norm": 1731.2054443359375,
+      "learning_rate": 1.28e-06,
+      "loss": 47.9539,
       "step": 6400
     },
     {
-      "epoch": 0.02589721110065167,
-      "grad_norm": 756.0393676757812,
-      "learning_rate": 1.2820000000000001e-05,
-      "loss": 186.7894,
+      "epoch": 0.012948605550325835,
+      "grad_norm": 750.8800048828125,
+      "learning_rate": 1.2820000000000002e-06,
+      "loss": 42.8562,
       "step": 6410
     },
     {
-      "epoch": 0.025937612366019304,
-      "grad_norm": 2048.561279296875,
-      "learning_rate": 1.2839999999999999e-05,
-      "loss": 182.1876,
+      "epoch": 0.012968806183009652,
+      "grad_norm": 646.1731567382812,
+      "learning_rate": 1.284e-06,
+      "loss": 32.3806,
       "step": 6420
     },
     {
-      "epoch": 0.025978013631386934,
-      "grad_norm": 805.5757446289062,
-      "learning_rate": 1.286e-05,
-      "loss": 196.0959,
+      "epoch": 0.012989006815693467,
+      "grad_norm": 306.4696044921875,
+      "learning_rate": 1.286e-06,
+      "loss": 24.9085,
       "step": 6430
     },
     {
-      "epoch": 0.026018414896754567,
-      "grad_norm": 5938.0439453125,
-      "learning_rate": 1.288e-05,
-      "loss": 254.1976,
+      "epoch": 0.013009207448377283,
+      "grad_norm": 169.34942626953125,
+      "learning_rate": 1.288e-06,
+      "loss": 32.6633,
       "step": 6440
     },
     {
-      "epoch": 0.026058816162122196,
-      "grad_norm": 1488.987060546875,
-      "learning_rate": 1.29e-05,
-      "loss": 250.4606,
+      "epoch": 0.013029408081061098,
+      "grad_norm": 469.367919921875,
+      "learning_rate": 1.2900000000000001e-06,
+      "loss": 27.3972,
       "step": 6450
     },
     {
-      "epoch": 0.02609921742748983,
-      "grad_norm": 1574.7130126953125,
-      "learning_rate": 1.2920000000000002e-05,
-      "loss": 157.1144,
+      "epoch": 0.013049608713744915,
+      "grad_norm": 553.5028686523438,
+      "learning_rate": 1.2920000000000003e-06,
+      "loss": 35.9573,
       "step": 6460
     },
     {
-      "epoch": 0.02613961869285746,
-      "grad_norm": 736.7561645507812,
-      "learning_rate": 1.294e-05,
-      "loss": 220.315,
+      "epoch": 0.01306980934642873,
+      "grad_norm": 262.7359924316406,
+      "learning_rate": 1.294e-06,
+      "loss": 30.1157,
       "step": 6470
     },
     {
-      "epoch": 0.026180019958225092,
-      "grad_norm": 972.5496826171875,
-      "learning_rate": 1.296e-05,
-      "loss": 257.4698,
+      "epoch": 0.013090009979112546,
+      "grad_norm": 839.3622436523438,
+      "learning_rate": 1.296e-06,
+      "loss": 38.7697,
       "step": 6480
     },
     {
-      "epoch": 0.02622042122359272,
-      "grad_norm": 4659.64794921875,
-      "learning_rate": 1.2980000000000001e-05,
-      "loss": 201.6754,
+      "epoch": 0.01311021061179636,
+      "grad_norm": 469.48895263671875,
+      "learning_rate": 1.2980000000000001e-06,
+      "loss": 33.9329,
       "step": 6490
     },
     {
-      "epoch": 0.026260822488960354,
-      "grad_norm": 803.3303833007812,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 227.2734,
+      "epoch": 0.013130411244480177,
+      "grad_norm": 435.9288635253906,
+      "learning_rate": 1.3e-06,
+      "loss": 30.6598,
       "step": 6500
     },
     {
-      "epoch": 0.026301223754327984,
-      "grad_norm": 900.7424926757812,
-      "learning_rate": 1.3020000000000002e-05,
-      "loss": 204.3565,
+      "epoch": 0.013150611877163992,
+      "grad_norm": 1100.7294921875,
+      "learning_rate": 1.3020000000000002e-06,
+      "loss": 38.6149,
       "step": 6510
     },
     {
-      "epoch": 0.026341625019695617,
-      "grad_norm": 880.3982543945312,
-      "learning_rate": 1.3039999999999999e-05,
-      "loss": 213.0109,
+      "epoch": 0.013170812509847809,
+      "grad_norm": 978.08349609375,
+      "learning_rate": 1.304e-06,
+      "loss": 29.158,
       "step": 6520
     },
     {
-      "epoch": 0.026382026285063247,
-      "grad_norm": 1765.91357421875,
-      "learning_rate": 1.306e-05,
-      "loss": 238.4321,
+      "epoch": 0.013191013142531623,
+      "grad_norm": 440.118408203125,
+      "learning_rate": 1.306e-06,
+      "loss": 51.669,
       "step": 6530
     },
     {
-      "epoch": 0.02642242755043088,
-      "grad_norm": 2665.921875,
-      "learning_rate": 1.308e-05,
-      "loss": 227.712,
+      "epoch": 0.01321121377521544,
+      "grad_norm": 217.9273223876953,
+      "learning_rate": 1.308e-06,
+      "loss": 24.4326,
       "step": 6540
     },
     {
-      "epoch": 0.02646282881579851,
-      "grad_norm": 2605.635009765625,
-      "learning_rate": 1.3100000000000002e-05,
-      "loss": 162.2536,
+      "epoch": 0.013231414407899255,
+      "grad_norm": 356.3488464355469,
+      "learning_rate": 1.3100000000000002e-06,
+      "loss": 26.9776,
       "step": 6550
     },
     {
-      "epoch": 0.026503230081166142,
-      "grad_norm": 793.8833618164062,
-      "learning_rate": 1.3120000000000001e-05,
-      "loss": 192.5132,
+      "epoch": 0.013251615040583071,
+      "grad_norm": 340.18060302734375,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 39.9926,
       "step": 6560
     },
     {
-      "epoch": 0.026543631346533772,
-      "grad_norm": 2946.634765625,
-      "learning_rate": 1.314e-05,
-      "loss": 159.2729,
+      "epoch": 0.013271815673266886,
+      "grad_norm": 984.8867797851562,
+      "learning_rate": 1.314e-06,
+      "loss": 54.8693,
       "step": 6570
     },
     {
-      "epoch": 0.026584032611901405,
-      "grad_norm": 910.0969848632812,
-      "learning_rate": 1.316e-05,
-      "loss": 191.1152,
+      "epoch": 0.013292016305950703,
+      "grad_norm": 1010.78466796875,
+      "learning_rate": 1.316e-06,
+      "loss": 23.7788,
       "step": 6580
     },
     {
-      "epoch": 0.026624433877269035,
-      "grad_norm": 808.4738159179688,
-      "learning_rate": 1.3180000000000001e-05,
-      "loss": 117.279,
+      "epoch": 0.013312216938634517,
+      "grad_norm": 733.6953125,
+      "learning_rate": 1.3180000000000001e-06,
+      "loss": 34.9994,
       "step": 6590
     },
     {
-      "epoch": 0.026664835142636668,
-      "grad_norm": 716.1993408203125,
-      "learning_rate": 1.32e-05,
-      "loss": 246.6991,
+      "epoch": 0.013332417571318334,
+      "grad_norm": 373.2271728515625,
+      "learning_rate": 1.32e-06,
+      "loss": 37.3207,
       "step": 6600
     },
     {
-      "epoch": 0.026705236408004297,
-      "grad_norm": 1362.393310546875,
-      "learning_rate": 1.3220000000000002e-05,
-      "loss": 233.1123,
+      "epoch": 0.013352618204002149,
+      "grad_norm": 171.13934326171875,
+      "learning_rate": 1.3220000000000002e-06,
+      "loss": 12.7663,
       "step": 6610
     },
     {
-      "epoch": 0.02674563767337193,
-      "grad_norm": 1083.3197021484375,
-      "learning_rate": 1.324e-05,
-      "loss": 169.5999,
+      "epoch": 0.013372818836685965,
+      "grad_norm": 357.4339599609375,
+      "learning_rate": 1.324e-06,
+      "loss": 35.8358,
       "step": 6620
     },
     {
-      "epoch": 0.02678603893873956,
-      "grad_norm": 754.7849731445312,
-      "learning_rate": 1.326e-05,
-      "loss": 306.0971,
+      "epoch": 0.01339301946936978,
+      "grad_norm": 220.25674438476562,
+      "learning_rate": 1.326e-06,
+      "loss": 15.3749,
       "step": 6630
     },
     {
-      "epoch": 0.026826440204107193,
-      "grad_norm": 968.374755859375,
-      "learning_rate": 1.3280000000000002e-05,
-      "loss": 159.3224,
+      "epoch": 0.013413220102053596,
+      "grad_norm": 462.05841064453125,
+      "learning_rate": 1.328e-06,
+      "loss": 38.8295,
       "step": 6640
     },
     {
-      "epoch": 0.026866841469474823,
-      "grad_norm": 670.0838623046875,
-      "learning_rate": 1.3300000000000001e-05,
-      "loss": 155.3162,
+      "epoch": 0.013433420734737411,
+      "grad_norm": 469.8040466308594,
+      "learning_rate": 1.3300000000000002e-06,
+      "loss": 22.9475,
       "step": 6650
     },
     {
-      "epoch": 0.026907242734842456,
-      "grad_norm": 2666.84326171875,
-      "learning_rate": 1.3320000000000001e-05,
-      "loss": 173.0536,
+      "epoch": 0.013453621367421228,
+      "grad_norm": 971.557373046875,
+      "learning_rate": 1.3320000000000003e-06,
+      "loss": 44.1314,
       "step": 6660
     },
     {
-      "epoch": 0.026947644000210085,
-      "grad_norm": 1368.1116943359375,
-      "learning_rate": 1.334e-05,
-      "loss": 197.1025,
+      "epoch": 0.013473822000105043,
+      "grad_norm": 298.933837890625,
+      "learning_rate": 1.334e-06,
+      "loss": 28.8329,
       "step": 6670
     },
     {
-      "epoch": 0.026988045265577718,
-      "grad_norm": 1038.586181640625,
-      "learning_rate": 1.336e-05,
-      "loss": 156.9807,
+      "epoch": 0.013494022632788859,
+      "grad_norm": 623.2842407226562,
+      "learning_rate": 1.336e-06,
+      "loss": 35.835,
       "step": 6680
     },
     {
-      "epoch": 0.027028446530945348,
-      "grad_norm": 1009.5262451171875,
-      "learning_rate": 1.338e-05,
-      "loss": 204.5245,
+      "epoch": 0.013514223265472674,
+      "grad_norm": 441.5866394042969,
+      "learning_rate": 1.3380000000000001e-06,
+      "loss": 19.0923,
       "step": 6690
     },
     {
-      "epoch": 0.02706884779631298,
-      "grad_norm": 1839.5418701171875,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 266.4659,
+      "epoch": 0.01353442389815649,
+      "grad_norm": 1859.7860107421875,
+      "learning_rate": 1.34e-06,
+      "loss": 35.8646,
       "step": 6700
     },
     {
-      "epoch": 0.02710924906168061,
-      "grad_norm": 829.3477783203125,
-      "learning_rate": 1.3420000000000002e-05,
-      "loss": 240.5496,
+      "epoch": 0.013554624530840305,
+      "grad_norm": 64.71363067626953,
+      "learning_rate": 1.3420000000000002e-06,
+      "loss": 15.234,
       "step": 6710
     },
     {
-      "epoch": 0.027149650327048244,
-      "grad_norm": 2405.634033203125,
-      "learning_rate": 1.344e-05,
-      "loss": 190.1635,
+      "epoch": 0.013574825163524122,
+      "grad_norm": 333.45794677734375,
+      "learning_rate": 1.344e-06,
+      "loss": 22.3284,
       "step": 6720
     },
     {
-      "epoch": 0.027190051592415873,
-      "grad_norm": 3415.678466796875,
-      "learning_rate": 1.346e-05,
-      "loss": 256.6028,
+      "epoch": 0.013595025796207937,
+      "grad_norm": 393.3450622558594,
+      "learning_rate": 1.3460000000000001e-06,
+      "loss": 54.5108,
       "step": 6730
     },
     {
-      "epoch": 0.027230452857783506,
-      "grad_norm": 1140.322998046875,
-      "learning_rate": 1.3480000000000001e-05,
-      "loss": 213.0321,
+      "epoch": 0.013615226428891753,
+      "grad_norm": 587.0012817382812,
+      "learning_rate": 1.348e-06,
+      "loss": 31.177,
       "step": 6740
     },
     {
-      "epoch": 0.027270854123151136,
-      "grad_norm": 1814.612548828125,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 289.2976,
+      "epoch": 0.013635427061575568,
+      "grad_norm": 309.5393371582031,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 13.8727,
       "step": 6750
     },
     {
-      "epoch": 0.02731125538851877,
-      "grad_norm": 5082.0244140625,
-      "learning_rate": 1.352e-05,
-      "loss": 213.1429,
+      "epoch": 0.013655627694259384,
+      "grad_norm": 468.2892150878906,
+      "learning_rate": 1.352e-06,
+      "loss": 16.0866,
       "step": 6760
     },
     {
-      "epoch": 0.0273516566538864,
-      "grad_norm": 791.2843627929688,
-      "learning_rate": 1.3539999999999999e-05,
-      "loss": 174.2937,
+      "epoch": 0.0136758283269432,
+      "grad_norm": 878.9414672851562,
+      "learning_rate": 1.354e-06,
+      "loss": 31.6945,
       "step": 6770
     },
     {
-      "epoch": 0.02739205791925403,
-      "grad_norm": 1486.5364990234375,
-      "learning_rate": 1.356e-05,
-      "loss": 211.7225,
+      "epoch": 0.013696028959627016,
+      "grad_norm": 963.8592529296875,
+      "learning_rate": 1.356e-06,
+      "loss": 26.1409,
       "step": 6780
     },
     {
-      "epoch": 0.02743245918462166,
-      "grad_norm": 2779.84765625,
-      "learning_rate": 1.358e-05,
-      "loss": 206.9393,
+      "epoch": 0.01371622959231083,
+      "grad_norm": 742.5006713867188,
+      "learning_rate": 1.3580000000000002e-06,
+      "loss": 20.0253,
       "step": 6790
     },
     {
-      "epoch": 0.027472860449989294,
-      "grad_norm": 2755.901611328125,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 244.6609,
+      "epoch": 0.013736430224994647,
+      "grad_norm": 302.4109802246094,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 30.4902,
       "step": 6800
     },
     {
-      "epoch": 0.027513261715356924,
-      "grad_norm": 1252.9466552734375,
-      "learning_rate": 1.362e-05,
-      "loss": 185.0354,
+      "epoch": 0.013756630857678462,
+      "grad_norm": 703.2691650390625,
+      "learning_rate": 1.362e-06,
+      "loss": 101.9395,
       "step": 6810
     },
     {
-      "epoch": 0.027553662980724557,
-      "grad_norm": 803.9773559570312,
-      "learning_rate": 1.364e-05,
-      "loss": 163.7829,
+      "epoch": 0.013776831490362278,
+      "grad_norm": 1015.4454956054688,
+      "learning_rate": 1.364e-06,
+      "loss": 22.2598,
       "step": 6820
     },
     {
-      "epoch": 0.027594064246092186,
-      "grad_norm": 3479.50341796875,
-      "learning_rate": 1.3660000000000001e-05,
-      "loss": 199.5576,
+      "epoch": 0.013797032123046093,
+      "grad_norm": 394.2761535644531,
+      "learning_rate": 1.3660000000000001e-06,
+      "loss": 36.6149,
       "step": 6830
     },
     {
-      "epoch": 0.02763446551145982,
-      "grad_norm": 4041.821533203125,
-      "learning_rate": 1.3680000000000001e-05,
-      "loss": 146.2625,
+      "epoch": 0.01381723275572991,
+      "grad_norm": 867.6441650390625,
+      "learning_rate": 1.368e-06,
+      "loss": 32.4753,
       "step": 6840
     },
     {
-      "epoch": 0.02767486677682745,
-      "grad_norm": 951.2683715820312,
-      "learning_rate": 1.3700000000000001e-05,
-      "loss": 242.1677,
+      "epoch": 0.013837433388413724,
+      "grad_norm": 386.0970458984375,
+      "learning_rate": 1.3700000000000002e-06,
+      "loss": 31.9341,
       "step": 6850
     },
     {
-      "epoch": 0.027715268042195082,
-      "grad_norm": 2521.80224609375,
-      "learning_rate": 1.3719999999999999e-05,
-      "loss": 162.2531,
+      "epoch": 0.013857634021097541,
+      "grad_norm": 650.8311767578125,
+      "learning_rate": 1.372e-06,
+      "loss": 25.4292,
       "step": 6860
     },
     {
-      "epoch": 0.02775566930756271,
-      "grad_norm": 2120.54150390625,
-      "learning_rate": 1.374e-05,
-      "loss": 218.3948,
+      "epoch": 0.013877834653781356,
+      "grad_norm": 281.85302734375,
+      "learning_rate": 1.374e-06,
+      "loss": 15.7737,
       "step": 6870
     },
     {
-      "epoch": 0.027796070572930345,
-      "grad_norm": 890.4383544921875,
-      "learning_rate": 1.376e-05,
-      "loss": 237.4137,
+      "epoch": 0.013898035286465172,
+      "grad_norm": 494.3153991699219,
+      "learning_rate": 1.376e-06,
+      "loss": 12.811,
       "step": 6880
     },
     {
-      "epoch": 0.027836471838297974,
-      "grad_norm": 1014.8067626953125,
-      "learning_rate": 1.3780000000000002e-05,
-      "loss": 271.9613,
+      "epoch": 0.013918235919148987,
+      "grad_norm": 103.57233428955078,
+      "learning_rate": 1.3780000000000002e-06,
+      "loss": 24.2835,
       "step": 6890
     },
     {
-      "epoch": 0.027876873103665607,
-      "grad_norm": 2228.795654296875,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 160.2129,
+      "epoch": 0.013938436551832804,
+      "grad_norm": 313.1036071777344,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 26.341,
       "step": 6900
     },
     {
-      "epoch": 0.027917274369033237,
-      "grad_norm": 1605.4246826171875,
-      "learning_rate": 1.382e-05,
-      "loss": 192.7938,
+      "epoch": 0.013958637184516618,
+      "grad_norm": 364.7337951660156,
+      "learning_rate": 1.382e-06,
+      "loss": 19.7548,
       "step": 6910
     },
     {
-      "epoch": 0.02795767563440087,
-      "grad_norm": 726.9761352539062,
-      "learning_rate": 1.384e-05,
-      "loss": 166.0171,
+      "epoch": 0.013978837817200435,
+      "grad_norm": 708.6334838867188,
+      "learning_rate": 1.384e-06,
+      "loss": 24.823,
       "step": 6920
     },
     {
-      "epoch": 0.0279980768997685,
-      "grad_norm": 1300.53076171875,
-      "learning_rate": 1.3860000000000001e-05,
-      "loss": 155.9882,
+      "epoch": 0.01399903844988425,
+      "grad_norm": 370.78387451171875,
+      "learning_rate": 1.3860000000000002e-06,
+      "loss": 22.2033,
       "step": 6930
     },
     {
-      "epoch": 0.028038478165136133,
-      "grad_norm": 1467.7305908203125,
-      "learning_rate": 1.3880000000000001e-05,
-      "loss": 175.5876,
+      "epoch": 0.014019239082568066,
+      "grad_norm": 584.7822875976562,
+      "learning_rate": 1.388e-06,
+      "loss": 21.2259,
       "step": 6940
     },
     {
-      "epoch": 0.028078879430503762,
-      "grad_norm": 653.6290283203125,
-      "learning_rate": 1.3900000000000002e-05,
-      "loss": 177.4526,
+      "epoch": 0.014039439715251881,
+      "grad_norm": 434.93896484375,
+      "learning_rate": 1.3900000000000002e-06,
+      "loss": 17.3879,
       "step": 6950
     },
     {
-      "epoch": 0.028119280695871395,
-      "grad_norm": 1209.3408203125,
-      "learning_rate": 1.3919999999999999e-05,
-      "loss": 139.5799,
+      "epoch": 0.014059640347935698,
+      "grad_norm": 1412.916015625,
+      "learning_rate": 1.392e-06,
+      "loss": 43.3578,
       "step": 6960
     },
     {
-      "epoch": 0.028159681961239025,
-      "grad_norm": 0.0,
-      "learning_rate": 1.394e-05,
-      "loss": 114.6302,
+      "epoch": 0.014079840980619512,
+      "grad_norm": 690.4659423828125,
+      "learning_rate": 1.3940000000000001e-06,
+      "loss": 29.9748,
       "step": 6970
     },
     {
-      "epoch": 0.028200083226606658,
-      "grad_norm": 2785.091796875,
-      "learning_rate": 1.396e-05,
-      "loss": 197.9756,
+      "epoch": 0.014100041613303329,
+      "grad_norm": 631.6376953125,
+      "learning_rate": 1.396e-06,
+      "loss": 27.4844,
       "step": 6980
     },
     {
-      "epoch": 0.028240484491974287,
-      "grad_norm": 1010.1685180664062,
-      "learning_rate": 1.3980000000000002e-05,
-      "loss": 194.4374,
+      "epoch": 0.014120242245987144,
+      "grad_norm": 136.48374938964844,
+      "learning_rate": 1.3980000000000002e-06,
+      "loss": 14.787,
       "step": 6990
     },
     {
-      "epoch": 0.02828088575734192,
-      "grad_norm": 810.1452026367188,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 185.4311,
+      "epoch": 0.01414044287867096,
+      "grad_norm": 419.7654724121094,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 29.0004,
       "step": 7000
     },
     {
-      "epoch": 0.02832128702270955,
-      "grad_norm": 793.3577270507812,
-      "learning_rate": 1.402e-05,
-      "loss": 186.9826,
+      "epoch": 0.014160643511354775,
+      "grad_norm": 468.74468994140625,
+      "learning_rate": 1.402e-06,
+      "loss": 19.4345,
       "step": 7010
     },
     {
-      "epoch": 0.028361688288077183,
-      "grad_norm": 1589.5145263671875,
-      "learning_rate": 1.4040000000000001e-05,
-      "loss": 224.4251,
+      "epoch": 0.014180844144038592,
+      "grad_norm": 468.9650573730469,
+      "learning_rate": 1.404e-06,
+      "loss": 26.9539,
       "step": 7020
     },
     {
-      "epoch": 0.028402089553444813,
-      "grad_norm": 730.669189453125,
-      "learning_rate": 1.4060000000000001e-05,
-      "loss": 129.3718,
+      "epoch": 0.014201044776722406,
+      "grad_norm": 256.4075622558594,
+      "learning_rate": 1.4060000000000002e-06,
+      "loss": 21.5345,
       "step": 7030
     },
     {
-      "epoch": 0.028442490818812446,
-      "grad_norm": 2354.304443359375,
-      "learning_rate": 1.408e-05,
-      "loss": 176.9131,
+      "epoch": 0.014221245409406223,
+      "grad_norm": 363.6224060058594,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 28.2582,
       "step": 7040
     },
     {
-      "epoch": 0.028482892084180075,
-      "grad_norm": 1103.234619140625,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 198.4438,
+      "epoch": 0.014241446042090038,
+      "grad_norm": 2213.649169921875,
+      "learning_rate": 1.41e-06,
+      "loss": 29.7299,
       "step": 7050
     },
     {
-      "epoch": 0.02852329334954771,
-      "grad_norm": 1036.2913818359375,
-      "learning_rate": 1.412e-05,
-      "loss": 228.0053,
+      "epoch": 0.014261646674773854,
+      "grad_norm": 525.2031860351562,
+      "learning_rate": 1.412e-06,
+      "loss": 15.0173,
       "step": 7060
     },
     {
-      "epoch": 0.028563694614915338,
-      "grad_norm": 3621.711181640625,
-      "learning_rate": 1.414e-05,
-      "loss": 140.6286,
+      "epoch": 0.014281847307457669,
+      "grad_norm": 571.482666015625,
+      "learning_rate": 1.4140000000000001e-06,
+      "loss": 22.69,
       "step": 7070
     },
     {
-      "epoch": 0.02860409588028297,
-      "grad_norm": 1558.1011962890625,
-      "learning_rate": 1.4160000000000002e-05,
-      "loss": 187.4032,
+      "epoch": 0.014302047940141486,
+      "grad_norm": 432.14385986328125,
+      "learning_rate": 1.416e-06,
+      "loss": 21.501,
       "step": 7080
     },
     {
-      "epoch": 0.0286444971456506,
-      "grad_norm": 2425.656494140625,
-      "learning_rate": 1.4180000000000001e-05,
-      "loss": 198.9401,
+      "epoch": 0.0143222485728253,
+      "grad_norm": 230.9308319091797,
+      "learning_rate": 1.4180000000000002e-06,
+      "loss": 23.7384,
       "step": 7090
     },
     {
-      "epoch": 0.028684898411018234,
-      "grad_norm": 878.8297729492188,
-      "learning_rate": 1.42e-05,
-      "loss": 248.0995,
+      "epoch": 0.014342449205509117,
+      "grad_norm": 689.4318237304688,
+      "learning_rate": 1.42e-06,
+      "loss": 16.4564,
       "step": 7100
     },
     {
-      "epoch": 0.028725299676385863,
-      "grad_norm": 791.4844360351562,
-      "learning_rate": 1.422e-05,
-      "loss": 223.9328,
+      "epoch": 0.014362649838192932,
+      "grad_norm": 833.6482543945312,
+      "learning_rate": 1.4220000000000001e-06,
+      "loss": 42.0199,
       "step": 7110
     },
     {
-      "epoch": 0.028765700941753496,
-      "grad_norm": 984.3999633789062,
-      "learning_rate": 1.4240000000000001e-05,
-      "loss": 207.1025,
+      "epoch": 0.014382850470876748,
+      "grad_norm": 263.74810791015625,
+      "learning_rate": 1.424e-06,
+      "loss": 16.752,
       "step": 7120
     },
     {
-      "epoch": 0.028806102207121126,
-      "grad_norm": 1250.86669921875,
-      "learning_rate": 1.426e-05,
-      "loss": 168.4282,
+      "epoch": 0.014403051103560563,
+      "grad_norm": 69.91686248779297,
+      "learning_rate": 1.4260000000000002e-06,
+      "loss": 11.7691,
       "step": 7130
     },
     {
-      "epoch": 0.02884650347248876,
-      "grad_norm": 993.6478881835938,
-      "learning_rate": 1.4280000000000002e-05,
-      "loss": 204.2866,
+      "epoch": 0.01442325173624438,
+      "grad_norm": 802.9617309570312,
+      "learning_rate": 1.4280000000000001e-06,
+      "loss": 29.6307,
       "step": 7140
     },
     {
-      "epoch": 0.02888690473785639,
-      "grad_norm": 1394.8699951171875,
-      "learning_rate": 1.43e-05,
-      "loss": 321.6906,
+      "epoch": 0.014443452368928194,
+      "grad_norm": 1388.6160888671875,
+      "learning_rate": 1.43e-06,
+      "loss": 17.46,
       "step": 7150
     },
     {
-      "epoch": 0.02892730600322402,
-      "grad_norm": 1557.498291015625,
-      "learning_rate": 1.432e-05,
-      "loss": 240.2873,
+      "epoch": 0.01446365300161201,
+      "grad_norm": 202.5712127685547,
+      "learning_rate": 1.432e-06,
+      "loss": 14.3374,
       "step": 7160
     },
     {
-      "epoch": 0.02896770726859165,
-      "grad_norm": 3150.471435546875,
-      "learning_rate": 1.434e-05,
-      "loss": 173.3683,
+      "epoch": 0.014483853634295826,
+      "grad_norm": 547.8307495117188,
+      "learning_rate": 1.4340000000000002e-06,
+      "loss": 30.6452,
       "step": 7170
     },
     {
-      "epoch": 0.029008108533959284,
-      "grad_norm": 40070.46875,
-      "learning_rate": 1.4360000000000001e-05,
-      "loss": 300.3157,
+      "epoch": 0.014504054266979642,
+      "grad_norm": 211.10414123535156,
+      "learning_rate": 1.436e-06,
+      "loss": 13.0712,
       "step": 7180
     },
     {
-      "epoch": 0.029048509799326914,
-      "grad_norm": 1962.33447265625,
-      "learning_rate": 1.4380000000000001e-05,
-      "loss": 240.5531,
+      "epoch": 0.014524254899663457,
+      "grad_norm": 756.9300537109375,
+      "learning_rate": 1.4380000000000003e-06,
+      "loss": 23.8358,
       "step": 7190
     },
     {
-      "epoch": 0.029088911064694547,
-      "grad_norm": 9524.1953125,
-      "learning_rate": 1.44e-05,
-      "loss": 197.5709,
+      "epoch": 0.014544455532347273,
+      "grad_norm": 378.86663818359375,
+      "learning_rate": 1.44e-06,
+      "loss": 26.3164,
       "step": 7200
     },
     {
-      "epoch": 0.029129312330062176,
-      "grad_norm": 1291.194091796875,
-      "learning_rate": 1.4420000000000001e-05,
-      "loss": 173.9855,
+      "epoch": 0.014564656165031088,
+      "grad_norm": 337.78131103515625,
+      "learning_rate": 1.4420000000000001e-06,
+      "loss": 30.7286,
       "step": 7210
     },
     {
-      "epoch": 0.02916971359542981,
-      "grad_norm": 508.72991943359375,
-      "learning_rate": 1.444e-05,
-      "loss": 198.4389,
+      "epoch": 0.014584856797714905,
+      "grad_norm": 1024.835205078125,
+      "learning_rate": 1.444e-06,
+      "loss": 29.8971,
       "step": 7220
     },
     {
-      "epoch": 0.02921011486079744,
-      "grad_norm": 3161.301513671875,
-      "learning_rate": 1.4460000000000002e-05,
-      "loss": 232.296,
+      "epoch": 0.01460505743039872,
+      "grad_norm": 938.3273315429688,
+      "learning_rate": 1.4460000000000002e-06,
+      "loss": 27.8823,
       "step": 7230
     },
     {
-      "epoch": 0.029250516126165072,
-      "grad_norm": 888.337646484375,
-      "learning_rate": 1.4480000000000002e-05,
-      "loss": 165.0789,
+      "epoch": 0.014625258063082536,
+      "grad_norm": 552.2940063476562,
+      "learning_rate": 1.4480000000000002e-06,
+      "loss": 25.7623,
       "step": 7240
     },
     {
-      "epoch": 0.029290917391532702,
-      "grad_norm": 1378.0032958984375,
-      "learning_rate": 1.45e-05,
-      "loss": 199.7313,
+      "epoch": 0.014645458695766351,
+      "grad_norm": 352.24383544921875,
+      "learning_rate": 1.45e-06,
+      "loss": 12.5614,
       "step": 7250
     },
     {
-      "epoch": 0.029331318656900335,
-      "grad_norm": 2439.7119140625,
-      "learning_rate": 1.452e-05,
-      "loss": 201.1323,
+      "epoch": 0.014665659328450167,
+      "grad_norm": 380.1072082519531,
+      "learning_rate": 1.452e-06,
+      "loss": 21.0116,
       "step": 7260
     },
     {
-      "epoch": 0.029371719922267964,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4540000000000001e-05,
-      "loss": 139.8308,
+      "epoch": 0.014685859961133982,
+      "grad_norm": 89.25300598144531,
+      "learning_rate": 1.4540000000000002e-06,
+      "loss": 37.3816,
       "step": 7270
     },
     {
-      "epoch": 0.029412121187635597,
-      "grad_norm": 697.4395751953125,
-      "learning_rate": 1.4560000000000001e-05,
-      "loss": 215.4804,
+      "epoch": 0.014706060593817799,
+      "grad_norm": 742.039306640625,
+      "learning_rate": 1.4560000000000001e-06,
+      "loss": 20.3916,
       "step": 7280
     },
     {
-      "epoch": 0.029452522453003227,
-      "grad_norm": 1335.80859375,
-      "learning_rate": 1.4580000000000003e-05,
-      "loss": 206.551,
+      "epoch": 0.014726261226501614,
+      "grad_norm": 408.6346435546875,
+      "learning_rate": 1.4580000000000003e-06,
+      "loss": 22.624,
       "step": 7290
     },
     {
-      "epoch": 0.02949292371837086,
-      "grad_norm": 3926.513427734375,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 161.3456,
+      "epoch": 0.01474646185918543,
+      "grad_norm": 489.7460632324219,
+      "learning_rate": 1.46e-06,
+      "loss": 51.9042,
       "step": 7300
     },
     {
-      "epoch": 0.02953332498373849,
-      "grad_norm": 2071.3857421875,
-      "learning_rate": 1.462e-05,
-      "loss": 245.7546,
+      "epoch": 0.014766662491869245,
+      "grad_norm": 378.1903991699219,
+      "learning_rate": 1.4620000000000001e-06,
+      "loss": 22.1225,
       "step": 7310
     },
     {
-      "epoch": 0.029573726249106123,
-      "grad_norm": 1162.638916015625,
-      "learning_rate": 1.464e-05,
-      "loss": 247.4758,
+      "epoch": 0.014786863124553061,
+      "grad_norm": 118.77698516845703,
+      "learning_rate": 1.464e-06,
+      "loss": 27.0962,
       "step": 7320
     },
     {
-      "epoch": 0.029614127514473752,
-      "grad_norm": 1904.095947265625,
-      "learning_rate": 1.4660000000000002e-05,
-      "loss": 209.4585,
+      "epoch": 0.014807063757236876,
+      "grad_norm": 497.64141845703125,
+      "learning_rate": 1.4660000000000002e-06,
+      "loss": 23.5584,
       "step": 7330
     },
     {
-      "epoch": 0.029654528779841385,
-      "grad_norm": 2012.481689453125,
-      "learning_rate": 1.4680000000000002e-05,
-      "loss": 177.5473,
+      "epoch": 0.014827264389920693,
+      "grad_norm": 26.968568801879883,
+      "learning_rate": 1.4680000000000002e-06,
+      "loss": 46.2282,
       "step": 7340
     },
     {
-      "epoch": 0.029694930045209015,
-      "grad_norm": 827.6367797851562,
-      "learning_rate": 1.47e-05,
-      "loss": 132.3128,
+      "epoch": 0.014847465022604507,
+      "grad_norm": 621.72314453125,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 33.6739,
       "step": 7350
     },
     {
-      "epoch": 0.029735331310576648,
-      "grad_norm": 1051.8994140625,
-      "learning_rate": 1.472e-05,
-      "loss": 201.6083,
+      "epoch": 0.014867665655288324,
+      "grad_norm": 1012.1883544921875,
+      "learning_rate": 1.472e-06,
+      "loss": 30.1342,
       "step": 7360
     },
     {
-      "epoch": 0.029775732575944278,
-      "grad_norm": 2179.918701171875,
-      "learning_rate": 1.4740000000000001e-05,
-      "loss": 142.6359,
+      "epoch": 0.014887866287972139,
+      "grad_norm": 164.61184692382812,
+      "learning_rate": 1.4740000000000002e-06,
+      "loss": 22.5415,
       "step": 7370
     },
     {
-      "epoch": 0.02981613384131191,
-      "grad_norm": 1439.194091796875,
-      "learning_rate": 1.4760000000000001e-05,
-      "loss": 170.9291,
+      "epoch": 0.014908066920655955,
+      "grad_norm": 382.0716552734375,
+      "learning_rate": 1.4760000000000001e-06,
+      "loss": 63.1735,
       "step": 7380
     },
     {
-      "epoch": 0.02985653510667954,
-      "grad_norm": 1795.2176513671875,
-      "learning_rate": 1.4779999999999999e-05,
-      "loss": 205.2162,
+      "epoch": 0.01492826755333977,
+      "grad_norm": 807.2562866210938,
+      "learning_rate": 1.478e-06,
+      "loss": 51.1082,
       "step": 7390
     },
     {
-      "epoch": 0.029896936372047173,
-      "grad_norm": 1061.8426513671875,
-      "learning_rate": 1.48e-05,
-      "loss": 291.1986,
+      "epoch": 0.014948468186023587,
+      "grad_norm": 600.788330078125,
+      "learning_rate": 1.48e-06,
+      "loss": 41.2442,
       "step": 7400
     },
     {
-      "epoch": 0.029937337637414803,
-      "grad_norm": 1072.127685546875,
-      "learning_rate": 1.482e-05,
-      "loss": 190.8879,
+      "epoch": 0.014968668818707401,
+      "grad_norm": 547.8676147460938,
+      "learning_rate": 1.4820000000000002e-06,
+      "loss": 33.4473,
       "step": 7410
     },
     {
-      "epoch": 0.029977738902782436,
-      "grad_norm": 1368.1405029296875,
-      "learning_rate": 1.4840000000000002e-05,
-      "loss": 196.2706,
+      "epoch": 0.014988869451391218,
+      "grad_norm": 785.7941284179688,
+      "learning_rate": 1.4840000000000001e-06,
+      "loss": 21.1182,
       "step": 7420
     },
     {
-      "epoch": 0.030018140168150065,
-      "grad_norm": 1639.9654541015625,
-      "learning_rate": 1.4860000000000002e-05,
-      "loss": 129.1992,
+      "epoch": 0.015009070084075033,
+      "grad_norm": 373.2097473144531,
+      "learning_rate": 1.4860000000000003e-06,
+      "loss": 19.4519,
       "step": 7430
     },
     {
-      "epoch": 0.0300585414335177,
-      "grad_norm": 1488.4791259765625,
-      "learning_rate": 1.488e-05,
-      "loss": 151.4941,
+      "epoch": 0.01502927071675885,
+      "grad_norm": 283.32928466796875,
+      "learning_rate": 1.488e-06,
+      "loss": 33.4474,
       "step": 7440
     },
     {
-      "epoch": 0.030098942698885328,
-      "grad_norm": 864.4747924804688,
-      "learning_rate": 1.49e-05,
-      "loss": 199.6265,
+      "epoch": 0.015049471349442664,
+      "grad_norm": 601.2964477539062,
+      "learning_rate": 1.4900000000000001e-06,
+      "loss": 37.3402,
       "step": 7450
     },
     {
-      "epoch": 0.03013934396425296,
-      "grad_norm": 1255.5771484375,
-      "learning_rate": 1.4920000000000001e-05,
-      "loss": 226.9639,
+      "epoch": 0.01506967198212648,
+      "grad_norm": 461.42108154296875,
+      "learning_rate": 1.492e-06,
+      "loss": 33.2678,
       "step": 7460
     },
     {
-      "epoch": 0.03017974522962059,
-      "grad_norm": 1635.7647705078125,
-      "learning_rate": 1.4940000000000001e-05,
-      "loss": 174.848,
+      "epoch": 0.015089872614810295,
+      "grad_norm": 301.6255187988281,
+      "learning_rate": 1.4940000000000002e-06,
+      "loss": 21.1947,
       "step": 7470
     },
     {
-      "epoch": 0.030220146494988224,
-      "grad_norm": 1959.9365234375,
-      "learning_rate": 1.4960000000000002e-05,
-      "loss": 215.0835,
+      "epoch": 0.015110073247494112,
+      "grad_norm": 528.6672973632812,
+      "learning_rate": 1.4960000000000002e-06,
+      "loss": 30.9436,
       "step": 7480
     },
     {
-      "epoch": 0.030260547760355853,
-      "grad_norm": 1300.870361328125,
-      "learning_rate": 1.4979999999999999e-05,
-      "loss": 187.4061,
+      "epoch": 0.015130273880177927,
+      "grad_norm": 429.56304931640625,
+      "learning_rate": 1.498e-06,
+      "loss": 43.7286,
       "step": 7490
     },
     {
-      "epoch": 0.030300949025723486,
-      "grad_norm": 5733.919921875,
-      "learning_rate": 1.5e-05,
-      "loss": 203.0896,
+      "epoch": 0.015150474512861743,
+      "grad_norm": 477.3526306152344,
+      "learning_rate": 1.5e-06,
+      "loss": 44.6997,
       "step": 7500
     },
     {
-      "epoch": 0.030341350291091116,
-      "grad_norm": 6152.53173828125,
-      "learning_rate": 1.502e-05,
-      "loss": 292.1695,
+      "epoch": 0.015170675145545558,
+      "grad_norm": 694.835205078125,
+      "learning_rate": 1.5020000000000002e-06,
+      "loss": 33.5782,
       "step": 7510
     },
     {
-      "epoch": 0.03038175155645875,
-      "grad_norm": 1118.759033203125,
-      "learning_rate": 1.5040000000000002e-05,
-      "loss": 179.8201,
+      "epoch": 0.015190875778229375,
+      "grad_norm": 437.0287780761719,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 14.402,
       "step": 7520
     },
     {
-      "epoch": 0.03042215282182638,
-      "grad_norm": 1003.1773681640625,
-      "learning_rate": 1.5060000000000001e-05,
-      "loss": 168.1782,
+      "epoch": 0.01521107641091319,
+      "grad_norm": 432.5226135253906,
+      "learning_rate": 1.5060000000000003e-06,
+      "loss": 54.8615,
       "step": 7530
     },
     {
-      "epoch": 0.030462554087194012,
-      "grad_norm": 1641.7908935546875,
-      "learning_rate": 1.508e-05,
-      "loss": 234.2666,
+      "epoch": 0.015231277043597006,
+      "grad_norm": 739.1048583984375,
+      "learning_rate": 1.508e-06,
+      "loss": 28.0186,
       "step": 7540
     },
     {
-      "epoch": 0.03050295535256164,
-      "grad_norm": 6260.85791015625,
-      "learning_rate": 1.51e-05,
-      "loss": 151.5416,
+      "epoch": 0.01525147767628082,
+      "grad_norm": 751.8126220703125,
+      "learning_rate": 1.5100000000000002e-06,
+      "loss": 14.3841,
       "step": 7550
     },
     {
-      "epoch": 0.030543356617929274,
-      "grad_norm": 1593.7576904296875,
-      "learning_rate": 1.5120000000000001e-05,
-      "loss": 159.2535,
+      "epoch": 0.015271678308964637,
+      "grad_norm": 699.0371704101562,
+      "learning_rate": 1.512e-06,
+      "loss": 26.7135,
       "step": 7560
     },
     {
-      "epoch": 0.030583757883296904,
-      "grad_norm": 1710.3023681640625,
-      "learning_rate": 1.514e-05,
-      "loss": 128.8493,
+      "epoch": 0.015291878941648452,
+      "grad_norm": 580.2321166992188,
+      "learning_rate": 1.5140000000000002e-06,
+      "loss": 43.0058,
       "step": 7570
     },
     {
-      "epoch": 0.030624159148664537,
-      "grad_norm": 1299.740966796875,
-      "learning_rate": 1.5160000000000002e-05,
-      "loss": 291.4667,
+      "epoch": 0.015312079574332269,
+      "grad_norm": 953.4240112304688,
+      "learning_rate": 1.5160000000000002e-06,
+      "loss": 30.3746,
       "step": 7580
     },
     {
-      "epoch": 0.030664560414032167,
-      "grad_norm": 772.3851318359375,
-      "learning_rate": 1.518e-05,
-      "loss": 254.7446,
+      "epoch": 0.015332280207016083,
+      "grad_norm": 498.05792236328125,
+      "learning_rate": 1.5180000000000001e-06,
+      "loss": 19.2205,
       "step": 7590
     },
     {
-      "epoch": 0.0307049616793998,
-      "grad_norm": 4841.77294921875,
-      "learning_rate": 1.52e-05,
-      "loss": 261.962,
+      "epoch": 0.0153524808396999,
+      "grad_norm": 510.3348083496094,
+      "learning_rate": 1.52e-06,
+      "loss": 15.5117,
       "step": 7600
     },
     {
-      "epoch": 0.03074536294476743,
-      "grad_norm": 1306.54296875,
-      "learning_rate": 1.5220000000000002e-05,
-      "loss": 251.6979,
+      "epoch": 0.015372681472383715,
+      "grad_norm": 1451.5672607421875,
+      "learning_rate": 1.5220000000000002e-06,
+      "loss": 29.8484,
       "step": 7610
     },
     {
-      "epoch": 0.030785764210135062,
-      "grad_norm": 1031.349365234375,
-      "learning_rate": 1.5240000000000001e-05,
-      "loss": 180.2815,
+      "epoch": 0.015392882105067531,
+      "grad_norm": 402.60919189453125,
+      "learning_rate": 1.5240000000000001e-06,
+      "loss": 28.366,
       "step": 7620
     },
     {
-      "epoch": 0.030826165475502692,
-      "grad_norm": 1182.68994140625,
-      "learning_rate": 1.5260000000000003e-05,
-      "loss": 202.774,
+      "epoch": 0.015413082737751346,
+      "grad_norm": 191.9103240966797,
+      "learning_rate": 1.5260000000000003e-06,
+      "loss": 35.3297,
       "step": 7630
     },
     {
-      "epoch": 0.030866566740870325,
-      "grad_norm": 524.015869140625,
-      "learning_rate": 1.528e-05,
-      "loss": 148.2353,
+      "epoch": 0.015433283370435162,
+      "grad_norm": 371.203125,
+      "learning_rate": 1.528e-06,
+      "loss": 10.1235,
       "step": 7640
     },
     {
-      "epoch": 0.030906968006237955,
-      "grad_norm": 1932.3292236328125,
-      "learning_rate": 1.53e-05,
-      "loss": 169.3441,
+      "epoch": 0.015453484003118977,
+      "grad_norm": 465.6565856933594,
+      "learning_rate": 1.5300000000000002e-06,
+      "loss": 21.8699,
       "step": 7650
     },
     {
-      "epoch": 0.030947369271605588,
-      "grad_norm": 840.219970703125,
-      "learning_rate": 1.5320000000000002e-05,
-      "loss": 129.2729,
+      "epoch": 0.015473684635802794,
+      "grad_norm": 703.3648681640625,
+      "learning_rate": 1.5320000000000001e-06,
+      "loss": 32.0473,
       "step": 7660
     },
     {
-      "epoch": 0.030987770536973217,
-      "grad_norm": 1622.7508544921875,
-      "learning_rate": 1.5340000000000002e-05,
-      "loss": 160.2887,
+      "epoch": 0.015493885268486609,
+      "grad_norm": 468.8616638183594,
+      "learning_rate": 1.5340000000000003e-06,
+      "loss": 21.0127,
       "step": 7670
     },
     {
-      "epoch": 0.03102817180234085,
-      "grad_norm": 2095.8623046875,
-      "learning_rate": 1.536e-05,
-      "loss": 245.4731,
+      "epoch": 0.015514085901170425,
+      "grad_norm": 1880.29248046875,
+      "learning_rate": 1.536e-06,
+      "loss": 26.7551,
       "step": 7680
     },
     {
-      "epoch": 0.03106857306770848,
-      "grad_norm": 3175.37646484375,
-      "learning_rate": 1.538e-05,
-      "loss": 208.5475,
+      "epoch": 0.01553428653385424,
+      "grad_norm": 274.2958068847656,
+      "learning_rate": 1.5380000000000001e-06,
+      "loss": 19.2041,
       "step": 7690
     },
     {
-      "epoch": 0.031108974333076113,
-      "grad_norm": 924.9345092773438,
-      "learning_rate": 1.54e-05,
-      "loss": 170.3805,
+      "epoch": 0.015554487166538056,
+      "grad_norm": 539.3201904296875,
+      "learning_rate": 1.54e-06,
+      "loss": 32.8616,
       "step": 7700
     },
     {
-      "epoch": 0.031149375598443742,
-      "grad_norm": 1739.8531494140625,
-      "learning_rate": 1.542e-05,
-      "loss": 194.0813,
+      "epoch": 0.015574687799221871,
+      "grad_norm": 87.47315216064453,
+      "learning_rate": 1.5420000000000002e-06,
+      "loss": 22.3236,
       "step": 7710
     },
     {
-      "epoch": 0.031189776863811376,
-      "grad_norm": 1644.390869140625,
-      "learning_rate": 1.544e-05,
-      "loss": 136.579,
+      "epoch": 0.015594888431905688,
+      "grad_norm": 319.38226318359375,
+      "learning_rate": 1.5440000000000002e-06,
+      "loss": 30.9488,
       "step": 7720
     },
     {
-      "epoch": 0.031230178129179005,
-      "grad_norm": 789.648193359375,
-      "learning_rate": 1.546e-05,
-      "loss": 187.856,
+      "epoch": 0.015615089064589503,
+      "grad_norm": 1636.626220703125,
+      "learning_rate": 1.546e-06,
+      "loss": 30.0025,
       "step": 7730
     },
     {
-      "epoch": 0.03127057939454664,
-      "grad_norm": 1162.065185546875,
-      "learning_rate": 1.548e-05,
-      "loss": 114.7405,
+      "epoch": 0.01563528969727332,
+      "grad_norm": 128.85711669921875,
+      "learning_rate": 1.548e-06,
+      "loss": 15.6842,
       "step": 7740
     },
     {
-      "epoch": 0.03131098065991427,
-      "grad_norm": 1140.320556640625,
-      "learning_rate": 1.55e-05,
-      "loss": 181.2212,
+      "epoch": 0.015655490329957136,
+      "grad_norm": 682.8628540039062,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 14.3482,
       "step": 7750
     },
     {
-      "epoch": 0.0313513819252819,
-      "grad_norm": 2704.695556640625,
-      "learning_rate": 1.552e-05,
-      "loss": 241.535,
+      "epoch": 0.01567569096264095,
+      "grad_norm": 274.7837829589844,
+      "learning_rate": 1.5520000000000001e-06,
+      "loss": 26.8116,
       "step": 7760
     },
     {
-      "epoch": 0.03139178319064953,
-      "grad_norm": 939.2614135742188,
-      "learning_rate": 1.554e-05,
-      "loss": 151.1776,
+      "epoch": 0.015695891595324765,
+      "grad_norm": 698.118408203125,
+      "learning_rate": 1.5540000000000003e-06,
+      "loss": 40.2702,
       "step": 7770
     },
     {
-      "epoch": 0.03143218445601716,
-      "grad_norm": 1043.3226318359375,
-      "learning_rate": 1.556e-05,
-      "loss": 200.2128,
+      "epoch": 0.01571609222800858,
+      "grad_norm": 505.4995422363281,
+      "learning_rate": 1.556e-06,
+      "loss": 19.0552,
       "step": 7780
     },
     {
-      "epoch": 0.031472585721384796,
-      "grad_norm": 1068.1412353515625,
-      "learning_rate": 1.558e-05,
-      "loss": 181.7895,
+      "epoch": 0.015736292860692398,
+      "grad_norm": 604.3401489257812,
+      "learning_rate": 1.5580000000000002e-06,
+      "loss": 25.9554,
       "step": 7790
     },
     {
-      "epoch": 0.03151298698675242,
-      "grad_norm": 1479.6229248046875,
-      "learning_rate": 1.56e-05,
-      "loss": 196.9196,
+      "epoch": 0.01575649349337621,
+      "grad_norm": 551.531494140625,
+      "learning_rate": 1.56e-06,
+      "loss": 31.4826,
       "step": 7800
     },
     {
-      "epoch": 0.031553388252120056,
-      "grad_norm": 701.58544921875,
-      "learning_rate": 1.5620000000000003e-05,
-      "loss": 137.6865,
+      "epoch": 0.015776694126060028,
+      "grad_norm": 597.856689453125,
+      "learning_rate": 1.5620000000000002e-06,
+      "loss": 33.1883,
       "step": 7810
     },
     {
-      "epoch": 0.03159378951748769,
-      "grad_norm": 1822.54345703125,
-      "learning_rate": 1.5640000000000003e-05,
-      "loss": 227.0767,
+      "epoch": 0.015796894758743844,
+      "grad_norm": 255.5839385986328,
+      "learning_rate": 1.5640000000000002e-06,
+      "loss": 16.2388,
       "step": 7820
     },
     {
-      "epoch": 0.03163419078285532,
-      "grad_norm": 546.575439453125,
-      "learning_rate": 1.566e-05,
-      "loss": 146.3474,
+      "epoch": 0.01581709539142766,
+      "grad_norm": 638.1707763671875,
+      "learning_rate": 1.566e-06,
+      "loss": 12.3245,
       "step": 7830
     },
     {
-      "epoch": 0.03167459204822295,
-      "grad_norm": 1540.7342529296875,
-      "learning_rate": 1.568e-05,
-      "loss": 192.144,
+      "epoch": 0.015837296024111474,
+      "grad_norm": 561.0966796875,
+      "learning_rate": 1.568e-06,
+      "loss": 34.3022,
       "step": 7840
     },
     {
-      "epoch": 0.03171499331359058,
-      "grad_norm": 1766.8616943359375,
-      "learning_rate": 1.5700000000000002e-05,
-      "loss": 194.4416,
+      "epoch": 0.01585749665679529,
+      "grad_norm": 181.12254333496094,
+      "learning_rate": 1.5700000000000002e-06,
+      "loss": 14.8665,
       "step": 7850
     },
     {
-      "epoch": 0.031755394578958214,
-      "grad_norm": 1534.951416015625,
-      "learning_rate": 1.5720000000000002e-05,
-      "loss": 158.2836,
+      "epoch": 0.015877697289479107,
+      "grad_norm": 320.9978332519531,
+      "learning_rate": 1.5720000000000002e-06,
+      "loss": 40.0919,
       "step": 7860
     },
     {
-      "epoch": 0.03179579584432585,
-      "grad_norm": 980.215576171875,
-      "learning_rate": 1.5740000000000002e-05,
-      "loss": 166.7142,
+      "epoch": 0.015897897922162924,
+      "grad_norm": 488.3760681152344,
+      "learning_rate": 1.5740000000000003e-06,
+      "loss": 23.0373,
       "step": 7870
     },
     {
-      "epoch": 0.03183619710969347,
-      "grad_norm": 5774.56396484375,
-      "learning_rate": 1.5759999999999998e-05,
-      "loss": 305.0332,
+      "epoch": 0.015918098554846737,
+      "grad_norm": 725.4952392578125,
+      "learning_rate": 1.576e-06,
+      "loss": 17.3022,
       "step": 7880
     },
     {
-      "epoch": 0.031876598375061106,
-      "grad_norm": 1329.786865234375,
-      "learning_rate": 1.578e-05,
-      "loss": 137.7941,
+      "epoch": 0.015938299187530553,
+      "grad_norm": 807.7362060546875,
+      "learning_rate": 1.5780000000000002e-06,
+      "loss": 29.6214,
       "step": 7890
     },
     {
-      "epoch": 0.03191699964042874,
-      "grad_norm": 1180.7379150390625,
-      "learning_rate": 1.58e-05,
-      "loss": 122.4153,
+      "epoch": 0.01595849982021437,
+      "grad_norm": 402.11663818359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 37.3078,
       "step": 7900
     },
     {
-      "epoch": 0.03195740090579637,
-      "grad_norm": 867.472900390625,
-      "learning_rate": 1.582e-05,
-      "loss": 210.3185,
+      "epoch": 0.015978700452898186,
+      "grad_norm": 416.4609680175781,
+      "learning_rate": 1.5820000000000003e-06,
+      "loss": 24.7349,
       "step": 7910
     },
     {
-      "epoch": 0.031997802171164,
-      "grad_norm": 1057.9029541015625,
-      "learning_rate": 1.584e-05,
-      "loss": 209.8075,
+      "epoch": 0.015998901085582,
+      "grad_norm": 794.6650390625,
+      "learning_rate": 1.5840000000000002e-06,
+      "loss": 38.0387,
       "step": 7920
     },
     {
-      "epoch": 0.03203820343653163,
-      "grad_norm": 1288.350341796875,
-      "learning_rate": 1.586e-05,
-      "loss": 198.4188,
+      "epoch": 0.016019101718265816,
+      "grad_norm": 593.4451293945312,
+      "learning_rate": 1.586e-06,
+      "loss": 23.7971,
       "step": 7930
     },
     {
-      "epoch": 0.032078604701899265,
-      "grad_norm": 0.0,
-      "learning_rate": 1.588e-05,
-      "loss": 145.5795,
+      "epoch": 0.016039302350949632,
+      "grad_norm": 258.35321044921875,
+      "learning_rate": 1.588e-06,
+      "loss": 27.5747,
       "step": 7940
     },
     {
-      "epoch": 0.0321190059672669,
-      "grad_norm": 983.6244506835938,
-      "learning_rate": 1.59e-05,
-      "loss": 227.0045,
+      "epoch": 0.01605950298363345,
+      "grad_norm": 773.8060302734375,
+      "learning_rate": 1.5900000000000002e-06,
+      "loss": 38.5026,
       "step": 7950
     },
     {
-      "epoch": 0.032159407232634524,
-      "grad_norm": 2391.579345703125,
-      "learning_rate": 1.592e-05,
-      "loss": 261.5729,
+      "epoch": 0.016079703616317262,
+      "grad_norm": 408.9706115722656,
+      "learning_rate": 1.5920000000000002e-06,
+      "loss": 16.2559,
       "step": 7960
     },
     {
-      "epoch": 0.03219980849800216,
-      "grad_norm": 2793.90234375,
-      "learning_rate": 1.594e-05,
-      "loss": 247.0734,
+      "epoch": 0.01609990424900108,
+      "grad_norm": 763.1421508789062,
+      "learning_rate": 1.594e-06,
+      "loss": 29.3452,
       "step": 7970
     },
     {
-      "epoch": 0.03224020976336979,
-      "grad_norm": 1532.1849365234375,
-      "learning_rate": 1.596e-05,
-      "loss": 199.9248,
+      "epoch": 0.016120104881684895,
+      "grad_norm": 139.5187530517578,
+      "learning_rate": 1.596e-06,
+      "loss": 21.015,
       "step": 7980
     },
     {
-      "epoch": 0.03228061102873742,
-      "grad_norm": 1343.6947021484375,
-      "learning_rate": 1.598e-05,
-      "loss": 167.6554,
+      "epoch": 0.01614030551436871,
+      "grad_norm": 920.0736083984375,
+      "learning_rate": 1.5980000000000002e-06,
+      "loss": 27.6155,
       "step": 7990
     },
     {
-      "epoch": 0.03232101229410505,
-      "grad_norm": 717.1405639648438,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 218.2069,
+      "epoch": 0.016160506147052525,
+      "grad_norm": 302.7886657714844,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 38.6215,
       "step": 8000
-    },
-    {
-      "epoch": 0.03236141355947268,
-      "grad_norm": 2144.71337890625,
-      "learning_rate": 1.6020000000000002e-05,
-      "loss": 210.7316,
-      "step": 8010
-    },
-    {
-      "epoch": 0.032401814824840315,
-      "grad_norm": 2510.060546875,
-      "learning_rate": 1.604e-05,
-      "loss": 300.955,
-      "step": 8020
-    },
-    {
-      "epoch": 0.03244221609020795,
-      "grad_norm": 1254.329833984375,
-      "learning_rate": 1.606e-05,
-      "loss": 224.6631,
-      "step": 8030
-    },
-    {
-      "epoch": 0.032482617355575574,
-      "grad_norm": 533.80517578125,
-      "learning_rate": 1.6080000000000002e-05,
-      "loss": 195.5154,
-      "step": 8040
-    },
-    {
-      "epoch": 0.03252301862094321,
-      "grad_norm": 932.7647705078125,
-      "learning_rate": 1.6100000000000002e-05,
-      "loss": 173.2508,
-      "step": 8050
-    },
-    {
-      "epoch": 0.03256341988631084,
-      "grad_norm": 1513.1566162109375,
-      "learning_rate": 1.612e-05,
-      "loss": 252.4677,
-      "step": 8060
-    },
-    {
-      "epoch": 0.03260382115167847,
-      "grad_norm": 8973.974609375,
-      "learning_rate": 1.6139999999999998e-05,
-      "loss": 160.9031,
-      "step": 8070
-    },
-    {
-      "epoch": 0.0326442224170461,
-      "grad_norm": 1346.521728515625,
-      "learning_rate": 1.616e-05,
-      "loss": 190.9977,
-      "step": 8080
-    },
-    {
-      "epoch": 0.03268462368241373,
-      "grad_norm": 696.8346557617188,
-      "learning_rate": 1.618e-05,
-      "loss": 148.2339,
-      "step": 8090
-    },
-    {
-      "epoch": 0.032725024947781366,
-      "grad_norm": 2406.046142578125,
-      "learning_rate": 1.62e-05,
-      "loss": 129.0067,
-      "step": 8100
-    },
-    {
-      "epoch": 0.032765426213149,
-      "grad_norm": 1443.768310546875,
-      "learning_rate": 1.622e-05,
-      "loss": 206.6445,
-      "step": 8110
-    },
-    {
-      "epoch": 0.032805827478516625,
-      "grad_norm": 2201.40087890625,
-      "learning_rate": 1.624e-05,
-      "loss": 188.9112,
-      "step": 8120
-    },
-    {
-      "epoch": 0.03284622874388426,
-      "grad_norm": 699.2227783203125,
-      "learning_rate": 1.626e-05,
-      "loss": 193.334,
-      "step": 8130
-    },
-    {
-      "epoch": 0.03288663000925189,
-      "grad_norm": 545.5186767578125,
-      "learning_rate": 1.628e-05,
-      "loss": 140.6468,
-      "step": 8140
-    },
-    {
-      "epoch": 0.032927031274619524,
-      "grad_norm": 611.3945922851562,
-      "learning_rate": 1.63e-05,
-      "loss": 131.4079,
-      "step": 8150
-    },
-    {
-      "epoch": 0.03296743253998715,
-      "grad_norm": 1427.6285400390625,
-      "learning_rate": 1.6320000000000003e-05,
-      "loss": 186.4774,
-      "step": 8160
-    },
-    {
-      "epoch": 0.03300783380535478,
-      "grad_norm": 769.5264892578125,
-      "learning_rate": 1.634e-05,
-      "loss": 140.9033,
-      "step": 8170
-    },
-    {
-      "epoch": 0.033048235070722416,
-      "grad_norm": 988.6732788085938,
-      "learning_rate": 1.636e-05,
-      "loss": 193.9264,
-      "step": 8180
-    },
-    {
-      "epoch": 0.03308863633609005,
-      "grad_norm": 1023.501708984375,
-      "learning_rate": 1.6380000000000002e-05,
-      "loss": 173.0703,
-      "step": 8190
-    },
-    {
-      "epoch": 0.033129037601457675,
-      "grad_norm": 1767.6136474609375,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 235.5256,
-      "step": 8200
-    },
-    {
-      "epoch": 0.03316943886682531,
-      "grad_norm": 1146.1134033203125,
-      "learning_rate": 1.6420000000000002e-05,
-      "loss": 174.4802,
-      "step": 8210
-    },
-    {
-      "epoch": 0.03320984013219294,
-      "grad_norm": 723.232666015625,
-      "learning_rate": 1.644e-05,
-      "loss": 126.8088,
-      "step": 8220
-    },
-    {
-      "epoch": 0.033250241397560575,
-      "grad_norm": 939.8434448242188,
-      "learning_rate": 1.646e-05,
-      "loss": 169.1382,
-      "step": 8230
-    },
-    {
-      "epoch": 0.0332906426629282,
-      "grad_norm": 986.3238525390625,
-      "learning_rate": 1.648e-05,
-      "loss": 111.5822,
-      "step": 8240
-    },
-    {
-      "epoch": 0.033331043928295834,
-      "grad_norm": 1208.2476806640625,
-      "learning_rate": 1.65e-05,
-      "loss": 191.2895,
-      "step": 8250
-    },
-    {
-      "epoch": 0.03337144519366347,
-      "grad_norm": 827.257568359375,
-      "learning_rate": 1.652e-05,
-      "loss": 175.8226,
-      "step": 8260
-    },
-    {
-      "epoch": 0.0334118464590311,
-      "grad_norm": 1213.089111328125,
-      "learning_rate": 1.654e-05,
-      "loss": 214.4393,
-      "step": 8270
-    },
-    {
-      "epoch": 0.033452247724398726,
-      "grad_norm": 985.6383666992188,
-      "learning_rate": 1.656e-05,
-      "loss": 240.4976,
-      "step": 8280
-    },
-    {
-      "epoch": 0.03349264898976636,
-      "grad_norm": 1016.156494140625,
-      "learning_rate": 1.658e-05,
-      "loss": 211.4707,
-      "step": 8290
-    },
-    {
-      "epoch": 0.03353305025513399,
-      "grad_norm": 590.7755737304688,
-      "learning_rate": 1.66e-05,
-      "loss": 173.2719,
-      "step": 8300
-    },
-    {
-      "epoch": 0.033573451520501625,
-      "grad_norm": 850.2467041015625,
-      "learning_rate": 1.662e-05,
-      "loss": 169.5091,
-      "step": 8310
-    },
-    {
-      "epoch": 0.03361385278586925,
-      "grad_norm": 6493.62109375,
-      "learning_rate": 1.664e-05,
-      "loss": 206.8252,
-      "step": 8320
-    },
-    {
-      "epoch": 0.033654254051236884,
-      "grad_norm": 721.207275390625,
-      "learning_rate": 1.666e-05,
-      "loss": 172.7568,
-      "step": 8330
-    },
-    {
-      "epoch": 0.03369465531660452,
-      "grad_norm": 941.2609252929688,
-      "learning_rate": 1.668e-05,
-      "loss": 169.4456,
-      "step": 8340
-    },
-    {
-      "epoch": 0.03373505658197215,
-      "grad_norm": 2292.378173828125,
-      "learning_rate": 1.6700000000000003e-05,
-      "loss": 217.5546,
-      "step": 8350
-    },
-    {
-      "epoch": 0.033775457847339777,
-      "grad_norm": 752.173583984375,
-      "learning_rate": 1.672e-05,
-      "loss": 186.4127,
-      "step": 8360
-    },
-    {
-      "epoch": 0.03381585911270741,
-      "grad_norm": 1046.0406494140625,
-      "learning_rate": 1.674e-05,
-      "loss": 156.5079,
-      "step": 8370
-    },
-    {
-      "epoch": 0.03385626037807504,
-      "grad_norm": 926.9135131835938,
-      "learning_rate": 1.6760000000000002e-05,
-      "loss": 145.1286,
-      "step": 8380
-    },
-    {
-      "epoch": 0.033896661643442676,
-      "grad_norm": 2465.574951171875,
-      "learning_rate": 1.6780000000000002e-05,
-      "loss": 230.1866,
-      "step": 8390
-    },
-    {
-      "epoch": 0.0339370629088103,
-      "grad_norm": 1376.998779296875,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 149.4985,
-      "step": 8400
-    },
-    {
-      "epoch": 0.033977464174177935,
-      "grad_norm": 1141.4237060546875,
-      "learning_rate": 1.6819999999999998e-05,
-      "loss": 199.0205,
-      "step": 8410
-    },
-    {
-      "epoch": 0.03401786543954557,
-      "grad_norm": 823.1524047851562,
-      "learning_rate": 1.684e-05,
-      "loss": 160.2142,
-      "step": 8420
-    },
-    {
-      "epoch": 0.0340582667049132,
-      "grad_norm": 557.1250610351562,
-      "learning_rate": 1.686e-05,
-      "loss": 177.5694,
-      "step": 8430
-    },
-    {
-      "epoch": 0.03409866797028083,
-      "grad_norm": 841.5819702148438,
-      "learning_rate": 1.688e-05,
-      "loss": 165.8316,
-      "step": 8440
-    },
-    {
-      "epoch": 0.03413906923564846,
-      "grad_norm": 949.8247680664062,
-      "learning_rate": 1.69e-05,
-      "loss": 109.1915,
-      "step": 8450
-    },
-    {
-      "epoch": 0.03417947050101609,
-      "grad_norm": 1175.8765869140625,
-      "learning_rate": 1.692e-05,
-      "loss": 186.9094,
-      "step": 8460
-    },
-    {
-      "epoch": 0.034219871766383726,
-      "grad_norm": 601.2833862304688,
-      "learning_rate": 1.694e-05,
-      "loss": 173.7073,
-      "step": 8470
-    },
-    {
-      "epoch": 0.03426027303175135,
-      "grad_norm": 1607.2193603515625,
-      "learning_rate": 1.696e-05,
-      "loss": 193.047,
-      "step": 8480
-    },
-    {
-      "epoch": 0.034300674297118985,
-      "grad_norm": 794.0247802734375,
-      "learning_rate": 1.698e-05,
-      "loss": 230.634,
-      "step": 8490
-    },
-    {
-      "epoch": 0.03434107556248662,
-      "grad_norm": 506.083740234375,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 194.7215,
-      "step": 8500
-    },
-    {
-      "epoch": 0.03438147682785425,
-      "grad_norm": 1354.748046875,
-      "learning_rate": 1.702e-05,
-      "loss": 237.6669,
-      "step": 8510
-    },
-    {
-      "epoch": 0.03442187809322188,
-      "grad_norm": 903.7907104492188,
-      "learning_rate": 1.704e-05,
-      "loss": 176.5095,
-      "step": 8520
-    },
-    {
-      "epoch": 0.03446227935858951,
-      "grad_norm": 1148.167236328125,
-      "learning_rate": 1.706e-05,
-      "loss": 178.5096,
-      "step": 8530
-    },
-    {
-      "epoch": 0.034502680623957144,
-      "grad_norm": 2201.17919921875,
-      "learning_rate": 1.7080000000000002e-05,
-      "loss": 149.1055,
-      "step": 8540
-    },
-    {
-      "epoch": 0.03454308188932478,
-      "grad_norm": 2470.737060546875,
-      "learning_rate": 1.7100000000000002e-05,
-      "loss": 239.2134,
-      "step": 8550
-    },
-    {
-      "epoch": 0.0345834831546924,
-      "grad_norm": 1102.5830078125,
-      "learning_rate": 1.712e-05,
-      "loss": 116.4302,
-      "step": 8560
-    },
-    {
-      "epoch": 0.034623884420060036,
-      "grad_norm": 1073.0867919921875,
-      "learning_rate": 1.7140000000000002e-05,
-      "loss": 208.7902,
-      "step": 8570
-    },
-    {
-      "epoch": 0.03466428568542767,
-      "grad_norm": 1288.13818359375,
-      "learning_rate": 1.7160000000000002e-05,
-      "loss": 177.9251,
-      "step": 8580
-    },
-    {
-      "epoch": 0.0347046869507953,
-      "grad_norm": 3063.13525390625,
-      "learning_rate": 1.718e-05,
-      "loss": 132.933,
-      "step": 8590
-    },
-    {
-      "epoch": 0.03474508821616293,
-      "grad_norm": 817.6849365234375,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 329.9039,
-      "step": 8600
-    },
-    {
-      "epoch": 0.03478548948153056,
-      "grad_norm": 871.9563598632812,
-      "learning_rate": 1.722e-05,
-      "loss": 198.9892,
-      "step": 8610
-    },
-    {
-      "epoch": 0.034825890746898194,
-      "grad_norm": 2624.832275390625,
-      "learning_rate": 1.724e-05,
-      "loss": 134.0866,
-      "step": 8620
-    },
-    {
-      "epoch": 0.03486629201226583,
-      "grad_norm": 1550.62548828125,
-      "learning_rate": 1.726e-05,
-      "loss": 195.2643,
-      "step": 8630
-    },
-    {
-      "epoch": 0.034906693277633453,
-      "grad_norm": 2074.622314453125,
-      "learning_rate": 1.728e-05,
-      "loss": 202.7179,
-      "step": 8640
-    },
-    {
-      "epoch": 0.03494709454300109,
-      "grad_norm": 1240.3668212890625,
-      "learning_rate": 1.73e-05,
-      "loss": 164.9288,
-      "step": 8650
-    },
-    {
-      "epoch": 0.03498749580836872,
-      "grad_norm": 1201.8397216796875,
-      "learning_rate": 1.732e-05,
-      "loss": 173.0605,
-      "step": 8660
-    },
-    {
-      "epoch": 0.03502789707373635,
-      "grad_norm": 1204.3741455078125,
-      "learning_rate": 1.734e-05,
-      "loss": 161.6967,
-      "step": 8670
-    },
-    {
-      "epoch": 0.03506829833910398,
-      "grad_norm": 1461.201904296875,
-      "learning_rate": 1.736e-05,
-      "loss": 91.053,
-      "step": 8680
-    },
-    {
-      "epoch": 0.03510869960447161,
-      "grad_norm": 669.2667846679688,
-      "learning_rate": 1.7380000000000003e-05,
-      "loss": 125.9261,
-      "step": 8690
-    },
-    {
-      "epoch": 0.035149100869839245,
-      "grad_norm": 902.3343505859375,
-      "learning_rate": 1.74e-05,
-      "loss": 250.4465,
-      "step": 8700
-    },
-    {
-      "epoch": 0.03518950213520688,
-      "grad_norm": 2767.30224609375,
-      "learning_rate": 1.742e-05,
-      "loss": 194.7519,
-      "step": 8710
-    },
-    {
-      "epoch": 0.035229903400574504,
-      "grad_norm": 778.5036010742188,
-      "learning_rate": 1.7440000000000002e-05,
-      "loss": 205.9209,
-      "step": 8720
-    },
-    {
-      "epoch": 0.03527030466594214,
-      "grad_norm": 2534.48681640625,
-      "learning_rate": 1.7460000000000002e-05,
-      "loss": 201.9174,
-      "step": 8730
-    },
-    {
-      "epoch": 0.03531070593130977,
-      "grad_norm": 3830.944580078125,
-      "learning_rate": 1.7480000000000002e-05,
-      "loss": 249.3413,
-      "step": 8740
-    },
-    {
-      "epoch": 0.0353511071966774,
-      "grad_norm": 1058.1517333984375,
-      "learning_rate": 1.75e-05,
-      "loss": 156.2294,
-      "step": 8750
-    },
-    {
-      "epoch": 0.03539150846204503,
-      "grad_norm": 1754.2557373046875,
-      "learning_rate": 1.752e-05,
-      "loss": 169.7632,
-      "step": 8760
-    },
-    {
-      "epoch": 0.03543190972741266,
-      "grad_norm": 631.4237060546875,
-      "learning_rate": 1.754e-05,
-      "loss": 143.3636,
-      "step": 8770
-    },
-    {
-      "epoch": 0.035472310992780295,
-      "grad_norm": 1334.2506103515625,
-      "learning_rate": 1.756e-05,
-      "loss": 172.2922,
-      "step": 8780
-    },
-    {
-      "epoch": 0.03551271225814793,
-      "grad_norm": 1203.375,
-      "learning_rate": 1.758e-05,
-      "loss": 174.6774,
-      "step": 8790
-    },
-    {
-      "epoch": 0.035553113523515555,
-      "grad_norm": 1072.609375,
-      "learning_rate": 1.76e-05,
-      "loss": 177.6043,
-      "step": 8800
-    },
-    {
-      "epoch": 0.03559351478888319,
-      "grad_norm": 1734.830810546875,
-      "learning_rate": 1.762e-05,
-      "loss": 213.749,
-      "step": 8810
-    },
-    {
-      "epoch": 0.03563391605425082,
-      "grad_norm": 1706.71875,
-      "learning_rate": 1.764e-05,
-      "loss": 141.3769,
-      "step": 8820
-    },
-    {
-      "epoch": 0.035674317319618454,
-      "grad_norm": 823.6603393554688,
-      "learning_rate": 1.766e-05,
-      "loss": 211.0458,
-      "step": 8830
-    },
-    {
-      "epoch": 0.03571471858498608,
-      "grad_norm": 1883.2462158203125,
-      "learning_rate": 1.7680000000000004e-05,
-      "loss": 147.5754,
-      "step": 8840
-    },
-    {
-      "epoch": 0.03575511985035371,
-      "grad_norm": 1167.08935546875,
-      "learning_rate": 1.77e-05,
-      "loss": 158.7962,
-      "step": 8850
-    },
-    {
-      "epoch": 0.035795521115721346,
-      "grad_norm": 653.7176513671875,
-      "learning_rate": 1.772e-05,
-      "loss": 162.9139,
-      "step": 8860
-    },
-    {
-      "epoch": 0.03583592238108898,
-      "grad_norm": 1610.84912109375,
-      "learning_rate": 1.774e-05,
-      "loss": 147.6589,
-      "step": 8870
-    },
-    {
-      "epoch": 0.035876323646456605,
-      "grad_norm": 764.0466918945312,
-      "learning_rate": 1.7760000000000003e-05,
-      "loss": 188.6739,
-      "step": 8880
-    },
-    {
-      "epoch": 0.03591672491182424,
-      "grad_norm": 1810.3558349609375,
-      "learning_rate": 1.7780000000000003e-05,
-      "loss": 205.71,
-      "step": 8890
-    },
-    {
-      "epoch": 0.03595712617719187,
-      "grad_norm": 968.5712280273438,
-      "learning_rate": 1.78e-05,
-      "loss": 164.1197,
-      "step": 8900
-    },
-    {
-      "epoch": 0.035997527442559504,
-      "grad_norm": 2819.0068359375,
-      "learning_rate": 1.7820000000000002e-05,
-      "loss": 162.7462,
-      "step": 8910
-    },
-    {
-      "epoch": 0.03603792870792713,
-      "grad_norm": 2304.89306640625,
-      "learning_rate": 1.7840000000000002e-05,
-      "loss": 166.9301,
-      "step": 8920
-    },
-    {
-      "epoch": 0.036078329973294763,
-      "grad_norm": 1451.4320068359375,
-      "learning_rate": 1.7860000000000002e-05,
-      "loss": 199.7026,
-      "step": 8930
-    },
-    {
-      "epoch": 0.0361187312386624,
-      "grad_norm": 1088.584228515625,
-      "learning_rate": 1.7879999999999998e-05,
-      "loss": 122.1127,
-      "step": 8940
-    },
-    {
-      "epoch": 0.03615913250403003,
-      "grad_norm": 1242.6817626953125,
-      "learning_rate": 1.79e-05,
-      "loss": 205.5091,
-      "step": 8950
-    },
-    {
-      "epoch": 0.036199533769397656,
-      "grad_norm": 3576.470458984375,
-      "learning_rate": 1.792e-05,
-      "loss": 222.864,
-      "step": 8960
-    },
-    {
-      "epoch": 0.03623993503476529,
-      "grad_norm": 7665.30419921875,
-      "learning_rate": 1.794e-05,
-      "loss": 177.141,
-      "step": 8970
-    },
-    {
-      "epoch": 0.03628033630013292,
-      "grad_norm": 874.79052734375,
-      "learning_rate": 1.796e-05,
-      "loss": 151.6607,
-      "step": 8980
-    },
-    {
-      "epoch": 0.036320737565500555,
-      "grad_norm": 2215.158203125,
-      "learning_rate": 1.798e-05,
-      "loss": 187.3096,
-      "step": 8990
-    },
-    {
-      "epoch": 0.03636113883086818,
-      "grad_norm": 2184.711669921875,
-      "learning_rate": 1.8e-05,
-      "loss": 192.0445,
-      "step": 9000
-    },
-    {
-      "epoch": 0.036401540096235814,
-      "grad_norm": 1615.7005615234375,
-      "learning_rate": 1.802e-05,
-      "loss": 200.9116,
-      "step": 9010
-    },
-    {
-      "epoch": 0.03644194136160345,
-      "grad_norm": 1396.12109375,
-      "learning_rate": 1.804e-05,
-      "loss": 116.5097,
-      "step": 9020
-    },
-    {
-      "epoch": 0.03648234262697108,
-      "grad_norm": 2052.5712890625,
-      "learning_rate": 1.8060000000000003e-05,
-      "loss": 91.6937,
-      "step": 9030
-    },
-    {
-      "epoch": 0.036522743892338706,
-      "grad_norm": 1198.88232421875,
-      "learning_rate": 1.808e-05,
-      "loss": 162.9016,
-      "step": 9040
-    },
-    {
-      "epoch": 0.03656314515770634,
-      "grad_norm": 1004.2339477539062,
-      "learning_rate": 1.81e-05,
-      "loss": 165.1726,
-      "step": 9050
-    },
-    {
-      "epoch": 0.03660354642307397,
-      "grad_norm": 2106.733642578125,
-      "learning_rate": 1.812e-05,
-      "loss": 293.9893,
-      "step": 9060
-    },
-    {
-      "epoch": 0.036643947688441605,
-      "grad_norm": 1582.510986328125,
-      "learning_rate": 1.8140000000000003e-05,
-      "loss": 190.1606,
-      "step": 9070
-    },
-    {
-      "epoch": 0.03668434895380923,
-      "grad_norm": 905.4036254882812,
-      "learning_rate": 1.8160000000000002e-05,
-      "loss": 141.9137,
-      "step": 9080
-    },
-    {
-      "epoch": 0.036724750219176865,
-      "grad_norm": 411.4434509277344,
-      "learning_rate": 1.818e-05,
-      "loss": 135.4074,
-      "step": 9090
-    },
-    {
-      "epoch": 0.0367651514845445,
-      "grad_norm": 2590.279541015625,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 237.3702,
-      "step": 9100
-    },
-    {
-      "epoch": 0.03680555274991213,
-      "grad_norm": 966.46435546875,
-      "learning_rate": 1.8220000000000002e-05,
-      "loss": 99.4366,
-      "step": 9110
-    },
-    {
-      "epoch": 0.03684595401527976,
-      "grad_norm": 1426.09375,
-      "learning_rate": 1.824e-05,
-      "loss": 128.6301,
-      "step": 9120
-    },
-    {
-      "epoch": 0.03688635528064739,
-      "grad_norm": 661.1119384765625,
-      "learning_rate": 1.826e-05,
-      "loss": 119.7455,
-      "step": 9130
-    },
-    {
-      "epoch": 0.03692675654601502,
-      "grad_norm": 1419.5643310546875,
-      "learning_rate": 1.828e-05,
-      "loss": 157.1782,
-      "step": 9140
-    },
-    {
-      "epoch": 0.03696715781138265,
-      "grad_norm": 875.4746704101562,
-      "learning_rate": 1.83e-05,
-      "loss": 167.5964,
-      "step": 9150
-    },
-    {
-      "epoch": 0.03700755907675028,
-      "grad_norm": 914.225341796875,
-      "learning_rate": 1.832e-05,
-      "loss": 229.4271,
-      "step": 9160
-    },
-    {
-      "epoch": 0.037047960342117915,
-      "grad_norm": 1139.4866943359375,
-      "learning_rate": 1.834e-05,
-      "loss": 176.9276,
-      "step": 9170
-    },
-    {
-      "epoch": 0.03708836160748555,
-      "grad_norm": 1400.506103515625,
-      "learning_rate": 1.8360000000000004e-05,
-      "loss": 137.9502,
-      "step": 9180
-    },
-    {
-      "epoch": 0.037128762872853174,
-      "grad_norm": 1892.8255615234375,
-      "learning_rate": 1.838e-05,
-      "loss": 177.446,
-      "step": 9190
-    },
-    {
-      "epoch": 0.03716916413822081,
-      "grad_norm": 6680.482421875,
-      "learning_rate": 1.84e-05,
-      "loss": 177.2692,
-      "step": 9200
-    },
-    {
-      "epoch": 0.03720956540358844,
-      "grad_norm": 5937.34521484375,
-      "learning_rate": 1.842e-05,
-      "loss": 209.457,
-      "step": 9210
-    },
-    {
-      "epoch": 0.037249966668956074,
-      "grad_norm": 724.9736938476562,
-      "learning_rate": 1.8440000000000003e-05,
-      "loss": 144.7552,
-      "step": 9220
-    },
-    {
-      "epoch": 0.0372903679343237,
-      "grad_norm": 1280.418212890625,
-      "learning_rate": 1.846e-05,
-      "loss": 177.4063,
-      "step": 9230
-    },
-    {
-      "epoch": 0.03733076919969133,
-      "grad_norm": 4052.574462890625,
-      "learning_rate": 1.848e-05,
-      "loss": 192.6805,
-      "step": 9240
-    },
-    {
-      "epoch": 0.037371170465058966,
-      "grad_norm": 1359.7706298828125,
-      "learning_rate": 1.85e-05,
-      "loss": 187.3014,
-      "step": 9250
-    },
-    {
-      "epoch": 0.0374115717304266,
-      "grad_norm": 3640.478515625,
-      "learning_rate": 1.8520000000000002e-05,
-      "loss": 134.8755,
-      "step": 9260
-    },
-    {
-      "epoch": 0.037451972995794225,
-      "grad_norm": 495.39373779296875,
-      "learning_rate": 1.8540000000000002e-05,
-      "loss": 199.3222,
-      "step": 9270
-    },
-    {
-      "epoch": 0.03749237426116186,
-      "grad_norm": 614.9301147460938,
-      "learning_rate": 1.856e-05,
-      "loss": 172.9295,
-      "step": 9280
-    },
-    {
-      "epoch": 0.03753277552652949,
-      "grad_norm": 1786.66552734375,
-      "learning_rate": 1.858e-05,
-      "loss": 170.4556,
-      "step": 9290
-    },
-    {
-      "epoch": 0.037573176791897124,
-      "grad_norm": 1209.83740234375,
-      "learning_rate": 1.86e-05,
-      "loss": 179.2815,
-      "step": 9300
-    },
-    {
-      "epoch": 0.03761357805726475,
-      "grad_norm": 935.8829345703125,
-      "learning_rate": 1.862e-05,
-      "loss": 153.5103,
-      "step": 9310
-    },
-    {
-      "epoch": 0.03765397932263238,
-      "grad_norm": 1041.441162109375,
-      "learning_rate": 1.864e-05,
-      "loss": 220.6227,
-      "step": 9320
-    },
-    {
-      "epoch": 0.037694380588000016,
-      "grad_norm": 1657.690673828125,
-      "learning_rate": 1.866e-05,
-      "loss": 238.1361,
-      "step": 9330
-    },
-    {
-      "epoch": 0.03773478185336765,
-      "grad_norm": 1077.7347412109375,
-      "learning_rate": 1.868e-05,
-      "loss": 156.283,
-      "step": 9340
-    },
-    {
-      "epoch": 0.037775183118735275,
-      "grad_norm": 834.681396484375,
-      "learning_rate": 1.87e-05,
-      "loss": 94.2577,
-      "step": 9350
-    },
-    {
-      "epoch": 0.03781558438410291,
-      "grad_norm": 1160.058349609375,
-      "learning_rate": 1.872e-05,
-      "loss": 231.7988,
-      "step": 9360
-    },
-    {
-      "epoch": 0.03785598564947054,
-      "grad_norm": 997.937255859375,
-      "learning_rate": 1.8740000000000004e-05,
-      "loss": 134.2654,
-      "step": 9370
-    },
-    {
-      "epoch": 0.037896386914838175,
-      "grad_norm": 882.4503784179688,
-      "learning_rate": 1.876e-05,
-      "loss": 193.2273,
-      "step": 9380
-    },
-    {
-      "epoch": 0.0379367881802058,
-      "grad_norm": 3181.698974609375,
-      "learning_rate": 1.878e-05,
-      "loss": 132.2644,
-      "step": 9390
-    },
-    {
-      "epoch": 0.037977189445573434,
-      "grad_norm": 725.4539184570312,
-      "learning_rate": 1.88e-05,
-      "loss": 153.5085,
-      "step": 9400
-    },
-    {
-      "epoch": 0.03801759071094107,
-      "grad_norm": 1430.8912353515625,
-      "learning_rate": 1.8820000000000003e-05,
-      "loss": 196.1231,
-      "step": 9410
-    },
-    {
-      "epoch": 0.0380579919763087,
-      "grad_norm": 1368.989501953125,
-      "learning_rate": 1.8840000000000003e-05,
-      "loss": 139.1376,
-      "step": 9420
-    },
-    {
-      "epoch": 0.038098393241676326,
-      "grad_norm": 838.9828491210938,
-      "learning_rate": 1.886e-05,
-      "loss": 131.1992,
-      "step": 9430
-    },
-    {
-      "epoch": 0.03813879450704396,
-      "grad_norm": 3527.458251953125,
-      "learning_rate": 1.888e-05,
-      "loss": 187.7699,
-      "step": 9440
-    },
-    {
-      "epoch": 0.03817919577241159,
-      "grad_norm": 2998.174072265625,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 175.4584,
-      "step": 9450
-    },
-    {
-      "epoch": 0.038219597037779225,
-      "grad_norm": 1297.8941650390625,
-      "learning_rate": 1.8920000000000002e-05,
-      "loss": 210.6279,
-      "step": 9460
-    },
-    {
-      "epoch": 0.03825999830314685,
-      "grad_norm": 3371.50244140625,
-      "learning_rate": 1.894e-05,
-      "loss": 217.5112,
-      "step": 9470
-    },
-    {
-      "epoch": 0.038300399568514484,
-      "grad_norm": 1407.463623046875,
-      "learning_rate": 1.896e-05,
-      "loss": 192.8969,
-      "step": 9480
-    },
-    {
-      "epoch": 0.03834080083388212,
-      "grad_norm": 1015.240966796875,
-      "learning_rate": 1.898e-05,
-      "loss": 173.7605,
-      "step": 9490
-    },
-    {
-      "epoch": 0.03838120209924975,
-      "grad_norm": 731.1430053710938,
-      "learning_rate": 1.9e-05,
-      "loss": 191.5719,
-      "step": 9500
-    },
-    {
-      "epoch": 0.03842160336461738,
-      "grad_norm": 849.3814086914062,
-      "learning_rate": 1.902e-05,
-      "loss": 91.8381,
-      "step": 9510
-    },
-    {
-      "epoch": 0.03846200462998501,
-      "grad_norm": 1715.9556884765625,
-      "learning_rate": 1.904e-05,
-      "loss": 175.3195,
-      "step": 9520
-    },
-    {
-      "epoch": 0.03850240589535264,
-      "grad_norm": 7559.9091796875,
-      "learning_rate": 1.906e-05,
-      "loss": 175.7841,
-      "step": 9530
-    },
-    {
-      "epoch": 0.038542807160720276,
-      "grad_norm": 689.8999633789062,
-      "learning_rate": 1.908e-05,
-      "loss": 116.3339,
-      "step": 9540
-    },
-    {
-      "epoch": 0.0385832084260879,
-      "grad_norm": 873.10009765625,
-      "learning_rate": 1.91e-05,
-      "loss": 140.4007,
-      "step": 9550
-    },
-    {
-      "epoch": 0.038623609691455535,
-      "grad_norm": 2358.655517578125,
-      "learning_rate": 1.9120000000000003e-05,
-      "loss": 168.361,
-      "step": 9560
-    },
-    {
-      "epoch": 0.03866401095682317,
-      "grad_norm": 5055.4892578125,
-      "learning_rate": 1.914e-05,
-      "loss": 174.0588,
-      "step": 9570
-    },
-    {
-      "epoch": 0.0387044122221908,
-      "grad_norm": 897.6202392578125,
-      "learning_rate": 1.916e-05,
-      "loss": 153.3665,
-      "step": 9580
-    },
-    {
-      "epoch": 0.03874481348755843,
-      "grad_norm": 729.1337280273438,
-      "learning_rate": 1.918e-05,
-      "loss": 131.9806,
-      "step": 9590
-    },
-    {
-      "epoch": 0.03878521475292606,
-      "grad_norm": 795.0892944335938,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 166.5195,
-      "step": 9600
-    },
-    {
-      "epoch": 0.03882561601829369,
-      "grad_norm": 644.804443359375,
-      "learning_rate": 1.9220000000000002e-05,
-      "loss": 129.34,
-      "step": 9610
-    },
-    {
-      "epoch": 0.038866017283661326,
-      "grad_norm": 1547.818359375,
-      "learning_rate": 1.924e-05,
-      "loss": 165.855,
-      "step": 9620
-    },
-    {
-      "epoch": 0.03890641854902895,
-      "grad_norm": 0.0,
-      "learning_rate": 1.9260000000000002e-05,
-      "loss": 181.7364,
-      "step": 9630
-    },
-    {
-      "epoch": 0.038946819814396585,
-      "grad_norm": 2220.852294921875,
-      "learning_rate": 1.9280000000000002e-05,
-      "loss": 154.0527,
-      "step": 9640
-    },
-    {
-      "epoch": 0.03898722107976422,
-      "grad_norm": 1048.51123046875,
-      "learning_rate": 1.93e-05,
-      "loss": 160.1465,
-      "step": 9650
-    },
-    {
-      "epoch": 0.03902762234513185,
-      "grad_norm": 0.0,
-      "learning_rate": 1.932e-05,
-      "loss": 149.3159,
-      "step": 9660
-    },
-    {
-      "epoch": 0.03906802361049948,
-      "grad_norm": 1180.42919921875,
-      "learning_rate": 1.934e-05,
-      "loss": 145.5787,
-      "step": 9670
-    },
-    {
-      "epoch": 0.03910842487586711,
-      "grad_norm": 1122.5113525390625,
-      "learning_rate": 1.936e-05,
-      "loss": 124.5202,
-      "step": 9680
-    },
-    {
-      "epoch": 0.039148826141234744,
-      "grad_norm": 1447.1502685546875,
-      "learning_rate": 1.938e-05,
-      "loss": 178.7178,
-      "step": 9690
-    },
-    {
-      "epoch": 0.03918922740660238,
-      "grad_norm": 1735.78125,
-      "learning_rate": 1.94e-05,
-      "loss": 184.0263,
-      "step": 9700
-    },
-    {
-      "epoch": 0.03922962867197,
-      "grad_norm": 660.9119873046875,
-      "learning_rate": 1.942e-05,
-      "loss": 127.7493,
-      "step": 9710
-    },
-    {
-      "epoch": 0.039270029937337636,
-      "grad_norm": 1763.235107421875,
-      "learning_rate": 1.944e-05,
-      "loss": 176.0283,
-      "step": 9720
-    },
-    {
-      "epoch": 0.03931043120270527,
-      "grad_norm": 2768.470947265625,
-      "learning_rate": 1.946e-05,
-      "loss": 107.1719,
-      "step": 9730
-    },
-    {
-      "epoch": 0.0393508324680729,
-      "grad_norm": 600.2669067382812,
-      "learning_rate": 1.948e-05,
-      "loss": 162.8209,
-      "step": 9740
-    },
-    {
-      "epoch": 0.03939123373344053,
-      "grad_norm": 3563.5947265625,
-      "learning_rate": 1.9500000000000003e-05,
-      "loss": 197.7512,
-      "step": 9750
-    },
-    {
-      "epoch": 0.03943163499880816,
-      "grad_norm": 2130.047607421875,
-      "learning_rate": 1.9520000000000003e-05,
-      "loss": 160.2494,
-      "step": 9760
-    },
-    {
-      "epoch": 0.039472036264175794,
-      "grad_norm": 2377.900146484375,
-      "learning_rate": 1.954e-05,
-      "loss": 171.3171,
-      "step": 9770
-    },
-    {
-      "epoch": 0.03951243752954343,
-      "grad_norm": 1052.0291748046875,
-      "learning_rate": 1.956e-05,
-      "loss": 113.5263,
-      "step": 9780
-    },
-    {
-      "epoch": 0.039552838794911054,
-      "grad_norm": 2078.976318359375,
-      "learning_rate": 1.9580000000000002e-05,
-      "loss": 191.9397,
-      "step": 9790
-    },
-    {
-      "epoch": 0.03959324006027869,
-      "grad_norm": 773.3158569335938,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 126.11,
-      "step": 9800
-    },
-    {
-      "epoch": 0.03963364132564632,
-      "grad_norm": 1939.7193603515625,
-      "learning_rate": 1.9620000000000002e-05,
-      "loss": 210.1035,
-      "step": 9810
-    },
-    {
-      "epoch": 0.03967404259101395,
-      "grad_norm": 531.8984985351562,
-      "learning_rate": 1.9640000000000002e-05,
-      "loss": 161.9592,
-      "step": 9820
-    },
-    {
-      "epoch": 0.03971444385638158,
-      "grad_norm": 441.3695983886719,
-      "learning_rate": 1.966e-05,
-      "loss": 131.7898,
-      "step": 9830
-    },
-    {
-      "epoch": 0.03975484512174921,
-      "grad_norm": 2530.7509765625,
-      "learning_rate": 1.968e-05,
-      "loss": 190.4342,
-      "step": 9840
-    },
-    {
-      "epoch": 0.039795246387116845,
-      "grad_norm": 1077.1409912109375,
-      "learning_rate": 1.97e-05,
-      "loss": 141.0169,
-      "step": 9850
-    },
-    {
-      "epoch": 0.03983564765248448,
-      "grad_norm": 1837.5992431640625,
-      "learning_rate": 1.972e-05,
-      "loss": 197.5706,
-      "step": 9860
-    },
-    {
-      "epoch": 0.039876048917852104,
-      "grad_norm": 1183.7669677734375,
-      "learning_rate": 1.974e-05,
-      "loss": 186.1175,
-      "step": 9870
-    },
-    {
-      "epoch": 0.03991645018321974,
-      "grad_norm": 584.4487915039062,
-      "learning_rate": 1.976e-05,
-      "loss": 128.3029,
-      "step": 9880
-    },
-    {
-      "epoch": 0.03995685144858737,
-      "grad_norm": 731.9237060546875,
-      "learning_rate": 1.978e-05,
-      "loss": 97.8851,
-      "step": 9890
-    },
-    {
-      "epoch": 0.039997252713955,
-      "grad_norm": 734.3873901367188,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 170.2385,
-      "step": 9900
-    },
-    {
-      "epoch": 0.04003765397932263,
-      "grad_norm": 530.0155029296875,
-      "learning_rate": 1.982e-05,
-      "loss": 140.1505,
-      "step": 9910
-    },
-    {
-      "epoch": 0.04007805524469026,
-      "grad_norm": 1368.81298828125,
-      "learning_rate": 1.984e-05,
-      "loss": 188.7803,
-      "step": 9920
-    },
-    {
-      "epoch": 0.040118456510057895,
-      "grad_norm": 2284.725341796875,
-      "learning_rate": 1.986e-05,
-      "loss": 213.4412,
-      "step": 9930
-    },
-    {
-      "epoch": 0.04015885777542553,
-      "grad_norm": 953.0332641601562,
-      "learning_rate": 1.9880000000000003e-05,
-      "loss": 81.9998,
-      "step": 9940
-    },
-    {
-      "epoch": 0.040199259040793155,
-      "grad_norm": 2036.2686767578125,
-      "learning_rate": 1.9900000000000003e-05,
-      "loss": 124.487,
-      "step": 9950
-    },
-    {
-      "epoch": 0.04023966030616079,
-      "grad_norm": 1137.9842529296875,
-      "learning_rate": 1.992e-05,
-      "loss": 166.7126,
-      "step": 9960
-    },
-    {
-      "epoch": 0.04028006157152842,
-      "grad_norm": 906.6007080078125,
-      "learning_rate": 1.994e-05,
-      "loss": 188.0849,
-      "step": 9970
-    },
-    {
-      "epoch": 0.040320462836896054,
-      "grad_norm": 1181.5977783203125,
-      "learning_rate": 1.9960000000000002e-05,
-      "loss": 168.7636,
-      "step": 9980
-    },
-    {
-      "epoch": 0.04036086410226368,
-      "grad_norm": 1022.9138793945312,
-      "learning_rate": 1.9980000000000002e-05,
-      "loss": 168.0031,
-      "step": 9990
-    },
-    {
-      "epoch": 0.04040126536763131,
-      "grad_norm": 717.4224243164062,
-      "learning_rate": 2e-05,
-      "loss": 122.0701,
-      "step": 10000
-    },
-    {
-      "epoch": 0.040441666632998946,
-      "grad_norm": 595.6052856445312,
-      "learning_rate": 2.002e-05,
-      "loss": 143.9377,
-      "step": 10010
-    },
-    {
-      "epoch": 0.04048206789836658,
-      "grad_norm": 1366.42431640625,
-      "learning_rate": 2.004e-05,
-      "loss": 167.5448,
-      "step": 10020
-    },
-    {
-      "epoch": 0.040522469163734205,
-      "grad_norm": 1101.7801513671875,
-      "learning_rate": 2.006e-05,
-      "loss": 192.1521,
-      "step": 10030
-    },
-    {
-      "epoch": 0.04056287042910184,
-      "grad_norm": 1699.7630615234375,
-      "learning_rate": 2.008e-05,
-      "loss": 154.8807,
-      "step": 10040
-    },
-    {
-      "epoch": 0.04060327169446947,
-      "grad_norm": 1256.3902587890625,
-      "learning_rate": 2.01e-05,
-      "loss": 134.1945,
-      "step": 10050
-    },
-    {
-      "epoch": 0.040643672959837104,
-      "grad_norm": 0.0,
-      "learning_rate": 2.012e-05,
-      "loss": 131.3899,
-      "step": 10060
-    },
-    {
-      "epoch": 0.04068407422520473,
-      "grad_norm": 1447.4774169921875,
-      "learning_rate": 2.014e-05,
-      "loss": 159.1337,
-      "step": 10070
-    },
-    {
-      "epoch": 0.040724475490572364,
-      "grad_norm": 4435.76904296875,
-      "learning_rate": 2.016e-05,
-      "loss": 198.5379,
-      "step": 10080
-    },
-    {
-      "epoch": 0.04076487675594,
-      "grad_norm": 852.9769287109375,
-      "learning_rate": 2.0180000000000003e-05,
-      "loss": 160.13,
-      "step": 10090
-    },
-    {
-      "epoch": 0.04080527802130763,
-      "grad_norm": 1151.4747314453125,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 162.3118,
-      "step": 10100
-    },
-    {
-      "epoch": 0.040845679286675256,
-      "grad_norm": 1879.0950927734375,
-      "learning_rate": 2.022e-05,
-      "loss": 172.1667,
-      "step": 10110
-    },
-    {
-      "epoch": 0.04088608055204289,
-      "grad_norm": 953.0318603515625,
-      "learning_rate": 2.024e-05,
-      "loss": 152.1001,
-      "step": 10120
-    },
-    {
-      "epoch": 0.04092648181741052,
-      "grad_norm": 834.5413208007812,
-      "learning_rate": 2.0260000000000003e-05,
-      "loss": 166.7469,
-      "step": 10130
-    },
-    {
-      "epoch": 0.040966883082778155,
-      "grad_norm": 1954.07080078125,
-      "learning_rate": 2.0280000000000002e-05,
-      "loss": 170.3608,
-      "step": 10140
-    },
-    {
-      "epoch": 0.04100728434814578,
-      "grad_norm": 1222.4248046875,
-      "learning_rate": 2.0300000000000002e-05,
-      "loss": 168.8605,
-      "step": 10150
-    },
-    {
-      "epoch": 0.041047685613513414,
-      "grad_norm": 2559.77099609375,
-      "learning_rate": 2.032e-05,
-      "loss": 169.7865,
-      "step": 10160
-    },
-    {
-      "epoch": 0.04108808687888105,
-      "grad_norm": 2988.0048828125,
-      "learning_rate": 2.0340000000000002e-05,
-      "loss": 175.7436,
-      "step": 10170
-    },
-    {
-      "epoch": 0.04112848814424868,
-      "grad_norm": 647.4199829101562,
-      "learning_rate": 2.036e-05,
-      "loss": 161.4738,
-      "step": 10180
-    },
-    {
-      "epoch": 0.041168889409616306,
-      "grad_norm": 1833.9415283203125,
-      "learning_rate": 2.038e-05,
-      "loss": 198.7381,
-      "step": 10190
-    },
-    {
-      "epoch": 0.04120929067498394,
-      "grad_norm": 734.6459350585938,
-      "learning_rate": 2.04e-05,
-      "loss": 145.9311,
-      "step": 10200
-    },
-    {
-      "epoch": 0.04124969194035157,
-      "grad_norm": 1672.8916015625,
-      "learning_rate": 2.042e-05,
-      "loss": 149.8653,
-      "step": 10210
-    },
-    {
-      "epoch": 0.041290093205719205,
-      "grad_norm": 1029.32421875,
-      "learning_rate": 2.044e-05,
-      "loss": 178.9316,
-      "step": 10220
-    },
-    {
-      "epoch": 0.04133049447108683,
-      "grad_norm": 714.8513793945312,
-      "learning_rate": 2.046e-05,
-      "loss": 112.8458,
-      "step": 10230
-    },
-    {
-      "epoch": 0.041370895736454465,
-      "grad_norm": 1445.0704345703125,
-      "learning_rate": 2.048e-05,
-      "loss": 164.2769,
-      "step": 10240
-    },
-    {
-      "epoch": 0.0414112970018221,
-      "grad_norm": 2015.292236328125,
-      "learning_rate": 2.05e-05,
-      "loss": 258.4405,
-      "step": 10250
-    },
-    {
-      "epoch": 0.04145169826718973,
-      "grad_norm": 820.0475463867188,
-      "learning_rate": 2.052e-05,
-      "loss": 139.5321,
-      "step": 10260
-    },
-    {
-      "epoch": 0.04149209953255736,
-      "grad_norm": 886.0709838867188,
-      "learning_rate": 2.054e-05,
-      "loss": 148.4284,
-      "step": 10270
-    },
-    {
-      "epoch": 0.04153250079792499,
-      "grad_norm": 1072.3577880859375,
-      "learning_rate": 2.0560000000000003e-05,
-      "loss": 192.8738,
-      "step": 10280
-    },
-    {
-      "epoch": 0.04157290206329262,
-      "grad_norm": 747.4786987304688,
-      "learning_rate": 2.0580000000000003e-05,
-      "loss": 101.708,
-      "step": 10290
-    },
-    {
-      "epoch": 0.041613303328660256,
-      "grad_norm": 5367.84521484375,
-      "learning_rate": 2.06e-05,
-      "loss": 194.0073,
-      "step": 10300
-    },
-    {
-      "epoch": 0.04165370459402788,
-      "grad_norm": 1179.1737060546875,
-      "learning_rate": 2.062e-05,
-      "loss": 121.6645,
-      "step": 10310
-    },
-    {
-      "epoch": 0.041694105859395515,
-      "grad_norm": 779.1863403320312,
-      "learning_rate": 2.0640000000000002e-05,
-      "loss": 156.9918,
-      "step": 10320
-    },
-    {
-      "epoch": 0.04173450712476315,
-      "grad_norm": 609.2747192382812,
-      "learning_rate": 2.0660000000000002e-05,
-      "loss": 121.481,
-      "step": 10330
-    },
-    {
-      "epoch": 0.04177490839013078,
-      "grad_norm": 856.9229736328125,
-      "learning_rate": 2.0680000000000002e-05,
-      "loss": 126.45,
-      "step": 10340
-    },
-    {
-      "epoch": 0.04181530965549841,
-      "grad_norm": 3409.32568359375,
-      "learning_rate": 2.07e-05,
-      "loss": 135.8212,
-      "step": 10350
-    },
-    {
-      "epoch": 0.04185571092086604,
-      "grad_norm": 2779.28369140625,
-      "learning_rate": 2.072e-05,
-      "loss": 180.5799,
-      "step": 10360
-    },
-    {
-      "epoch": 0.041896112186233674,
-      "grad_norm": 2200.98583984375,
-      "learning_rate": 2.074e-05,
-      "loss": 228.0418,
-      "step": 10370
-    },
-    {
-      "epoch": 0.04193651345160131,
-      "grad_norm": 864.7232055664062,
-      "learning_rate": 2.076e-05,
-      "loss": 200.8099,
-      "step": 10380
-    },
-    {
-      "epoch": 0.04197691471696893,
-      "grad_norm": 373.90771484375,
-      "learning_rate": 2.078e-05,
-      "loss": 123.1325,
-      "step": 10390
-    },
-    {
-      "epoch": 0.042017315982336566,
-      "grad_norm": 865.7789916992188,
-      "learning_rate": 2.08e-05,
-      "loss": 177.7623,
-      "step": 10400
-    },
-    {
-      "epoch": 0.0420577172477042,
-      "grad_norm": 1141.6817626953125,
-      "learning_rate": 2.082e-05,
-      "loss": 150.2677,
-      "step": 10410
-    },
-    {
-      "epoch": 0.04209811851307183,
-      "grad_norm": 720.0677490234375,
-      "learning_rate": 2.084e-05,
-      "loss": 146.5872,
-      "step": 10420
-    },
-    {
-      "epoch": 0.04213851977843946,
-      "grad_norm": 1222.44775390625,
-      "learning_rate": 2.086e-05,
-      "loss": 162.8006,
-      "step": 10430
-    },
-    {
-      "epoch": 0.04217892104380709,
-      "grad_norm": 909.986083984375,
-      "learning_rate": 2.0880000000000003e-05,
-      "loss": 193.938,
-      "step": 10440
-    },
-    {
-      "epoch": 0.042219322309174724,
-      "grad_norm": 1559.7752685546875,
-      "learning_rate": 2.09e-05,
-      "loss": 167.0518,
-      "step": 10450
-    },
-    {
-      "epoch": 0.04225972357454236,
-      "grad_norm": 2482.876953125,
-      "learning_rate": 2.092e-05,
-      "loss": 138.0267,
-      "step": 10460
-    },
-    {
-      "epoch": 0.04230012483990998,
-      "grad_norm": 1091.695556640625,
-      "learning_rate": 2.0940000000000003e-05,
-      "loss": 103.3883,
-      "step": 10470
-    },
-    {
-      "epoch": 0.042340526105277616,
-      "grad_norm": 806.2134399414062,
-      "learning_rate": 2.0960000000000003e-05,
-      "loss": 124.8749,
-      "step": 10480
-    },
-    {
-      "epoch": 0.04238092737064525,
-      "grad_norm": 1276.008544921875,
-      "learning_rate": 2.098e-05,
-      "loss": 163.3479,
-      "step": 10490
-    },
-    {
-      "epoch": 0.04242132863601288,
-      "grad_norm": 4489.3603515625,
-      "learning_rate": 2.1e-05,
-      "loss": 171.6303,
-      "step": 10500
-    },
-    {
-      "epoch": 0.04246172990138051,
-      "grad_norm": 838.2408447265625,
-      "learning_rate": 2.1020000000000002e-05,
-      "loss": 150.1332,
-      "step": 10510
-    },
-    {
-      "epoch": 0.04250213116674814,
-      "grad_norm": 1553.7894287109375,
-      "learning_rate": 2.1040000000000002e-05,
-      "loss": 161.144,
-      "step": 10520
-    },
-    {
-      "epoch": 0.042542532432115775,
-      "grad_norm": 1126.568115234375,
-      "learning_rate": 2.106e-05,
-      "loss": 152.1566,
-      "step": 10530
-    },
-    {
-      "epoch": 0.04258293369748341,
-      "grad_norm": 832.1701049804688,
-      "learning_rate": 2.1079999999999998e-05,
-      "loss": 139.9062,
-      "step": 10540
-    },
-    {
-      "epoch": 0.042623334962851034,
-      "grad_norm": 519.9462890625,
-      "learning_rate": 2.11e-05,
-      "loss": 108.9907,
-      "step": 10550
-    },
-    {
-      "epoch": 0.04266373622821867,
-      "grad_norm": 2303.817626953125,
-      "learning_rate": 2.112e-05,
-      "loss": 150.1262,
-      "step": 10560
-    },
-    {
-      "epoch": 0.0427041374935863,
-      "grad_norm": 1118.6610107421875,
-      "learning_rate": 2.114e-05,
-      "loss": 127.0093,
-      "step": 10570
-    },
-    {
-      "epoch": 0.04274453875895393,
-      "grad_norm": 1355.4498291015625,
-      "learning_rate": 2.116e-05,
-      "loss": 165.2866,
-      "step": 10580
-    },
-    {
-      "epoch": 0.04278494002432156,
-      "grad_norm": 2492.5751953125,
-      "learning_rate": 2.118e-05,
-      "loss": 167.1242,
-      "step": 10590
-    },
-    {
-      "epoch": 0.04282534128968919,
-      "grad_norm": 3182.501953125,
-      "learning_rate": 2.12e-05,
-      "loss": 146.6266,
-      "step": 10600
-    },
-    {
-      "epoch": 0.042865742555056825,
-      "grad_norm": 1109.6849365234375,
-      "learning_rate": 2.122e-05,
-      "loss": 195.9611,
-      "step": 10610
-    },
-    {
-      "epoch": 0.04290614382042446,
-      "grad_norm": 2110.990234375,
-      "learning_rate": 2.124e-05,
-      "loss": 199.2886,
-      "step": 10620
-    },
-    {
-      "epoch": 0.042946545085792084,
-      "grad_norm": 1315.0323486328125,
-      "learning_rate": 2.1260000000000003e-05,
-      "loss": 136.6857,
-      "step": 10630
-    },
-    {
-      "epoch": 0.04298694635115972,
-      "grad_norm": 1708.447998046875,
-      "learning_rate": 2.128e-05,
-      "loss": 173.377,
-      "step": 10640
-    },
-    {
-      "epoch": 0.04302734761652735,
-      "grad_norm": 1813.0264892578125,
-      "learning_rate": 2.13e-05,
-      "loss": 166.6691,
-      "step": 10650
-    },
-    {
-      "epoch": 0.043067748881894984,
-      "grad_norm": 1098.9163818359375,
-      "learning_rate": 2.1320000000000003e-05,
-      "loss": 187.8283,
-      "step": 10660
-    },
-    {
-      "epoch": 0.04310815014726261,
-      "grad_norm": 608.4671630859375,
-      "learning_rate": 2.1340000000000002e-05,
-      "loss": 141.6912,
-      "step": 10670
-    },
-    {
-      "epoch": 0.04314855141263024,
-      "grad_norm": 830.4765625,
-      "learning_rate": 2.1360000000000002e-05,
-      "loss": 142.6794,
-      "step": 10680
-    },
-    {
-      "epoch": 0.043188952677997876,
-      "grad_norm": 600.3438110351562,
-      "learning_rate": 2.138e-05,
-      "loss": 130.3432,
-      "step": 10690
-    },
-    {
-      "epoch": 0.04322935394336551,
-      "grad_norm": 1370.6168212890625,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 209.9451,
-      "step": 10700
-    },
-    {
-      "epoch": 0.043269755208733135,
-      "grad_norm": 622.1041870117188,
-      "learning_rate": 2.142e-05,
-      "loss": 169.206,
-      "step": 10710
-    },
-    {
-      "epoch": 0.04331015647410077,
-      "grad_norm": 2445.648681640625,
-      "learning_rate": 2.144e-05,
-      "loss": 200.1026,
-      "step": 10720
-    },
-    {
-      "epoch": 0.0433505577394684,
-      "grad_norm": 1902.2943115234375,
-      "learning_rate": 2.146e-05,
-      "loss": 187.5802,
-      "step": 10730
-    },
-    {
-      "epoch": 0.043390959004836034,
-      "grad_norm": 1008.3157348632812,
-      "learning_rate": 2.148e-05,
-      "loss": 143.9581,
-      "step": 10740
-    },
-    {
-      "epoch": 0.04343136027020366,
-      "grad_norm": 594.2481079101562,
-      "learning_rate": 2.15e-05,
-      "loss": 186.3464,
-      "step": 10750
-    },
-    {
-      "epoch": 0.04347176153557129,
-      "grad_norm": 13106.4541015625,
-      "learning_rate": 2.152e-05,
-      "loss": 218.3369,
-      "step": 10760
-    },
-    {
-      "epoch": 0.043512162800938926,
-      "grad_norm": 2128.98291015625,
-      "learning_rate": 2.154e-05,
-      "loss": 170.2695,
-      "step": 10770
-    },
-    {
-      "epoch": 0.04355256406630656,
-      "grad_norm": 1214.33642578125,
-      "learning_rate": 2.1560000000000004e-05,
-      "loss": 147.123,
-      "step": 10780
-    },
-    {
-      "epoch": 0.043592965331674186,
-      "grad_norm": 1065.6256103515625,
-      "learning_rate": 2.158e-05,
-      "loss": 168.7387,
-      "step": 10790
-    },
-    {
-      "epoch": 0.04363336659704182,
-      "grad_norm": 1667.8955078125,
-      "learning_rate": 2.16e-05,
-      "loss": 152.4974,
-      "step": 10800
-    },
-    {
-      "epoch": 0.04367376786240945,
-      "grad_norm": 1360.904052734375,
-      "learning_rate": 2.162e-05,
-      "loss": 254.4096,
-      "step": 10810
-    },
-    {
-      "epoch": 0.043714169127777085,
-      "grad_norm": 1592.132080078125,
-      "learning_rate": 2.1640000000000003e-05,
-      "loss": 174.9611,
-      "step": 10820
-    },
-    {
-      "epoch": 0.04375457039314471,
-      "grad_norm": 776.6605834960938,
-      "learning_rate": 2.166e-05,
-      "loss": 184.3046,
-      "step": 10830
-    },
-    {
-      "epoch": 0.043794971658512344,
-      "grad_norm": 2525.784423828125,
-      "learning_rate": 2.168e-05,
-      "loss": 164.588,
-      "step": 10840
-    },
-    {
-      "epoch": 0.04383537292387998,
-      "grad_norm": 1608.6549072265625,
-      "learning_rate": 2.1700000000000002e-05,
-      "loss": 196.8271,
-      "step": 10850
-    },
-    {
-      "epoch": 0.04387577418924761,
-      "grad_norm": 892.496826171875,
-      "learning_rate": 2.1720000000000002e-05,
-      "loss": 148.6461,
-      "step": 10860
-    },
-    {
-      "epoch": 0.043916175454615236,
-      "grad_norm": 1432.486572265625,
-      "learning_rate": 2.1740000000000002e-05,
-      "loss": 182.0831,
-      "step": 10870
-    },
-    {
-      "epoch": 0.04395657671998287,
-      "grad_norm": 1320.6700439453125,
-      "learning_rate": 2.176e-05,
-      "loss": 180.9173,
-      "step": 10880
-    },
-    {
-      "epoch": 0.0439969779853505,
-      "grad_norm": 1471.670166015625,
-      "learning_rate": 2.178e-05,
-      "loss": 178.3413,
-      "step": 10890
-    },
-    {
-      "epoch": 0.044037379250718135,
-      "grad_norm": 1130.7198486328125,
-      "learning_rate": 2.18e-05,
-      "loss": 148.5332,
-      "step": 10900
-    },
-    {
-      "epoch": 0.04407778051608576,
-      "grad_norm": 1092.3216552734375,
-      "learning_rate": 2.182e-05,
-      "loss": 114.7338,
-      "step": 10910
-    },
-    {
-      "epoch": 0.044118181781453394,
-      "grad_norm": 1079.041259765625,
-      "learning_rate": 2.184e-05,
-      "loss": 132.0338,
-      "step": 10920
-    },
-    {
-      "epoch": 0.04415858304682103,
-      "grad_norm": 2612.80517578125,
-      "learning_rate": 2.186e-05,
-      "loss": 138.605,
-      "step": 10930
-    },
-    {
-      "epoch": 0.04419898431218866,
-      "grad_norm": 1243.3662109375,
-      "learning_rate": 2.188e-05,
-      "loss": 157.8075,
-      "step": 10940
-    },
-    {
-      "epoch": 0.04423938557755629,
-      "grad_norm": 1165.291015625,
-      "learning_rate": 2.19e-05,
-      "loss": 123.9359,
-      "step": 10950
-    },
-    {
-      "epoch": 0.04427978684292392,
-      "grad_norm": 970.86083984375,
-      "learning_rate": 2.192e-05,
-      "loss": 99.9449,
-      "step": 10960
-    },
-    {
-      "epoch": 0.04432018810829155,
-      "grad_norm": 282.0602722167969,
-      "learning_rate": 2.1940000000000003e-05,
-      "loss": 126.6935,
-      "step": 10970
-    },
-    {
-      "epoch": 0.044360589373659186,
-      "grad_norm": 2742.96044921875,
-      "learning_rate": 2.196e-05,
-      "loss": 216.6111,
-      "step": 10980
-    },
-    {
-      "epoch": 0.04440099063902681,
-      "grad_norm": 725.5316772460938,
-      "learning_rate": 2.198e-05,
-      "loss": 283.1829,
-      "step": 10990
-    },
-    {
-      "epoch": 0.044441391904394445,
-      "grad_norm": 1572.414794921875,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 171.7036,
-      "step": 11000
-    },
-    {
-      "epoch": 0.04448179316976208,
-      "grad_norm": 0.0,
-      "learning_rate": 2.2020000000000003e-05,
-      "loss": 158.3066,
-      "step": 11010
-    },
-    {
-      "epoch": 0.04452219443512971,
-      "grad_norm": 1373.786376953125,
-      "learning_rate": 2.2040000000000002e-05,
-      "loss": 125.9849,
-      "step": 11020
-    },
-    {
-      "epoch": 0.04456259570049734,
-      "grad_norm": 1483.981689453125,
-      "learning_rate": 2.206e-05,
-      "loss": 176.718,
-      "step": 11030
-    },
-    {
-      "epoch": 0.04460299696586497,
-      "grad_norm": 997.575927734375,
-      "learning_rate": 2.2080000000000002e-05,
-      "loss": 164.0399,
-      "step": 11040
-    },
-    {
-      "epoch": 0.0446433982312326,
-      "grad_norm": 550.5761108398438,
-      "learning_rate": 2.2100000000000002e-05,
-      "loss": 135.1882,
-      "step": 11050
-    },
-    {
-      "epoch": 0.044683799496600236,
-      "grad_norm": 733.0974731445312,
-      "learning_rate": 2.212e-05,
-      "loss": 112.1112,
-      "step": 11060
-    },
-    {
-      "epoch": 0.04472420076196786,
-      "grad_norm": 856.7572021484375,
-      "learning_rate": 2.214e-05,
-      "loss": 113.8155,
-      "step": 11070
-    },
-    {
-      "epoch": 0.044764602027335496,
-      "grad_norm": 1171.519287109375,
-      "learning_rate": 2.216e-05,
-      "loss": 144.4537,
-      "step": 11080
-    },
-    {
-      "epoch": 0.04480500329270313,
-      "grad_norm": 1287.0703125,
-      "learning_rate": 2.218e-05,
-      "loss": 122.8268,
-      "step": 11090
-    },
-    {
-      "epoch": 0.04484540455807076,
-      "grad_norm": 1202.9896240234375,
-      "learning_rate": 2.22e-05,
-      "loss": 206.9014,
-      "step": 11100
-    },
-    {
-      "epoch": 0.04488580582343839,
-      "grad_norm": 1963.7164306640625,
-      "learning_rate": 2.222e-05,
-      "loss": 128.2789,
-      "step": 11110
-    },
-    {
-      "epoch": 0.04492620708880602,
-      "grad_norm": 2504.82958984375,
-      "learning_rate": 2.224e-05,
-      "loss": 124.5799,
-      "step": 11120
-    },
-    {
-      "epoch": 0.044966608354173654,
-      "grad_norm": 1163.331298828125,
-      "learning_rate": 2.226e-05,
-      "loss": 91.1706,
-      "step": 11130
-    },
-    {
-      "epoch": 0.04500700961954129,
-      "grad_norm": 10312.1435546875,
-      "learning_rate": 2.228e-05,
-      "loss": 200.2307,
-      "step": 11140
-    },
-    {
-      "epoch": 0.04504741088490891,
-      "grad_norm": 1264.3104248046875,
-      "learning_rate": 2.23e-05,
-      "loss": 161.8118,
-      "step": 11150
-    },
-    {
-      "epoch": 0.045087812150276546,
-      "grad_norm": 8854.8544921875,
-      "learning_rate": 2.2320000000000003e-05,
-      "loss": 162.5124,
-      "step": 11160
-    },
-    {
-      "epoch": 0.04512821341564418,
-      "grad_norm": 4647.8427734375,
-      "learning_rate": 2.234e-05,
-      "loss": 137.7809,
-      "step": 11170
-    },
-    {
-      "epoch": 0.04516861468101181,
-      "grad_norm": 1007.0010375976562,
-      "learning_rate": 2.236e-05,
-      "loss": 128.0276,
-      "step": 11180
-    },
-    {
-      "epoch": 0.04520901594637944,
-      "grad_norm": 2929.648681640625,
-      "learning_rate": 2.2380000000000003e-05,
-      "loss": 132.0453,
-      "step": 11190
-    },
-    {
-      "epoch": 0.04524941721174707,
-      "grad_norm": 2017.7265625,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 242.607,
-      "step": 11200
-    },
-    {
-      "epoch": 0.045289818477114704,
-      "grad_norm": 783.3958129882812,
-      "learning_rate": 2.2420000000000002e-05,
-      "loss": 146.9575,
-      "step": 11210
-    },
-    {
-      "epoch": 0.04533021974248234,
-      "grad_norm": 1278.9561767578125,
-      "learning_rate": 2.244e-05,
-      "loss": 183.9778,
-      "step": 11220
-    },
-    {
-      "epoch": 0.045370621007849964,
-      "grad_norm": 879.9774169921875,
-      "learning_rate": 2.2460000000000002e-05,
-      "loss": 139.0227,
-      "step": 11230
-    },
-    {
-      "epoch": 0.0454110222732176,
-      "grad_norm": 1722.348876953125,
-      "learning_rate": 2.248e-05,
-      "loss": 229.9668,
-      "step": 11240
-    },
-    {
-      "epoch": 0.04545142353858523,
-      "grad_norm": 652.4034423828125,
-      "learning_rate": 2.25e-05,
-      "loss": 129.4687,
-      "step": 11250
-    },
-    {
-      "epoch": 0.04549182480395286,
-      "grad_norm": 1721.231689453125,
-      "learning_rate": 2.252e-05,
-      "loss": 168.8096,
-      "step": 11260
-    },
-    {
-      "epoch": 0.04553222606932049,
-      "grad_norm": 1650.2803955078125,
-      "learning_rate": 2.254e-05,
-      "loss": 162.424,
-      "step": 11270
-    },
-    {
-      "epoch": 0.04557262733468812,
-      "grad_norm": 954.37890625,
-      "learning_rate": 2.256e-05,
-      "loss": 197.3434,
-      "step": 11280
-    },
-    {
-      "epoch": 0.045613028600055755,
-      "grad_norm": 4419.76123046875,
-      "learning_rate": 2.258e-05,
-      "loss": 210.9756,
-      "step": 11290
-    },
-    {
-      "epoch": 0.04565342986542339,
-      "grad_norm": 1498.54296875,
-      "learning_rate": 2.26e-05,
-      "loss": 134.9217,
-      "step": 11300
-    },
-    {
-      "epoch": 0.045693831130791014,
-      "grad_norm": 987.8052368164062,
-      "learning_rate": 2.2620000000000004e-05,
-      "loss": 163.9031,
-      "step": 11310
-    },
-    {
-      "epoch": 0.04573423239615865,
-      "grad_norm": 1421.3912353515625,
-      "learning_rate": 2.264e-05,
-      "loss": 175.6454,
-      "step": 11320
-    },
-    {
-      "epoch": 0.04577463366152628,
-      "grad_norm": 4952.677734375,
-      "learning_rate": 2.266e-05,
-      "loss": 191.8143,
-      "step": 11330
-    },
-    {
-      "epoch": 0.04581503492689391,
-      "grad_norm": 1982.3878173828125,
-      "learning_rate": 2.268e-05,
-      "loss": 163.27,
-      "step": 11340
-    },
-    {
-      "epoch": 0.04585543619226154,
-      "grad_norm": 1355.781005859375,
-      "learning_rate": 2.2700000000000003e-05,
-      "loss": 169.8846,
-      "step": 11350
-    },
-    {
-      "epoch": 0.04589583745762917,
-      "grad_norm": 1866.7637939453125,
-      "learning_rate": 2.2720000000000003e-05,
-      "loss": 195.6275,
-      "step": 11360
-    },
-    {
-      "epoch": 0.045936238722996806,
-      "grad_norm": 1269.4019775390625,
-      "learning_rate": 2.274e-05,
-      "loss": 136.8505,
-      "step": 11370
-    },
-    {
-      "epoch": 0.04597663998836444,
-      "grad_norm": 712.604736328125,
-      "learning_rate": 2.2760000000000002e-05,
-      "loss": 142.4693,
-      "step": 11380
-    },
-    {
-      "epoch": 0.046017041253732065,
-      "grad_norm": 730.568115234375,
-      "learning_rate": 2.2780000000000002e-05,
-      "loss": 123.7075,
-      "step": 11390
-    },
-    {
-      "epoch": 0.0460574425190997,
-      "grad_norm": 748.4280395507812,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 143.0049,
-      "step": 11400
-    },
-    {
-      "epoch": 0.04609784378446733,
-      "grad_norm": 1697.1856689453125,
-      "learning_rate": 2.282e-05,
-      "loss": 126.3755,
-      "step": 11410
-    },
-    {
-      "epoch": 0.046138245049834964,
-      "grad_norm": 2124.771240234375,
-      "learning_rate": 2.284e-05,
-      "loss": 172.3426,
-      "step": 11420
-    },
-    {
-      "epoch": 0.04617864631520259,
-      "grad_norm": 1425.5914306640625,
-      "learning_rate": 2.286e-05,
-      "loss": 152.4624,
-      "step": 11430
-    },
-    {
-      "epoch": 0.04621904758057022,
-      "grad_norm": 1039.5628662109375,
-      "learning_rate": 2.288e-05,
-      "loss": 168.2912,
-      "step": 11440
-    },
-    {
-      "epoch": 0.046259448845937856,
-      "grad_norm": 956.6918334960938,
-      "learning_rate": 2.29e-05,
-      "loss": 153.5313,
-      "step": 11450
-    },
-    {
-      "epoch": 0.04629985011130549,
-      "grad_norm": 1411.84423828125,
-      "learning_rate": 2.292e-05,
-      "loss": 180.7848,
-      "step": 11460
-    },
-    {
-      "epoch": 0.046340251376673115,
-      "grad_norm": 1310.616455078125,
-      "learning_rate": 2.294e-05,
-      "loss": 156.0493,
-      "step": 11470
-    },
-    {
-      "epoch": 0.04638065264204075,
-      "grad_norm": 1483.5418701171875,
-      "learning_rate": 2.296e-05,
-      "loss": 194.3713,
-      "step": 11480
-    },
-    {
-      "epoch": 0.04642105390740838,
-      "grad_norm": 1165.32763671875,
-      "learning_rate": 2.298e-05,
-      "loss": 143.955,
-      "step": 11490
-    },
-    {
-      "epoch": 0.046461455172776014,
-      "grad_norm": 1113.7464599609375,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 151.1558,
-      "step": 11500
-    },
-    {
-      "epoch": 0.04650185643814364,
-      "grad_norm": 591.7924194335938,
-      "learning_rate": 2.302e-05,
-      "loss": 147.8064,
-      "step": 11510
-    },
-    {
-      "epoch": 0.046542257703511274,
-      "grad_norm": 375.6275329589844,
-      "learning_rate": 2.304e-05,
-      "loss": 147.5092,
-      "step": 11520
-    },
-    {
-      "epoch": 0.04658265896887891,
-      "grad_norm": 1058.679443359375,
-      "learning_rate": 2.306e-05,
-      "loss": 129.6008,
-      "step": 11530
-    },
-    {
-      "epoch": 0.04662306023424654,
-      "grad_norm": 2693.062255859375,
-      "learning_rate": 2.3080000000000003e-05,
-      "loss": 168.9607,
-      "step": 11540
-    },
-    {
-      "epoch": 0.046663461499614166,
-      "grad_norm": 1006.8905029296875,
-      "learning_rate": 2.3100000000000002e-05,
-      "loss": 186.7646,
-      "step": 11550
-    },
-    {
-      "epoch": 0.0467038627649818,
-      "grad_norm": 1659.6192626953125,
-      "learning_rate": 2.312e-05,
-      "loss": 245.8637,
-      "step": 11560
-    },
-    {
-      "epoch": 0.04674426403034943,
-      "grad_norm": 1359.0103759765625,
-      "learning_rate": 2.3140000000000002e-05,
-      "loss": 150.9513,
-      "step": 11570
-    },
-    {
-      "epoch": 0.046784665295717065,
-      "grad_norm": 627.9051513671875,
-      "learning_rate": 2.3160000000000002e-05,
-      "loss": 96.2488,
-      "step": 11580
-    },
-    {
-      "epoch": 0.04682506656108469,
-      "grad_norm": 1331.7342529296875,
-      "learning_rate": 2.318e-05,
-      "loss": 139.2599,
-      "step": 11590
-    },
-    {
-      "epoch": 0.046865467826452324,
-      "grad_norm": 1107.905517578125,
-      "learning_rate": 2.32e-05,
-      "loss": 127.3733,
-      "step": 11600
-    },
-    {
-      "epoch": 0.04690586909181996,
-      "grad_norm": 1476.7293701171875,
-      "learning_rate": 2.322e-05,
-      "loss": 88.5844,
-      "step": 11610
-    },
-    {
-      "epoch": 0.04694627035718759,
-      "grad_norm": 1438.66748046875,
-      "learning_rate": 2.324e-05,
-      "loss": 146.773,
-      "step": 11620
-    },
-    {
-      "epoch": 0.046986671622555216,
-      "grad_norm": 1060.174560546875,
-      "learning_rate": 2.326e-05,
-      "loss": 185.9057,
-      "step": 11630
-    },
-    {
-      "epoch": 0.04702707288792285,
-      "grad_norm": 1116.5933837890625,
-      "learning_rate": 2.328e-05,
-      "loss": 161.711,
-      "step": 11640
-    },
-    {
-      "epoch": 0.04706747415329048,
-      "grad_norm": 1230.9957275390625,
-      "learning_rate": 2.3300000000000004e-05,
-      "loss": 161.603,
-      "step": 11650
-    },
-    {
-      "epoch": 0.047107875418658116,
-      "grad_norm": 1050.322265625,
-      "learning_rate": 2.332e-05,
-      "loss": 87.1483,
-      "step": 11660
-    },
-    {
-      "epoch": 0.04714827668402574,
-      "grad_norm": 561.3192138671875,
-      "learning_rate": 2.334e-05,
-      "loss": 72.6746,
-      "step": 11670
-    },
-    {
-      "epoch": 0.047188677949393375,
-      "grad_norm": 768.3349609375,
-      "learning_rate": 2.336e-05,
-      "loss": 245.4826,
-      "step": 11680
-    },
-    {
-      "epoch": 0.04722907921476101,
-      "grad_norm": 1956.8682861328125,
-      "learning_rate": 2.3380000000000003e-05,
-      "loss": 246.9039,
-      "step": 11690
-    },
-    {
-      "epoch": 0.04726948048012864,
-      "grad_norm": 1342.6962890625,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 143.455,
-      "step": 11700
-    },
-    {
-      "epoch": 0.04730988174549627,
-      "grad_norm": 1271.6512451171875,
-      "learning_rate": 2.342e-05,
-      "loss": 119.4163,
-      "step": 11710
-    },
-    {
-      "epoch": 0.0473502830108639,
-      "grad_norm": 973.38427734375,
-      "learning_rate": 2.344e-05,
-      "loss": 142.8418,
-      "step": 11720
-    },
-    {
-      "epoch": 0.04739068427623153,
-      "grad_norm": 725.7868041992188,
-      "learning_rate": 2.3460000000000002e-05,
-      "loss": 151.6831,
-      "step": 11730
-    },
-    {
-      "epoch": 0.047431085541599166,
-      "grad_norm": 1243.775146484375,
-      "learning_rate": 2.3480000000000002e-05,
-      "loss": 129.1701,
-      "step": 11740
-    },
-    {
-      "epoch": 0.04747148680696679,
-      "grad_norm": 1194.3792724609375,
-      "learning_rate": 2.35e-05,
-      "loss": 179.6937,
-      "step": 11750
-    },
-    {
-      "epoch": 0.047511888072334425,
-      "grad_norm": 836.063720703125,
-      "learning_rate": 2.3520000000000002e-05,
-      "loss": 96.7801,
-      "step": 11760
-    },
-    {
-      "epoch": 0.04755228933770206,
-      "grad_norm": 6927.7421875,
-      "learning_rate": 2.354e-05,
-      "loss": 170.788,
-      "step": 11770
-    },
-    {
-      "epoch": 0.04759269060306969,
-      "grad_norm": 1191.6163330078125,
-      "learning_rate": 2.356e-05,
-      "loss": 95.2363,
-      "step": 11780
-    },
-    {
-      "epoch": 0.04763309186843732,
-      "grad_norm": 4237.19384765625,
-      "learning_rate": 2.358e-05,
-      "loss": 149.3932,
-      "step": 11790
-    },
-    {
-      "epoch": 0.04767349313380495,
-      "grad_norm": 770.5933227539062,
-      "learning_rate": 2.36e-05,
-      "loss": 163.007,
-      "step": 11800
-    },
-    {
-      "epoch": 0.047713894399172584,
-      "grad_norm": 393.3005065917969,
-      "learning_rate": 2.362e-05,
-      "loss": 169.8844,
-      "step": 11810
-    },
-    {
-      "epoch": 0.04775429566454022,
-      "grad_norm": 1076.961181640625,
-      "learning_rate": 2.364e-05,
-      "loss": 144.8054,
-      "step": 11820
-    },
-    {
-      "epoch": 0.04779469692990784,
-      "grad_norm": 881.7391967773438,
-      "learning_rate": 2.366e-05,
-      "loss": 157.4215,
-      "step": 11830
-    },
-    {
-      "epoch": 0.047835098195275476,
-      "grad_norm": 1243.5374755859375,
-      "learning_rate": 2.3680000000000004e-05,
-      "loss": 159.1609,
-      "step": 11840
-    },
-    {
-      "epoch": 0.04787549946064311,
-      "grad_norm": 1366.5406494140625,
-      "learning_rate": 2.37e-05,
-      "loss": 132.554,
-      "step": 11850
-    },
-    {
-      "epoch": 0.04791590072601074,
-      "grad_norm": 1306.282470703125,
-      "learning_rate": 2.372e-05,
-      "loss": 133.127,
-      "step": 11860
-    },
-    {
-      "epoch": 0.04795630199137837,
-      "grad_norm": 1843.0394287109375,
-      "learning_rate": 2.374e-05,
-      "loss": 167.0897,
-      "step": 11870
-    },
-    {
-      "epoch": 0.047996703256746,
-      "grad_norm": 1241.4002685546875,
-      "learning_rate": 2.3760000000000003e-05,
-      "loss": 195.5516,
-      "step": 11880
-    },
-    {
-      "epoch": 0.048037104522113634,
-      "grad_norm": 1342.593017578125,
-      "learning_rate": 2.3780000000000003e-05,
-      "loss": 188.283,
-      "step": 11890
-    },
-    {
-      "epoch": 0.04807750578748127,
-      "grad_norm": 6172.35498046875,
-      "learning_rate": 2.38e-05,
-      "loss": 202.552,
-      "step": 11900
-    },
-    {
-      "epoch": 0.04811790705284889,
-      "grad_norm": 979.6281127929688,
-      "learning_rate": 2.3820000000000002e-05,
-      "loss": 126.3583,
-      "step": 11910
-    },
-    {
-      "epoch": 0.048158308318216526,
-      "grad_norm": 3780.92236328125,
-      "learning_rate": 2.3840000000000002e-05,
-      "loss": 125.3818,
-      "step": 11920
-    },
-    {
-      "epoch": 0.04819870958358416,
-      "grad_norm": 1145.655029296875,
-      "learning_rate": 2.3860000000000002e-05,
-      "loss": 135.4183,
-      "step": 11930
-    },
-    {
-      "epoch": 0.04823911084895179,
-      "grad_norm": 3658.722412109375,
-      "learning_rate": 2.3880000000000002e-05,
-      "loss": 180.5494,
-      "step": 11940
-    },
-    {
-      "epoch": 0.04827951211431942,
-      "grad_norm": 1273.83740234375,
-      "learning_rate": 2.39e-05,
-      "loss": 208.1243,
-      "step": 11950
-    },
-    {
-      "epoch": 0.04831991337968705,
-      "grad_norm": 701.6416015625,
-      "learning_rate": 2.392e-05,
-      "loss": 119.6717,
-      "step": 11960
-    },
-    {
-      "epoch": 0.048360314645054685,
-      "grad_norm": 598.38818359375,
-      "learning_rate": 2.394e-05,
-      "loss": 106.7641,
-      "step": 11970
-    },
-    {
-      "epoch": 0.04840071591042232,
-      "grad_norm": 697.3867797851562,
-      "learning_rate": 2.396e-05,
-      "loss": 157.0546,
-      "step": 11980
-    },
-    {
-      "epoch": 0.048441117175789944,
-      "grad_norm": 2313.9013671875,
-      "learning_rate": 2.398e-05,
-      "loss": 199.3208,
-      "step": 11990
-    },
-    {
-      "epoch": 0.04848151844115758,
-      "grad_norm": 712.7728881835938,
-      "learning_rate": 2.4e-05,
-      "loss": 163.3351,
-      "step": 12000
-    },
-    {
-      "epoch": 0.04852191970652521,
-      "grad_norm": 2287.64794921875,
-      "learning_rate": 2.402e-05,
-      "loss": 156.7126,
-      "step": 12010
-    },
-    {
-      "epoch": 0.04856232097189284,
-      "grad_norm": 1171.4212646484375,
-      "learning_rate": 2.404e-05,
-      "loss": 154.6671,
-      "step": 12020
-    },
-    {
-      "epoch": 0.04860272223726047,
-      "grad_norm": 928.2921142578125,
-      "learning_rate": 2.4060000000000003e-05,
-      "loss": 189.5635,
-      "step": 12030
-    },
-    {
-      "epoch": 0.0486431235026281,
-      "grad_norm": 2594.64794921875,
-      "learning_rate": 2.408e-05,
-      "loss": 122.9152,
-      "step": 12040
-    },
-    {
-      "epoch": 0.048683524767995735,
-      "grad_norm": 625.3611450195312,
-      "learning_rate": 2.41e-05,
-      "loss": 157.0619,
-      "step": 12050
-    },
-    {
-      "epoch": 0.04872392603336337,
-      "grad_norm": 1715.5015869140625,
-      "learning_rate": 2.412e-05,
-      "loss": 146.0929,
-      "step": 12060
-    },
-    {
-      "epoch": 0.048764327298730994,
-      "grad_norm": 1481.7435302734375,
-      "learning_rate": 2.4140000000000003e-05,
-      "loss": 150.0942,
-      "step": 12070
-    },
-    {
-      "epoch": 0.04880472856409863,
-      "grad_norm": 1766.0885009765625,
-      "learning_rate": 2.4160000000000002e-05,
-      "loss": 122.9469,
-      "step": 12080
-    },
-    {
-      "epoch": 0.04884512982946626,
-      "grad_norm": 1821.820556640625,
-      "learning_rate": 2.418e-05,
-      "loss": 236.4261,
-      "step": 12090
-    },
-    {
-      "epoch": 0.048885531094833894,
-      "grad_norm": 1121.892333984375,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 194.8495,
-      "step": 12100
-    },
-    {
-      "epoch": 0.04892593236020152,
-      "grad_norm": 2687.8505859375,
-      "learning_rate": 2.4220000000000002e-05,
-      "loss": 199.661,
-      "step": 12110
-    },
-    {
-      "epoch": 0.04896633362556915,
-      "grad_norm": 1251.5821533203125,
-      "learning_rate": 2.4240000000000002e-05,
-      "loss": 188.0581,
-      "step": 12120
-    },
-    {
-      "epoch": 0.049006734890936786,
-      "grad_norm": 1508.91455078125,
-      "learning_rate": 2.426e-05,
-      "loss": 155.5656,
-      "step": 12130
-    },
-    {
-      "epoch": 0.04904713615630442,
-      "grad_norm": 1015.0963745117188,
-      "learning_rate": 2.428e-05,
-      "loss": 156.3022,
-      "step": 12140
-    },
-    {
-      "epoch": 0.049087537421672045,
-      "grad_norm": 992.3895874023438,
-      "learning_rate": 2.43e-05,
-      "loss": 173.4455,
-      "step": 12150
-    },
-    {
-      "epoch": 0.04912793868703968,
-      "grad_norm": 5113.12451171875,
-      "learning_rate": 2.432e-05,
-      "loss": 201.295,
-      "step": 12160
-    },
-    {
-      "epoch": 0.04916833995240731,
-      "grad_norm": 1270.9464111328125,
-      "learning_rate": 2.434e-05,
-      "loss": 185.7967,
-      "step": 12170
-    },
-    {
-      "epoch": 0.04920874121777494,
-      "grad_norm": 1565.5892333984375,
-      "learning_rate": 2.4360000000000004e-05,
-      "loss": 170.4479,
-      "step": 12180
-    },
-    {
-      "epoch": 0.04924914248314257,
-      "grad_norm": 1304.619140625,
-      "learning_rate": 2.438e-05,
-      "loss": 173.9784,
-      "step": 12190
-    },
-    {
-      "epoch": 0.0492895437485102,
-      "grad_norm": 1918.267822265625,
-      "learning_rate": 2.44e-05,
-      "loss": 100.1017,
-      "step": 12200
-    },
-    {
-      "epoch": 0.049329945013877836,
-      "grad_norm": 3872.838134765625,
-      "learning_rate": 2.442e-05,
-      "loss": 128.8565,
-      "step": 12210
-    },
-    {
-      "epoch": 0.04937034627924546,
-      "grad_norm": 1211.8031005859375,
-      "learning_rate": 2.4440000000000003e-05,
-      "loss": 130.2835,
-      "step": 12220
-    },
-    {
-      "epoch": 0.049410747544613096,
-      "grad_norm": 1228.3138427734375,
-      "learning_rate": 2.4460000000000003e-05,
-      "loss": 191.7344,
-      "step": 12230
-    },
-    {
-      "epoch": 0.04945114880998073,
-      "grad_norm": 931.3762817382812,
-      "learning_rate": 2.448e-05,
-      "loss": 188.0092,
-      "step": 12240
-    },
-    {
-      "epoch": 0.04949155007534836,
-      "grad_norm": 3841.882568359375,
-      "learning_rate": 2.45e-05,
-      "loss": 176.4354,
-      "step": 12250
-    },
-    {
-      "epoch": 0.04953195134071599,
-      "grad_norm": 868.33935546875,
-      "learning_rate": 2.4520000000000002e-05,
-      "loss": 211.7608,
-      "step": 12260
-    },
-    {
-      "epoch": 0.04957235260608362,
-      "grad_norm": 1211.6044921875,
-      "learning_rate": 2.4540000000000002e-05,
-      "loss": 130.6806,
-      "step": 12270
-    },
-    {
-      "epoch": 0.049612753871451254,
-      "grad_norm": 1216.9310302734375,
-      "learning_rate": 2.4560000000000002e-05,
-      "loss": 163.6193,
-      "step": 12280
-    },
-    {
-      "epoch": 0.04965315513681889,
-      "grad_norm": 745.621337890625,
-      "learning_rate": 2.4580000000000002e-05,
-      "loss": 108.5664,
-      "step": 12290
-    },
-    {
-      "epoch": 0.04969355640218651,
-      "grad_norm": 737.6925048828125,
-      "learning_rate": 2.46e-05,
-      "loss": 163.5928,
-      "step": 12300
-    },
-    {
-      "epoch": 0.049733957667554146,
-      "grad_norm": 1154.443115234375,
-      "learning_rate": 2.462e-05,
-      "loss": 114.6066,
-      "step": 12310
-    },
-    {
-      "epoch": 0.04977435893292178,
-      "grad_norm": 1484.0311279296875,
-      "learning_rate": 2.464e-05,
-      "loss": 141.8092,
-      "step": 12320
-    },
-    {
-      "epoch": 0.04981476019828941,
-      "grad_norm": 1149.4248046875,
-      "learning_rate": 2.466e-05,
-      "loss": 142.7456,
-      "step": 12330
-    },
-    {
-      "epoch": 0.04985516146365704,
-      "grad_norm": 1364.304931640625,
-      "learning_rate": 2.468e-05,
-      "loss": 207.4074,
-      "step": 12340
-    },
-    {
-      "epoch": 0.04989556272902467,
-      "grad_norm": 802.7201538085938,
-      "learning_rate": 2.47e-05,
-      "loss": 162.2597,
-      "step": 12350
-    },
-    {
-      "epoch": 0.049935963994392304,
-      "grad_norm": 1634.9937744140625,
-      "learning_rate": 2.472e-05,
-      "loss": 158.4684,
-      "step": 12360
-    },
-    {
-      "epoch": 0.04997636525975994,
-      "grad_norm": 2859.273681640625,
-      "learning_rate": 2.4740000000000004e-05,
-      "loss": 171.345,
-      "step": 12370
-    },
-    {
-      "epoch": 0.050016766525127564,
-      "grad_norm": 1695.544189453125,
-      "learning_rate": 2.476e-05,
-      "loss": 153.0457,
-      "step": 12380
-    },
-    {
-      "epoch": 0.0500571677904952,
-      "grad_norm": 1252.4989013671875,
-      "learning_rate": 2.478e-05,
-      "loss": 223.5171,
-      "step": 12390
-    },
-    {
-      "epoch": 0.05009756905586283,
-      "grad_norm": 0.0,
-      "learning_rate": 2.48e-05,
-      "loss": 124.2808,
-      "step": 12400
-    },
-    {
-      "epoch": 0.05013797032123046,
-      "grad_norm": 1869.185546875,
-      "learning_rate": 2.4820000000000003e-05,
-      "loss": 176.1387,
-      "step": 12410
-    },
-    {
-      "epoch": 0.05017837158659809,
-      "grad_norm": 2657.873291015625,
-      "learning_rate": 2.4840000000000003e-05,
-      "loss": 136.5703,
-      "step": 12420
-    },
-    {
-      "epoch": 0.05021877285196572,
-      "grad_norm": 1434.21337890625,
-      "learning_rate": 2.486e-05,
-      "loss": 185.8033,
-      "step": 12430
-    },
-    {
-      "epoch": 0.050259174117333355,
-      "grad_norm": 4622.36328125,
-      "learning_rate": 2.488e-05,
-      "loss": 238.0834,
-      "step": 12440
-    },
-    {
-      "epoch": 0.05029957538270099,
-      "grad_norm": 2024.818359375,
-      "learning_rate": 2.4900000000000002e-05,
-      "loss": 134.9255,
-      "step": 12450
-    },
-    {
-      "epoch": 0.050339976648068614,
-      "grad_norm": 2336.821044921875,
-      "learning_rate": 2.4920000000000002e-05,
-      "loss": 112.3334,
-      "step": 12460
-    },
-    {
-      "epoch": 0.05038037791343625,
-      "grad_norm": 617.8527221679688,
-      "learning_rate": 2.4940000000000002e-05,
-      "loss": 69.1087,
-      "step": 12470
-    },
-    {
-      "epoch": 0.05042077917880388,
-      "grad_norm": 871.863037109375,
-      "learning_rate": 2.496e-05,
-      "loss": 201.0841,
-      "step": 12480
-    },
-    {
-      "epoch": 0.05046118044417151,
-      "grad_norm": 1000.557373046875,
-      "learning_rate": 2.498e-05,
-      "loss": 167.9705,
-      "step": 12490
-    },
-    {
-      "epoch": 0.05050158170953914,
-      "grad_norm": 1009.2303466796875,
-      "learning_rate": 2.5e-05,
-      "loss": 119.5033,
-      "step": 12500
-    },
-    {
-      "epoch": 0.05054198297490677,
-      "grad_norm": 2307.47607421875,
-      "learning_rate": 2.5019999999999998e-05,
-      "loss": 138.7849,
-      "step": 12510
-    },
-    {
-      "epoch": 0.050582384240274406,
-      "grad_norm": 2979.96533203125,
-      "learning_rate": 2.504e-05,
-      "loss": 132.3597,
-      "step": 12520
-    },
-    {
-      "epoch": 0.05062278550564204,
-      "grad_norm": 1333.7869873046875,
-      "learning_rate": 2.506e-05,
-      "loss": 114.2453,
-      "step": 12530
-    },
-    {
-      "epoch": 0.050663186771009665,
-      "grad_norm": 2323.83935546875,
-      "learning_rate": 2.5080000000000004e-05,
-      "loss": 199.8365,
-      "step": 12540
-    },
-    {
-      "epoch": 0.0507035880363773,
-      "grad_norm": 988.7455444335938,
-      "learning_rate": 2.51e-05,
-      "loss": 158.2298,
-      "step": 12550
-    },
-    {
-      "epoch": 0.05074398930174493,
-      "grad_norm": 1229.513427734375,
-      "learning_rate": 2.512e-05,
-      "loss": 177.6321,
-      "step": 12560
-    },
-    {
-      "epoch": 0.050784390567112564,
-      "grad_norm": 598.7608032226562,
-      "learning_rate": 2.5140000000000003e-05,
-      "loss": 205.6968,
-      "step": 12570
-    },
-    {
-      "epoch": 0.05082479183248019,
-      "grad_norm": 476.6854553222656,
-      "learning_rate": 2.516e-05,
-      "loss": 135.5016,
-      "step": 12580
-    },
-    {
-      "epoch": 0.05086519309784782,
-      "grad_norm": 441.1506652832031,
-      "learning_rate": 2.5180000000000003e-05,
-      "loss": 183.0943,
-      "step": 12590
-    },
-    {
-      "epoch": 0.050905594363215456,
-      "grad_norm": 861.925537109375,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 110.7362,
-      "step": 12600
-    },
-    {
-      "epoch": 0.05094599562858309,
-      "grad_norm": 1135.808837890625,
-      "learning_rate": 2.522e-05,
-      "loss": 157.3768,
-      "step": 12610
-    },
-    {
-      "epoch": 0.050986396893950715,
-      "grad_norm": 764.6640014648438,
-      "learning_rate": 2.5240000000000002e-05,
-      "loss": 197.2682,
-      "step": 12620
-    },
-    {
-      "epoch": 0.05102679815931835,
-      "grad_norm": 1185.298583984375,
-      "learning_rate": 2.526e-05,
-      "loss": 154.5542,
-      "step": 12630
-    },
-    {
-      "epoch": 0.05106719942468598,
-      "grad_norm": 572.9112548828125,
-      "learning_rate": 2.5280000000000005e-05,
-      "loss": 185.382,
-      "step": 12640
-    },
-    {
-      "epoch": 0.051107600690053615,
-      "grad_norm": 0.0,
-      "learning_rate": 2.5300000000000002e-05,
-      "loss": 155.8,
-      "step": 12650
-    },
-    {
-      "epoch": 0.05114800195542124,
-      "grad_norm": 1676.220703125,
-      "learning_rate": 2.5319999999999998e-05,
-      "loss": 153.2075,
-      "step": 12660
-    },
-    {
-      "epoch": 0.051188403220788874,
-      "grad_norm": 3056.052001953125,
-      "learning_rate": 2.534e-05,
-      "loss": 160.0424,
-      "step": 12670
-    },
-    {
-      "epoch": 0.05122880448615651,
-      "grad_norm": 3853.407958984375,
-      "learning_rate": 2.536e-05,
-      "loss": 142.0041,
-      "step": 12680
-    },
-    {
-      "epoch": 0.05126920575152414,
-      "grad_norm": 1022.783447265625,
-      "learning_rate": 2.5380000000000004e-05,
-      "loss": 173.1114,
-      "step": 12690
-    },
-    {
-      "epoch": 0.051309607016891766,
-      "grad_norm": 609.7467041015625,
-      "learning_rate": 2.54e-05,
-      "loss": 124.7601,
-      "step": 12700
-    },
-    {
-      "epoch": 0.0513500082822594,
-      "grad_norm": 871.4917602539062,
-      "learning_rate": 2.542e-05,
-      "loss": 107.3807,
-      "step": 12710
-    },
-    {
-      "epoch": 0.05139040954762703,
-      "grad_norm": 1300.36279296875,
-      "learning_rate": 2.5440000000000004e-05,
-      "loss": 128.8927,
-      "step": 12720
-    },
-    {
-      "epoch": 0.051430810812994665,
-      "grad_norm": 1378.396484375,
-      "learning_rate": 2.546e-05,
-      "loss": 179.4028,
-      "step": 12730
-    },
-    {
-      "epoch": 0.05147121207836229,
-      "grad_norm": 5043.6845703125,
-      "learning_rate": 2.5480000000000003e-05,
-      "loss": 145.2417,
-      "step": 12740
-    },
-    {
-      "epoch": 0.051511613343729924,
-      "grad_norm": 956.9037475585938,
-      "learning_rate": 2.5500000000000003e-05,
-      "loss": 122.089,
-      "step": 12750
-    },
-    {
-      "epoch": 0.05155201460909756,
-      "grad_norm": 655.8726806640625,
-      "learning_rate": 2.552e-05,
-      "loss": 172.3547,
-      "step": 12760
-    },
-    {
-      "epoch": 0.05159241587446519,
-      "grad_norm": 6090.908203125,
-      "learning_rate": 2.5540000000000003e-05,
-      "loss": 200.5886,
-      "step": 12770
-    },
-    {
-      "epoch": 0.051632817139832816,
-      "grad_norm": 760.5556640625,
-      "learning_rate": 2.556e-05,
-      "loss": 168.6275,
-      "step": 12780
-    },
-    {
-      "epoch": 0.05167321840520045,
-      "grad_norm": 1935.444091796875,
-      "learning_rate": 2.5580000000000002e-05,
-      "loss": 173.0487,
-      "step": 12790
-    },
-    {
-      "epoch": 0.05171361967056808,
-      "grad_norm": 1355.29931640625,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 169.9729,
-      "step": 12800
-    },
-    {
-      "epoch": 0.051754020935935716,
-      "grad_norm": 1155.65380859375,
-      "learning_rate": 2.562e-05,
-      "loss": 230.3313,
-      "step": 12810
-    },
-    {
-      "epoch": 0.05179442220130334,
-      "grad_norm": 1020.41064453125,
-      "learning_rate": 2.5640000000000002e-05,
-      "loss": 151.9028,
-      "step": 12820
-    },
-    {
-      "epoch": 0.051834823466670975,
-      "grad_norm": 1026.9921875,
-      "learning_rate": 2.566e-05,
-      "loss": 194.6138,
-      "step": 12830
-    },
-    {
-      "epoch": 0.05187522473203861,
-      "grad_norm": 859.4483032226562,
-      "learning_rate": 2.5679999999999998e-05,
-      "loss": 137.587,
-      "step": 12840
-    },
-    {
-      "epoch": 0.05191562599740624,
-      "grad_norm": 1088.9954833984375,
-      "learning_rate": 2.57e-05,
-      "loss": 137.7595,
-      "step": 12850
-    },
-    {
-      "epoch": 0.05195602726277387,
-      "grad_norm": 576.4170532226562,
-      "learning_rate": 2.572e-05,
-      "loss": 112.6303,
-      "step": 12860
-    },
-    {
-      "epoch": 0.0519964285281415,
-      "grad_norm": 1509.91943359375,
-      "learning_rate": 2.5740000000000004e-05,
-      "loss": 152.7388,
-      "step": 12870
-    },
-    {
-      "epoch": 0.05203682979350913,
-      "grad_norm": 1924.8671875,
-      "learning_rate": 2.576e-05,
-      "loss": 150.297,
-      "step": 12880
-    },
-    {
-      "epoch": 0.052077231058876766,
-      "grad_norm": 4206.75048828125,
-      "learning_rate": 2.5779999999999997e-05,
-      "loss": 153.3789,
-      "step": 12890
-    },
-    {
-      "epoch": 0.05211763232424439,
-      "grad_norm": 732.2725219726562,
-      "learning_rate": 2.58e-05,
-      "loss": 146.1245,
-      "step": 12900
-    },
-    {
-      "epoch": 0.052158033589612025,
-      "grad_norm": 804.0054321289062,
-      "learning_rate": 2.582e-05,
-      "loss": 155.7175,
-      "step": 12910
-    },
-    {
-      "epoch": 0.05219843485497966,
-      "grad_norm": 1622.7066650390625,
-      "learning_rate": 2.5840000000000003e-05,
-      "loss": 181.8288,
-      "step": 12920
-    },
-    {
-      "epoch": 0.05223883612034729,
-      "grad_norm": 1736.3221435546875,
-      "learning_rate": 2.586e-05,
-      "loss": 156.0094,
-      "step": 12930
-    },
-    {
-      "epoch": 0.05227923738571492,
-      "grad_norm": 1899.74072265625,
-      "learning_rate": 2.588e-05,
-      "loss": 161.6015,
-      "step": 12940
-    },
-    {
-      "epoch": 0.05231963865108255,
-      "grad_norm": 2343.95556640625,
-      "learning_rate": 2.5900000000000003e-05,
-      "loss": 170.2185,
-      "step": 12950
-    },
-    {
-      "epoch": 0.052360039916450184,
-      "grad_norm": 1888.6707763671875,
-      "learning_rate": 2.592e-05,
-      "loss": 204.9056,
-      "step": 12960
-    },
-    {
-      "epoch": 0.05240044118181782,
-      "grad_norm": 498.68780517578125,
-      "learning_rate": 2.5940000000000002e-05,
-      "loss": 142.0052,
-      "step": 12970
-    },
-    {
-      "epoch": 0.05244084244718544,
-      "grad_norm": 838.1019287109375,
-      "learning_rate": 2.5960000000000002e-05,
-      "loss": 113.003,
-      "step": 12980
-    },
-    {
-      "epoch": 0.052481243712553076,
-      "grad_norm": 636.7835083007812,
-      "learning_rate": 2.598e-05,
-      "loss": 96.236,
-      "step": 12990
-    },
-    {
-      "epoch": 0.05252164497792071,
-      "grad_norm": 1207.0146484375,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 128.6797,
-      "step": 13000
-    },
-    {
-      "epoch": 0.05256204624328834,
-      "grad_norm": 2183.191650390625,
-      "learning_rate": 2.602e-05,
-      "loss": 135.0814,
-      "step": 13010
-    },
-    {
-      "epoch": 0.05260244750865597,
-      "grad_norm": 1018.5341796875,
-      "learning_rate": 2.6040000000000005e-05,
-      "loss": 83.9305,
-      "step": 13020
-    },
-    {
-      "epoch": 0.0526428487740236,
-      "grad_norm": 437.8675537109375,
-      "learning_rate": 2.606e-05,
-      "loss": 141.287,
-      "step": 13030
-    },
-    {
-      "epoch": 0.052683250039391234,
-      "grad_norm": 2055.59228515625,
-      "learning_rate": 2.6079999999999998e-05,
-      "loss": 131.6824,
-      "step": 13040
-    },
-    {
-      "epoch": 0.05272365130475887,
-      "grad_norm": 1526.861572265625,
-      "learning_rate": 2.61e-05,
-      "loss": 104.1964,
-      "step": 13050
-    },
-    {
-      "epoch": 0.05276405257012649,
-      "grad_norm": 1107.270263671875,
-      "learning_rate": 2.612e-05,
-      "loss": 119.3528,
-      "step": 13060
-    },
-    {
-      "epoch": 0.052804453835494126,
-      "grad_norm": 0.0,
-      "learning_rate": 2.6140000000000004e-05,
-      "loss": 209.7389,
-      "step": 13070
-    },
-    {
-      "epoch": 0.05284485510086176,
-      "grad_norm": 887.7672729492188,
-      "learning_rate": 2.616e-05,
-      "loss": 134.3163,
-      "step": 13080
-    },
-    {
-      "epoch": 0.05288525636622939,
-      "grad_norm": 600.7372436523438,
-      "learning_rate": 2.618e-05,
-      "loss": 120.8824,
-      "step": 13090
-    },
-    {
-      "epoch": 0.05292565763159702,
-      "grad_norm": 2771.222900390625,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 116.1366,
-      "step": 13100
-    },
-    {
-      "epoch": 0.05296605889696465,
-      "grad_norm": 1079.399658203125,
-      "learning_rate": 2.622e-05,
-      "loss": 120.0783,
-      "step": 13110
-    },
-    {
-      "epoch": 0.053006460162332285,
-      "grad_norm": 3635.591796875,
-      "learning_rate": 2.6240000000000003e-05,
-      "loss": 114.0672,
-      "step": 13120
-    },
-    {
-      "epoch": 0.05304686142769992,
-      "grad_norm": 967.0631713867188,
-      "learning_rate": 2.6260000000000003e-05,
-      "loss": 197.697,
-      "step": 13130
-    },
-    {
-      "epoch": 0.053087262693067544,
-      "grad_norm": 540.1041259765625,
-      "learning_rate": 2.628e-05,
-      "loss": 169.7453,
-      "step": 13140
-    },
-    {
-      "epoch": 0.05312766395843518,
-      "grad_norm": 1378.034912109375,
-      "learning_rate": 2.6300000000000002e-05,
-      "loss": 152.1042,
-      "step": 13150
-    },
-    {
-      "epoch": 0.05316806522380281,
-      "grad_norm": 2139.15234375,
-      "learning_rate": 2.632e-05,
-      "loss": 219.7844,
-      "step": 13160
-    },
-    {
-      "epoch": 0.05320846648917044,
-      "grad_norm": 2143.552734375,
-      "learning_rate": 2.6340000000000002e-05,
-      "loss": 152.3353,
-      "step": 13170
-    },
-    {
-      "epoch": 0.05324886775453807,
-      "grad_norm": 1372.249755859375,
-      "learning_rate": 2.6360000000000002e-05,
-      "loss": 216.3558,
-      "step": 13180
-    },
-    {
-      "epoch": 0.0532892690199057,
-      "grad_norm": 723.7531127929688,
-      "learning_rate": 2.6379999999999998e-05,
-      "loss": 106.8604,
-      "step": 13190
-    },
-    {
-      "epoch": 0.053329670285273335,
-      "grad_norm": 1488.490234375,
-      "learning_rate": 2.64e-05,
-      "loss": 128.0357,
-      "step": 13200
-    },
-    {
-      "epoch": 0.05337007155064097,
-      "grad_norm": 962.85498046875,
-      "learning_rate": 2.642e-05,
-      "loss": 214.9135,
-      "step": 13210
-    },
-    {
-      "epoch": 0.053410472816008595,
-      "grad_norm": 916.577392578125,
-      "learning_rate": 2.6440000000000004e-05,
-      "loss": 100.3119,
-      "step": 13220
-    },
-    {
-      "epoch": 0.05345087408137623,
-      "grad_norm": 520.3284912109375,
-      "learning_rate": 2.646e-05,
-      "loss": 159.4224,
-      "step": 13230
-    },
-    {
-      "epoch": 0.05349127534674386,
-      "grad_norm": 1644.9588623046875,
-      "learning_rate": 2.648e-05,
-      "loss": 153.6036,
-      "step": 13240
-    },
-    {
-      "epoch": 0.053531676612111494,
-      "grad_norm": 834.4176025390625,
-      "learning_rate": 2.6500000000000004e-05,
-      "loss": 179.9941,
-      "step": 13250
-    },
-    {
-      "epoch": 0.05357207787747912,
-      "grad_norm": 753.1478881835938,
-      "learning_rate": 2.652e-05,
-      "loss": 114.1454,
-      "step": 13260
-    },
-    {
-      "epoch": 0.05361247914284675,
-      "grad_norm": 1375.6973876953125,
-      "learning_rate": 2.6540000000000003e-05,
-      "loss": 121.5828,
-      "step": 13270
-    },
-    {
-      "epoch": 0.053652880408214386,
-      "grad_norm": 837.9177856445312,
-      "learning_rate": 2.6560000000000003e-05,
-      "loss": 162.0316,
-      "step": 13280
-    },
-    {
-      "epoch": 0.05369328167358202,
-      "grad_norm": 1957.6124267578125,
-      "learning_rate": 2.658e-05,
-      "loss": 123.4925,
-      "step": 13290
-    },
-    {
-      "epoch": 0.053733682938949645,
-      "grad_norm": 2115.5908203125,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 163.0786,
-      "step": 13300
-    },
-    {
-      "epoch": 0.05377408420431728,
-      "grad_norm": 767.2696533203125,
-      "learning_rate": 2.662e-05,
-      "loss": 99.3506,
-      "step": 13310
-    },
-    {
-      "epoch": 0.05381448546968491,
-      "grad_norm": 1096.5877685546875,
-      "learning_rate": 2.6640000000000002e-05,
-      "loss": 130.1019,
-      "step": 13320
-    },
-    {
-      "epoch": 0.053854886735052544,
-      "grad_norm": 990.4303588867188,
-      "learning_rate": 2.6660000000000002e-05,
-      "loss": 218.8272,
-      "step": 13330
-    },
-    {
-      "epoch": 0.05389528800042017,
-      "grad_norm": 1289.1722412109375,
-      "learning_rate": 2.668e-05,
-      "loss": 181.8528,
-      "step": 13340
-    },
-    {
-      "epoch": 0.0539356892657878,
-      "grad_norm": 779.4696655273438,
-      "learning_rate": 2.6700000000000002e-05,
-      "loss": 129.5975,
-      "step": 13350
-    },
-    {
-      "epoch": 0.053976090531155436,
-      "grad_norm": 928.5990600585938,
-      "learning_rate": 2.672e-05,
-      "loss": 186.8758,
-      "step": 13360
-    },
-    {
-      "epoch": 0.05401649179652307,
-      "grad_norm": 3074.220947265625,
-      "learning_rate": 2.6740000000000005e-05,
-      "loss": 182.7136,
-      "step": 13370
-    },
-    {
-      "epoch": 0.054056893061890696,
-      "grad_norm": 1353.44970703125,
-      "learning_rate": 2.676e-05,
-      "loss": 115.3092,
-      "step": 13380
-    },
-    {
-      "epoch": 0.05409729432725833,
-      "grad_norm": 1833.181396484375,
-      "learning_rate": 2.678e-05,
-      "loss": 136.1055,
-      "step": 13390
-    },
-    {
-      "epoch": 0.05413769559262596,
-      "grad_norm": 671.771484375,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 101.5414,
-      "step": 13400
-    },
-    {
-      "epoch": 0.054178096857993595,
-      "grad_norm": 1407.6787109375,
-      "learning_rate": 2.682e-05,
-      "loss": 140.1277,
-      "step": 13410
-    },
-    {
-      "epoch": 0.05421849812336122,
-      "grad_norm": 999.6378784179688,
-      "learning_rate": 2.6840000000000004e-05,
-      "loss": 173.1913,
-      "step": 13420
-    },
-    {
-      "epoch": 0.054258899388728854,
-      "grad_norm": 1126.044189453125,
-      "learning_rate": 2.686e-05,
-      "loss": 100.9075,
-      "step": 13430
-    },
-    {
-      "epoch": 0.05429930065409649,
-      "grad_norm": 995.6368408203125,
-      "learning_rate": 2.688e-05,
-      "loss": 152.3883,
-      "step": 13440
-    },
-    {
-      "epoch": 0.05433970191946412,
-      "grad_norm": 1857.345947265625,
-      "learning_rate": 2.6900000000000003e-05,
-      "loss": 164.2901,
-      "step": 13450
-    },
-    {
-      "epoch": 0.054380103184831746,
-      "grad_norm": 1917.1810302734375,
-      "learning_rate": 2.692e-05,
-      "loss": 117.7432,
-      "step": 13460
-    },
-    {
-      "epoch": 0.05442050445019938,
-      "grad_norm": 1812.2779541015625,
-      "learning_rate": 2.694e-05,
-      "loss": 158.6927,
-      "step": 13470
-    },
-    {
-      "epoch": 0.05446090571556701,
-      "grad_norm": 823.9198608398438,
-      "learning_rate": 2.6960000000000003e-05,
-      "loss": 147.1588,
-      "step": 13480
-    },
-    {
-      "epoch": 0.054501306980934645,
-      "grad_norm": 1157.1685791015625,
-      "learning_rate": 2.698e-05,
-      "loss": 167.8175,
-      "step": 13490
-    },
-    {
-      "epoch": 0.05454170824630227,
-      "grad_norm": 1123.951416015625,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 194.7915,
-      "step": 13500
-    },
-    {
-      "epoch": 0.054582109511669905,
-      "grad_norm": 1402.1549072265625,
-      "learning_rate": 2.7020000000000002e-05,
-      "loss": 137.6079,
-      "step": 13510
-    },
-    {
-      "epoch": 0.05462251077703754,
-      "grad_norm": 786.1826782226562,
-      "learning_rate": 2.704e-05,
-      "loss": 113.2719,
-      "step": 13520
-    },
-    {
-      "epoch": 0.05466291204240517,
-      "grad_norm": 1365.093017578125,
-      "learning_rate": 2.7060000000000002e-05,
-      "loss": 104.7612,
-      "step": 13530
-    },
-    {
-      "epoch": 0.0547033133077728,
-      "grad_norm": 1461.1785888671875,
-      "learning_rate": 2.7079999999999998e-05,
-      "loss": 168.2114,
-      "step": 13540
-    },
-    {
-      "epoch": 0.05474371457314043,
-      "grad_norm": 1377.26806640625,
-      "learning_rate": 2.7100000000000005e-05,
-      "loss": 184.7103,
-      "step": 13550
-    },
-    {
-      "epoch": 0.05478411583850806,
-      "grad_norm": 1345.474853515625,
-      "learning_rate": 2.712e-05,
-      "loss": 137.9565,
-      "step": 13560
-    },
-    {
-      "epoch": 0.054824517103875696,
-      "grad_norm": 1087.7122802734375,
-      "learning_rate": 2.7139999999999998e-05,
-      "loss": 152.1091,
-      "step": 13570
-    },
-    {
-      "epoch": 0.05486491836924332,
-      "grad_norm": 0.0,
-      "learning_rate": 2.716e-05,
-      "loss": 82.2364,
-      "step": 13580
-    },
-    {
-      "epoch": 0.054905319634610955,
-      "grad_norm": 730.9200439453125,
-      "learning_rate": 2.718e-05,
-      "loss": 165.8833,
-      "step": 13590
-    },
-    {
-      "epoch": 0.05494572089997859,
-      "grad_norm": 1368.674072265625,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 121.1773,
-      "step": 13600
-    },
-    {
-      "epoch": 0.05498612216534622,
-      "grad_norm": 1869.849853515625,
-      "learning_rate": 2.722e-05,
-      "loss": 231.0817,
-      "step": 13610
-    },
-    {
-      "epoch": 0.05502652343071385,
-      "grad_norm": 976.0435180664062,
-      "learning_rate": 2.724e-05,
-      "loss": 111.1706,
-      "step": 13620
-    },
-    {
-      "epoch": 0.05506692469608148,
-      "grad_norm": 1996.230224609375,
-      "learning_rate": 2.7260000000000003e-05,
-      "loss": 189.7502,
-      "step": 13630
-    },
-    {
-      "epoch": 0.05510732596144911,
-      "grad_norm": 971.2109985351562,
-      "learning_rate": 2.728e-05,
-      "loss": 117.1644,
-      "step": 13640
-    },
-    {
-      "epoch": 0.055147727226816746,
-      "grad_norm": 1315.8665771484375,
-      "learning_rate": 2.7300000000000003e-05,
-      "loss": 185.0801,
-      "step": 13650
-    },
-    {
-      "epoch": 0.05518812849218437,
-      "grad_norm": 600.6631469726562,
-      "learning_rate": 2.7320000000000003e-05,
-      "loss": 157.1564,
-      "step": 13660
-    },
-    {
-      "epoch": 0.055228529757552006,
-      "grad_norm": 888.3346557617188,
-      "learning_rate": 2.734e-05,
-      "loss": 117.6915,
-      "step": 13670
-    },
-    {
-      "epoch": 0.05526893102291964,
-      "grad_norm": 808.2348022460938,
-      "learning_rate": 2.7360000000000002e-05,
-      "loss": 99.7357,
-      "step": 13680
-    },
-    {
-      "epoch": 0.05530933228828727,
-      "grad_norm": 1063.2569580078125,
-      "learning_rate": 2.738e-05,
-      "loss": 143.7845,
-      "step": 13690
-    },
-    {
-      "epoch": 0.0553497335536549,
-      "grad_norm": 1069.0965576171875,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 155.6282,
-      "step": 13700
-    },
-    {
-      "epoch": 0.05539013481902253,
-      "grad_norm": 1055.5029296875,
-      "learning_rate": 2.7420000000000002e-05,
-      "loss": 173.6117,
-      "step": 13710
-    },
-    {
-      "epoch": 0.055430536084390164,
-      "grad_norm": 940.4775390625,
-      "learning_rate": 2.7439999999999998e-05,
-      "loss": 120.3455,
-      "step": 13720
-    },
-    {
-      "epoch": 0.0554709373497578,
-      "grad_norm": 1010.4285888671875,
-      "learning_rate": 2.746e-05,
-      "loss": 142.2922,
-      "step": 13730
-    },
-    {
-      "epoch": 0.05551133861512542,
-      "grad_norm": 1600.6968994140625,
-      "learning_rate": 2.748e-05,
-      "loss": 157.9479,
-      "step": 13740
-    },
-    {
-      "epoch": 0.055551739880493056,
-      "grad_norm": 1101.0772705078125,
-      "learning_rate": 2.7500000000000004e-05,
-      "loss": 137.5087,
-      "step": 13750
-    },
-    {
-      "epoch": 0.05559214114586069,
-      "grad_norm": 2070.27978515625,
-      "learning_rate": 2.752e-05,
-      "loss": 125.7709,
-      "step": 13760
-    },
-    {
-      "epoch": 0.05563254241122832,
-      "grad_norm": 869.279052734375,
-      "learning_rate": 2.754e-05,
-      "loss": 135.6122,
-      "step": 13770
-    },
-    {
-      "epoch": 0.05567294367659595,
-      "grad_norm": 635.1331787109375,
-      "learning_rate": 2.7560000000000004e-05,
-      "loss": 120.2631,
-      "step": 13780
-    },
-    {
-      "epoch": 0.05571334494196358,
-      "grad_norm": 712.9791870117188,
-      "learning_rate": 2.758e-05,
-      "loss": 132.9366,
-      "step": 13790
-    },
-    {
-      "epoch": 0.055753746207331215,
-      "grad_norm": 1173.3760986328125,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 174.3566,
-      "step": 13800
-    },
-    {
-      "epoch": 0.05579414747269885,
-      "grad_norm": 1058.1973876953125,
-      "learning_rate": 2.762e-05,
-      "loss": 101.473,
-      "step": 13810
-    },
-    {
-      "epoch": 0.055834548738066474,
-      "grad_norm": 1767.0400390625,
-      "learning_rate": 2.764e-05,
-      "loss": 149.575,
-      "step": 13820
-    },
-    {
-      "epoch": 0.05587495000343411,
-      "grad_norm": 438.71697998046875,
-      "learning_rate": 2.7660000000000003e-05,
-      "loss": 150.794,
-      "step": 13830
-    },
-    {
-      "epoch": 0.05591535126880174,
-      "grad_norm": 1495.567626953125,
-      "learning_rate": 2.768e-05,
-      "loss": 116.9854,
-      "step": 13840
-    },
-    {
-      "epoch": 0.05595575253416937,
-      "grad_norm": 1169.4012451171875,
-      "learning_rate": 2.7700000000000002e-05,
-      "loss": 158.6589,
-      "step": 13850
-    },
-    {
-      "epoch": 0.055996153799537,
-      "grad_norm": 1633.1995849609375,
-      "learning_rate": 2.7720000000000002e-05,
-      "loss": 126.1669,
-      "step": 13860
-    },
-    {
-      "epoch": 0.05603655506490463,
-      "grad_norm": 1052.2532958984375,
-      "learning_rate": 2.774e-05,
-      "loss": 180.1025,
-      "step": 13870
-    },
-    {
-      "epoch": 0.056076956330272265,
-      "grad_norm": 1021.2620239257812,
-      "learning_rate": 2.7760000000000002e-05,
-      "loss": 136.4397,
-      "step": 13880
-    },
-    {
-      "epoch": 0.0561173575956399,
-      "grad_norm": 886.8902587890625,
-      "learning_rate": 2.778e-05,
-      "loss": 91.0113,
-      "step": 13890
-    },
-    {
-      "epoch": 0.056157758861007524,
-      "grad_norm": 1001.5513305664062,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 110.4852,
-      "step": 13900
-    },
-    {
-      "epoch": 0.05619816012637516,
-      "grad_norm": 668.9522094726562,
-      "learning_rate": 2.782e-05,
-      "loss": 126.04,
-      "step": 13910
-    },
-    {
-      "epoch": 0.05623856139174279,
-      "grad_norm": 3181.50146484375,
-      "learning_rate": 2.7839999999999998e-05,
-      "loss": 175.8362,
-      "step": 13920
-    },
-    {
-      "epoch": 0.05627896265711042,
-      "grad_norm": 1067.25732421875,
-      "learning_rate": 2.7860000000000004e-05,
-      "loss": 124.6091,
-      "step": 13930
-    },
-    {
-      "epoch": 0.05631936392247805,
-      "grad_norm": 758.8274536132812,
-      "learning_rate": 2.788e-05,
-      "loss": 129.1508,
-      "step": 13940
-    },
-    {
-      "epoch": 0.05635976518784568,
-      "grad_norm": 845.1571044921875,
-      "learning_rate": 2.7900000000000004e-05,
-      "loss": 164.5396,
-      "step": 13950
-    },
-    {
-      "epoch": 0.056400166453213316,
-      "grad_norm": 812.546875,
-      "learning_rate": 2.792e-05,
-      "loss": 144.2103,
-      "step": 13960
-    },
-    {
-      "epoch": 0.05644056771858095,
-      "grad_norm": 1005.88720703125,
-      "learning_rate": 2.794e-05,
-      "loss": 131.6613,
-      "step": 13970
-    },
-    {
-      "epoch": 0.056480968983948575,
-      "grad_norm": 2054.029541015625,
-      "learning_rate": 2.7960000000000003e-05,
-      "loss": 127.4749,
-      "step": 13980
-    },
-    {
-      "epoch": 0.05652137024931621,
-      "grad_norm": 466.14068603515625,
-      "learning_rate": 2.798e-05,
-      "loss": 163.5435,
-      "step": 13990
-    },
-    {
-      "epoch": 0.05656177151468384,
-      "grad_norm": 599.2301025390625,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 107.817,
-      "step": 14000
-    },
-    {
-      "epoch": 0.056602172780051474,
-      "grad_norm": 2105.31640625,
-      "learning_rate": 2.8020000000000003e-05,
-      "loss": 197.3428,
-      "step": 14010
-    },
-    {
-      "epoch": 0.0566425740454191,
-      "grad_norm": 953.76025390625,
-      "learning_rate": 2.804e-05,
-      "loss": 147.2875,
-      "step": 14020
-    },
-    {
-      "epoch": 0.05668297531078673,
-      "grad_norm": 495.1427001953125,
-      "learning_rate": 2.8060000000000002e-05,
-      "loss": 132.651,
-      "step": 14030
-    },
-    {
-      "epoch": 0.056723376576154366,
-      "grad_norm": 1009.431884765625,
-      "learning_rate": 2.8080000000000002e-05,
-      "loss": 145.7106,
-      "step": 14040
-    },
-    {
-      "epoch": 0.056763777841522,
-      "grad_norm": 939.3583374023438,
-      "learning_rate": 2.8100000000000005e-05,
-      "loss": 139.3796,
-      "step": 14050
-    },
-    {
-      "epoch": 0.056804179106889625,
-      "grad_norm": 1285.1646728515625,
-      "learning_rate": 2.8120000000000002e-05,
-      "loss": 142.4715,
-      "step": 14060
-    },
-    {
-      "epoch": 0.05684458037225726,
-      "grad_norm": 1945.31494140625,
-      "learning_rate": 2.8139999999999998e-05,
-      "loss": 181.1451,
-      "step": 14070
-    },
-    {
-      "epoch": 0.05688498163762489,
-      "grad_norm": 811.0607299804688,
-      "learning_rate": 2.816e-05,
-      "loss": 126.362,
-      "step": 14080
-    },
-    {
-      "epoch": 0.056925382902992525,
-      "grad_norm": 4073.4345703125,
-      "learning_rate": 2.818e-05,
-      "loss": 116.4024,
-      "step": 14090
-    },
-    {
-      "epoch": 0.05696578416836015,
-      "grad_norm": 3792.585205078125,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 192.2604,
-      "step": 14100
-    },
-    {
-      "epoch": 0.057006185433727784,
-      "grad_norm": 2066.12060546875,
-      "learning_rate": 2.822e-05,
-      "loss": 119.7598,
-      "step": 14110
-    },
-    {
-      "epoch": 0.05704658669909542,
-      "grad_norm": 938.1848754882812,
-      "learning_rate": 2.824e-05,
-      "loss": 127.1674,
-      "step": 14120
-    },
-    {
-      "epoch": 0.05708698796446305,
-      "grad_norm": 1010.5850830078125,
-      "learning_rate": 2.8260000000000004e-05,
-      "loss": 144.3951,
-      "step": 14130
-    },
-    {
-      "epoch": 0.057127389229830676,
-      "grad_norm": 2251.332275390625,
-      "learning_rate": 2.828e-05,
-      "loss": 202.1318,
-      "step": 14140
-    },
-    {
-      "epoch": 0.05716779049519831,
-      "grad_norm": 701.1483154296875,
-      "learning_rate": 2.83e-05,
-      "loss": 138.087,
-      "step": 14150
-    },
-    {
-      "epoch": 0.05720819176056594,
-      "grad_norm": 646.7645263671875,
-      "learning_rate": 2.8320000000000003e-05,
-      "loss": 114.6259,
-      "step": 14160
-    },
-    {
-      "epoch": 0.057248593025933575,
-      "grad_norm": 1258.2159423828125,
-      "learning_rate": 2.834e-05,
-      "loss": 134.9526,
-      "step": 14170
-    },
-    {
-      "epoch": 0.0572889942913012,
-      "grad_norm": 994.9586181640625,
-      "learning_rate": 2.8360000000000003e-05,
-      "loss": 116.3534,
-      "step": 14180
-    },
-    {
-      "epoch": 0.057329395556668834,
-      "grad_norm": 1028.46240234375,
-      "learning_rate": 2.8380000000000003e-05,
-      "loss": 140.937,
-      "step": 14190
-    },
-    {
-      "epoch": 0.05736979682203647,
-      "grad_norm": 814.48681640625,
-      "learning_rate": 2.84e-05,
-      "loss": 117.722,
-      "step": 14200
-    },
-    {
-      "epoch": 0.0574101980874041,
-      "grad_norm": 1542.7169189453125,
-      "learning_rate": 2.8420000000000002e-05,
-      "loss": 115.8539,
-      "step": 14210
-    },
-    {
-      "epoch": 0.057450599352771727,
-      "grad_norm": 576.6505737304688,
-      "learning_rate": 2.844e-05,
-      "loss": 111.9559,
-      "step": 14220
-    },
-    {
-      "epoch": 0.05749100061813936,
-      "grad_norm": 1681.9095458984375,
-      "learning_rate": 2.8460000000000002e-05,
-      "loss": 145.2461,
-      "step": 14230
-    },
-    {
-      "epoch": 0.05753140188350699,
-      "grad_norm": 1178.3514404296875,
-      "learning_rate": 2.8480000000000002e-05,
-      "loss": 178.6759,
-      "step": 14240
-    },
-    {
-      "epoch": 0.057571803148874626,
-      "grad_norm": 1500.439208984375,
-      "learning_rate": 2.8499999999999998e-05,
-      "loss": 131.998,
-      "step": 14250
-    },
-    {
-      "epoch": 0.05761220441424225,
-      "grad_norm": 1533.505126953125,
-      "learning_rate": 2.852e-05,
-      "loss": 147.4367,
-      "step": 14260
-    },
-    {
-      "epoch": 0.057652605679609885,
-      "grad_norm": 495.7720947265625,
-      "learning_rate": 2.854e-05,
-      "loss": 115.4109,
-      "step": 14270
-    },
-    {
-      "epoch": 0.05769300694497752,
-      "grad_norm": 561.2579956054688,
-      "learning_rate": 2.8560000000000004e-05,
-      "loss": 147.6145,
-      "step": 14280
-    },
-    {
-      "epoch": 0.05773340821034515,
-      "grad_norm": 571.2040405273438,
-      "learning_rate": 2.858e-05,
-      "loss": 197.6198,
-      "step": 14290
-    },
-    {
-      "epoch": 0.05777380947571278,
-      "grad_norm": 2524.6904296875,
-      "learning_rate": 2.86e-05,
-      "loss": 171.4188,
-      "step": 14300
-    },
-    {
-      "epoch": 0.05781421074108041,
-      "grad_norm": 2208.06396484375,
-      "learning_rate": 2.8620000000000004e-05,
-      "loss": 124.7192,
-      "step": 14310
-    },
-    {
-      "epoch": 0.05785461200644804,
-      "grad_norm": 1544.6385498046875,
-      "learning_rate": 2.864e-05,
-      "loss": 145.8727,
-      "step": 14320
-    },
-    {
-      "epoch": 0.057895013271815676,
-      "grad_norm": 953.2141723632812,
-      "learning_rate": 2.8660000000000003e-05,
-      "loss": 98.7387,
-      "step": 14330
-    },
-    {
-      "epoch": 0.0579354145371833,
-      "grad_norm": 497.53021240234375,
-      "learning_rate": 2.868e-05,
-      "loss": 167.9052,
-      "step": 14340
-    },
-    {
-      "epoch": 0.057975815802550935,
-      "grad_norm": 441.49951171875,
-      "learning_rate": 2.87e-05,
-      "loss": 166.814,
-      "step": 14350
-    },
-    {
-      "epoch": 0.05801621706791857,
-      "grad_norm": 2393.28271484375,
-      "learning_rate": 2.8720000000000003e-05,
-      "loss": 155.3981,
-      "step": 14360
-    },
-    {
-      "epoch": 0.0580566183332862,
-      "grad_norm": 5085.328125,
-      "learning_rate": 2.874e-05,
-      "loss": 126.6876,
-      "step": 14370
-    },
-    {
-      "epoch": 0.05809701959865383,
-      "grad_norm": 1251.988525390625,
-      "learning_rate": 2.8760000000000002e-05,
-      "loss": 161.9475,
-      "step": 14380
-    },
-    {
-      "epoch": 0.05813742086402146,
-      "grad_norm": 2122.52392578125,
-      "learning_rate": 2.8780000000000002e-05,
-      "loss": 177.3616,
-      "step": 14390
-    },
-    {
-      "epoch": 0.058177822129389094,
-      "grad_norm": 886.3876342773438,
-      "learning_rate": 2.88e-05,
-      "loss": 102.448,
-      "step": 14400
-    },
-    {
-      "epoch": 0.05821822339475673,
-      "grad_norm": 1496.39697265625,
-      "learning_rate": 2.8820000000000002e-05,
-      "loss": 177.4485,
-      "step": 14410
-    },
-    {
-      "epoch": 0.05825862466012435,
-      "grad_norm": 1387.0819091796875,
-      "learning_rate": 2.8840000000000002e-05,
-      "loss": 134.0211,
-      "step": 14420
-    },
-    {
-      "epoch": 0.058299025925491986,
-      "grad_norm": 571.4561157226562,
-      "learning_rate": 2.8860000000000005e-05,
-      "loss": 109.2302,
-      "step": 14430
-    },
-    {
-      "epoch": 0.05833942719085962,
-      "grad_norm": 679.628173828125,
-      "learning_rate": 2.888e-05,
-      "loss": 124.9287,
-      "step": 14440
-    },
-    {
-      "epoch": 0.05837982845622725,
-      "grad_norm": 1030.914306640625,
-      "learning_rate": 2.8899999999999998e-05,
-      "loss": 141.9723,
-      "step": 14450
-    },
-    {
-      "epoch": 0.05842022972159488,
-      "grad_norm": 1213.34033203125,
-      "learning_rate": 2.8920000000000004e-05,
-      "loss": 133.645,
-      "step": 14460
-    },
-    {
-      "epoch": 0.05846063098696251,
-      "grad_norm": 867.3865356445312,
-      "learning_rate": 2.894e-05,
-      "loss": 109.0037,
-      "step": 14470
-    },
-    {
-      "epoch": 0.058501032252330144,
-      "grad_norm": 1019.241943359375,
-      "learning_rate": 2.8960000000000004e-05,
-      "loss": 179.6844,
-      "step": 14480
-    },
-    {
-      "epoch": 0.05854143351769778,
-      "grad_norm": 3133.170166015625,
-      "learning_rate": 2.898e-05,
-      "loss": 141.274,
-      "step": 14490
-    },
-    {
-      "epoch": 0.058581834783065403,
-      "grad_norm": 675.6682739257812,
-      "learning_rate": 2.9e-05,
-      "loss": 116.3539,
-      "step": 14500
-    },
-    {
-      "epoch": 0.05862223604843304,
-      "grad_norm": 915.9588012695312,
-      "learning_rate": 2.9020000000000003e-05,
-      "loss": 95.1154,
-      "step": 14510
-    },
-    {
-      "epoch": 0.05866263731380067,
-      "grad_norm": 1290.10546875,
-      "learning_rate": 2.904e-05,
-      "loss": 179.5605,
-      "step": 14520
-    },
-    {
-      "epoch": 0.0587030385791683,
-      "grad_norm": 712.0923461914062,
-      "learning_rate": 2.9060000000000003e-05,
-      "loss": 146.6162,
-      "step": 14530
-    },
-    {
-      "epoch": 0.05874343984453593,
-      "grad_norm": 893.6064453125,
-      "learning_rate": 2.9080000000000003e-05,
-      "loss": 104.5347,
-      "step": 14540
-    },
-    {
-      "epoch": 0.05878384110990356,
-      "grad_norm": 1025.2301025390625,
-      "learning_rate": 2.91e-05,
-      "loss": 104.5477,
-      "step": 14550
-    },
-    {
-      "epoch": 0.058824242375271195,
-      "grad_norm": 2338.9072265625,
-      "learning_rate": 2.9120000000000002e-05,
-      "loss": 142.9934,
-      "step": 14560
-    },
-    {
-      "epoch": 0.05886464364063883,
-      "grad_norm": 633.3606567382812,
-      "learning_rate": 2.9140000000000002e-05,
-      "loss": 98.545,
-      "step": 14570
-    },
-    {
-      "epoch": 0.058905044906006454,
-      "grad_norm": 1361.783935546875,
-      "learning_rate": 2.9160000000000005e-05,
-      "loss": 113.2945,
-      "step": 14580
-    },
-    {
-      "epoch": 0.05894544617137409,
-      "grad_norm": 1828.3387451171875,
-      "learning_rate": 2.9180000000000002e-05,
-      "loss": 119.5862,
-      "step": 14590
-    },
-    {
-      "epoch": 0.05898584743674172,
-      "grad_norm": 2555.387939453125,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 177.2158,
-      "step": 14600
-    },
-    {
-      "epoch": 0.05902624870210935,
-      "grad_norm": 2425.751220703125,
-      "learning_rate": 2.922e-05,
-      "loss": 118.8401,
-      "step": 14610
-    },
-    {
-      "epoch": 0.05906664996747698,
-      "grad_norm": 758.7974243164062,
-      "learning_rate": 2.924e-05,
-      "loss": 136.7834,
-      "step": 14620
-    },
-    {
-      "epoch": 0.05910705123284461,
-      "grad_norm": 2093.38720703125,
-      "learning_rate": 2.9260000000000004e-05,
-      "loss": 91.4101,
-      "step": 14630
-    },
-    {
-      "epoch": 0.059147452498212245,
-      "grad_norm": 1140.317138671875,
-      "learning_rate": 2.928e-05,
-      "loss": 142.2483,
-      "step": 14640
-    },
-    {
-      "epoch": 0.05918785376357988,
-      "grad_norm": 764.8568115234375,
-      "learning_rate": 2.93e-05,
-      "loss": 105.0373,
-      "step": 14650
-    },
-    {
-      "epoch": 0.059228255028947505,
-      "grad_norm": 562.0804443359375,
-      "learning_rate": 2.9320000000000004e-05,
-      "loss": 140.6483,
-      "step": 14660
-    },
-    {
-      "epoch": 0.05926865629431514,
-      "grad_norm": 1717.2381591796875,
-      "learning_rate": 2.934e-05,
-      "loss": 120.7989,
-      "step": 14670
-    },
-    {
-      "epoch": 0.05930905755968277,
-      "grad_norm": 1295.689208984375,
-      "learning_rate": 2.9360000000000003e-05,
-      "loss": 145.4503,
-      "step": 14680
-    },
-    {
-      "epoch": 0.059349458825050404,
-      "grad_norm": 1297.196533203125,
-      "learning_rate": 2.9380000000000003e-05,
-      "loss": 127.0184,
-      "step": 14690
-    },
-    {
-      "epoch": 0.05938986009041803,
-      "grad_norm": 483.24420166015625,
-      "learning_rate": 2.94e-05,
-      "loss": 153.2117,
-      "step": 14700
-    },
-    {
-      "epoch": 0.05943026135578566,
-      "grad_norm": 777.18408203125,
-      "learning_rate": 2.9420000000000003e-05,
-      "loss": 126.1312,
-      "step": 14710
-    },
-    {
-      "epoch": 0.059470662621153296,
-      "grad_norm": 1387.6092529296875,
-      "learning_rate": 2.944e-05,
-      "loss": 158.3849,
-      "step": 14720
-    },
-    {
-      "epoch": 0.05951106388652093,
-      "grad_norm": 3654.914306640625,
-      "learning_rate": 2.946e-05,
-      "loss": 177.4717,
-      "step": 14730
-    },
-    {
-      "epoch": 0.059551465151888555,
-      "grad_norm": 2103.89599609375,
-      "learning_rate": 2.9480000000000002e-05,
-      "loss": 133.8275,
-      "step": 14740
-    },
-    {
-      "epoch": 0.05959186641725619,
-      "grad_norm": 506.49200439453125,
-      "learning_rate": 2.95e-05,
-      "loss": 121.0761,
-      "step": 14750
-    },
-    {
-      "epoch": 0.05963226768262382,
-      "grad_norm": 1055.931396484375,
-      "learning_rate": 2.9520000000000002e-05,
-      "loss": 143.0577,
-      "step": 14760
-    },
-    {
-      "epoch": 0.059672668947991454,
-      "grad_norm": 745.9114379882812,
-      "learning_rate": 2.9540000000000002e-05,
-      "loss": 140.8658,
-      "step": 14770
-    },
-    {
-      "epoch": 0.05971307021335908,
-      "grad_norm": 809.5819702148438,
-      "learning_rate": 2.9559999999999998e-05,
-      "loss": 104.4415,
-      "step": 14780
-    },
-    {
-      "epoch": 0.059753471478726713,
-      "grad_norm": 1688.5380859375,
-      "learning_rate": 2.958e-05,
-      "loss": 152.457,
-      "step": 14790
-    },
-    {
-      "epoch": 0.05979387274409435,
-      "grad_norm": 910.4462280273438,
-      "learning_rate": 2.96e-05,
-      "loss": 149.1814,
-      "step": 14800
-    },
-    {
-      "epoch": 0.05983427400946198,
-      "grad_norm": 753.3027954101562,
-      "learning_rate": 2.9620000000000004e-05,
-      "loss": 142.1045,
-      "step": 14810
-    },
-    {
-      "epoch": 0.059874675274829606,
-      "grad_norm": 545.16552734375,
-      "learning_rate": 2.964e-05,
-      "loss": 120.4504,
-      "step": 14820
-    },
-    {
-      "epoch": 0.05991507654019724,
-      "grad_norm": 1327.6309814453125,
-      "learning_rate": 2.9659999999999997e-05,
-      "loss": 109.4865,
-      "step": 14830
-    },
-    {
-      "epoch": 0.05995547780556487,
-      "grad_norm": 867.6209106445312,
-      "learning_rate": 2.9680000000000004e-05,
-      "loss": 128.9373,
-      "step": 14840
-    },
-    {
-      "epoch": 0.059995879070932505,
-      "grad_norm": 552.6983032226562,
-      "learning_rate": 2.97e-05,
-      "loss": 148.3395,
-      "step": 14850
-    },
-    {
-      "epoch": 0.06003628033630013,
-      "grad_norm": 1947.0716552734375,
-      "learning_rate": 2.9720000000000003e-05,
-      "loss": 115.2734,
-      "step": 14860
-    },
-    {
-      "epoch": 0.060076681601667764,
-      "grad_norm": 793.457763671875,
-      "learning_rate": 2.974e-05,
-      "loss": 112.6054,
-      "step": 14870
-    },
-    {
-      "epoch": 0.0601170828670354,
-      "grad_norm": 849.2798461914062,
-      "learning_rate": 2.976e-05,
-      "loss": 126.2524,
-      "step": 14880
-    },
-    {
-      "epoch": 0.06015748413240303,
-      "grad_norm": 1418.7293701171875,
-      "learning_rate": 2.9780000000000003e-05,
-      "loss": 135.3141,
-      "step": 14890
-    },
-    {
-      "epoch": 0.060197885397770656,
-      "grad_norm": 787.8895263671875,
-      "learning_rate": 2.98e-05,
-      "loss": 150.2098,
-      "step": 14900
-    },
-    {
-      "epoch": 0.06023828666313829,
-      "grad_norm": 1818.129150390625,
-      "learning_rate": 2.9820000000000002e-05,
-      "loss": 142.2445,
-      "step": 14910
-    },
-    {
-      "epoch": 0.06027868792850592,
-      "grad_norm": 910.6742553710938,
-      "learning_rate": 2.9840000000000002e-05,
-      "loss": 178.141,
-      "step": 14920
-    },
-    {
-      "epoch": 0.060319089193873555,
-      "grad_norm": 2399.4091796875,
-      "learning_rate": 2.986e-05,
-      "loss": 175.9365,
-      "step": 14930
-    },
-    {
-      "epoch": 0.06035949045924118,
-      "grad_norm": 1148.7041015625,
-      "learning_rate": 2.9880000000000002e-05,
-      "loss": 145.5652,
-      "step": 14940
-    },
-    {
-      "epoch": 0.060399891724608815,
-      "grad_norm": 1392.08740234375,
-      "learning_rate": 2.9900000000000002e-05,
-      "loss": 139.3894,
-      "step": 14950
-    },
-    {
-      "epoch": 0.06044029298997645,
-      "grad_norm": 878.8516235351562,
-      "learning_rate": 2.9920000000000005e-05,
-      "loss": 65.7636,
-      "step": 14960
-    },
-    {
-      "epoch": 0.06048069425534408,
-      "grad_norm": 2386.329833984375,
-      "learning_rate": 2.994e-05,
-      "loss": 186.3342,
-      "step": 14970
-    },
-    {
-      "epoch": 0.06052109552071171,
-      "grad_norm": 365.4548645019531,
-      "learning_rate": 2.9959999999999998e-05,
-      "loss": 117.0936,
-      "step": 14980
-    },
-    {
-      "epoch": 0.06056149678607934,
-      "grad_norm": 557.1009521484375,
-      "learning_rate": 2.998e-05,
-      "loss": 98.8324,
-      "step": 14990
-    },
-    {
-      "epoch": 0.06060189805144697,
-      "grad_norm": 1404.343017578125,
-      "learning_rate": 3e-05,
-      "loss": 129.4244,
-      "step": 15000
-    },
-    {
-      "epoch": 0.060642299316814606,
-      "grad_norm": 1246.394775390625,
-      "learning_rate": 3.0020000000000004e-05,
-      "loss": 126.248,
-      "step": 15010
-    },
-    {
-      "epoch": 0.06068270058218223,
-      "grad_norm": 1066.50048828125,
-      "learning_rate": 3.004e-05,
-      "loss": 124.5184,
-      "step": 15020
-    },
-    {
-      "epoch": 0.060723101847549865,
-      "grad_norm": 705.3773803710938,
-      "learning_rate": 3.006e-05,
-      "loss": 133.4564,
-      "step": 15030
-    },
-    {
-      "epoch": 0.0607635031129175,
-      "grad_norm": 874.49072265625,
-      "learning_rate": 3.0080000000000003e-05,
-      "loss": 111.9202,
-      "step": 15040
-    },
-    {
-      "epoch": 0.06080390437828513,
-      "grad_norm": 725.5115356445312,
-      "learning_rate": 3.01e-05,
-      "loss": 116.5958,
-      "step": 15050
-    },
-    {
-      "epoch": 0.06084430564365276,
-      "grad_norm": 1046.1822509765625,
-      "learning_rate": 3.0120000000000003e-05,
-      "loss": 108.4798,
-      "step": 15060
-    },
-    {
-      "epoch": 0.06088470690902039,
-      "grad_norm": 1189.979248046875,
-      "learning_rate": 3.0140000000000003e-05,
-      "loss": 141.0663,
-      "step": 15070
-    },
-    {
-      "epoch": 0.060925108174388024,
-      "grad_norm": 3112.551513671875,
-      "learning_rate": 3.016e-05,
-      "loss": 172.9219,
-      "step": 15080
-    },
-    {
-      "epoch": 0.06096550943975566,
-      "grad_norm": 2147.462890625,
-      "learning_rate": 3.0180000000000002e-05,
-      "loss": 146.7627,
-      "step": 15090
-    },
-    {
-      "epoch": 0.06100591070512328,
-      "grad_norm": 1260.1737060546875,
-      "learning_rate": 3.02e-05,
-      "loss": 151.8984,
-      "step": 15100
-    },
-    {
-      "epoch": 0.061046311970490916,
-      "grad_norm": 1158.15771484375,
-      "learning_rate": 3.0220000000000005e-05,
-      "loss": 116.2214,
-      "step": 15110
-    },
-    {
-      "epoch": 0.06108671323585855,
-      "grad_norm": 3242.16259765625,
-      "learning_rate": 3.0240000000000002e-05,
-      "loss": 153.6299,
-      "step": 15120
-    },
-    {
-      "epoch": 0.06112711450122618,
-      "grad_norm": 837.7076416015625,
-      "learning_rate": 3.0259999999999998e-05,
-      "loss": 84.5595,
-      "step": 15130
-    },
-    {
-      "epoch": 0.06116751576659381,
-      "grad_norm": 2217.368408203125,
-      "learning_rate": 3.028e-05,
-      "loss": 154.1678,
-      "step": 15140
-    },
-    {
-      "epoch": 0.06120791703196144,
-      "grad_norm": 753.1808471679688,
-      "learning_rate": 3.03e-05,
-      "loss": 130.1928,
-      "step": 15150
-    },
-    {
-      "epoch": 0.061248318297329074,
-      "grad_norm": 1219.47607421875,
-      "learning_rate": 3.0320000000000004e-05,
-      "loss": 169.753,
-      "step": 15160
-    },
-    {
-      "epoch": 0.06128871956269671,
-      "grad_norm": 6474.40478515625,
-      "learning_rate": 3.034e-05,
-      "loss": 149.4557,
-      "step": 15170
-    },
-    {
-      "epoch": 0.06132912082806433,
-      "grad_norm": 5858.99658203125,
-      "learning_rate": 3.036e-05,
-      "loss": 191.529,
-      "step": 15180
-    },
-    {
-      "epoch": 0.061369522093431966,
-      "grad_norm": 1600.056396484375,
-      "learning_rate": 3.0380000000000004e-05,
-      "loss": 130.8423,
-      "step": 15190
-    },
-    {
-      "epoch": 0.0614099233587996,
-      "grad_norm": 814.158935546875,
-      "learning_rate": 3.04e-05,
-      "loss": 148.7461,
-      "step": 15200
-    },
-    {
-      "epoch": 0.061450324624167225,
-      "grad_norm": 844.9783935546875,
-      "learning_rate": 3.0420000000000004e-05,
-      "loss": 148.264,
-      "step": 15210
-    },
-    {
-      "epoch": 0.06149072588953486,
-      "grad_norm": 1373.4136962890625,
-      "learning_rate": 3.0440000000000003e-05,
-      "loss": 172.1749,
-      "step": 15220
-    },
-    {
-      "epoch": 0.06153112715490249,
-      "grad_norm": 897.79248046875,
-      "learning_rate": 3.046e-05,
-      "loss": 113.2348,
-      "step": 15230
-    },
-    {
-      "epoch": 0.061571528420270125,
-      "grad_norm": 1343.7520751953125,
-      "learning_rate": 3.0480000000000003e-05,
-      "loss": 98.1409,
-      "step": 15240
-    },
-    {
-      "epoch": 0.06161192968563775,
-      "grad_norm": 828.697998046875,
-      "learning_rate": 3.05e-05,
-      "loss": 162.0212,
-      "step": 15250
-    },
-    {
-      "epoch": 0.061652330951005384,
-      "grad_norm": 366.743896484375,
-      "learning_rate": 3.0520000000000006e-05,
-      "loss": 75.6308,
-      "step": 15260
-    },
-    {
-      "epoch": 0.06169273221637302,
-      "grad_norm": 2019.9920654296875,
-      "learning_rate": 3.054e-05,
-      "loss": 237.3679,
-      "step": 15270
-    },
-    {
-      "epoch": 0.06173313348174065,
-      "grad_norm": 1361.7977294921875,
-      "learning_rate": 3.056e-05,
-      "loss": 108.2089,
-      "step": 15280
-    },
-    {
-      "epoch": 0.061773534747108276,
-      "grad_norm": 1101.6763916015625,
-      "learning_rate": 3.058e-05,
-      "loss": 90.1339,
-      "step": 15290
-    },
-    {
-      "epoch": 0.06181393601247591,
-      "grad_norm": 1785.1817626953125,
-      "learning_rate": 3.06e-05,
-      "loss": 142.6872,
-      "step": 15300
-    },
-    {
-      "epoch": 0.06185433727784354,
-      "grad_norm": 1439.0386962890625,
-      "learning_rate": 3.062e-05,
-      "loss": 135.7678,
-      "step": 15310
-    },
-    {
-      "epoch": 0.061894738543211175,
-      "grad_norm": 1499.1324462890625,
-      "learning_rate": 3.0640000000000005e-05,
-      "loss": 123.4259,
-      "step": 15320
-    },
-    {
-      "epoch": 0.0619351398085788,
-      "grad_norm": 1074.73681640625,
-      "learning_rate": 3.066e-05,
-      "loss": 98.1504,
-      "step": 15330
-    },
-    {
-      "epoch": 0.061975541073946434,
-      "grad_norm": 1967.539306640625,
-      "learning_rate": 3.0680000000000004e-05,
-      "loss": 151.5911,
-      "step": 15340
-    },
-    {
-      "epoch": 0.06201594233931407,
-      "grad_norm": 485.7192077636719,
-      "learning_rate": 3.07e-05,
-      "loss": 137.3458,
-      "step": 15350
-    },
-    {
-      "epoch": 0.0620563436046817,
-      "grad_norm": 753.7666015625,
-      "learning_rate": 3.072e-05,
-      "loss": 103.7652,
-      "step": 15360
-    },
-    {
-      "epoch": 0.06209674487004933,
-      "grad_norm": 1353.95556640625,
-      "learning_rate": 3.074e-05,
-      "loss": 117.3775,
-      "step": 15370
-    },
-    {
-      "epoch": 0.06213714613541696,
-      "grad_norm": 438.0009765625,
-      "learning_rate": 3.076e-05,
-      "loss": 115.8443,
-      "step": 15380
-    },
-    {
-      "epoch": 0.06217754740078459,
-      "grad_norm": 1873.119140625,
-      "learning_rate": 3.078e-05,
-      "loss": 134.3721,
-      "step": 15390
-    },
-    {
-      "epoch": 0.062217948666152226,
-      "grad_norm": 1559.0321044921875,
-      "learning_rate": 3.08e-05,
-      "loss": 146.9015,
-      "step": 15400
-    },
-    {
-      "epoch": 0.06225834993151985,
-      "grad_norm": 953.1947021484375,
-      "learning_rate": 3.082e-05,
-      "loss": 146.0465,
-      "step": 15410
-    },
-    {
-      "epoch": 0.062298751196887485,
-      "grad_norm": 1465.51806640625,
-      "learning_rate": 3.084e-05,
-      "loss": 127.4744,
-      "step": 15420
-    },
-    {
-      "epoch": 0.06233915246225512,
-      "grad_norm": 871.5877685546875,
-      "learning_rate": 3.086e-05,
-      "loss": 110.5899,
-      "step": 15430
-    },
-    {
-      "epoch": 0.06237955372762275,
-      "grad_norm": 2501.28564453125,
-      "learning_rate": 3.088e-05,
-      "loss": 202.4409,
-      "step": 15440
-    },
-    {
-      "epoch": 0.06241995499299038,
-      "grad_norm": 1591.2879638671875,
-      "learning_rate": 3.09e-05,
-      "loss": 129.12,
-      "step": 15450
-    },
-    {
-      "epoch": 0.06246035625835801,
-      "grad_norm": 838.3685302734375,
-      "learning_rate": 3.092e-05,
-      "loss": 102.4406,
-      "step": 15460
-    },
-    {
-      "epoch": 0.06250075752372564,
-      "grad_norm": 1001.7224731445312,
-      "learning_rate": 3.0940000000000005e-05,
-      "loss": 108.1954,
-      "step": 15470
-    },
-    {
-      "epoch": 0.06254115878909328,
-      "grad_norm": 1513.2293701171875,
-      "learning_rate": 3.096e-05,
-      "loss": 103.5236,
-      "step": 15480
-    },
-    {
-      "epoch": 0.0625815600544609,
-      "grad_norm": 1613.0614013671875,
-      "learning_rate": 3.0980000000000005e-05,
-      "loss": 159.1191,
-      "step": 15490
-    },
-    {
-      "epoch": 0.06262196131982854,
-      "grad_norm": 1024.2308349609375,
-      "learning_rate": 3.1e-05,
-      "loss": 131.2479,
-      "step": 15500
-    },
-    {
-      "epoch": 0.06266236258519617,
-      "grad_norm": 2692.8251953125,
-      "learning_rate": 3.102e-05,
-      "loss": 125.2973,
-      "step": 15510
-    },
-    {
-      "epoch": 0.0627027638505638,
-      "grad_norm": 1083.9881591796875,
-      "learning_rate": 3.104e-05,
-      "loss": 121.1962,
-      "step": 15520
-    },
-    {
-      "epoch": 0.06274316511593143,
-      "grad_norm": 1520.9947509765625,
-      "learning_rate": 3.106e-05,
-      "loss": 127.8408,
-      "step": 15530
-    },
-    {
-      "epoch": 0.06278356638129906,
-      "grad_norm": 871.9234008789062,
-      "learning_rate": 3.108e-05,
-      "loss": 107.3415,
-      "step": 15540
-    },
-    {
-      "epoch": 0.06282396764666669,
-      "grad_norm": 1313.199462890625,
-      "learning_rate": 3.1100000000000004e-05,
-      "loss": 124.2825,
-      "step": 15550
-    },
-    {
-      "epoch": 0.06286436891203433,
-      "grad_norm": 997.6826782226562,
-      "learning_rate": 3.112e-05,
-      "loss": 116.8165,
-      "step": 15560
-    },
-    {
-      "epoch": 0.06290477017740195,
-      "grad_norm": 713.2454833984375,
-      "learning_rate": 3.1140000000000003e-05,
-      "loss": 142.1327,
-      "step": 15570
-    },
-    {
-      "epoch": 0.06294517144276959,
-      "grad_norm": 841.623779296875,
-      "learning_rate": 3.116e-05,
-      "loss": 136.2796,
-      "step": 15580
-    },
-    {
-      "epoch": 0.06298557270813722,
-      "grad_norm": 1948.3192138671875,
-      "learning_rate": 3.118e-05,
-      "loss": 192.9212,
-      "step": 15590
-    },
-    {
-      "epoch": 0.06302597397350485,
-      "grad_norm": 1726.7506103515625,
-      "learning_rate": 3.12e-05,
-      "loss": 172.2863,
-      "step": 15600
-    },
-    {
-      "epoch": 0.06306637523887249,
-      "grad_norm": 1170.4307861328125,
-      "learning_rate": 3.122e-05,
-      "loss": 130.6975,
-      "step": 15610
-    },
-    {
-      "epoch": 0.06310677650424011,
-      "grad_norm": 1774.3511962890625,
-      "learning_rate": 3.1240000000000006e-05,
-      "loss": 137.2435,
-      "step": 15620
-    },
-    {
-      "epoch": 0.06314717776960774,
-      "grad_norm": 4738.48193359375,
-      "learning_rate": 3.126e-05,
-      "loss": 140.0143,
-      "step": 15630
-    },
-    {
-      "epoch": 0.06318757903497538,
-      "grad_norm": 793.2022094726562,
-      "learning_rate": 3.1280000000000005e-05,
-      "loss": 163.8472,
-      "step": 15640
-    },
-    {
-      "epoch": 0.063227980300343,
-      "grad_norm": 0.0,
-      "learning_rate": 3.13e-05,
-      "loss": 113.3165,
-      "step": 15650
-    },
-    {
-      "epoch": 0.06326838156571064,
-      "grad_norm": 1195.0482177734375,
-      "learning_rate": 3.132e-05,
-      "loss": 151.147,
-      "step": 15660
-    },
-    {
-      "epoch": 0.06330878283107827,
-      "grad_norm": 1180.14453125,
-      "learning_rate": 3.134e-05,
-      "loss": 100.6629,
-      "step": 15670
-    },
-    {
-      "epoch": 0.0633491840964459,
-      "grad_norm": 926.8880615234375,
-      "learning_rate": 3.136e-05,
-      "loss": 161.5127,
-      "step": 15680
-    },
-    {
-      "epoch": 0.06338958536181354,
-      "grad_norm": 1517.244873046875,
-      "learning_rate": 3.138e-05,
-      "loss": 164.7377,
-      "step": 15690
-    },
-    {
-      "epoch": 0.06342998662718116,
-      "grad_norm": 942.4938354492188,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 127.0913,
-      "step": 15700
-    },
-    {
-      "epoch": 0.06347038789254879,
-      "grad_norm": 519.3255004882812,
-      "learning_rate": 3.142e-05,
-      "loss": 110.9436,
-      "step": 15710
-    },
-    {
-      "epoch": 0.06351078915791643,
-      "grad_norm": 931.0354614257812,
-      "learning_rate": 3.1440000000000004e-05,
-      "loss": 155.0716,
-      "step": 15720
-    },
-    {
-      "epoch": 0.06355119042328405,
-      "grad_norm": 2942.558837890625,
-      "learning_rate": 3.146e-05,
-      "loss": 106.2107,
-      "step": 15730
-    },
-    {
-      "epoch": 0.0635915916886517,
-      "grad_norm": 1383.4794921875,
-      "learning_rate": 3.1480000000000004e-05,
-      "loss": 146.2436,
-      "step": 15740
-    },
-    {
-      "epoch": 0.06363199295401932,
-      "grad_norm": 3386.990966796875,
-      "learning_rate": 3.15e-05,
-      "loss": 148.2886,
-      "step": 15750
-    },
-    {
-      "epoch": 0.06367239421938695,
-      "grad_norm": 1014.3907470703125,
-      "learning_rate": 3.1519999999999996e-05,
-      "loss": 120.5053,
-      "step": 15760
-    },
-    {
-      "epoch": 0.06371279548475459,
-      "grad_norm": 762.3274536132812,
-      "learning_rate": 3.154e-05,
-      "loss": 139.9023,
-      "step": 15770
-    },
-    {
-      "epoch": 0.06375319675012221,
-      "grad_norm": 1421.413330078125,
-      "learning_rate": 3.156e-05,
-      "loss": 146.9265,
-      "step": 15780
-    },
-    {
-      "epoch": 0.06379359801548984,
-      "grad_norm": 1463.6063232421875,
-      "learning_rate": 3.1580000000000006e-05,
-      "loss": 134.0383,
-      "step": 15790
-    },
-    {
-      "epoch": 0.06383399928085748,
-      "grad_norm": 1137.2342529296875,
-      "learning_rate": 3.16e-05,
-      "loss": 161.4233,
-      "step": 15800
-    },
-    {
-      "epoch": 0.0638744005462251,
-      "grad_norm": 1729.1817626953125,
-      "learning_rate": 3.162e-05,
-      "loss": 109.8035,
-      "step": 15810
-    },
-    {
-      "epoch": 0.06391480181159274,
-      "grad_norm": 757.51025390625,
-      "learning_rate": 3.164e-05,
-      "loss": 83.0763,
-      "step": 15820
-    },
-    {
-      "epoch": 0.06395520307696037,
-      "grad_norm": 5324.74169921875,
-      "learning_rate": 3.166e-05,
-      "loss": 132.4072,
-      "step": 15830
-    },
-    {
-      "epoch": 0.063995604342328,
-      "grad_norm": 1629.523681640625,
-      "learning_rate": 3.168e-05,
-      "loss": 84.4823,
-      "step": 15840
-    },
-    {
-      "epoch": 0.06403600560769564,
-      "grad_norm": 1089.279296875,
-      "learning_rate": 3.1700000000000005e-05,
-      "loss": 130.4876,
-      "step": 15850
-    },
-    {
-      "epoch": 0.06407640687306326,
-      "grad_norm": 1170.0994873046875,
-      "learning_rate": 3.172e-05,
-      "loss": 144.4939,
-      "step": 15860
-    },
-    {
-      "epoch": 0.06411680813843089,
-      "grad_norm": 1094.1689453125,
-      "learning_rate": 3.1740000000000004e-05,
-      "loss": 124.1748,
-      "step": 15870
-    },
-    {
-      "epoch": 0.06415720940379853,
-      "grad_norm": 742.95703125,
-      "learning_rate": 3.176e-05,
-      "loss": 85.3714,
-      "step": 15880
-    },
-    {
-      "epoch": 0.06419761066916616,
-      "grad_norm": 1239.1507568359375,
-      "learning_rate": 3.1780000000000004e-05,
-      "loss": 215.5378,
-      "step": 15890
-    },
-    {
-      "epoch": 0.0642380119345338,
-      "grad_norm": 734.71484375,
-      "learning_rate": 3.18e-05,
-      "loss": 109.2592,
-      "step": 15900
-    },
-    {
-      "epoch": 0.06427841319990142,
-      "grad_norm": 1894.3482666015625,
-      "learning_rate": 3.182e-05,
-      "loss": 115.1967,
-      "step": 15910
-    },
-    {
-      "epoch": 0.06431881446526905,
-      "grad_norm": 626.6444091796875,
-      "learning_rate": 3.184e-05,
-      "loss": 85.9778,
-      "step": 15920
-    },
-    {
-      "epoch": 0.06435921573063669,
-      "grad_norm": 1511.507568359375,
-      "learning_rate": 3.186e-05,
-      "loss": 182.5304,
-      "step": 15930
-    },
-    {
-      "epoch": 0.06439961699600431,
-      "grad_norm": 693.2582397460938,
-      "learning_rate": 3.188e-05,
-      "loss": 90.9457,
-      "step": 15940
-    },
-    {
-      "epoch": 0.06444001826137194,
-      "grad_norm": 683.79150390625,
-      "learning_rate": 3.19e-05,
-      "loss": 98.913,
-      "step": 15950
-    },
-    {
-      "epoch": 0.06448041952673958,
-      "grad_norm": 1719.1845703125,
-      "learning_rate": 3.192e-05,
-      "loss": 129.3339,
-      "step": 15960
-    },
-    {
-      "epoch": 0.0645208207921072,
-      "grad_norm": 1166.0206298828125,
-      "learning_rate": 3.194e-05,
-      "loss": 153.576,
-      "step": 15970
-    },
-    {
-      "epoch": 0.06456122205747485,
-      "grad_norm": 1580.4215087890625,
-      "learning_rate": 3.196e-05,
-      "loss": 170.2588,
-      "step": 15980
-    },
-    {
-      "epoch": 0.06460162332284247,
-      "grad_norm": 984.2314453125,
-      "learning_rate": 3.198e-05,
-      "loss": 156.6156,
-      "step": 15990
-    },
-    {
-      "epoch": 0.0646420245882101,
-      "grad_norm": 1595.7958984375,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 174.3226,
-      "step": 16000
-    },
-    {
-      "epoch": 0.06468242585357774,
-      "grad_norm": 12306.052734375,
-      "learning_rate": 3.202e-05,
-      "loss": 203.855,
-      "step": 16010
-    },
-    {
-      "epoch": 0.06472282711894536,
-      "grad_norm": 2370.59765625,
-      "learning_rate": 3.2040000000000005e-05,
-      "loss": 144.5134,
-      "step": 16020
-    },
-    {
-      "epoch": 0.06476322838431299,
-      "grad_norm": 1784.508056640625,
-      "learning_rate": 3.206e-05,
-      "loss": 159.992,
-      "step": 16030
-    },
-    {
-      "epoch": 0.06480362964968063,
-      "grad_norm": 894.4352416992188,
-      "learning_rate": 3.208e-05,
-      "loss": 162.522,
-      "step": 16040
-    },
-    {
-      "epoch": 0.06484403091504826,
-      "grad_norm": 1397.84814453125,
-      "learning_rate": 3.21e-05,
-      "loss": 113.1159,
-      "step": 16050
-    },
-    {
-      "epoch": 0.0648844321804159,
-      "grad_norm": 2406.892578125,
-      "learning_rate": 3.212e-05,
-      "loss": 104.4846,
-      "step": 16060
-    },
-    {
-      "epoch": 0.06492483344578352,
-      "grad_norm": 1434.818115234375,
-      "learning_rate": 3.214e-05,
-      "loss": 142.6461,
-      "step": 16070
-    },
-    {
-      "epoch": 0.06496523471115115,
-      "grad_norm": 3077.646240234375,
-      "learning_rate": 3.2160000000000004e-05,
-      "loss": 101.5211,
-      "step": 16080
-    },
-    {
-      "epoch": 0.06500563597651879,
-      "grad_norm": 1358.7418212890625,
-      "learning_rate": 3.218e-05,
-      "loss": 129.6278,
-      "step": 16090
-    },
-    {
-      "epoch": 0.06504603724188641,
-      "grad_norm": 11825.310546875,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 145.8424,
-      "step": 16100
-    },
-    {
-      "epoch": 0.06508643850725404,
-      "grad_norm": 1621.6112060546875,
-      "learning_rate": 3.222e-05,
-      "loss": 117.4145,
-      "step": 16110
-    },
-    {
-      "epoch": 0.06512683977262168,
-      "grad_norm": 1896.356689453125,
-      "learning_rate": 3.224e-05,
-      "loss": 103.7541,
-      "step": 16120
-    },
-    {
-      "epoch": 0.0651672410379893,
-      "grad_norm": 8822.8671875,
-      "learning_rate": 3.226e-05,
-      "loss": 174.6907,
-      "step": 16130
-    },
-    {
-      "epoch": 0.06520764230335695,
-      "grad_norm": 1795.896728515625,
-      "learning_rate": 3.2279999999999996e-05,
-      "loss": 187.9359,
-      "step": 16140
-    },
-    {
-      "epoch": 0.06524804356872457,
-      "grad_norm": 870.3344116210938,
-      "learning_rate": 3.2300000000000006e-05,
-      "loss": 113.2782,
-      "step": 16150
-    },
-    {
-      "epoch": 0.0652884448340922,
-      "grad_norm": 1710.718505859375,
-      "learning_rate": 3.232e-05,
-      "loss": 114.0318,
-      "step": 16160
-    },
-    {
-      "epoch": 0.06532884609945984,
-      "grad_norm": 879.309326171875,
-      "learning_rate": 3.2340000000000005e-05,
-      "loss": 159.0129,
-      "step": 16170
-    },
-    {
-      "epoch": 0.06536924736482747,
-      "grad_norm": 1179.8787841796875,
-      "learning_rate": 3.236e-05,
-      "loss": 108.0792,
-      "step": 16180
-    },
-    {
-      "epoch": 0.06540964863019509,
-      "grad_norm": 1053.493896484375,
-      "learning_rate": 3.238e-05,
-      "loss": 162.0999,
-      "step": 16190
-    },
-    {
-      "epoch": 0.06545004989556273,
-      "grad_norm": 1169.240478515625,
-      "learning_rate": 3.24e-05,
-      "loss": 117.8543,
-      "step": 16200
-    },
-    {
-      "epoch": 0.06549045116093036,
-      "grad_norm": 1077.4576416015625,
-      "learning_rate": 3.242e-05,
-      "loss": 161.1472,
-      "step": 16210
-    },
-    {
-      "epoch": 0.065530852426298,
-      "grad_norm": 827.2643432617188,
-      "learning_rate": 3.244e-05,
-      "loss": 122.5282,
-      "step": 16220
-    },
-    {
-      "epoch": 0.06557125369166562,
-      "grad_norm": 1683.487060546875,
-      "learning_rate": 3.2460000000000004e-05,
-      "loss": 163.201,
-      "step": 16230
-    },
-    {
-      "epoch": 0.06561165495703325,
-      "grad_norm": 1506.306884765625,
-      "learning_rate": 3.248e-05,
-      "loss": 123.3369,
-      "step": 16240
-    },
-    {
-      "epoch": 0.06565205622240089,
-      "grad_norm": 410.9195861816406,
-      "learning_rate": 3.2500000000000004e-05,
-      "loss": 102.7201,
-      "step": 16250
-    },
-    {
-      "epoch": 0.06569245748776852,
-      "grad_norm": 484.0295104980469,
-      "learning_rate": 3.252e-05,
-      "loss": 191.6438,
-      "step": 16260
-    },
-    {
-      "epoch": 0.06573285875313614,
-      "grad_norm": 649.029541015625,
-      "learning_rate": 3.2540000000000004e-05,
-      "loss": 139.1572,
-      "step": 16270
-    },
-    {
-      "epoch": 0.06577326001850378,
-      "grad_norm": 2727.947509765625,
-      "learning_rate": 3.256e-05,
-      "loss": 198.9751,
-      "step": 16280
-    },
-    {
-      "epoch": 0.06581366128387141,
-      "grad_norm": 879.7364501953125,
-      "learning_rate": 3.2579999999999996e-05,
-      "loss": 119.9532,
-      "step": 16290
-    },
-    {
-      "epoch": 0.06585406254923905,
-      "grad_norm": 2567.528564453125,
-      "learning_rate": 3.26e-05,
-      "loss": 182.2825,
-      "step": 16300
-    },
-    {
-      "epoch": 0.06589446381460667,
-      "grad_norm": 1746.7139892578125,
-      "learning_rate": 3.262e-05,
-      "loss": 109.7434,
-      "step": 16310
-    },
-    {
-      "epoch": 0.0659348650799743,
-      "grad_norm": 1511.91845703125,
-      "learning_rate": 3.2640000000000006e-05,
-      "loss": 117.7299,
-      "step": 16320
-    },
-    {
-      "epoch": 0.06597526634534194,
-      "grad_norm": 626.0679321289062,
-      "learning_rate": 3.266e-05,
-      "loss": 154.5593,
-      "step": 16330
-    },
-    {
-      "epoch": 0.06601566761070957,
-      "grad_norm": 1950.873779296875,
-      "learning_rate": 3.268e-05,
-      "loss": 134.659,
-      "step": 16340
-    },
-    {
-      "epoch": 0.06605606887607719,
-      "grad_norm": 2815.42919921875,
-      "learning_rate": 3.27e-05,
-      "loss": 189.5417,
-      "step": 16350
-    },
-    {
-      "epoch": 0.06609647014144483,
-      "grad_norm": 791.9197387695312,
-      "learning_rate": 3.272e-05,
-      "loss": 104.6571,
-      "step": 16360
-    },
-    {
-      "epoch": 0.06613687140681246,
-      "grad_norm": 961.1111450195312,
-      "learning_rate": 3.274e-05,
-      "loss": 110.3847,
-      "step": 16370
-    },
-    {
-      "epoch": 0.0661772726721801,
-      "grad_norm": 1224.700927734375,
-      "learning_rate": 3.2760000000000005e-05,
-      "loss": 129.7902,
-      "step": 16380
-    },
-    {
-      "epoch": 0.06621767393754772,
-      "grad_norm": 892.2734985351562,
-      "learning_rate": 3.278e-05,
-      "loss": 131.0276,
-      "step": 16390
-    },
-    {
-      "epoch": 0.06625807520291535,
-      "grad_norm": 2211.9521484375,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 102.1123,
-      "step": 16400
-    },
-    {
-      "epoch": 0.06629847646828299,
-      "grad_norm": 2223.035888671875,
-      "learning_rate": 3.282e-05,
-      "loss": 130.2351,
-      "step": 16410
-    },
-    {
-      "epoch": 0.06633887773365062,
-      "grad_norm": 896.9553833007812,
-      "learning_rate": 3.2840000000000004e-05,
-      "loss": 113.7288,
-      "step": 16420
-    },
-    {
-      "epoch": 0.06637927899901824,
-      "grad_norm": 0.0,
-      "learning_rate": 3.286e-05,
-      "loss": 70.063,
-      "step": 16430
-    },
-    {
-      "epoch": 0.06641968026438588,
-      "grad_norm": 665.3837890625,
-      "learning_rate": 3.288e-05,
-      "loss": 102.8864,
-      "step": 16440
-    },
-    {
-      "epoch": 0.06646008152975351,
-      "grad_norm": 529.5450439453125,
-      "learning_rate": 3.29e-05,
-      "loss": 107.0171,
-      "step": 16450
-    },
-    {
-      "epoch": 0.06650048279512115,
-      "grad_norm": 445.30950927734375,
-      "learning_rate": 3.292e-05,
-      "loss": 118.4387,
-      "step": 16460
-    },
-    {
-      "epoch": 0.06654088406048878,
-      "grad_norm": 1200.6650390625,
-      "learning_rate": 3.2940000000000006e-05,
-      "loss": 115.6011,
-      "step": 16470
-    },
-    {
-      "epoch": 0.0665812853258564,
-      "grad_norm": 997.8995971679688,
-      "learning_rate": 3.296e-05,
-      "loss": 101.1121,
-      "step": 16480
-    },
-    {
-      "epoch": 0.06662168659122404,
-      "grad_norm": 1181.9188232421875,
-      "learning_rate": 3.298e-05,
-      "loss": 94.8608,
-      "step": 16490
-    },
-    {
-      "epoch": 0.06666208785659167,
-      "grad_norm": 816.2405395507812,
-      "learning_rate": 3.3e-05,
-      "loss": 125.1284,
-      "step": 16500
-    },
-    {
-      "epoch": 0.0667024891219593,
-      "grad_norm": 703.4663696289062,
-      "learning_rate": 3.302e-05,
-      "loss": 123.1175,
-      "step": 16510
-    },
-    {
-      "epoch": 0.06674289038732693,
-      "grad_norm": 671.1204833984375,
-      "learning_rate": 3.304e-05,
-      "loss": 133.0177,
-      "step": 16520
-    },
-    {
-      "epoch": 0.06678329165269456,
-      "grad_norm": 635.96728515625,
-      "learning_rate": 3.3060000000000005e-05,
-      "loss": 117.4494,
-      "step": 16530
-    },
-    {
-      "epoch": 0.0668236929180622,
-      "grad_norm": 0.0,
-      "learning_rate": 3.308e-05,
-      "loss": 84.0806,
-      "step": 16540
-    },
-    {
-      "epoch": 0.06686409418342983,
-      "grad_norm": 1110.686279296875,
-      "learning_rate": 3.3100000000000005e-05,
-      "loss": 149.0371,
-      "step": 16550
-    },
-    {
-      "epoch": 0.06690449544879745,
-      "grad_norm": 1048.002197265625,
-      "learning_rate": 3.312e-05,
-      "loss": 118.8765,
-      "step": 16560
-    },
-    {
-      "epoch": 0.06694489671416509,
-      "grad_norm": 532.6653442382812,
-      "learning_rate": 3.314e-05,
-      "loss": 115.488,
-      "step": 16570
-    },
-    {
-      "epoch": 0.06698529797953272,
-      "grad_norm": 1357.1689453125,
-      "learning_rate": 3.316e-05,
-      "loss": 121.7834,
-      "step": 16580
-    },
-    {
-      "epoch": 0.06702569924490034,
-      "grad_norm": 3220.8486328125,
-      "learning_rate": 3.318e-05,
-      "loss": 146.6197,
-      "step": 16590
-    },
-    {
-      "epoch": 0.06706610051026798,
-      "grad_norm": 809.4137573242188,
-      "learning_rate": 3.32e-05,
-      "loss": 157.1161,
-      "step": 16600
-    },
-    {
-      "epoch": 0.06710650177563561,
-      "grad_norm": 479.6476135253906,
-      "learning_rate": 3.3220000000000004e-05,
-      "loss": 123.665,
-      "step": 16610
-    },
-    {
-      "epoch": 0.06714690304100325,
-      "grad_norm": 1147.013916015625,
-      "learning_rate": 3.324e-05,
-      "loss": 151.3696,
-      "step": 16620
-    },
-    {
-      "epoch": 0.06718730430637088,
-      "grad_norm": 1325.5657958984375,
-      "learning_rate": 3.3260000000000003e-05,
-      "loss": 228.0247,
-      "step": 16630
-    },
-    {
-      "epoch": 0.0672277055717385,
-      "grad_norm": 1421.17578125,
-      "learning_rate": 3.328e-05,
-      "loss": 174.1318,
-      "step": 16640
-    },
-    {
-      "epoch": 0.06726810683710614,
-      "grad_norm": 872.6542358398438,
-      "learning_rate": 3.33e-05,
-      "loss": 118.2037,
-      "step": 16650
-    },
-    {
-      "epoch": 0.06730850810247377,
-      "grad_norm": 936.040283203125,
-      "learning_rate": 3.332e-05,
-      "loss": 113.6235,
-      "step": 16660
-    },
-    {
-      "epoch": 0.0673489093678414,
-      "grad_norm": 829.9392700195312,
-      "learning_rate": 3.3339999999999996e-05,
-      "loss": 124.7012,
-      "step": 16670
-    },
-    {
-      "epoch": 0.06738931063320903,
-      "grad_norm": 829.8602294921875,
-      "learning_rate": 3.336e-05,
-      "loss": 122.1225,
-      "step": 16680
-    },
-    {
-      "epoch": 0.06742971189857666,
-      "grad_norm": 591.7499389648438,
-      "learning_rate": 3.338e-05,
-      "loss": 117.5648,
-      "step": 16690
-    },
-    {
-      "epoch": 0.0674701131639443,
-      "grad_norm": 585.739013671875,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 120.4557,
-      "step": 16700
-    },
-    {
-      "epoch": 0.06751051442931193,
-      "grad_norm": 1151.9459228515625,
-      "learning_rate": 3.342e-05,
-      "loss": 122.6646,
-      "step": 16710
-    },
-    {
-      "epoch": 0.06755091569467955,
-      "grad_norm": 1100.1302490234375,
-      "learning_rate": 3.344e-05,
-      "loss": 124.8112,
-      "step": 16720
-    },
-    {
-      "epoch": 0.06759131696004719,
-      "grad_norm": 1049.39501953125,
-      "learning_rate": 3.346e-05,
-      "loss": 131.1703,
-      "step": 16730
-    },
-    {
-      "epoch": 0.06763171822541482,
-      "grad_norm": 976.2811889648438,
-      "learning_rate": 3.348e-05,
-      "loss": 104.992,
-      "step": 16740
-    },
-    {
-      "epoch": 0.06767211949078245,
-      "grad_norm": 1059.683837890625,
-      "learning_rate": 3.35e-05,
-      "loss": 109.1284,
-      "step": 16750
-    },
-    {
-      "epoch": 0.06771252075615009,
-      "grad_norm": 1005.8837280273438,
-      "learning_rate": 3.3520000000000004e-05,
-      "loss": 107.3349,
-      "step": 16760
-    },
-    {
-      "epoch": 0.06775292202151771,
-      "grad_norm": 850.0352783203125,
-      "learning_rate": 3.354e-05,
-      "loss": 126.588,
-      "step": 16770
-    },
-    {
-      "epoch": 0.06779332328688535,
-      "grad_norm": 1083.05712890625,
-      "learning_rate": 3.3560000000000004e-05,
-      "loss": 139.6427,
-      "step": 16780
-    },
-    {
-      "epoch": 0.06783372455225298,
-      "grad_norm": 1554.07080078125,
-      "learning_rate": 3.358e-05,
-      "loss": 145.7689,
-      "step": 16790
-    },
-    {
-      "epoch": 0.0678741258176206,
-      "grad_norm": 3450.517333984375,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 140.9184,
-      "step": 16800
-    },
-    {
-      "epoch": 0.06791452708298824,
-      "grad_norm": 1706.325439453125,
-      "learning_rate": 3.362e-05,
-      "loss": 199.2792,
-      "step": 16810
-    },
-    {
-      "epoch": 0.06795492834835587,
-      "grad_norm": 1185.1922607421875,
-      "learning_rate": 3.3639999999999996e-05,
-      "loss": 140.3882,
-      "step": 16820
-    },
-    {
-      "epoch": 0.0679953296137235,
-      "grad_norm": 1169.992919921875,
-      "learning_rate": 3.366e-05,
-      "loss": 123.5538,
-      "step": 16830
-    },
-    {
-      "epoch": 0.06803573087909114,
-      "grad_norm": 1750.4180908203125,
-      "learning_rate": 3.368e-05,
-      "loss": 159.0448,
-      "step": 16840
-    },
-    {
-      "epoch": 0.06807613214445876,
-      "grad_norm": 1771.9908447265625,
-      "learning_rate": 3.3700000000000006e-05,
-      "loss": 153.4333,
-      "step": 16850
-    },
-    {
-      "epoch": 0.0681165334098264,
-      "grad_norm": 2022.061279296875,
-      "learning_rate": 3.372e-05,
-      "loss": 139.1972,
-      "step": 16860
-    },
-    {
-      "epoch": 0.06815693467519403,
-      "grad_norm": 1393.0201416015625,
-      "learning_rate": 3.374e-05,
-      "loss": 152.1366,
-      "step": 16870
-    },
-    {
-      "epoch": 0.06819733594056165,
-      "grad_norm": 2371.81494140625,
-      "learning_rate": 3.376e-05,
-      "loss": 142.447,
-      "step": 16880
-    },
-    {
-      "epoch": 0.0682377372059293,
-      "grad_norm": 1049.164794921875,
-      "learning_rate": 3.378e-05,
-      "loss": 115.6586,
-      "step": 16890
-    },
-    {
-      "epoch": 0.06827813847129692,
-      "grad_norm": 1043.5615234375,
-      "learning_rate": 3.38e-05,
-      "loss": 174.1537,
-      "step": 16900
-    },
-    {
-      "epoch": 0.06831853973666455,
-      "grad_norm": 1559.866455078125,
-      "learning_rate": 3.3820000000000005e-05,
-      "loss": 108.5037,
-      "step": 16910
-    },
-    {
-      "epoch": 0.06835894100203219,
-      "grad_norm": 2484.740478515625,
-      "learning_rate": 3.384e-05,
-      "loss": 184.837,
-      "step": 16920
-    },
-    {
-      "epoch": 0.06839934226739981,
-      "grad_norm": 775.7838745117188,
-      "learning_rate": 3.3860000000000004e-05,
-      "loss": 126.9209,
-      "step": 16930
-    },
-    {
-      "epoch": 0.06843974353276745,
-      "grad_norm": 471.7488708496094,
-      "learning_rate": 3.388e-05,
-      "loss": 135.7176,
-      "step": 16940
-    },
-    {
-      "epoch": 0.06848014479813508,
-      "grad_norm": 969.9213256835938,
-      "learning_rate": 3.3900000000000004e-05,
-      "loss": 95.2846,
-      "step": 16950
-    },
-    {
-      "epoch": 0.0685205460635027,
-      "grad_norm": 1296.1826171875,
-      "learning_rate": 3.392e-05,
-      "loss": 159.3641,
-      "step": 16960
-    },
-    {
-      "epoch": 0.06856094732887034,
-      "grad_norm": 2220.822021484375,
-      "learning_rate": 3.394e-05,
-      "loss": 171.076,
-      "step": 16970
-    },
-    {
-      "epoch": 0.06860134859423797,
-      "grad_norm": 764.9531860351562,
-      "learning_rate": 3.396e-05,
-      "loss": 139.6084,
-      "step": 16980
-    },
-    {
-      "epoch": 0.0686417498596056,
-      "grad_norm": 2453.2900390625,
-      "learning_rate": 3.398e-05,
-      "loss": 162.632,
-      "step": 16990
-    },
-    {
-      "epoch": 0.06868215112497324,
-      "grad_norm": 1016.1632690429688,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 236.2789,
-      "step": 17000
-    },
-    {
-      "epoch": 0.06872255239034086,
-      "grad_norm": 1268.531494140625,
-      "learning_rate": 3.402e-05,
-      "loss": 119.6548,
-      "step": 17010
-    },
-    {
-      "epoch": 0.0687629536557085,
-      "grad_norm": 0.0,
-      "learning_rate": 3.404e-05,
-      "loss": 115.3437,
-      "step": 17020
-    },
-    {
-      "epoch": 0.06880335492107613,
-      "grad_norm": 867.4444580078125,
-      "learning_rate": 3.406e-05,
-      "loss": 173.6102,
-      "step": 17030
-    },
-    {
-      "epoch": 0.06884375618644376,
-      "grad_norm": 1306.5975341796875,
-      "learning_rate": 3.408e-05,
-      "loss": 147.4384,
-      "step": 17040
-    },
-    {
-      "epoch": 0.0688841574518114,
-      "grad_norm": 701.0490112304688,
-      "learning_rate": 3.41e-05,
-      "loss": 138.4343,
-      "step": 17050
-    },
-    {
-      "epoch": 0.06892455871717902,
-      "grad_norm": 1092.9249267578125,
-      "learning_rate": 3.412e-05,
-      "loss": 112.9486,
-      "step": 17060
-    },
-    {
-      "epoch": 0.06896495998254665,
-      "grad_norm": 1403.239990234375,
-      "learning_rate": 3.414e-05,
-      "loss": 106.8123,
-      "step": 17070
-    },
-    {
-      "epoch": 0.06900536124791429,
-      "grad_norm": 2368.7109375,
-      "learning_rate": 3.4160000000000005e-05,
-      "loss": 90.9559,
-      "step": 17080
-    },
-    {
-      "epoch": 0.06904576251328191,
-      "grad_norm": 723.0379028320312,
-      "learning_rate": 3.418e-05,
-      "loss": 92.075,
-      "step": 17090
-    },
-    {
-      "epoch": 0.06908616377864955,
-      "grad_norm": 1454.345703125,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 122.138,
-      "step": 17100
-    },
-    {
-      "epoch": 0.06912656504401718,
-      "grad_norm": 849.4160766601562,
-      "learning_rate": 3.422e-05,
-      "loss": 132.3787,
-      "step": 17110
-    },
-    {
-      "epoch": 0.0691669663093848,
-      "grad_norm": 1275.969482421875,
-      "learning_rate": 3.424e-05,
-      "loss": 80.5995,
-      "step": 17120
-    },
-    {
-      "epoch": 0.06920736757475245,
-      "grad_norm": 977.9779052734375,
-      "learning_rate": 3.426e-05,
-      "loss": 126.0984,
-      "step": 17130
-    },
-    {
-      "epoch": 0.06924776884012007,
-      "grad_norm": 1691.71826171875,
-      "learning_rate": 3.4280000000000004e-05,
-      "loss": 123.841,
-      "step": 17140
-    },
-    {
-      "epoch": 0.0692881701054877,
-      "grad_norm": 3872.140869140625,
-      "learning_rate": 3.430000000000001e-05,
-      "loss": 148.4945,
-      "step": 17150
-    },
-    {
-      "epoch": 0.06932857137085534,
-      "grad_norm": 3100.976318359375,
-      "learning_rate": 3.4320000000000003e-05,
-      "loss": 117.4309,
-      "step": 17160
-    },
-    {
-      "epoch": 0.06936897263622296,
-      "grad_norm": 1637.591552734375,
-      "learning_rate": 3.434e-05,
-      "loss": 119.0649,
-      "step": 17170
-    },
-    {
-      "epoch": 0.0694093739015906,
-      "grad_norm": 8878.0693359375,
-      "learning_rate": 3.436e-05,
-      "loss": 140.0768,
-      "step": 17180
-    },
-    {
-      "epoch": 0.06944977516695823,
-      "grad_norm": 959.3743286132812,
-      "learning_rate": 3.438e-05,
-      "loss": 172.569,
-      "step": 17190
-    },
-    {
-      "epoch": 0.06949017643232586,
-      "grad_norm": 520.9686889648438,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 93.1335,
-      "step": 17200
-    },
-    {
-      "epoch": 0.0695305776976935,
-      "grad_norm": 1169.4296875,
-      "learning_rate": 3.442e-05,
-      "loss": 109.4803,
-      "step": 17210
-    },
-    {
-      "epoch": 0.06957097896306112,
-      "grad_norm": 2471.82568359375,
-      "learning_rate": 3.444e-05,
-      "loss": 149.8675,
-      "step": 17220
-    },
-    {
-      "epoch": 0.06961138022842875,
-      "grad_norm": 1193.904541015625,
-      "learning_rate": 3.4460000000000005e-05,
-      "loss": 137.5065,
-      "step": 17230
-    },
-    {
-      "epoch": 0.06965178149379639,
-      "grad_norm": 796.9132080078125,
-      "learning_rate": 3.448e-05,
-      "loss": 89.8475,
-      "step": 17240
-    },
-    {
-      "epoch": 0.06969218275916401,
-      "grad_norm": 688.8414916992188,
-      "learning_rate": 3.45e-05,
-      "loss": 125.9576,
-      "step": 17250
-    },
-    {
-      "epoch": 0.06973258402453165,
-      "grad_norm": 480.16009521484375,
-      "learning_rate": 3.452e-05,
-      "loss": 78.7464,
-      "step": 17260
-    },
-    {
-      "epoch": 0.06977298528989928,
-      "grad_norm": 783.6867065429688,
-      "learning_rate": 3.454e-05,
-      "loss": 108.8815,
-      "step": 17270
-    },
-    {
-      "epoch": 0.06981338655526691,
-      "grad_norm": 0.0,
-      "learning_rate": 3.456e-05,
-      "loss": 116.7318,
-      "step": 17280
-    },
-    {
-      "epoch": 0.06985378782063455,
-      "grad_norm": 972.9898681640625,
-      "learning_rate": 3.4580000000000004e-05,
-      "loss": 163.7325,
-      "step": 17290
-    },
-    {
-      "epoch": 0.06989418908600217,
-      "grad_norm": 2305.32470703125,
-      "learning_rate": 3.46e-05,
-      "loss": 102.6796,
-      "step": 17300
-    },
-    {
-      "epoch": 0.0699345903513698,
-      "grad_norm": 985.5175170898438,
-      "learning_rate": 3.4620000000000004e-05,
-      "loss": 133.6453,
-      "step": 17310
-    },
-    {
-      "epoch": 0.06997499161673744,
-      "grad_norm": 966.200439453125,
-      "learning_rate": 3.464e-05,
-      "loss": 133.9981,
-      "step": 17320
-    },
-    {
-      "epoch": 0.07001539288210507,
-      "grad_norm": 582.17041015625,
-      "learning_rate": 3.4660000000000004e-05,
-      "loss": 119.5202,
-      "step": 17330
-    },
-    {
-      "epoch": 0.0700557941474727,
-      "grad_norm": 841.1046752929688,
-      "learning_rate": 3.468e-05,
-      "loss": 133.1583,
-      "step": 17340
-    },
-    {
-      "epoch": 0.07009619541284033,
-      "grad_norm": 675.8413696289062,
-      "learning_rate": 3.4699999999999996e-05,
-      "loss": 139.8855,
-      "step": 17350
-    },
-    {
-      "epoch": 0.07013659667820796,
-      "grad_norm": 957.8898315429688,
-      "learning_rate": 3.472e-05,
-      "loss": 137.3781,
-      "step": 17360
-    },
-    {
-      "epoch": 0.0701769979435756,
-      "grad_norm": 3106.9208984375,
-      "learning_rate": 3.474e-05,
-      "loss": 167.9274,
-      "step": 17370
-    },
-    {
-      "epoch": 0.07021739920894322,
-      "grad_norm": 1493.3551025390625,
-      "learning_rate": 3.4760000000000006e-05,
-      "loss": 163.2431,
-      "step": 17380
-    },
-    {
-      "epoch": 0.07025780047431085,
-      "grad_norm": 993.3938598632812,
-      "learning_rate": 3.478e-05,
-      "loss": 114.7312,
-      "step": 17390
-    },
-    {
-      "epoch": 0.07029820173967849,
-      "grad_norm": 1034.116943359375,
-      "learning_rate": 3.48e-05,
-      "loss": 70.717,
-      "step": 17400
-    },
-    {
-      "epoch": 0.07033860300504612,
-      "grad_norm": 921.6234741210938,
-      "learning_rate": 3.482e-05,
-      "loss": 185.619,
-      "step": 17410
-    },
-    {
-      "epoch": 0.07037900427041376,
-      "grad_norm": 1703.0343017578125,
-      "learning_rate": 3.484e-05,
-      "loss": 142.7865,
-      "step": 17420
-    },
-    {
-      "epoch": 0.07041940553578138,
-      "grad_norm": 1368.336181640625,
-      "learning_rate": 3.486e-05,
-      "loss": 116.2519,
-      "step": 17430
-    },
-    {
-      "epoch": 0.07045980680114901,
-      "grad_norm": 1559.697265625,
-      "learning_rate": 3.4880000000000005e-05,
-      "loss": 110.2219,
-      "step": 17440
-    },
-    {
-      "epoch": 0.07050020806651665,
-      "grad_norm": 3143.30517578125,
-      "learning_rate": 3.49e-05,
-      "loss": 91.7808,
-      "step": 17450
-    },
-    {
-      "epoch": 0.07054060933188427,
-      "grad_norm": 1277.235595703125,
-      "learning_rate": 3.4920000000000004e-05,
-      "loss": 143.4617,
-      "step": 17460
-    },
-    {
-      "epoch": 0.0705810105972519,
-      "grad_norm": 2107.630615234375,
-      "learning_rate": 3.494e-05,
-      "loss": 141.62,
-      "step": 17470
-    },
-    {
-      "epoch": 0.07062141186261954,
-      "grad_norm": 916.9269409179688,
-      "learning_rate": 3.4960000000000004e-05,
-      "loss": 124.4125,
-      "step": 17480
-    },
-    {
-      "epoch": 0.07066181312798717,
-      "grad_norm": 495.5371398925781,
-      "learning_rate": 3.498e-05,
-      "loss": 113.1787,
-      "step": 17490
-    },
-    {
-      "epoch": 0.0707022143933548,
-      "grad_norm": 722.7784423828125,
-      "learning_rate": 3.5e-05,
-      "loss": 113.6147,
-      "step": 17500
-    },
-    {
-      "epoch": 0.07074261565872243,
-      "grad_norm": 8005.68017578125,
-      "learning_rate": 3.502e-05,
-      "loss": 160.4929,
-      "step": 17510
-    },
-    {
-      "epoch": 0.07078301692409006,
-      "grad_norm": 836.4596557617188,
-      "learning_rate": 3.504e-05,
-      "loss": 93.0578,
-      "step": 17520
-    },
-    {
-      "epoch": 0.0708234181894577,
-      "grad_norm": 1631.076171875,
-      "learning_rate": 3.5060000000000007e-05,
-      "loss": 126.6171,
-      "step": 17530
-    },
-    {
-      "epoch": 0.07086381945482532,
-      "grad_norm": 639.418212890625,
-      "learning_rate": 3.508e-05,
-      "loss": 105.7042,
-      "step": 17540
-    },
-    {
-      "epoch": 0.07090422072019295,
-      "grad_norm": 1241.2578125,
-      "learning_rate": 3.51e-05,
-      "loss": 201.5257,
-      "step": 17550
-    },
-    {
-      "epoch": 0.07094462198556059,
-      "grad_norm": 835.3104248046875,
-      "learning_rate": 3.512e-05,
-      "loss": 159.556,
-      "step": 17560
-    },
-    {
-      "epoch": 0.07098502325092822,
-      "grad_norm": 565.4751586914062,
-      "learning_rate": 3.514e-05,
-      "loss": 93.872,
-      "step": 17570
-    },
-    {
-      "epoch": 0.07102542451629586,
-      "grad_norm": 599.105712890625,
-      "learning_rate": 3.516e-05,
-      "loss": 132.7931,
-      "step": 17580
-    },
-    {
-      "epoch": 0.07106582578166348,
-      "grad_norm": 1342.3175048828125,
-      "learning_rate": 3.518e-05,
-      "loss": 121.4404,
-      "step": 17590
-    },
-    {
-      "epoch": 0.07110622704703111,
-      "grad_norm": 853.9281616210938,
-      "learning_rate": 3.52e-05,
-      "loss": 185.0158,
-      "step": 17600
-    },
-    {
-      "epoch": 0.07114662831239875,
-      "grad_norm": 1526.38427734375,
-      "learning_rate": 3.5220000000000005e-05,
-      "loss": 130.5014,
-      "step": 17610
-    },
-    {
-      "epoch": 0.07118702957776638,
-      "grad_norm": 0.0,
-      "learning_rate": 3.524e-05,
-      "loss": 155.2073,
-      "step": 17620
-    },
-    {
-      "epoch": 0.071227430843134,
-      "grad_norm": 722.3214111328125,
-      "learning_rate": 3.5260000000000005e-05,
-      "loss": 98.7197,
-      "step": 17630
-    },
-    {
-      "epoch": 0.07126783210850164,
-      "grad_norm": 769.6561889648438,
-      "learning_rate": 3.528e-05,
-      "loss": 98.4344,
-      "step": 17640
-    },
-    {
-      "epoch": 0.07130823337386927,
-      "grad_norm": 1916.5062255859375,
-      "learning_rate": 3.53e-05,
-      "loss": 169.392,
-      "step": 17650
-    },
-    {
-      "epoch": 0.07134863463923691,
-      "grad_norm": 790.8156127929688,
-      "learning_rate": 3.532e-05,
-      "loss": 112.2699,
-      "step": 17660
-    },
-    {
-      "epoch": 0.07138903590460453,
-      "grad_norm": 0.0,
-      "learning_rate": 3.5340000000000004e-05,
-      "loss": 153.1781,
-      "step": 17670
-    },
-    {
-      "epoch": 0.07142943716997216,
-      "grad_norm": 1370.72216796875,
-      "learning_rate": 3.536000000000001e-05,
-      "loss": 170.2639,
-      "step": 17680
-    },
-    {
-      "epoch": 0.0714698384353398,
-      "grad_norm": 466.6986999511719,
-      "learning_rate": 3.5380000000000003e-05,
-      "loss": 140.3446,
-      "step": 17690
-    },
-    {
-      "epoch": 0.07151023970070743,
-      "grad_norm": 1512.586181640625,
-      "learning_rate": 3.54e-05,
-      "loss": 123.2379,
-      "step": 17700
-    },
-    {
-      "epoch": 0.07155064096607505,
-      "grad_norm": 1187.423828125,
-      "learning_rate": 3.542e-05,
-      "loss": 121.9325,
-      "step": 17710
-    },
-    {
-      "epoch": 0.07159104223144269,
-      "grad_norm": 1491.8526611328125,
-      "learning_rate": 3.544e-05,
-      "loss": 97.88,
-      "step": 17720
-    },
-    {
-      "epoch": 0.07163144349681032,
-      "grad_norm": 819.60400390625,
-      "learning_rate": 3.546e-05,
-      "loss": 136.5916,
-      "step": 17730
-    },
-    {
-      "epoch": 0.07167184476217796,
-      "grad_norm": 650.4725952148438,
-      "learning_rate": 3.548e-05,
-      "loss": 100.4026,
-      "step": 17740
-    },
-    {
-      "epoch": 0.07171224602754558,
-      "grad_norm": 822.773681640625,
-      "learning_rate": 3.55e-05,
-      "loss": 103.9474,
-      "step": 17750
-    },
-    {
-      "epoch": 0.07175264729291321,
-      "grad_norm": 879.6776123046875,
-      "learning_rate": 3.5520000000000006e-05,
-      "loss": 90.2404,
-      "step": 17760
-    },
-    {
-      "epoch": 0.07179304855828085,
-      "grad_norm": 1487.169921875,
-      "learning_rate": 3.554e-05,
-      "loss": 172.1061,
-      "step": 17770
-    },
-    {
-      "epoch": 0.07183344982364848,
-      "grad_norm": 953.589111328125,
-      "learning_rate": 3.5560000000000005e-05,
-      "loss": 150.0098,
-      "step": 17780
-    },
-    {
-      "epoch": 0.0718738510890161,
-      "grad_norm": 688.1337890625,
-      "learning_rate": 3.558e-05,
-      "loss": 78.3569,
-      "step": 17790
-    },
-    {
-      "epoch": 0.07191425235438374,
-      "grad_norm": 789.90283203125,
-      "learning_rate": 3.56e-05,
-      "loss": 106.5888,
-      "step": 17800
-    },
-    {
-      "epoch": 0.07195465361975137,
-      "grad_norm": 934.8272094726562,
-      "learning_rate": 3.562e-05,
-      "loss": 111.0842,
-      "step": 17810
-    },
-    {
-      "epoch": 0.07199505488511901,
-      "grad_norm": 336.4629211425781,
-      "learning_rate": 3.5640000000000004e-05,
-      "loss": 99.3697,
-      "step": 17820
-    },
-    {
-      "epoch": 0.07203545615048663,
-      "grad_norm": 3090.107666015625,
-      "learning_rate": 3.566e-05,
-      "loss": 97.4285,
-      "step": 17830
-    },
-    {
-      "epoch": 0.07207585741585426,
-      "grad_norm": 1428.7044677734375,
-      "learning_rate": 3.5680000000000004e-05,
-      "loss": 118.1064,
-      "step": 17840
-    },
-    {
-      "epoch": 0.0721162586812219,
-      "grad_norm": 2355.272705078125,
-      "learning_rate": 3.57e-05,
-      "loss": 151.7737,
-      "step": 17850
-    },
-    {
-      "epoch": 0.07215665994658953,
-      "grad_norm": 1052.816650390625,
-      "learning_rate": 3.5720000000000004e-05,
-      "loss": 86.3124,
-      "step": 17860
-    },
-    {
-      "epoch": 0.07219706121195715,
-      "grad_norm": 683.0614624023438,
-      "learning_rate": 3.574e-05,
-      "loss": 120.8411,
-      "step": 17870
-    },
-    {
-      "epoch": 0.0722374624773248,
-      "grad_norm": 1161.9776611328125,
-      "learning_rate": 3.5759999999999996e-05,
-      "loss": 145.4579,
-      "step": 17880
-    },
-    {
-      "epoch": 0.07227786374269242,
-      "grad_norm": 1213.32861328125,
-      "learning_rate": 3.578e-05,
-      "loss": 160.1318,
-      "step": 17890
-    },
-    {
-      "epoch": 0.07231826500806006,
-      "grad_norm": 1079.72119140625,
-      "learning_rate": 3.58e-05,
-      "loss": 125.1568,
-      "step": 17900
-    },
-    {
-      "epoch": 0.07235866627342769,
-      "grad_norm": 960.3634643554688,
-      "learning_rate": 3.5820000000000006e-05,
-      "loss": 116.5516,
-      "step": 17910
-    },
-    {
-      "epoch": 0.07239906753879531,
-      "grad_norm": 1030.158447265625,
-      "learning_rate": 3.584e-05,
-      "loss": 78.2271,
-      "step": 17920
-    },
-    {
-      "epoch": 0.07243946880416295,
-      "grad_norm": 812.413818359375,
-      "learning_rate": 3.586e-05,
-      "loss": 122.26,
-      "step": 17930
-    },
-    {
-      "epoch": 0.07247987006953058,
-      "grad_norm": 2242.89794921875,
-      "learning_rate": 3.588e-05,
-      "loss": 150.4362,
-      "step": 17940
-    },
-    {
-      "epoch": 0.0725202713348982,
-      "grad_norm": 1707.64306640625,
-      "learning_rate": 3.59e-05,
-      "loss": 107.4575,
-      "step": 17950
-    },
-    {
-      "epoch": 0.07256067260026584,
-      "grad_norm": 1091.1968994140625,
-      "learning_rate": 3.592e-05,
-      "loss": 176.9033,
-      "step": 17960
-    },
-    {
-      "epoch": 0.07260107386563347,
-      "grad_norm": 660.7951049804688,
-      "learning_rate": 3.594e-05,
-      "loss": 120.4594,
-      "step": 17970
-    },
-    {
-      "epoch": 0.07264147513100111,
-      "grad_norm": 1081.6932373046875,
-      "learning_rate": 3.596e-05,
-      "loss": 101.9581,
-      "step": 17980
-    },
-    {
-      "epoch": 0.07268187639636874,
-      "grad_norm": 1247.1197509765625,
-      "learning_rate": 3.5980000000000004e-05,
-      "loss": 90.3047,
-      "step": 17990
-    },
-    {
-      "epoch": 0.07272227766173636,
-      "grad_norm": 2352.89208984375,
-      "learning_rate": 3.6e-05,
-      "loss": 121.5598,
-      "step": 18000
-    },
-    {
-      "epoch": 0.072762678927104,
-      "grad_norm": 774.28466796875,
-      "learning_rate": 3.6020000000000004e-05,
-      "loss": 126.1789,
-      "step": 18010
-    },
-    {
-      "epoch": 0.07280308019247163,
-      "grad_norm": 821.19384765625,
-      "learning_rate": 3.604e-05,
-      "loss": 100.4537,
-      "step": 18020
-    },
-    {
-      "epoch": 0.07284348145783925,
-      "grad_norm": 919.6599731445312,
-      "learning_rate": 3.606e-05,
-      "loss": 85.4199,
-      "step": 18030
-    },
-    {
-      "epoch": 0.0728838827232069,
-      "grad_norm": 805.6433715820312,
-      "learning_rate": 3.608e-05,
-      "loss": 76.5743,
-      "step": 18040
-    },
-    {
-      "epoch": 0.07292428398857452,
-      "grad_norm": 1283.3468017578125,
-      "learning_rate": 3.61e-05,
-      "loss": 128.92,
-      "step": 18050
-    },
-    {
-      "epoch": 0.07296468525394216,
-      "grad_norm": 2695.972900390625,
-      "learning_rate": 3.6120000000000007e-05,
-      "loss": 131.2778,
-      "step": 18060
-    },
-    {
-      "epoch": 0.07300508651930979,
-      "grad_norm": 2971.26904296875,
-      "learning_rate": 3.614e-05,
-      "loss": 103.6507,
-      "step": 18070
-    },
-    {
-      "epoch": 0.07304548778467741,
-      "grad_norm": 1631.7698974609375,
-      "learning_rate": 3.616e-05,
-      "loss": 101.8744,
-      "step": 18080
-    },
-    {
-      "epoch": 0.07308588905004505,
-      "grad_norm": 707.8176879882812,
-      "learning_rate": 3.618e-05,
-      "loss": 106.0584,
-      "step": 18090
-    },
-    {
-      "epoch": 0.07312629031541268,
-      "grad_norm": 497.0185546875,
-      "learning_rate": 3.62e-05,
-      "loss": 74.8409,
-      "step": 18100
-    },
-    {
-      "epoch": 0.0731666915807803,
-      "grad_norm": 1669.726806640625,
-      "learning_rate": 3.622e-05,
-      "loss": 105.9762,
-      "step": 18110
-    },
-    {
-      "epoch": 0.07320709284614794,
-      "grad_norm": 557.0554809570312,
-      "learning_rate": 3.624e-05,
-      "loss": 116.4344,
-      "step": 18120
-    },
-    {
-      "epoch": 0.07324749411151557,
-      "grad_norm": 2284.02099609375,
-      "learning_rate": 3.626e-05,
-      "loss": 122.7274,
-      "step": 18130
-    },
-    {
-      "epoch": 0.07328789537688321,
-      "grad_norm": 1483.209228515625,
-      "learning_rate": 3.6280000000000005e-05,
-      "loss": 100.2296,
-      "step": 18140
-    },
-    {
-      "epoch": 0.07332829664225084,
-      "grad_norm": 2553.381591796875,
-      "learning_rate": 3.63e-05,
-      "loss": 134.3823,
-      "step": 18150
-    },
-    {
-      "epoch": 0.07336869790761846,
-      "grad_norm": 465.40289306640625,
-      "learning_rate": 3.6320000000000005e-05,
-      "loss": 104.473,
-      "step": 18160
-    },
-    {
-      "epoch": 0.0734090991729861,
-      "grad_norm": 2143.05419921875,
-      "learning_rate": 3.634e-05,
-      "loss": 133.1543,
-      "step": 18170
-    },
-    {
-      "epoch": 0.07344950043835373,
-      "grad_norm": 854.1031494140625,
-      "learning_rate": 3.636e-05,
-      "loss": 139.8475,
-      "step": 18180
-    },
-    {
-      "epoch": 0.07348990170372136,
-      "grad_norm": 812.4703979492188,
-      "learning_rate": 3.638e-05,
-      "loss": 124.2388,
-      "step": 18190
-    },
-    {
-      "epoch": 0.073530302969089,
-      "grad_norm": 693.6560668945312,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 125.4126,
-      "step": 18200
-    },
-    {
-      "epoch": 0.07357070423445662,
-      "grad_norm": 1367.3717041015625,
-      "learning_rate": 3.642000000000001e-05,
-      "loss": 141.6495,
-      "step": 18210
-    },
-    {
-      "epoch": 0.07361110549982426,
-      "grad_norm": 1077.287353515625,
-      "learning_rate": 3.6440000000000003e-05,
-      "loss": 130.6979,
-      "step": 18220
-    },
-    {
-      "epoch": 0.07365150676519189,
-      "grad_norm": 2508.4072265625,
-      "learning_rate": 3.646e-05,
-      "loss": 125.5967,
-      "step": 18230
-    },
-    {
-      "epoch": 0.07369190803055951,
-      "grad_norm": 2781.412841796875,
-      "learning_rate": 3.648e-05,
-      "loss": 149.0504,
-      "step": 18240
-    },
-    {
-      "epoch": 0.07373230929592715,
-      "grad_norm": 1022.107177734375,
-      "learning_rate": 3.65e-05,
-      "loss": 128.4223,
-      "step": 18250
-    },
-    {
-      "epoch": 0.07377271056129478,
-      "grad_norm": 700.0909423828125,
-      "learning_rate": 3.652e-05,
-      "loss": 95.3364,
-      "step": 18260
-    },
-    {
-      "epoch": 0.0738131118266624,
-      "grad_norm": 992.6243896484375,
-      "learning_rate": 3.654e-05,
-      "loss": 130.7981,
-      "step": 18270
-    },
-    {
-      "epoch": 0.07385351309203005,
-      "grad_norm": 1268.7449951171875,
-      "learning_rate": 3.656e-05,
-      "loss": 116.5302,
-      "step": 18280
-    },
-    {
-      "epoch": 0.07389391435739767,
-      "grad_norm": 915.7352905273438,
-      "learning_rate": 3.6580000000000006e-05,
-      "loss": 141.5618,
-      "step": 18290
-    },
-    {
-      "epoch": 0.0739343156227653,
-      "grad_norm": 3412.534912109375,
-      "learning_rate": 3.66e-05,
-      "loss": 188.1361,
-      "step": 18300
-    },
-    {
-      "epoch": 0.07397471688813294,
-      "grad_norm": 1475.6259765625,
-      "learning_rate": 3.6620000000000005e-05,
-      "loss": 125.9999,
-      "step": 18310
-    },
-    {
-      "epoch": 0.07401511815350056,
-      "grad_norm": 838.94140625,
-      "learning_rate": 3.664e-05,
-      "loss": 142.8819,
-      "step": 18320
-    },
-    {
-      "epoch": 0.0740555194188682,
-      "grad_norm": 2300.423583984375,
-      "learning_rate": 3.666e-05,
-      "loss": 168.7538,
-      "step": 18330
-    },
-    {
-      "epoch": 0.07409592068423583,
-      "grad_norm": 2023.9940185546875,
-      "learning_rate": 3.668e-05,
-      "loss": 126.4141,
-      "step": 18340
-    },
-    {
-      "epoch": 0.07413632194960346,
-      "grad_norm": 884.4683227539062,
-      "learning_rate": 3.6700000000000004e-05,
-      "loss": 64.7707,
-      "step": 18350
-    },
-    {
-      "epoch": 0.0741767232149711,
-      "grad_norm": 598.0726928710938,
-      "learning_rate": 3.672000000000001e-05,
-      "loss": 90.2243,
-      "step": 18360
-    },
-    {
-      "epoch": 0.07421712448033872,
-      "grad_norm": 852.4734497070312,
-      "learning_rate": 3.6740000000000004e-05,
-      "loss": 193.19,
-      "step": 18370
-    },
-    {
-      "epoch": 0.07425752574570635,
-      "grad_norm": 926.2811889648438,
-      "learning_rate": 3.676e-05,
-      "loss": 113.1817,
-      "step": 18380
-    },
-    {
-      "epoch": 0.07429792701107399,
-      "grad_norm": 840.8629760742188,
-      "learning_rate": 3.6780000000000004e-05,
-      "loss": 128.7378,
-      "step": 18390
-    },
-    {
-      "epoch": 0.07433832827644161,
-      "grad_norm": 1284.182373046875,
-      "learning_rate": 3.68e-05,
-      "loss": 119.9407,
-      "step": 18400
-    },
-    {
-      "epoch": 0.07437872954180925,
-      "grad_norm": 3427.597900390625,
-      "learning_rate": 3.682e-05,
-      "loss": 168.6104,
-      "step": 18410
-    },
-    {
-      "epoch": 0.07441913080717688,
-      "grad_norm": 1045.6478271484375,
-      "learning_rate": 3.684e-05,
-      "loss": 92.3087,
-      "step": 18420
-    },
-    {
-      "epoch": 0.07445953207254451,
-      "grad_norm": 1453.0216064453125,
-      "learning_rate": 3.686e-05,
-      "loss": 128.912,
-      "step": 18430
-    },
-    {
-      "epoch": 0.07449993333791215,
-      "grad_norm": 1257.60986328125,
-      "learning_rate": 3.6880000000000006e-05,
-      "loss": 67.0141,
-      "step": 18440
-    },
-    {
-      "epoch": 0.07454033460327977,
-      "grad_norm": 900.5206298828125,
-      "learning_rate": 3.69e-05,
-      "loss": 94.6233,
-      "step": 18450
-    },
-    {
-      "epoch": 0.0745807358686474,
-      "grad_norm": 769.1461181640625,
-      "learning_rate": 3.692e-05,
-      "loss": 128.7582,
-      "step": 18460
-    },
-    {
-      "epoch": 0.07462113713401504,
-      "grad_norm": 1010.8888549804688,
-      "learning_rate": 3.694e-05,
-      "loss": 115.287,
-      "step": 18470
-    },
-    {
-      "epoch": 0.07466153839938267,
-      "grad_norm": 974.634765625,
-      "learning_rate": 3.696e-05,
-      "loss": 97.2538,
-      "step": 18480
-    },
-    {
-      "epoch": 0.0747019396647503,
-      "grad_norm": 998.1504516601562,
-      "learning_rate": 3.698e-05,
-      "loss": 123.2764,
-      "step": 18490
-    },
-    {
-      "epoch": 0.07474234093011793,
-      "grad_norm": 1043.8447265625,
-      "learning_rate": 3.7e-05,
-      "loss": 105.7509,
-      "step": 18500
-    },
-    {
-      "epoch": 0.07478274219548556,
-      "grad_norm": 551.1446533203125,
-      "learning_rate": 3.702e-05,
-      "loss": 92.3631,
-      "step": 18510
-    },
-    {
-      "epoch": 0.0748231434608532,
-      "grad_norm": 677.5746459960938,
-      "learning_rate": 3.7040000000000005e-05,
-      "loss": 92.7502,
-      "step": 18520
-    },
-    {
-      "epoch": 0.07486354472622082,
-      "grad_norm": 935.2184448242188,
-      "learning_rate": 3.706e-05,
-      "loss": 94.4942,
-      "step": 18530
-    },
-    {
-      "epoch": 0.07490394599158845,
-      "grad_norm": 4195.58837890625,
-      "learning_rate": 3.7080000000000004e-05,
-      "loss": 180.4082,
-      "step": 18540
-    },
-    {
-      "epoch": 0.07494434725695609,
-      "grad_norm": 553.2954711914062,
-      "learning_rate": 3.71e-05,
-      "loss": 136.8525,
-      "step": 18550
-    },
-    {
-      "epoch": 0.07498474852232372,
-      "grad_norm": 1196.5914306640625,
-      "learning_rate": 3.712e-05,
-      "loss": 87.8535,
-      "step": 18560
-    },
-    {
-      "epoch": 0.07502514978769136,
-      "grad_norm": 925.3016967773438,
-      "learning_rate": 3.714e-05,
-      "loss": 99.2028,
-      "step": 18570
-    },
-    {
-      "epoch": 0.07506555105305898,
-      "grad_norm": 743.458984375,
-      "learning_rate": 3.716e-05,
-      "loss": 115.926,
-      "step": 18580
-    },
-    {
-      "epoch": 0.07510595231842661,
-      "grad_norm": 452.0498962402344,
-      "learning_rate": 3.7180000000000007e-05,
-      "loss": 72.2173,
-      "step": 18590
-    },
-    {
-      "epoch": 0.07514635358379425,
-      "grad_norm": 1277.8033447265625,
-      "learning_rate": 3.72e-05,
-      "loss": 196.1923,
-      "step": 18600
-    },
-    {
-      "epoch": 0.07518675484916187,
-      "grad_norm": 613.0692138671875,
-      "learning_rate": 3.722e-05,
-      "loss": 165.0734,
-      "step": 18610
-    },
-    {
-      "epoch": 0.0752271561145295,
-      "grad_norm": 707.7522583007812,
-      "learning_rate": 3.724e-05,
-      "loss": 146.3568,
-      "step": 18620
-    },
-    {
-      "epoch": 0.07526755737989714,
-      "grad_norm": 892.3648681640625,
-      "learning_rate": 3.726e-05,
-      "loss": 119.203,
-      "step": 18630
-    },
-    {
-      "epoch": 0.07530795864526477,
-      "grad_norm": 763.5734252929688,
-      "learning_rate": 3.728e-05,
-      "loss": 154.0482,
-      "step": 18640
-    },
-    {
-      "epoch": 0.0753483599106324,
-      "grad_norm": 779.2640991210938,
-      "learning_rate": 3.73e-05,
-      "loss": 73.6996,
-      "step": 18650
-    },
-    {
-      "epoch": 0.07538876117600003,
-      "grad_norm": 2393.424560546875,
-      "learning_rate": 3.732e-05,
-      "loss": 129.6301,
-      "step": 18660
-    },
-    {
-      "epoch": 0.07542916244136766,
-      "grad_norm": 2221.367919921875,
-      "learning_rate": 3.7340000000000005e-05,
-      "loss": 132.7365,
-      "step": 18670
-    },
-    {
-      "epoch": 0.0754695637067353,
-      "grad_norm": 1009.402587890625,
-      "learning_rate": 3.736e-05,
-      "loss": 110.5086,
-      "step": 18680
-    },
-    {
-      "epoch": 0.07550996497210292,
-      "grad_norm": 5788.82275390625,
-      "learning_rate": 3.7380000000000005e-05,
-      "loss": 93.0837,
-      "step": 18690
-    },
-    {
-      "epoch": 0.07555036623747055,
-      "grad_norm": 989.4839477539062,
-      "learning_rate": 3.74e-05,
-      "loss": 120.5923,
-      "step": 18700
-    },
-    {
-      "epoch": 0.07559076750283819,
-      "grad_norm": 1187.691650390625,
-      "learning_rate": 3.742e-05,
-      "loss": 191.3571,
-      "step": 18710
-    },
-    {
-      "epoch": 0.07563116876820582,
-      "grad_norm": 1042.4420166015625,
-      "learning_rate": 3.744e-05,
-      "loss": 123.3321,
-      "step": 18720
-    },
-    {
-      "epoch": 0.07567157003357346,
-      "grad_norm": 593.2315063476562,
-      "learning_rate": 3.7460000000000004e-05,
-      "loss": 124.6118,
-      "step": 18730
-    },
-    {
-      "epoch": 0.07571197129894108,
-      "grad_norm": 1148.8316650390625,
-      "learning_rate": 3.748000000000001e-05,
-      "loss": 129.3611,
-      "step": 18740
-    },
-    {
-      "epoch": 0.07575237256430871,
-      "grad_norm": 721.3247680664062,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 115.121,
-      "step": 18750
-    },
-    {
-      "epoch": 0.07579277382967635,
-      "grad_norm": 1153.5718994140625,
-      "learning_rate": 3.752e-05,
-      "loss": 96.7754,
-      "step": 18760
-    },
-    {
-      "epoch": 0.07583317509504398,
-      "grad_norm": 559.9420166015625,
-      "learning_rate": 3.754e-05,
-      "loss": 117.4526,
-      "step": 18770
-    },
-    {
-      "epoch": 0.0758735763604116,
-      "grad_norm": 1740.1900634765625,
-      "learning_rate": 3.756e-05,
-      "loss": 103.682,
-      "step": 18780
-    },
-    {
-      "epoch": 0.07591397762577924,
-      "grad_norm": 0.0,
-      "learning_rate": 3.758e-05,
-      "loss": 74.3376,
-      "step": 18790
-    },
-    {
-      "epoch": 0.07595437889114687,
-      "grad_norm": 1305.6412353515625,
-      "learning_rate": 3.76e-05,
-      "loss": 114.3259,
-      "step": 18800
-    },
-    {
-      "epoch": 0.07599478015651451,
-      "grad_norm": 1425.0380859375,
-      "learning_rate": 3.762e-05,
-      "loss": 171.961,
-      "step": 18810
-    },
-    {
-      "epoch": 0.07603518142188213,
-      "grad_norm": 1266.49072265625,
-      "learning_rate": 3.7640000000000006e-05,
-      "loss": 166.5889,
-      "step": 18820
-    },
-    {
-      "epoch": 0.07607558268724976,
-      "grad_norm": 1901.879638671875,
-      "learning_rate": 3.766e-05,
-      "loss": 132.9882,
-      "step": 18830
-    },
-    {
-      "epoch": 0.0761159839526174,
-      "grad_norm": 690.49951171875,
-      "learning_rate": 3.7680000000000005e-05,
-      "loss": 84.5623,
-      "step": 18840
-    },
-    {
-      "epoch": 0.07615638521798503,
-      "grad_norm": 650.4954833984375,
-      "learning_rate": 3.77e-05,
-      "loss": 135.0686,
-      "step": 18850
-    },
-    {
-      "epoch": 0.07619678648335265,
-      "grad_norm": 2350.57373046875,
-      "learning_rate": 3.772e-05,
-      "loss": 133.9776,
-      "step": 18860
-    },
-    {
-      "epoch": 0.07623718774872029,
-      "grad_norm": 1563.426513671875,
-      "learning_rate": 3.774e-05,
-      "loss": 101.9517,
-      "step": 18870
-    },
-    {
-      "epoch": 0.07627758901408792,
-      "grad_norm": 797.4249267578125,
-      "learning_rate": 3.776e-05,
-      "loss": 120.9917,
-      "step": 18880
-    },
-    {
-      "epoch": 0.07631799027945556,
-      "grad_norm": 1120.25732421875,
-      "learning_rate": 3.778000000000001e-05,
-      "loss": 115.9363,
-      "step": 18890
-    },
-    {
-      "epoch": 0.07635839154482318,
-      "grad_norm": 1334.6461181640625,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 147.4801,
-      "step": 18900
-    },
-    {
-      "epoch": 0.07639879281019081,
-      "grad_norm": 841.9830322265625,
-      "learning_rate": 3.782e-05,
-      "loss": 127.5146,
-      "step": 18910
-    },
-    {
-      "epoch": 0.07643919407555845,
-      "grad_norm": 951.7286987304688,
-      "learning_rate": 3.7840000000000004e-05,
-      "loss": 222.0843,
-      "step": 18920
-    },
-    {
-      "epoch": 0.07647959534092608,
-      "grad_norm": 1344.3798828125,
-      "learning_rate": 3.786e-05,
-      "loss": 118.1746,
-      "step": 18930
-    },
-    {
-      "epoch": 0.0765199966062937,
-      "grad_norm": 763.527587890625,
-      "learning_rate": 3.788e-05,
-      "loss": 74.0978,
-      "step": 18940
-    },
-    {
-      "epoch": 0.07656039787166134,
-      "grad_norm": 800.6107177734375,
-      "learning_rate": 3.79e-05,
-      "loss": 105.7953,
-      "step": 18950
-    },
-    {
-      "epoch": 0.07660079913702897,
-      "grad_norm": 1480.68505859375,
-      "learning_rate": 3.792e-05,
-      "loss": 127.3082,
-      "step": 18960
-    },
-    {
-      "epoch": 0.07664120040239661,
-      "grad_norm": 334.6526184082031,
-      "learning_rate": 3.7940000000000006e-05,
-      "loss": 85.6625,
-      "step": 18970
-    },
-    {
-      "epoch": 0.07668160166776423,
-      "grad_norm": 1548.9031982421875,
-      "learning_rate": 3.796e-05,
-      "loss": 131.035,
-      "step": 18980
-    },
-    {
-      "epoch": 0.07672200293313186,
-      "grad_norm": 1739.5743408203125,
-      "learning_rate": 3.7980000000000006e-05,
-      "loss": 172.2465,
-      "step": 18990
-    },
-    {
-      "epoch": 0.0767624041984995,
-      "grad_norm": 3091.291015625,
-      "learning_rate": 3.8e-05,
-      "loss": 160.8068,
-      "step": 19000
-    },
-    {
-      "epoch": 0.07680280546386713,
-      "grad_norm": 1500.0257568359375,
-      "learning_rate": 3.802e-05,
-      "loss": 76.5043,
-      "step": 19010
-    },
-    {
-      "epoch": 0.07684320672923475,
-      "grad_norm": 1217.782958984375,
-      "learning_rate": 3.804e-05,
-      "loss": 102.2421,
-      "step": 19020
-    },
-    {
-      "epoch": 0.0768836079946024,
-      "grad_norm": 1387.4803466796875,
-      "learning_rate": 3.806e-05,
-      "loss": 94.5826,
-      "step": 19030
-    },
-    {
-      "epoch": 0.07692400925997002,
-      "grad_norm": 1300.2745361328125,
-      "learning_rate": 3.808e-05,
-      "loss": 130.5162,
-      "step": 19040
-    },
-    {
-      "epoch": 0.07696441052533766,
-      "grad_norm": 1954.7742919921875,
-      "learning_rate": 3.8100000000000005e-05,
-      "loss": 157.192,
-      "step": 19050
-    },
-    {
-      "epoch": 0.07700481179070529,
-      "grad_norm": 2402.571533203125,
-      "learning_rate": 3.812e-05,
-      "loss": 181.6468,
-      "step": 19060
-    },
-    {
-      "epoch": 0.07704521305607291,
-      "grad_norm": 1889.8359375,
-      "learning_rate": 3.8140000000000004e-05,
-      "loss": 112.0516,
-      "step": 19070
-    },
-    {
-      "epoch": 0.07708561432144055,
-      "grad_norm": 834.8114624023438,
-      "learning_rate": 3.816e-05,
-      "loss": 104.1592,
-      "step": 19080
-    },
-    {
-      "epoch": 0.07712601558680818,
-      "grad_norm": 1373.436279296875,
-      "learning_rate": 3.818e-05,
-      "loss": 121.8585,
-      "step": 19090
-    },
-    {
-      "epoch": 0.0771664168521758,
-      "grad_norm": 1009.5123901367188,
-      "learning_rate": 3.82e-05,
-      "loss": 156.3473,
-      "step": 19100
-    },
-    {
-      "epoch": 0.07720681811754344,
-      "grad_norm": 986.4491577148438,
-      "learning_rate": 3.822e-05,
-      "loss": 123.8396,
-      "step": 19110
-    },
-    {
-      "epoch": 0.07724721938291107,
-      "grad_norm": 690.7348022460938,
-      "learning_rate": 3.8240000000000007e-05,
-      "loss": 81.2586,
-      "step": 19120
-    },
-    {
-      "epoch": 0.07728762064827871,
-      "grad_norm": 603.4257202148438,
-      "learning_rate": 3.826e-05,
-      "loss": 148.1277,
-      "step": 19130
-    },
-    {
-      "epoch": 0.07732802191364634,
-      "grad_norm": 610.265869140625,
-      "learning_rate": 3.828e-05,
-      "loss": 107.6773,
-      "step": 19140
-    },
-    {
-      "epoch": 0.07736842317901396,
-      "grad_norm": 596.2022094726562,
-      "learning_rate": 3.83e-05,
-      "loss": 81.1043,
-      "step": 19150
-    },
-    {
-      "epoch": 0.0774088244443816,
-      "grad_norm": 630.5447387695312,
-      "learning_rate": 3.832e-05,
-      "loss": 75.0851,
-      "step": 19160
-    },
-    {
-      "epoch": 0.07744922570974923,
-      "grad_norm": 967.669921875,
-      "learning_rate": 3.834e-05,
-      "loss": 68.293,
-      "step": 19170
-    },
-    {
-      "epoch": 0.07748962697511685,
-      "grad_norm": 7127.17236328125,
-      "learning_rate": 3.836e-05,
-      "loss": 117.3339,
-      "step": 19180
-    },
-    {
-      "epoch": 0.0775300282404845,
-      "grad_norm": 978.7310180664062,
-      "learning_rate": 3.838e-05,
-      "loss": 169.6352,
-      "step": 19190
-    },
-    {
-      "epoch": 0.07757042950585212,
-      "grad_norm": 1465.735595703125,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 133.1634,
-      "step": 19200
-    },
-    {
-      "epoch": 0.07761083077121976,
-      "grad_norm": 1282.474609375,
-      "learning_rate": 3.842e-05,
-      "loss": 142.1115,
-      "step": 19210
-    },
-    {
-      "epoch": 0.07765123203658739,
-      "grad_norm": 576.253173828125,
-      "learning_rate": 3.8440000000000005e-05,
-      "loss": 167.8867,
-      "step": 19220
-    },
-    {
-      "epoch": 0.07769163330195501,
-      "grad_norm": 911.6886596679688,
-      "learning_rate": 3.846e-05,
-      "loss": 106.331,
-      "step": 19230
-    },
-    {
-      "epoch": 0.07773203456732265,
-      "grad_norm": 1108.574462890625,
-      "learning_rate": 3.848e-05,
-      "loss": 159.7625,
-      "step": 19240
-    },
-    {
-      "epoch": 0.07777243583269028,
-      "grad_norm": 1084.5352783203125,
-      "learning_rate": 3.85e-05,
-      "loss": 102.5427,
-      "step": 19250
-    },
-    {
-      "epoch": 0.0778128370980579,
-      "grad_norm": 2015.353515625,
-      "learning_rate": 3.8520000000000004e-05,
-      "loss": 152.4607,
-      "step": 19260
-    },
-    {
-      "epoch": 0.07785323836342554,
-      "grad_norm": 2406.21484375,
-      "learning_rate": 3.854000000000001e-05,
-      "loss": 115.7677,
-      "step": 19270
-    },
-    {
-      "epoch": 0.07789363962879317,
-      "grad_norm": 1180.568115234375,
-      "learning_rate": 3.8560000000000004e-05,
-      "loss": 87.3702,
-      "step": 19280
-    },
-    {
-      "epoch": 0.07793404089416081,
-      "grad_norm": 546.580810546875,
-      "learning_rate": 3.858e-05,
-      "loss": 71.9833,
-      "step": 19290
-    },
-    {
-      "epoch": 0.07797444215952844,
-      "grad_norm": 1000.6842651367188,
-      "learning_rate": 3.86e-05,
-      "loss": 127.7328,
-      "step": 19300
-    },
-    {
-      "epoch": 0.07801484342489606,
-      "grad_norm": 717.3246459960938,
-      "learning_rate": 3.862e-05,
-      "loss": 95.6857,
-      "step": 19310
-    },
-    {
-      "epoch": 0.0780552446902637,
-      "grad_norm": 4112.51220703125,
-      "learning_rate": 3.864e-05,
-      "loss": 199.442,
-      "step": 19320
-    },
-    {
-      "epoch": 0.07809564595563133,
-      "grad_norm": 972.7607421875,
-      "learning_rate": 3.866e-05,
-      "loss": 168.3066,
-      "step": 19330
-    },
-    {
-      "epoch": 0.07813604722099896,
-      "grad_norm": 1725.4652099609375,
-      "learning_rate": 3.868e-05,
-      "loss": 125.3838,
-      "step": 19340
-    },
-    {
-      "epoch": 0.0781764484863666,
-      "grad_norm": 991.9950561523438,
-      "learning_rate": 3.8700000000000006e-05,
-      "loss": 110.0063,
-      "step": 19350
-    },
-    {
-      "epoch": 0.07821684975173422,
-      "grad_norm": 570.45751953125,
-      "learning_rate": 3.872e-05,
-      "loss": 148.7821,
-      "step": 19360
-    },
-    {
-      "epoch": 0.07825725101710186,
-      "grad_norm": 998.0093994140625,
-      "learning_rate": 3.8740000000000005e-05,
-      "loss": 106.5317,
-      "step": 19370
-    },
-    {
-      "epoch": 0.07829765228246949,
-      "grad_norm": 2785.006103515625,
-      "learning_rate": 3.876e-05,
-      "loss": 174.5811,
-      "step": 19380
-    },
-    {
-      "epoch": 0.07833805354783711,
-      "grad_norm": 1203.6185302734375,
-      "learning_rate": 3.878e-05,
-      "loss": 159.747,
-      "step": 19390
-    },
-    {
-      "epoch": 0.07837845481320475,
-      "grad_norm": 1574.8548583984375,
-      "learning_rate": 3.88e-05,
-      "loss": 161.0951,
-      "step": 19400
-    },
-    {
-      "epoch": 0.07841885607857238,
-      "grad_norm": 3679.686279296875,
-      "learning_rate": 3.882e-05,
-      "loss": 140.7081,
-      "step": 19410
-    },
-    {
-      "epoch": 0.07845925734394,
-      "grad_norm": 3366.86669921875,
-      "learning_rate": 3.884e-05,
-      "loss": 129.2223,
-      "step": 19420
-    },
-    {
-      "epoch": 0.07849965860930765,
-      "grad_norm": 979.6845703125,
-      "learning_rate": 3.8860000000000004e-05,
-      "loss": 123.001,
-      "step": 19430
-    },
-    {
-      "epoch": 0.07854005987467527,
-      "grad_norm": 706.135498046875,
-      "learning_rate": 3.888e-05,
-      "loss": 107.5821,
-      "step": 19440
-    },
-    {
-      "epoch": 0.07858046114004291,
-      "grad_norm": 3030.436279296875,
-      "learning_rate": 3.8900000000000004e-05,
-      "loss": 136.2499,
-      "step": 19450
-    },
-    {
-      "epoch": 0.07862086240541054,
-      "grad_norm": 341.5601501464844,
-      "learning_rate": 3.892e-05,
-      "loss": 87.6859,
-      "step": 19460
-    },
-    {
-      "epoch": 0.07866126367077816,
-      "grad_norm": 890.729248046875,
-      "learning_rate": 3.894e-05,
-      "loss": 130.7491,
-      "step": 19470
-    },
-    {
-      "epoch": 0.0787016649361458,
-      "grad_norm": 2620.10400390625,
-      "learning_rate": 3.896e-05,
-      "loss": 100.2728,
-      "step": 19480
-    },
-    {
-      "epoch": 0.07874206620151343,
-      "grad_norm": 2374.006103515625,
-      "learning_rate": 3.898e-05,
-      "loss": 102.7446,
-      "step": 19490
-    },
-    {
-      "epoch": 0.07878246746688106,
-      "grad_norm": 661.5973510742188,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 103.1776,
-      "step": 19500
-    },
-    {
-      "epoch": 0.0788228687322487,
-      "grad_norm": 496.9710693359375,
-      "learning_rate": 3.902e-05,
-      "loss": 104.6827,
-      "step": 19510
-    },
-    {
-      "epoch": 0.07886326999761632,
-      "grad_norm": 950.9691772460938,
-      "learning_rate": 3.9040000000000006e-05,
-      "loss": 104.4576,
-      "step": 19520
-    },
-    {
-      "epoch": 0.07890367126298396,
-      "grad_norm": 1480.8165283203125,
-      "learning_rate": 3.906e-05,
-      "loss": 123.9244,
-      "step": 19530
-    },
-    {
-      "epoch": 0.07894407252835159,
-      "grad_norm": 1668.803955078125,
-      "learning_rate": 3.908e-05,
-      "loss": 114.0788,
-      "step": 19540
-    },
-    {
-      "epoch": 0.07898447379371921,
-      "grad_norm": 998.9266357421875,
-      "learning_rate": 3.91e-05,
-      "loss": 116.9195,
-      "step": 19550
-    },
-    {
-      "epoch": 0.07902487505908685,
-      "grad_norm": 685.7608032226562,
-      "learning_rate": 3.912e-05,
-      "loss": 120.8969,
-      "step": 19560
-    },
-    {
-      "epoch": 0.07906527632445448,
-      "grad_norm": 1155.7894287109375,
-      "learning_rate": 3.914e-05,
-      "loss": 121.0664,
-      "step": 19570
-    },
-    {
-      "epoch": 0.07910567758982211,
-      "grad_norm": 959.234130859375,
-      "learning_rate": 3.9160000000000005e-05,
-      "loss": 96.3062,
-      "step": 19580
-    },
-    {
-      "epoch": 0.07914607885518975,
-      "grad_norm": 1889.412109375,
-      "learning_rate": 3.918e-05,
-      "loss": 81.9331,
-      "step": 19590
-    },
-    {
-      "epoch": 0.07918648012055737,
-      "grad_norm": 966.55078125,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 164.067,
-      "step": 19600
-    },
-    {
-      "epoch": 0.07922688138592501,
-      "grad_norm": 645.4921875,
-      "learning_rate": 3.922e-05,
-      "loss": 126.4012,
-      "step": 19610
-    },
-    {
-      "epoch": 0.07926728265129264,
-      "grad_norm": 1248.7474365234375,
-      "learning_rate": 3.9240000000000004e-05,
-      "loss": 109.2031,
-      "step": 19620
-    },
-    {
-      "epoch": 0.07930768391666027,
-      "grad_norm": 2941.3408203125,
-      "learning_rate": 3.926e-05,
-      "loss": 100.5554,
-      "step": 19630
-    },
-    {
-      "epoch": 0.0793480851820279,
-      "grad_norm": 1487.9503173828125,
-      "learning_rate": 3.9280000000000003e-05,
-      "loss": 120.6536,
-      "step": 19640
-    },
-    {
-      "epoch": 0.07938848644739553,
-      "grad_norm": 956.5473022460938,
-      "learning_rate": 3.9300000000000007e-05,
-      "loss": 112.047,
-      "step": 19650
-    },
-    {
-      "epoch": 0.07942888771276316,
-      "grad_norm": 675.1141967773438,
-      "learning_rate": 3.932e-05,
-      "loss": 121.3393,
-      "step": 19660
-    },
-    {
-      "epoch": 0.0794692889781308,
-      "grad_norm": 1062.9857177734375,
-      "learning_rate": 3.9340000000000006e-05,
-      "loss": 118.6263,
-      "step": 19670
-    },
-    {
-      "epoch": 0.07950969024349842,
-      "grad_norm": 796.38525390625,
-      "learning_rate": 3.936e-05,
-      "loss": 93.5364,
-      "step": 19680
-    },
-    {
-      "epoch": 0.07955009150886606,
-      "grad_norm": 712.4339599609375,
-      "learning_rate": 3.938e-05,
-      "loss": 169.9139,
-      "step": 19690
-    },
-    {
-      "epoch": 0.07959049277423369,
-      "grad_norm": 348.9084777832031,
-      "learning_rate": 3.94e-05,
-      "loss": 98.5714,
-      "step": 19700
-    },
-    {
-      "epoch": 0.07963089403960132,
-      "grad_norm": 1002.7638549804688,
-      "learning_rate": 3.942e-05,
-      "loss": 101.6811,
-      "step": 19710
-    },
-    {
-      "epoch": 0.07967129530496896,
-      "grad_norm": 1141.4864501953125,
-      "learning_rate": 3.944e-05,
-      "loss": 99.0602,
-      "step": 19720
-    },
-    {
-      "epoch": 0.07971169657033658,
-      "grad_norm": 1789.052001953125,
-      "learning_rate": 3.9460000000000005e-05,
-      "loss": 146.9852,
-      "step": 19730
-    },
-    {
-      "epoch": 0.07975209783570421,
-      "grad_norm": 1004.76611328125,
-      "learning_rate": 3.948e-05,
-      "loss": 108.5757,
-      "step": 19740
-    },
-    {
-      "epoch": 0.07979249910107185,
-      "grad_norm": 1779.8487548828125,
-      "learning_rate": 3.9500000000000005e-05,
-      "loss": 92.2654,
-      "step": 19750
-    },
-    {
-      "epoch": 0.07983290036643947,
-      "grad_norm": 1891.206787109375,
-      "learning_rate": 3.952e-05,
-      "loss": 130.2949,
-      "step": 19760
-    },
-    {
-      "epoch": 0.07987330163180711,
-      "grad_norm": 1210.8375244140625,
-      "learning_rate": 3.954e-05,
-      "loss": 149.1875,
-      "step": 19770
-    },
-    {
-      "epoch": 0.07991370289717474,
-      "grad_norm": 1035.046630859375,
-      "learning_rate": 3.956e-05,
-      "loss": 90.7728,
-      "step": 19780
-    },
-    {
-      "epoch": 0.07995410416254237,
-      "grad_norm": 1166.8427734375,
-      "learning_rate": 3.958e-05,
-      "loss": 113.0428,
-      "step": 19790
-    },
-    {
-      "epoch": 0.07999450542791,
-      "grad_norm": 1342.7369384765625,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 84.0289,
-      "step": 19800
-    },
-    {
-      "epoch": 0.08003490669327763,
-      "grad_norm": 930.572509765625,
-      "learning_rate": 3.9620000000000004e-05,
-      "loss": 137.8586,
-      "step": 19810
-    },
-    {
-      "epoch": 0.08007530795864526,
-      "grad_norm": 537.0020141601562,
-      "learning_rate": 3.964e-05,
-      "loss": 136.4454,
-      "step": 19820
-    },
-    {
-      "epoch": 0.0801157092240129,
-      "grad_norm": 8671.7001953125,
-      "learning_rate": 3.966e-05,
-      "loss": 194.5568,
-      "step": 19830
-    },
-    {
-      "epoch": 0.08015611048938052,
-      "grad_norm": 1121.135986328125,
-      "learning_rate": 3.968e-05,
-      "loss": 171.3159,
-      "step": 19840
-    },
-    {
-      "epoch": 0.08019651175474816,
-      "grad_norm": 875.9639282226562,
-      "learning_rate": 3.97e-05,
-      "loss": 138.4204,
-      "step": 19850
-    },
-    {
-      "epoch": 0.08023691302011579,
-      "grad_norm": 1031.24462890625,
-      "learning_rate": 3.972e-05,
-      "loss": 70.7438,
-      "step": 19860
-    },
-    {
-      "epoch": 0.08027731428548342,
-      "grad_norm": 844.333251953125,
-      "learning_rate": 3.974e-05,
-      "loss": 86.4556,
-      "step": 19870
-    },
-    {
-      "epoch": 0.08031771555085106,
-      "grad_norm": 775.6649780273438,
-      "learning_rate": 3.9760000000000006e-05,
-      "loss": 141.9916,
-      "step": 19880
-    },
-    {
-      "epoch": 0.08035811681621868,
-      "grad_norm": 579.8970947265625,
-      "learning_rate": 3.978e-05,
-      "loss": 131.3396,
-      "step": 19890
-    },
-    {
-      "epoch": 0.08039851808158631,
-      "grad_norm": 0.0,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 98.0002,
-      "step": 19900
-    },
-    {
-      "epoch": 0.08043891934695395,
-      "grad_norm": 1350.4500732421875,
-      "learning_rate": 3.982e-05,
-      "loss": 221.3071,
-      "step": 19910
-    },
-    {
-      "epoch": 0.08047932061232158,
-      "grad_norm": 809.1064453125,
-      "learning_rate": 3.984e-05,
-      "loss": 82.145,
-      "step": 19920
-    },
-    {
-      "epoch": 0.08051972187768922,
-      "grad_norm": 1032.8472900390625,
-      "learning_rate": 3.986e-05,
-      "loss": 90.5167,
-      "step": 19930
-    },
-    {
-      "epoch": 0.08056012314305684,
-      "grad_norm": 548.9085693359375,
-      "learning_rate": 3.988e-05,
-      "loss": 82.506,
-      "step": 19940
-    },
-    {
-      "epoch": 0.08060052440842447,
-      "grad_norm": 828.6427001953125,
-      "learning_rate": 3.99e-05,
-      "loss": 103.581,
-      "step": 19950
-    },
-    {
-      "epoch": 0.08064092567379211,
-      "grad_norm": 1131.4522705078125,
-      "learning_rate": 3.9920000000000004e-05,
-      "loss": 124.3697,
-      "step": 19960
-    },
-    {
-      "epoch": 0.08068132693915973,
-      "grad_norm": 3095.481689453125,
-      "learning_rate": 3.994e-05,
-      "loss": 180.1047,
-      "step": 19970
-    },
-    {
-      "epoch": 0.08072172820452736,
-      "grad_norm": 1326.9815673828125,
-      "learning_rate": 3.9960000000000004e-05,
-      "loss": 111.703,
-      "step": 19980
-    },
-    {
-      "epoch": 0.080762129469895,
-      "grad_norm": 563.6680297851562,
-      "learning_rate": 3.998e-05,
-      "loss": 86.8286,
-      "step": 19990
-    },
-    {
-      "epoch": 0.08080253073526263,
-      "grad_norm": 867.2445068359375,
-      "learning_rate": 4e-05,
-      "loss": 120.7213,
-      "step": 20000
-    },
-    {
-      "epoch": 0.08084293200063027,
-      "grad_norm": 3826.01513671875,
-      "learning_rate": 4.002e-05,
-      "loss": 188.3878,
-      "step": 20010
-    },
-    {
-      "epoch": 0.08088333326599789,
-      "grad_norm": 626.7747192382812,
-      "learning_rate": 4.004e-05,
-      "loss": 102.346,
-      "step": 20020
-    },
-    {
-      "epoch": 0.08092373453136552,
-      "grad_norm": 447.5019836425781,
-      "learning_rate": 4.0060000000000006e-05,
-      "loss": 112.2855,
-      "step": 20030
-    },
-    {
-      "epoch": 0.08096413579673316,
-      "grad_norm": 1387.30029296875,
-      "learning_rate": 4.008e-05,
-      "loss": 144.4808,
-      "step": 20040
-    },
-    {
-      "epoch": 0.08100453706210078,
-      "grad_norm": 711.4352416992188,
-      "learning_rate": 4.0100000000000006e-05,
-      "loss": 123.1025,
-      "step": 20050
-    },
-    {
-      "epoch": 0.08104493832746841,
-      "grad_norm": 696.3189086914062,
-      "learning_rate": 4.012e-05,
-      "loss": 125.8491,
-      "step": 20060
-    },
-    {
-      "epoch": 0.08108533959283605,
-      "grad_norm": 1703.4642333984375,
-      "learning_rate": 4.014e-05,
-      "loss": 180.9456,
-      "step": 20070
-    },
-    {
-      "epoch": 0.08112574085820368,
-      "grad_norm": 931.850341796875,
-      "learning_rate": 4.016e-05,
-      "loss": 157.8773,
-      "step": 20080
-    },
-    {
-      "epoch": 0.08116614212357132,
-      "grad_norm": 1737.6793212890625,
-      "learning_rate": 4.018e-05,
-      "loss": 108.9433,
-      "step": 20090
-    },
-    {
-      "epoch": 0.08120654338893894,
-      "grad_norm": 1597.1063232421875,
-      "learning_rate": 4.02e-05,
-      "loss": 116.3655,
-      "step": 20100
-    },
-    {
-      "epoch": 0.08124694465430657,
-      "grad_norm": 1292.120849609375,
-      "learning_rate": 4.0220000000000005e-05,
-      "loss": 128.3085,
-      "step": 20110
-    },
-    {
-      "epoch": 0.08128734591967421,
-      "grad_norm": 784.8877563476562,
-      "learning_rate": 4.024e-05,
-      "loss": 142.1472,
-      "step": 20120
-    },
-    {
-      "epoch": 0.08132774718504183,
-      "grad_norm": 518.2711181640625,
-      "learning_rate": 4.0260000000000004e-05,
-      "loss": 138.0721,
-      "step": 20130
-    },
-    {
-      "epoch": 0.08136814845040946,
-      "grad_norm": 3041.567138671875,
-      "learning_rate": 4.028e-05,
-      "loss": 143.0653,
-      "step": 20140
-    },
-    {
-      "epoch": 0.0814085497157771,
-      "grad_norm": 393.4328918457031,
-      "learning_rate": 4.0300000000000004e-05,
-      "loss": 123.1964,
-      "step": 20150
-    },
-    {
-      "epoch": 0.08144895098114473,
-      "grad_norm": 433.4530334472656,
-      "learning_rate": 4.032e-05,
-      "loss": 108.3037,
-      "step": 20160
-    },
-    {
-      "epoch": 0.08148935224651237,
-      "grad_norm": 1056.2037353515625,
-      "learning_rate": 4.034e-05,
-      "loss": 104.124,
-      "step": 20170
-    },
-    {
-      "epoch": 0.08152975351188,
-      "grad_norm": 1283.22607421875,
-      "learning_rate": 4.0360000000000007e-05,
-      "loss": 146.2523,
-      "step": 20180
-    },
-    {
-      "epoch": 0.08157015477724762,
-      "grad_norm": 1303.0574951171875,
-      "learning_rate": 4.038e-05,
-      "loss": 108.7274,
-      "step": 20190
-    },
-    {
-      "epoch": 0.08161055604261526,
-      "grad_norm": 748.68115234375,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 118.8553,
-      "step": 20200
-    },
-    {
-      "epoch": 0.08165095730798289,
-      "grad_norm": 559.082763671875,
-      "learning_rate": 4.042e-05,
-      "loss": 104.5537,
-      "step": 20210
-    },
-    {
-      "epoch": 0.08169135857335051,
-      "grad_norm": 883.0214233398438,
-      "learning_rate": 4.044e-05,
-      "loss": 119.6957,
-      "step": 20220
-    },
-    {
-      "epoch": 0.08173175983871815,
-      "grad_norm": 3064.196533203125,
-      "learning_rate": 4.046e-05,
-      "loss": 167.4422,
-      "step": 20230
-    },
-    {
-      "epoch": 0.08177216110408578,
-      "grad_norm": 0.0,
-      "learning_rate": 4.048e-05,
-      "loss": 128.549,
-      "step": 20240
-    },
-    {
-      "epoch": 0.08181256236945342,
-      "grad_norm": 755.3316040039062,
-      "learning_rate": 4.05e-05,
-      "loss": 108.4741,
-      "step": 20250
-    },
-    {
-      "epoch": 0.08185296363482104,
-      "grad_norm": 1562.85498046875,
-      "learning_rate": 4.0520000000000005e-05,
-      "loss": 111.5061,
-      "step": 20260
-    },
-    {
-      "epoch": 0.08189336490018867,
-      "grad_norm": 871.8521118164062,
-      "learning_rate": 4.054e-05,
-      "loss": 110.9914,
-      "step": 20270
-    },
-    {
-      "epoch": 0.08193376616555631,
-      "grad_norm": 772.6279907226562,
-      "learning_rate": 4.0560000000000005e-05,
-      "loss": 145.3599,
-      "step": 20280
-    },
-    {
-      "epoch": 0.08197416743092394,
-      "grad_norm": 529.8480224609375,
-      "learning_rate": 4.058e-05,
-      "loss": 101.3806,
-      "step": 20290
-    },
-    {
-      "epoch": 0.08201456869629156,
-      "grad_norm": 2998.20556640625,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 111.4782,
-      "step": 20300
-    },
-    {
-      "epoch": 0.0820549699616592,
-      "grad_norm": 1405.43505859375,
-      "learning_rate": 4.062e-05,
-      "loss": 163.5577,
-      "step": 20310
-    },
-    {
-      "epoch": 0.08209537122702683,
-      "grad_norm": 1239.15087890625,
-      "learning_rate": 4.064e-05,
-      "loss": 94.2629,
-      "step": 20320
-    },
-    {
-      "epoch": 0.08213577249239447,
-      "grad_norm": 1255.966552734375,
-      "learning_rate": 4.066e-05,
-      "loss": 126.8316,
-      "step": 20330
-    },
-    {
-      "epoch": 0.0821761737577621,
-      "grad_norm": 737.30126953125,
-      "learning_rate": 4.0680000000000004e-05,
-      "loss": 113.6555,
-      "step": 20340
-    },
-    {
-      "epoch": 0.08221657502312972,
-      "grad_norm": 2296.812744140625,
-      "learning_rate": 4.07e-05,
-      "loss": 109.0417,
-      "step": 20350
-    },
-    {
-      "epoch": 0.08225697628849736,
-      "grad_norm": 1186.671875,
-      "learning_rate": 4.072e-05,
-      "loss": 86.3001,
-      "step": 20360
-    },
-    {
-      "epoch": 0.08229737755386499,
-      "grad_norm": 1619.2152099609375,
-      "learning_rate": 4.074e-05,
-      "loss": 148.3378,
-      "step": 20370
-    },
-    {
-      "epoch": 0.08233777881923261,
-      "grad_norm": 3074.595703125,
-      "learning_rate": 4.076e-05,
-      "loss": 112.1198,
-      "step": 20380
-    },
-    {
-      "epoch": 0.08237818008460025,
-      "grad_norm": 2610.1181640625,
-      "learning_rate": 4.078e-05,
-      "loss": 118.642,
-      "step": 20390
-    },
-    {
-      "epoch": 0.08241858134996788,
-      "grad_norm": 944.0482788085938,
-      "learning_rate": 4.08e-05,
-      "loss": 118.4644,
-      "step": 20400
-    },
-    {
-      "epoch": 0.08245898261533552,
-      "grad_norm": 2997.534423828125,
-      "learning_rate": 4.0820000000000006e-05,
-      "loss": 95.5434,
-      "step": 20410
-    },
-    {
-      "epoch": 0.08249938388070314,
-      "grad_norm": 1543.2197265625,
-      "learning_rate": 4.084e-05,
-      "loss": 155.0003,
-      "step": 20420
-    },
-    {
-      "epoch": 0.08253978514607077,
-      "grad_norm": 749.4830932617188,
-      "learning_rate": 4.0860000000000005e-05,
-      "loss": 98.9934,
-      "step": 20430
-    },
-    {
-      "epoch": 0.08258018641143841,
-      "grad_norm": 2779.212646484375,
-      "learning_rate": 4.088e-05,
-      "loss": 115.3504,
-      "step": 20440
-    },
-    {
-      "epoch": 0.08262058767680604,
-      "grad_norm": 2938.130615234375,
-      "learning_rate": 4.09e-05,
-      "loss": 121.3548,
-      "step": 20450
-    },
-    {
-      "epoch": 0.08266098894217366,
-      "grad_norm": 364.253173828125,
-      "learning_rate": 4.092e-05,
-      "loss": 118.8423,
-      "step": 20460
-    },
-    {
-      "epoch": 0.0827013902075413,
-      "grad_norm": 956.0880737304688,
-      "learning_rate": 4.094e-05,
-      "loss": 162.3844,
-      "step": 20470
-    },
-    {
-      "epoch": 0.08274179147290893,
-      "grad_norm": 651.1348266601562,
-      "learning_rate": 4.096e-05,
-      "loss": 101.0112,
-      "step": 20480
-    },
-    {
-      "epoch": 0.08278219273827657,
-      "grad_norm": 1472.0135498046875,
-      "learning_rate": 4.0980000000000004e-05,
-      "loss": 140.2291,
-      "step": 20490
-    },
-    {
-      "epoch": 0.0828225940036442,
-      "grad_norm": 1709.4990234375,
-      "learning_rate": 4.1e-05,
-      "loss": 156.1177,
-      "step": 20500
-    },
-    {
-      "epoch": 0.08286299526901182,
-      "grad_norm": 1078.7275390625,
-      "learning_rate": 4.1020000000000004e-05,
-      "loss": 149.2032,
-      "step": 20510
-    },
-    {
-      "epoch": 0.08290339653437946,
-      "grad_norm": 1183.4559326171875,
-      "learning_rate": 4.104e-05,
-      "loss": 85.2116,
-      "step": 20520
-    },
-    {
-      "epoch": 0.08294379779974709,
-      "grad_norm": 787.4295043945312,
-      "learning_rate": 4.106e-05,
-      "loss": 143.7207,
-      "step": 20530
-    },
-    {
-      "epoch": 0.08298419906511471,
-      "grad_norm": 497.1183776855469,
-      "learning_rate": 4.108e-05,
-      "loss": 89.7213,
-      "step": 20540
-    },
-    {
-      "epoch": 0.08302460033048235,
-      "grad_norm": 1972.8258056640625,
-      "learning_rate": 4.11e-05,
-      "loss": 115.1941,
-      "step": 20550
-    },
-    {
-      "epoch": 0.08306500159584998,
-      "grad_norm": 969.597900390625,
-      "learning_rate": 4.1120000000000006e-05,
-      "loss": 102.9652,
-      "step": 20560
-    },
-    {
-      "epoch": 0.08310540286121762,
-      "grad_norm": 698.786376953125,
-      "learning_rate": 4.114e-05,
-      "loss": 63.844,
-      "step": 20570
-    },
-    {
-      "epoch": 0.08314580412658525,
-      "grad_norm": 1212.814697265625,
-      "learning_rate": 4.1160000000000006e-05,
-      "loss": 149.4642,
-      "step": 20580
-    },
-    {
-      "epoch": 0.08318620539195287,
-      "grad_norm": 552.6875610351562,
-      "learning_rate": 4.118e-05,
-      "loss": 140.7715,
-      "step": 20590
-    },
-    {
-      "epoch": 0.08322660665732051,
-      "grad_norm": 1048.22607421875,
-      "learning_rate": 4.12e-05,
-      "loss": 154.9678,
-      "step": 20600
-    },
-    {
-      "epoch": 0.08326700792268814,
-      "grad_norm": 1216.959716796875,
-      "learning_rate": 4.122e-05,
-      "loss": 167.41,
-      "step": 20610
-    },
-    {
-      "epoch": 0.08330740918805576,
-      "grad_norm": 2889.177490234375,
-      "learning_rate": 4.124e-05,
-      "loss": 142.2961,
-      "step": 20620
-    },
-    {
-      "epoch": 0.0833478104534234,
-      "grad_norm": 1590.9716796875,
-      "learning_rate": 4.126e-05,
-      "loss": 158.046,
-      "step": 20630
-    },
-    {
-      "epoch": 0.08338821171879103,
-      "grad_norm": 1097.50146484375,
-      "learning_rate": 4.1280000000000005e-05,
-      "loss": 132.7002,
-      "step": 20640
-    },
-    {
-      "epoch": 0.08342861298415867,
-      "grad_norm": 2981.930908203125,
-      "learning_rate": 4.13e-05,
-      "loss": 95.1602,
-      "step": 20650
-    },
-    {
-      "epoch": 0.0834690142495263,
-      "grad_norm": 2784.6552734375,
-      "learning_rate": 4.1320000000000004e-05,
-      "loss": 173.26,
-      "step": 20660
-    },
-    {
-      "epoch": 0.08350941551489392,
-      "grad_norm": 1329.2828369140625,
-      "learning_rate": 4.134e-05,
-      "loss": 162.6604,
-      "step": 20670
-    },
-    {
-      "epoch": 0.08354981678026156,
-      "grad_norm": 808.767822265625,
-      "learning_rate": 4.1360000000000004e-05,
-      "loss": 112.3144,
-      "step": 20680
-    },
-    {
-      "epoch": 0.08359021804562919,
-      "grad_norm": 1342.0172119140625,
-      "learning_rate": 4.138e-05,
-      "loss": 131.2333,
-      "step": 20690
-    },
-    {
-      "epoch": 0.08363061931099681,
-      "grad_norm": 848.967529296875,
-      "learning_rate": 4.14e-05,
-      "loss": 147.4289,
-      "step": 20700
-    },
-    {
-      "epoch": 0.08367102057636445,
-      "grad_norm": 566.413818359375,
-      "learning_rate": 4.142000000000001e-05,
-      "loss": 150.0216,
-      "step": 20710
-    },
-    {
-      "epoch": 0.08371142184173208,
-      "grad_norm": 707.4857788085938,
-      "learning_rate": 4.144e-05,
-      "loss": 113.268,
-      "step": 20720
-    },
-    {
-      "epoch": 0.08375182310709972,
-      "grad_norm": 851.73779296875,
-      "learning_rate": 4.1460000000000006e-05,
-      "loss": 128.7113,
-      "step": 20730
-    },
-    {
-      "epoch": 0.08379222437246735,
-      "grad_norm": 28336.75390625,
-      "learning_rate": 4.148e-05,
-      "loss": 211.0017,
-      "step": 20740
-    },
-    {
-      "epoch": 0.08383262563783497,
-      "grad_norm": 1380.7489013671875,
-      "learning_rate": 4.15e-05,
-      "loss": 72.6195,
-      "step": 20750
-    },
-    {
-      "epoch": 0.08387302690320261,
-      "grad_norm": 1030.243408203125,
-      "learning_rate": 4.152e-05,
-      "loss": 127.0354,
-      "step": 20760
-    },
-    {
-      "epoch": 0.08391342816857024,
-      "grad_norm": 7085.2041015625,
-      "learning_rate": 4.154e-05,
-      "loss": 119.6303,
-      "step": 20770
-    },
-    {
-      "epoch": 0.08395382943393787,
-      "grad_norm": 783.239013671875,
-      "learning_rate": 4.156e-05,
-      "loss": 123.3007,
-      "step": 20780
-    },
-    {
-      "epoch": 0.0839942306993055,
-      "grad_norm": 1146.0455322265625,
-      "learning_rate": 4.1580000000000005e-05,
-      "loss": 111.6334,
-      "step": 20790
-    },
-    {
-      "epoch": 0.08403463196467313,
-      "grad_norm": 3808.298583984375,
-      "learning_rate": 4.16e-05,
-      "loss": 155.187,
-      "step": 20800
-    },
-    {
-      "epoch": 0.08407503323004077,
-      "grad_norm": 2816.82275390625,
-      "learning_rate": 4.1620000000000005e-05,
-      "loss": 146.4871,
-      "step": 20810
-    },
-    {
-      "epoch": 0.0841154344954084,
-      "grad_norm": 1327.615966796875,
-      "learning_rate": 4.164e-05,
-      "loss": 99.5535,
-      "step": 20820
-    },
-    {
-      "epoch": 0.08415583576077602,
-      "grad_norm": 759.77490234375,
-      "learning_rate": 4.1660000000000004e-05,
-      "loss": 135.0864,
-      "step": 20830
-    },
-    {
-      "epoch": 0.08419623702614366,
-      "grad_norm": 1578.577392578125,
-      "learning_rate": 4.168e-05,
-      "loss": 134.005,
-      "step": 20840
-    },
-    {
-      "epoch": 0.08423663829151129,
-      "grad_norm": 845.4844970703125,
-      "learning_rate": 4.17e-05,
-      "loss": 72.3538,
-      "step": 20850
-    },
-    {
-      "epoch": 0.08427703955687892,
-      "grad_norm": 681.817626953125,
-      "learning_rate": 4.172e-05,
-      "loss": 89.2945,
-      "step": 20860
-    },
-    {
-      "epoch": 0.08431744082224656,
-      "grad_norm": 610.7493896484375,
-      "learning_rate": 4.1740000000000004e-05,
-      "loss": 88.0488,
-      "step": 20870
-    },
-    {
-      "epoch": 0.08435784208761418,
-      "grad_norm": 1551.756103515625,
-      "learning_rate": 4.176000000000001e-05,
-      "loss": 144.3311,
-      "step": 20880
-    },
-    {
-      "epoch": 0.08439824335298182,
-      "grad_norm": 941.2823486328125,
-      "learning_rate": 4.178e-05,
-      "loss": 108.5786,
-      "step": 20890
-    },
-    {
-      "epoch": 0.08443864461834945,
-      "grad_norm": 2806.956787109375,
-      "learning_rate": 4.18e-05,
-      "loss": 139.1996,
-      "step": 20900
-    },
-    {
-      "epoch": 0.08447904588371707,
-      "grad_norm": 2764.781982421875,
-      "learning_rate": 4.182e-05,
-      "loss": 202.5234,
-      "step": 20910
-    },
-    {
-      "epoch": 0.08451944714908471,
-      "grad_norm": 4884.42431640625,
-      "learning_rate": 4.184e-05,
-      "loss": 113.7894,
-      "step": 20920
-    },
-    {
-      "epoch": 0.08455984841445234,
-      "grad_norm": 752.3662719726562,
-      "learning_rate": 4.186e-05,
-      "loss": 108.6684,
-      "step": 20930
-    },
-    {
-      "epoch": 0.08460024967981997,
-      "grad_norm": 997.4862060546875,
-      "learning_rate": 4.1880000000000006e-05,
-      "loss": 119.868,
-      "step": 20940
-    },
-    {
-      "epoch": 0.0846406509451876,
-      "grad_norm": 1489.1739501953125,
-      "learning_rate": 4.19e-05,
-      "loss": 118.152,
-      "step": 20950
-    },
-    {
-      "epoch": 0.08468105221055523,
-      "grad_norm": 509.3271789550781,
-      "learning_rate": 4.1920000000000005e-05,
-      "loss": 95.0759,
-      "step": 20960
-    },
-    {
-      "epoch": 0.08472145347592287,
-      "grad_norm": 1403.126708984375,
-      "learning_rate": 4.194e-05,
-      "loss": 130.0431,
-      "step": 20970
-    },
-    {
-      "epoch": 0.0847618547412905,
-      "grad_norm": 2763.188232421875,
-      "learning_rate": 4.196e-05,
-      "loss": 82.064,
-      "step": 20980
-    },
-    {
-      "epoch": 0.08480225600665812,
-      "grad_norm": 643.8973388671875,
-      "learning_rate": 4.198e-05,
-      "loss": 120.0969,
-      "step": 20990
-    },
-    {
-      "epoch": 0.08484265727202576,
-      "grad_norm": 953.9916381835938,
-      "learning_rate": 4.2e-05,
-      "loss": 99.3082,
-      "step": 21000
-    },
-    {
-      "epoch": 0.08488305853739339,
-      "grad_norm": 7278.77392578125,
-      "learning_rate": 4.202e-05,
-      "loss": 185.1893,
-      "step": 21010
-    },
-    {
-      "epoch": 0.08492345980276102,
-      "grad_norm": 4063.332763671875,
-      "learning_rate": 4.2040000000000004e-05,
-      "loss": 172.6903,
-      "step": 21020
-    },
-    {
-      "epoch": 0.08496386106812866,
-      "grad_norm": 355.11962890625,
-      "learning_rate": 4.206e-05,
-      "loss": 97.8671,
-      "step": 21030
-    },
-    {
-      "epoch": 0.08500426233349628,
-      "grad_norm": 803.3870239257812,
-      "learning_rate": 4.2080000000000004e-05,
-      "loss": 90.9247,
-      "step": 21040
-    },
-    {
-      "epoch": 0.08504466359886392,
-      "grad_norm": 1197.3392333984375,
-      "learning_rate": 4.21e-05,
-      "loss": 111.2428,
-      "step": 21050
-    },
-    {
-      "epoch": 0.08508506486423155,
-      "grad_norm": 1060.21728515625,
-      "learning_rate": 4.212e-05,
-      "loss": 80.9763,
-      "step": 21060
-    },
-    {
-      "epoch": 0.08512546612959918,
-      "grad_norm": 553.6648559570312,
-      "learning_rate": 4.214e-05,
-      "loss": 164.6878,
-      "step": 21070
-    },
-    {
-      "epoch": 0.08516586739496682,
-      "grad_norm": 1732.870361328125,
-      "learning_rate": 4.2159999999999996e-05,
-      "loss": 127.5546,
-      "step": 21080
-    },
-    {
-      "epoch": 0.08520626866033444,
-      "grad_norm": 3130.48828125,
-      "learning_rate": 4.2180000000000006e-05,
-      "loss": 125.8488,
-      "step": 21090
-    },
-    {
-      "epoch": 0.08524666992570207,
-      "grad_norm": 1232.9683837890625,
-      "learning_rate": 4.22e-05,
-      "loss": 110.2355,
-      "step": 21100
-    },
-    {
-      "epoch": 0.08528707119106971,
-      "grad_norm": 1068.6566162109375,
-      "learning_rate": 4.2220000000000006e-05,
-      "loss": 116.9769,
-      "step": 21110
-    },
-    {
-      "epoch": 0.08532747245643733,
-      "grad_norm": 687.2569580078125,
-      "learning_rate": 4.224e-05,
-      "loss": 124.6508,
-      "step": 21120
-    },
-    {
-      "epoch": 0.08536787372180497,
-      "grad_norm": 680.8721313476562,
-      "learning_rate": 4.226e-05,
-      "loss": 123.4505,
-      "step": 21130
-    },
-    {
-      "epoch": 0.0854082749871726,
-      "grad_norm": 1629.7750244140625,
-      "learning_rate": 4.228e-05,
-      "loss": 165.3711,
-      "step": 21140
-    },
-    {
-      "epoch": 0.08544867625254023,
-      "grad_norm": 1123.4427490234375,
-      "learning_rate": 4.23e-05,
-      "loss": 104.645,
-      "step": 21150
-    },
-    {
-      "epoch": 0.08548907751790787,
-      "grad_norm": 684.88134765625,
-      "learning_rate": 4.232e-05,
-      "loss": 87.5831,
-      "step": 21160
-    },
-    {
-      "epoch": 0.08552947878327549,
-      "grad_norm": 1038.5843505859375,
-      "learning_rate": 4.2340000000000005e-05,
-      "loss": 111.1774,
-      "step": 21170
-    },
-    {
-      "epoch": 0.08556988004864312,
-      "grad_norm": 441.3775634765625,
-      "learning_rate": 4.236e-05,
-      "loss": 97.0562,
-      "step": 21180
-    },
-    {
-      "epoch": 0.08561028131401076,
-      "grad_norm": 1054.5010986328125,
-      "learning_rate": 4.2380000000000004e-05,
-      "loss": 137.0658,
-      "step": 21190
-    },
-    {
-      "epoch": 0.08565068257937838,
-      "grad_norm": 2301.26708984375,
-      "learning_rate": 4.24e-05,
-      "loss": 121.703,
-      "step": 21200
-    },
-    {
-      "epoch": 0.08569108384474602,
-      "grad_norm": 1787.3651123046875,
-      "learning_rate": 4.2420000000000004e-05,
-      "loss": 168.9554,
-      "step": 21210
-    },
-    {
-      "epoch": 0.08573148511011365,
-      "grad_norm": 3590.007568359375,
-      "learning_rate": 4.244e-05,
-      "loss": 162.3379,
-      "step": 21220
-    },
-    {
-      "epoch": 0.08577188637548128,
-      "grad_norm": 1609.46142578125,
-      "learning_rate": 4.246e-05,
-      "loss": 124.5303,
-      "step": 21230
-    },
-    {
-      "epoch": 0.08581228764084892,
-      "grad_norm": 558.2980346679688,
-      "learning_rate": 4.248e-05,
-      "loss": 95.647,
-      "step": 21240
-    },
-    {
-      "epoch": 0.08585268890621654,
-      "grad_norm": 1266.0831298828125,
-      "learning_rate": 4.25e-05,
-      "loss": 95.554,
-      "step": 21250
-    },
-    {
-      "epoch": 0.08589309017158417,
-      "grad_norm": 1427.5911865234375,
-      "learning_rate": 4.2520000000000006e-05,
-      "loss": 136.6243,
-      "step": 21260
-    },
-    {
-      "epoch": 0.08593349143695181,
-      "grad_norm": 1910.5550537109375,
-      "learning_rate": 4.254e-05,
-      "loss": 104.6323,
-      "step": 21270
-    },
-    {
-      "epoch": 0.08597389270231943,
-      "grad_norm": 2682.699951171875,
-      "learning_rate": 4.256e-05,
-      "loss": 123.6776,
-      "step": 21280
-    },
-    {
-      "epoch": 0.08601429396768706,
-      "grad_norm": 7804.3984375,
-      "learning_rate": 4.258e-05,
-      "loss": 165.5531,
-      "step": 21290
-    },
-    {
-      "epoch": 0.0860546952330547,
-      "grad_norm": 790.6614990234375,
-      "learning_rate": 4.26e-05,
-      "loss": 114.8768,
-      "step": 21300
-    },
-    {
-      "epoch": 0.08609509649842233,
-      "grad_norm": 1748.3701171875,
-      "learning_rate": 4.262e-05,
-      "loss": 120.4578,
-      "step": 21310
-    },
-    {
-      "epoch": 0.08613549776378997,
-      "grad_norm": 948.3206176757812,
-      "learning_rate": 4.2640000000000005e-05,
-      "loss": 100.5346,
-      "step": 21320
-    },
-    {
-      "epoch": 0.0861758990291576,
-      "grad_norm": 695.2459106445312,
-      "learning_rate": 4.266e-05,
-      "loss": 97.4773,
-      "step": 21330
-    },
-    {
-      "epoch": 0.08621630029452522,
-      "grad_norm": 2172.08837890625,
-      "learning_rate": 4.2680000000000005e-05,
-      "loss": 132.5773,
-      "step": 21340
-    },
-    {
-      "epoch": 0.08625670155989286,
-      "grad_norm": 867.505859375,
-      "learning_rate": 4.27e-05,
-      "loss": 104.199,
-      "step": 21350
-    },
-    {
-      "epoch": 0.08629710282526049,
-      "grad_norm": 1258.5723876953125,
-      "learning_rate": 4.2720000000000004e-05,
-      "loss": 99.1206,
-      "step": 21360
-    },
-    {
-      "epoch": 0.08633750409062811,
-      "grad_norm": 974.3964233398438,
-      "learning_rate": 4.274e-05,
-      "loss": 98.0185,
-      "step": 21370
-    },
-    {
-      "epoch": 0.08637790535599575,
-      "grad_norm": 1084.6614990234375,
-      "learning_rate": 4.276e-05,
-      "loss": 95.9101,
-      "step": 21380
-    },
-    {
-      "epoch": 0.08641830662136338,
-      "grad_norm": 1883.6488037109375,
-      "learning_rate": 4.278e-05,
-      "loss": 125.3519,
-      "step": 21390
-    },
-    {
-      "epoch": 0.08645870788673102,
-      "grad_norm": 3283.783935546875,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 166.7042,
-      "step": 21400
-    },
-    {
-      "epoch": 0.08649910915209864,
-      "grad_norm": 942.7132568359375,
-      "learning_rate": 4.282000000000001e-05,
-      "loss": 154.2377,
-      "step": 21410
-    },
-    {
-      "epoch": 0.08653951041746627,
-      "grad_norm": 1186.5281982421875,
-      "learning_rate": 4.284e-05,
-      "loss": 104.8457,
-      "step": 21420
-    },
-    {
-      "epoch": 0.08657991168283391,
-      "grad_norm": 1021.5708618164062,
-      "learning_rate": 4.286e-05,
-      "loss": 100.3719,
-      "step": 21430
-    },
-    {
-      "epoch": 0.08662031294820154,
-      "grad_norm": 1118.336669921875,
-      "learning_rate": 4.288e-05,
-      "loss": 145.7167,
-      "step": 21440
-    },
-    {
-      "epoch": 0.08666071421356916,
-      "grad_norm": 686.393310546875,
-      "learning_rate": 4.29e-05,
-      "loss": 107.4845,
-      "step": 21450
-    },
-    {
-      "epoch": 0.0867011154789368,
-      "grad_norm": 10137.349609375,
-      "learning_rate": 4.292e-05,
-      "loss": 131.5519,
-      "step": 21460
-    },
-    {
-      "epoch": 0.08674151674430443,
-      "grad_norm": 4428.59521484375,
-      "learning_rate": 4.2940000000000006e-05,
-      "loss": 163.3146,
-      "step": 21470
-    },
-    {
-      "epoch": 0.08678191800967207,
-      "grad_norm": 816.3525390625,
-      "learning_rate": 4.296e-05,
-      "loss": 69.2457,
-      "step": 21480
-    },
-    {
-      "epoch": 0.0868223192750397,
-      "grad_norm": 1754.2156982421875,
-      "learning_rate": 4.2980000000000005e-05,
-      "loss": 163.8832,
-      "step": 21490
-    },
-    {
-      "epoch": 0.08686272054040732,
-      "grad_norm": 1116.2850341796875,
-      "learning_rate": 4.3e-05,
-      "loss": 86.4771,
-      "step": 21500
-    },
-    {
-      "epoch": 0.08690312180577496,
-      "grad_norm": 542.83984375,
-      "learning_rate": 4.3020000000000005e-05,
-      "loss": 121.8374,
-      "step": 21510
-    },
-    {
-      "epoch": 0.08694352307114259,
-      "grad_norm": 1956.363037109375,
-      "learning_rate": 4.304e-05,
-      "loss": 108.1872,
-      "step": 21520
-    },
-    {
-      "epoch": 0.08698392433651021,
-      "grad_norm": 1243.8492431640625,
-      "learning_rate": 4.306e-05,
-      "loss": 115.3768,
-      "step": 21530
-    },
-    {
-      "epoch": 0.08702432560187785,
-      "grad_norm": 1239.7283935546875,
-      "learning_rate": 4.308e-05,
-      "loss": 81.8431,
-      "step": 21540
-    },
-    {
-      "epoch": 0.08706472686724548,
-      "grad_norm": 759.3988037109375,
-      "learning_rate": 4.3100000000000004e-05,
-      "loss": 128.8443,
-      "step": 21550
-    },
-    {
-      "epoch": 0.08710512813261312,
-      "grad_norm": 1724.4061279296875,
-      "learning_rate": 4.312000000000001e-05,
-      "loss": 132.2185,
-      "step": 21560
-    },
-    {
-      "epoch": 0.08714552939798074,
-      "grad_norm": 815.0044555664062,
-      "learning_rate": 4.3140000000000004e-05,
-      "loss": 76.6797,
-      "step": 21570
-    },
-    {
-      "epoch": 0.08718593066334837,
-      "grad_norm": 458.9404602050781,
-      "learning_rate": 4.316e-05,
-      "loss": 115.9924,
-      "step": 21580
-    },
-    {
-      "epoch": 0.08722633192871601,
-      "grad_norm": 553.5025024414062,
-      "learning_rate": 4.318e-05,
-      "loss": 121.1555,
-      "step": 21590
-    },
-    {
-      "epoch": 0.08726673319408364,
-      "grad_norm": 1283.981689453125,
-      "learning_rate": 4.32e-05,
-      "loss": 134.989,
-      "step": 21600
-    },
-    {
-      "epoch": 0.08730713445945126,
-      "grad_norm": 438.9365234375,
-      "learning_rate": 4.3219999999999996e-05,
-      "loss": 96.8193,
-      "step": 21610
-    },
-    {
-      "epoch": 0.0873475357248189,
-      "grad_norm": 963.8663940429688,
-      "learning_rate": 4.324e-05,
-      "loss": 109.2355,
-      "step": 21620
-    },
-    {
-      "epoch": 0.08738793699018653,
-      "grad_norm": 389.078369140625,
-      "learning_rate": 4.326e-05,
-      "loss": 179.7937,
-      "step": 21630
-    },
-    {
-      "epoch": 0.08742833825555417,
-      "grad_norm": 1283.707763671875,
-      "learning_rate": 4.3280000000000006e-05,
-      "loss": 93.5443,
-      "step": 21640
-    },
-    {
-      "epoch": 0.0874687395209218,
-      "grad_norm": 1297.939208984375,
-      "learning_rate": 4.33e-05,
-      "loss": 119.5012,
-      "step": 21650
-    },
-    {
-      "epoch": 0.08750914078628942,
-      "grad_norm": 1461.7098388671875,
-      "learning_rate": 4.332e-05,
-      "loss": 112.2618,
-      "step": 21660
-    },
-    {
-      "epoch": 0.08754954205165706,
-      "grad_norm": 370.33056640625,
-      "learning_rate": 4.334e-05,
-      "loss": 118.6215,
-      "step": 21670
-    },
-    {
-      "epoch": 0.08758994331702469,
-      "grad_norm": 1230.4083251953125,
-      "learning_rate": 4.336e-05,
-      "loss": 83.1036,
-      "step": 21680
-    },
-    {
-      "epoch": 0.08763034458239231,
-      "grad_norm": 713.216796875,
-      "learning_rate": 4.338e-05,
-      "loss": 116.6274,
-      "step": 21690
-    },
-    {
-      "epoch": 0.08767074584775995,
-      "grad_norm": 333.86663818359375,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 138.7234,
-      "step": 21700
-    },
-    {
-      "epoch": 0.08771114711312758,
-      "grad_norm": 1813.9105224609375,
-      "learning_rate": 4.342e-05,
-      "loss": 124.1437,
-      "step": 21710
-    },
-    {
-      "epoch": 0.08775154837849522,
-      "grad_norm": 1551.122802734375,
-      "learning_rate": 4.3440000000000004e-05,
-      "loss": 117.1614,
-      "step": 21720
-    },
-    {
-      "epoch": 0.08779194964386285,
-      "grad_norm": 556.9833984375,
-      "learning_rate": 4.346e-05,
-      "loss": 85.8124,
-      "step": 21730
-    },
-    {
-      "epoch": 0.08783235090923047,
-      "grad_norm": 1249.522705078125,
-      "learning_rate": 4.3480000000000004e-05,
-      "loss": 82.4492,
-      "step": 21740
-    },
-    {
-      "epoch": 0.08787275217459811,
-      "grad_norm": 644.7138671875,
-      "learning_rate": 4.35e-05,
-      "loss": 103.3423,
-      "step": 21750
-    },
-    {
-      "epoch": 0.08791315343996574,
-      "grad_norm": 893.0758056640625,
-      "learning_rate": 4.352e-05,
-      "loss": 140.0829,
-      "step": 21760
-    },
-    {
-      "epoch": 0.08795355470533336,
-      "grad_norm": 527.2681274414062,
-      "learning_rate": 4.354e-05,
-      "loss": 99.5734,
-      "step": 21770
-    },
-    {
-      "epoch": 0.087993955970701,
-      "grad_norm": 623.2860717773438,
-      "learning_rate": 4.356e-05,
-      "loss": 75.2396,
-      "step": 21780
-    },
-    {
-      "epoch": 0.08803435723606863,
-      "grad_norm": 590.4302978515625,
-      "learning_rate": 4.3580000000000006e-05,
-      "loss": 117.1983,
-      "step": 21790
-    },
-    {
-      "epoch": 0.08807475850143627,
-      "grad_norm": 945.4894409179688,
-      "learning_rate": 4.36e-05,
-      "loss": 156.6693,
-      "step": 21800
-    },
-    {
-      "epoch": 0.0881151597668039,
-      "grad_norm": 892.6826782226562,
-      "learning_rate": 4.362e-05,
-      "loss": 128.0761,
-      "step": 21810
-    },
-    {
-      "epoch": 0.08815556103217152,
-      "grad_norm": 1329.8853759765625,
-      "learning_rate": 4.364e-05,
-      "loss": 81.0603,
-      "step": 21820
-    },
-    {
-      "epoch": 0.08819596229753916,
-      "grad_norm": 3241.863525390625,
-      "learning_rate": 4.366e-05,
-      "loss": 114.7735,
-      "step": 21830
-    },
-    {
-      "epoch": 0.08823636356290679,
-      "grad_norm": 621.0053100585938,
-      "learning_rate": 4.368e-05,
-      "loss": 99.9302,
-      "step": 21840
-    },
-    {
-      "epoch": 0.08827676482827441,
-      "grad_norm": 571.9194946289062,
-      "learning_rate": 4.3700000000000005e-05,
-      "loss": 60.8999,
-      "step": 21850
-    },
-    {
-      "epoch": 0.08831716609364205,
-      "grad_norm": 1223.1873779296875,
-      "learning_rate": 4.372e-05,
-      "loss": 115.3964,
-      "step": 21860
-    },
-    {
-      "epoch": 0.08835756735900968,
-      "grad_norm": 1025.819580078125,
-      "learning_rate": 4.3740000000000005e-05,
-      "loss": 115.1883,
-      "step": 21870
-    },
-    {
-      "epoch": 0.08839796862437732,
-      "grad_norm": 1899.594970703125,
-      "learning_rate": 4.376e-05,
-      "loss": 118.3364,
-      "step": 21880
-    },
-    {
-      "epoch": 0.08843836988974495,
-      "grad_norm": 1690.5460205078125,
-      "learning_rate": 4.3780000000000004e-05,
-      "loss": 88.396,
-      "step": 21890
-    },
-    {
-      "epoch": 0.08847877115511257,
-      "grad_norm": 1255.8304443359375,
-      "learning_rate": 4.38e-05,
-      "loss": 95.6884,
-      "step": 21900
-    },
-    {
-      "epoch": 0.08851917242048021,
-      "grad_norm": 2578.5478515625,
-      "learning_rate": 4.382e-05,
-      "loss": 143.0776,
-      "step": 21910
-    },
-    {
-      "epoch": 0.08855957368584784,
-      "grad_norm": 4696.71826171875,
-      "learning_rate": 4.384e-05,
-      "loss": 95.2831,
-      "step": 21920
-    },
-    {
-      "epoch": 0.08859997495121547,
-      "grad_norm": 3200.905517578125,
-      "learning_rate": 4.3860000000000004e-05,
-      "loss": 98.7196,
-      "step": 21930
-    },
-    {
-      "epoch": 0.0886403762165831,
-      "grad_norm": 771.8877563476562,
-      "learning_rate": 4.388000000000001e-05,
-      "loss": 109.9122,
-      "step": 21940
-    },
-    {
-      "epoch": 0.08868077748195073,
-      "grad_norm": 2375.28515625,
-      "learning_rate": 4.39e-05,
-      "loss": 97.5147,
-      "step": 21950
-    },
-    {
-      "epoch": 0.08872117874731837,
-      "grad_norm": 1200.1995849609375,
-      "learning_rate": 4.392e-05,
-      "loss": 163.8434,
-      "step": 21960
-    },
-    {
-      "epoch": 0.088761580012686,
-      "grad_norm": 743.7747192382812,
-      "learning_rate": 4.394e-05,
-      "loss": 152.2877,
-      "step": 21970
-    },
-    {
-      "epoch": 0.08880198127805362,
-      "grad_norm": 748.3530883789062,
-      "learning_rate": 4.396e-05,
-      "loss": 102.9261,
-      "step": 21980
-    },
-    {
-      "epoch": 0.08884238254342126,
-      "grad_norm": 0.0,
-      "learning_rate": 4.398e-05,
-      "loss": 86.7398,
-      "step": 21990
-    },
-    {
-      "epoch": 0.08888278380878889,
-      "grad_norm": 641.658935546875,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 115.4044,
-      "step": 22000
-    },
-    {
-      "epoch": 0.08892318507415652,
-      "grad_norm": 757.2327880859375,
-      "learning_rate": 4.402e-05,
-      "loss": 106.9586,
-      "step": 22010
-    },
-    {
-      "epoch": 0.08896358633952416,
-      "grad_norm": 1741.6318359375,
-      "learning_rate": 4.4040000000000005e-05,
-      "loss": 61.3795,
-      "step": 22020
-    },
-    {
-      "epoch": 0.08900398760489178,
-      "grad_norm": 3420.192138671875,
-      "learning_rate": 4.406e-05,
-      "loss": 150.3777,
-      "step": 22030
-    },
-    {
-      "epoch": 0.08904438887025942,
-      "grad_norm": 785.1651000976562,
-      "learning_rate": 4.4080000000000005e-05,
-      "loss": 112.2344,
-      "step": 22040
-    },
-    {
-      "epoch": 0.08908479013562705,
-      "grad_norm": 2430.542236328125,
-      "learning_rate": 4.41e-05,
-      "loss": 84.5093,
-      "step": 22050
-    },
-    {
-      "epoch": 0.08912519140099467,
-      "grad_norm": 1095.4105224609375,
-      "learning_rate": 4.412e-05,
-      "loss": 129.3121,
-      "step": 22060
-    },
-    {
-      "epoch": 0.08916559266636231,
-      "grad_norm": 654.0252075195312,
-      "learning_rate": 4.414e-05,
-      "loss": 105.0435,
-      "step": 22070
-    },
-    {
-      "epoch": 0.08920599393172994,
-      "grad_norm": 6289.986328125,
-      "learning_rate": 4.4160000000000004e-05,
-      "loss": 141.9109,
-      "step": 22080
-    },
-    {
-      "epoch": 0.08924639519709757,
-      "grad_norm": 3365.2255859375,
-      "learning_rate": 4.418000000000001e-05,
-      "loss": 164.1694,
-      "step": 22090
-    },
-    {
-      "epoch": 0.0892867964624652,
-      "grad_norm": 1369.172607421875,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 133.5227,
-      "step": 22100
-    },
-    {
-      "epoch": 0.08932719772783283,
-      "grad_norm": 2439.97265625,
-      "learning_rate": 4.422e-05,
-      "loss": 126.4849,
-      "step": 22110
-    },
-    {
-      "epoch": 0.08936759899320047,
-      "grad_norm": 1391.3780517578125,
-      "learning_rate": 4.424e-05,
-      "loss": 103.1816,
-      "step": 22120
-    },
-    {
-      "epoch": 0.0894080002585681,
-      "grad_norm": 2001.207275390625,
-      "learning_rate": 4.426e-05,
-      "loss": 124.9317,
-      "step": 22130
-    },
-    {
-      "epoch": 0.08944840152393572,
-      "grad_norm": 1090.6702880859375,
-      "learning_rate": 4.428e-05,
-      "loss": 86.8185,
-      "step": 22140
-    },
-    {
-      "epoch": 0.08948880278930336,
-      "grad_norm": 1352.2353515625,
-      "learning_rate": 4.43e-05,
-      "loss": 123.4582,
-      "step": 22150
-    },
-    {
-      "epoch": 0.08952920405467099,
-      "grad_norm": 1216.326171875,
-      "learning_rate": 4.432e-05,
-      "loss": 77.8633,
-      "step": 22160
-    },
-    {
-      "epoch": 0.08956960532003862,
-      "grad_norm": 805.3572998046875,
-      "learning_rate": 4.4340000000000006e-05,
-      "loss": 112.9714,
-      "step": 22170
-    },
-    {
-      "epoch": 0.08961000658540626,
-      "grad_norm": 558.133544921875,
-      "learning_rate": 4.436e-05,
-      "loss": 89.9262,
-      "step": 22180
-    },
-    {
-      "epoch": 0.08965040785077388,
-      "grad_norm": 2130.873046875,
-      "learning_rate": 4.438e-05,
-      "loss": 112.3018,
-      "step": 22190
-    },
-    {
-      "epoch": 0.08969080911614152,
-      "grad_norm": 1761.7283935546875,
-      "learning_rate": 4.44e-05,
-      "loss": 111.0985,
-      "step": 22200
-    },
-    {
-      "epoch": 0.08973121038150915,
-      "grad_norm": 915.045166015625,
-      "learning_rate": 4.442e-05,
-      "loss": 107.6649,
-      "step": 22210
-    },
-    {
-      "epoch": 0.08977161164687678,
-      "grad_norm": 1191.688232421875,
-      "learning_rate": 4.444e-05,
-      "loss": 165.1702,
-      "step": 22220
-    },
-    {
-      "epoch": 0.08981201291224442,
-      "grad_norm": 496.5788879394531,
-      "learning_rate": 4.4460000000000005e-05,
-      "loss": 82.1613,
-      "step": 22230
-    },
-    {
-      "epoch": 0.08985241417761204,
-      "grad_norm": 500.4427185058594,
-      "learning_rate": 4.448e-05,
-      "loss": 154.052,
-      "step": 22240
-    },
-    {
-      "epoch": 0.08989281544297967,
-      "grad_norm": 1123.475341796875,
-      "learning_rate": 4.4500000000000004e-05,
-      "loss": 103.6876,
-      "step": 22250
-    },
-    {
-      "epoch": 0.08993321670834731,
-      "grad_norm": 763.8755493164062,
-      "learning_rate": 4.452e-05,
-      "loss": 105.7942,
-      "step": 22260
-    },
-    {
-      "epoch": 0.08997361797371493,
-      "grad_norm": 659.6030883789062,
-      "learning_rate": 4.4540000000000004e-05,
-      "loss": 111.4345,
-      "step": 22270
-    },
-    {
-      "epoch": 0.09001401923908257,
-      "grad_norm": 649.8712158203125,
-      "learning_rate": 4.456e-05,
-      "loss": 78.0137,
-      "step": 22280
-    },
-    {
-      "epoch": 0.0900544205044502,
-      "grad_norm": 510.55853271484375,
-      "learning_rate": 4.458e-05,
-      "loss": 142.013,
-      "step": 22290
-    },
-    {
-      "epoch": 0.09009482176981783,
-      "grad_norm": 3399.17236328125,
-      "learning_rate": 4.46e-05,
-      "loss": 159.2714,
-      "step": 22300
-    },
-    {
-      "epoch": 0.09013522303518547,
-      "grad_norm": 998.1200561523438,
-      "learning_rate": 4.462e-05,
-      "loss": 100.7152,
-      "step": 22310
-    },
-    {
-      "epoch": 0.09017562430055309,
-      "grad_norm": 531.780517578125,
-      "learning_rate": 4.4640000000000006e-05,
-      "loss": 168.3419,
-      "step": 22320
-    },
-    {
-      "epoch": 0.09021602556592072,
-      "grad_norm": 910.6287841796875,
-      "learning_rate": 4.466e-05,
-      "loss": 98.8065,
-      "step": 22330
-    },
-    {
-      "epoch": 0.09025642683128836,
-      "grad_norm": 1744.2012939453125,
-      "learning_rate": 4.468e-05,
-      "loss": 105.1678,
-      "step": 22340
-    },
-    {
-      "epoch": 0.09029682809665598,
-      "grad_norm": 597.9017944335938,
-      "learning_rate": 4.47e-05,
-      "loss": 112.0376,
-      "step": 22350
-    },
-    {
-      "epoch": 0.09033722936202362,
-      "grad_norm": 769.6776733398438,
-      "learning_rate": 4.472e-05,
-      "loss": 94.9958,
-      "step": 22360
-    },
-    {
-      "epoch": 0.09037763062739125,
-      "grad_norm": 1181.265625,
-      "learning_rate": 4.474e-05,
-      "loss": 110.0878,
-      "step": 22370
-    },
-    {
-      "epoch": 0.09041803189275888,
-      "grad_norm": 757.1105346679688,
-      "learning_rate": 4.4760000000000005e-05,
-      "loss": 100.3439,
-      "step": 22380
-    },
-    {
-      "epoch": 0.09045843315812652,
-      "grad_norm": 717.3568115234375,
-      "learning_rate": 4.478e-05,
-      "loss": 134.2481,
-      "step": 22390
-    },
-    {
-      "epoch": 0.09049883442349414,
-      "grad_norm": 892.8013305664062,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 101.8901,
-      "step": 22400
-    },
-    {
-      "epoch": 0.09053923568886177,
-      "grad_norm": 796.65380859375,
-      "learning_rate": 4.482e-05,
-      "loss": 107.8212,
-      "step": 22410
-    },
-    {
-      "epoch": 0.09057963695422941,
-      "grad_norm": 599.0886840820312,
-      "learning_rate": 4.4840000000000004e-05,
-      "loss": 112.4707,
-      "step": 22420
-    },
-    {
-      "epoch": 0.09062003821959703,
-      "grad_norm": 384.7815246582031,
-      "learning_rate": 4.486e-05,
-      "loss": 103.7676,
-      "step": 22430
-    },
-    {
-      "epoch": 0.09066043948496467,
-      "grad_norm": 0.0,
-      "learning_rate": 4.488e-05,
-      "loss": 83.413,
-      "step": 22440
-    },
-    {
-      "epoch": 0.0907008407503323,
-      "grad_norm": 1596.82373046875,
-      "learning_rate": 4.49e-05,
-      "loss": 125.9112,
-      "step": 22450
-    },
-    {
-      "epoch": 0.09074124201569993,
-      "grad_norm": 8902.1123046875,
-      "learning_rate": 4.4920000000000004e-05,
-      "loss": 158.18,
-      "step": 22460
-    },
-    {
-      "epoch": 0.09078164328106757,
-      "grad_norm": 809.1474609375,
-      "learning_rate": 4.494000000000001e-05,
-      "loss": 129.158,
-      "step": 22470
-    },
-    {
-      "epoch": 0.0908220445464352,
-      "grad_norm": 657.6831665039062,
-      "learning_rate": 4.496e-05,
-      "loss": 121.9886,
-      "step": 22480
-    },
-    {
-      "epoch": 0.09086244581180282,
-      "grad_norm": 866.2117309570312,
-      "learning_rate": 4.498e-05,
-      "loss": 123.8151,
-      "step": 22490
-    },
-    {
-      "epoch": 0.09090284707717046,
-      "grad_norm": 585.983642578125,
-      "learning_rate": 4.5e-05,
-      "loss": 74.8665,
-      "step": 22500
-    },
-    {
-      "epoch": 0.09094324834253809,
-      "grad_norm": 1273.0323486328125,
-      "learning_rate": 4.502e-05,
-      "loss": 96.7427,
-      "step": 22510
-    },
-    {
-      "epoch": 0.09098364960790573,
-      "grad_norm": 1038.0701904296875,
-      "learning_rate": 4.504e-05,
-      "loss": 85.5748,
-      "step": 22520
-    },
-    {
-      "epoch": 0.09102405087327335,
-      "grad_norm": 1610.0634765625,
-      "learning_rate": 4.506e-05,
-      "loss": 92.2955,
-      "step": 22530
-    },
-    {
-      "epoch": 0.09106445213864098,
-      "grad_norm": 2549.61572265625,
-      "learning_rate": 4.508e-05,
-      "loss": 147.1345,
-      "step": 22540
-    },
-    {
-      "epoch": 0.09110485340400862,
-      "grad_norm": 931.28173828125,
-      "learning_rate": 4.5100000000000005e-05,
-      "loss": 82.7087,
-      "step": 22550
-    },
-    {
-      "epoch": 0.09114525466937624,
-      "grad_norm": 765.2784423828125,
-      "learning_rate": 4.512e-05,
-      "loss": 135.216,
-      "step": 22560
-    },
-    {
-      "epoch": 0.09118565593474387,
-      "grad_norm": 597.312255859375,
-      "learning_rate": 4.5140000000000005e-05,
-      "loss": 147.5036,
-      "step": 22570
-    },
-    {
-      "epoch": 0.09122605720011151,
-      "grad_norm": 756.5013427734375,
-      "learning_rate": 4.516e-05,
-      "loss": 128.549,
-      "step": 22580
-    },
-    {
-      "epoch": 0.09126645846547914,
-      "grad_norm": 719.6322631835938,
-      "learning_rate": 4.518e-05,
-      "loss": 80.9833,
-      "step": 22590
-    },
-    {
-      "epoch": 0.09130685973084678,
-      "grad_norm": 721.7781982421875,
-      "learning_rate": 4.52e-05,
-      "loss": 144.421,
-      "step": 22600
-    },
-    {
-      "epoch": 0.0913472609962144,
-      "grad_norm": 1193.63232421875,
-      "learning_rate": 4.5220000000000004e-05,
-      "loss": 99.2124,
-      "step": 22610
-    },
-    {
-      "epoch": 0.09138766226158203,
-      "grad_norm": 886.3824462890625,
-      "learning_rate": 4.524000000000001e-05,
-      "loss": 117.6559,
-      "step": 22620
-    },
-    {
-      "epoch": 0.09142806352694967,
-      "grad_norm": 869.9031372070312,
-      "learning_rate": 4.5260000000000004e-05,
-      "loss": 128.7948,
-      "step": 22630
-    },
-    {
-      "epoch": 0.0914684647923173,
-      "grad_norm": 323.6915283203125,
-      "learning_rate": 4.528e-05,
-      "loss": 103.8211,
-      "step": 22640
-    },
-    {
-      "epoch": 0.09150886605768492,
-      "grad_norm": 648.2799072265625,
-      "learning_rate": 4.53e-05,
-      "loss": 124.0231,
-      "step": 22650
-    },
-    {
-      "epoch": 0.09154926732305256,
-      "grad_norm": 3163.934814453125,
-      "learning_rate": 4.532e-05,
-      "loss": 133.5165,
-      "step": 22660
-    },
-    {
-      "epoch": 0.09158966858842019,
-      "grad_norm": 694.5257568359375,
-      "learning_rate": 4.534e-05,
-      "loss": 117.9862,
-      "step": 22670
-    },
-    {
-      "epoch": 0.09163006985378783,
-      "grad_norm": 1048.819091796875,
-      "learning_rate": 4.536e-05,
-      "loss": 110.8099,
-      "step": 22680
-    },
-    {
-      "epoch": 0.09167047111915545,
-      "grad_norm": 760.2977905273438,
-      "learning_rate": 4.538e-05,
-      "loss": 93.8862,
-      "step": 22690
-    },
-    {
-      "epoch": 0.09171087238452308,
-      "grad_norm": 1749.94580078125,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 141.0014,
-      "step": 22700
-    },
-    {
-      "epoch": 0.09175127364989072,
-      "grad_norm": 0.0,
-      "learning_rate": 4.542e-05,
-      "loss": 100.4336,
-      "step": 22710
-    },
-    {
-      "epoch": 0.09179167491525834,
-      "grad_norm": 520.8677978515625,
-      "learning_rate": 4.5440000000000005e-05,
-      "loss": 85.8024,
-      "step": 22720
-    },
-    {
-      "epoch": 0.09183207618062597,
-      "grad_norm": 1674.249755859375,
-      "learning_rate": 4.546e-05,
-      "loss": 138.2154,
-      "step": 22730
-    },
-    {
-      "epoch": 0.09187247744599361,
-      "grad_norm": 917.3933715820312,
-      "learning_rate": 4.548e-05,
-      "loss": 103.3672,
-      "step": 22740
-    },
-    {
-      "epoch": 0.09191287871136124,
-      "grad_norm": 1019.2123413085938,
-      "learning_rate": 4.55e-05,
-      "loss": 56.5289,
-      "step": 22750
-    },
-    {
-      "epoch": 0.09195327997672888,
-      "grad_norm": 2317.489501953125,
-      "learning_rate": 4.5520000000000005e-05,
-      "loss": 155.0266,
-      "step": 22760
-    },
-    {
-      "epoch": 0.0919936812420965,
-      "grad_norm": 819.0970458984375,
-      "learning_rate": 4.554000000000001e-05,
-      "loss": 116.7189,
-      "step": 22770
-    },
-    {
-      "epoch": 0.09203408250746413,
-      "grad_norm": 1225.271240234375,
-      "learning_rate": 4.5560000000000004e-05,
-      "loss": 111.6631,
-      "step": 22780
-    },
-    {
-      "epoch": 0.09207448377283177,
-      "grad_norm": 1159.3927001953125,
-      "learning_rate": 4.558e-05,
-      "loss": 66.4148,
-      "step": 22790
-    },
-    {
-      "epoch": 0.0921148850381994,
-      "grad_norm": 1292.975341796875,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 105.9983,
-      "step": 22800
-    },
-    {
-      "epoch": 0.09215528630356702,
-      "grad_norm": 1175.0252685546875,
-      "learning_rate": 4.562e-05,
-      "loss": 143.92,
-      "step": 22810
-    },
-    {
-      "epoch": 0.09219568756893466,
-      "grad_norm": 759.6773681640625,
-      "learning_rate": 4.564e-05,
-      "loss": 90.7138,
-      "step": 22820
-    },
-    {
-      "epoch": 0.09223608883430229,
-      "grad_norm": 857.7040405273438,
-      "learning_rate": 4.566e-05,
-      "loss": 92.1247,
-      "step": 22830
-    },
-    {
-      "epoch": 0.09227649009966993,
-      "grad_norm": 1580.8248291015625,
-      "learning_rate": 4.568e-05,
-      "loss": 128.0727,
-      "step": 22840
-    },
-    {
-      "epoch": 0.09231689136503755,
-      "grad_norm": 567.8139038085938,
-      "learning_rate": 4.5700000000000006e-05,
-      "loss": 97.4495,
-      "step": 22850
-    },
-    {
-      "epoch": 0.09235729263040518,
-      "grad_norm": 1576.912841796875,
-      "learning_rate": 4.572e-05,
-      "loss": 128.2519,
-      "step": 22860
-    },
-    {
-      "epoch": 0.09239769389577282,
-      "grad_norm": 716.1073608398438,
-      "learning_rate": 4.574e-05,
-      "loss": 94.8568,
-      "step": 22870
-    },
-    {
-      "epoch": 0.09243809516114045,
-      "grad_norm": 638.1775512695312,
-      "learning_rate": 4.576e-05,
-      "loss": 66.2251,
-      "step": 22880
-    },
-    {
-      "epoch": 0.09247849642650807,
-      "grad_norm": 2102.469482421875,
-      "learning_rate": 4.578e-05,
-      "loss": 128.2732,
-      "step": 22890
-    },
-    {
-      "epoch": 0.09251889769187571,
-      "grad_norm": 682.6776123046875,
-      "learning_rate": 4.58e-05,
-      "loss": 102.548,
-      "step": 22900
-    },
-    {
-      "epoch": 0.09255929895724334,
-      "grad_norm": 586.2388916015625,
-      "learning_rate": 4.5820000000000005e-05,
-      "loss": 148.1805,
-      "step": 22910
-    },
-    {
-      "epoch": 0.09259970022261098,
-      "grad_norm": 1975.264404296875,
-      "learning_rate": 4.584e-05,
-      "loss": 143.5229,
-      "step": 22920
-    },
-    {
-      "epoch": 0.0926401014879786,
-      "grad_norm": 758.1207885742188,
-      "learning_rate": 4.5860000000000005e-05,
-      "loss": 128.5342,
-      "step": 22930
-    },
-    {
-      "epoch": 0.09268050275334623,
-      "grad_norm": 2084.8974609375,
-      "learning_rate": 4.588e-05,
-      "loss": 94.417,
-      "step": 22940
-    },
-    {
-      "epoch": 0.09272090401871387,
-      "grad_norm": 830.3482055664062,
-      "learning_rate": 4.5900000000000004e-05,
-      "loss": 97.3431,
-      "step": 22950
-    },
-    {
-      "epoch": 0.0927613052840815,
-      "grad_norm": 562.5604858398438,
-      "learning_rate": 4.592e-05,
-      "loss": 115.6859,
-      "step": 22960
-    },
-    {
-      "epoch": 0.09280170654944912,
-      "grad_norm": 1079.0665283203125,
-      "learning_rate": 4.594e-05,
-      "loss": 118.9365,
-      "step": 22970
-    },
-    {
-      "epoch": 0.09284210781481676,
-      "grad_norm": 1732.04638671875,
-      "learning_rate": 4.596e-05,
-      "loss": 86.5984,
-      "step": 22980
-    },
-    {
-      "epoch": 0.09288250908018439,
-      "grad_norm": 1780.00537109375,
-      "learning_rate": 4.5980000000000004e-05,
-      "loss": 113.6839,
-      "step": 22990
-    },
-    {
-      "epoch": 0.09292291034555203,
-      "grad_norm": 1192.3056640625,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 89.4596,
-      "step": 23000
-    },
-    {
-      "epoch": 0.09296331161091966,
-      "grad_norm": 1134.75830078125,
-      "learning_rate": 4.602e-05,
-      "loss": 87.713,
-      "step": 23010
-    },
-    {
-      "epoch": 0.09300371287628728,
-      "grad_norm": 1336.7357177734375,
-      "learning_rate": 4.604e-05,
-      "loss": 65.8475,
-      "step": 23020
-    },
-    {
-      "epoch": 0.09304411414165492,
-      "grad_norm": 815.8203125,
-      "learning_rate": 4.606e-05,
-      "loss": 148.7902,
-      "step": 23030
-    },
-    {
-      "epoch": 0.09308451540702255,
-      "grad_norm": 1082.975830078125,
-      "learning_rate": 4.608e-05,
-      "loss": 114.2601,
-      "step": 23040
-    },
-    {
-      "epoch": 0.09312491667239017,
-      "grad_norm": 778.52587890625,
-      "learning_rate": 4.61e-05,
-      "loss": 284.198,
-      "step": 23050
-    },
-    {
-      "epoch": 0.09316531793775781,
-      "grad_norm": 591.0042114257812,
-      "learning_rate": 4.612e-05,
-      "loss": 106.8146,
-      "step": 23060
-    },
-    {
-      "epoch": 0.09320571920312544,
-      "grad_norm": 2606.027099609375,
-      "learning_rate": 4.614e-05,
-      "loss": 109.8302,
-      "step": 23070
-    },
-    {
-      "epoch": 0.09324612046849308,
-      "grad_norm": 1184.9754638671875,
-      "learning_rate": 4.6160000000000005e-05,
-      "loss": 118.0777,
-      "step": 23080
-    },
-    {
-      "epoch": 0.0932865217338607,
-      "grad_norm": 1047.3983154296875,
-      "learning_rate": 4.618e-05,
-      "loss": 119.2703,
-      "step": 23090
-    },
-    {
-      "epoch": 0.09332692299922833,
-      "grad_norm": 522.4287719726562,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 115.0015,
-      "step": 23100
-    },
-    {
-      "epoch": 0.09336732426459597,
-      "grad_norm": 387.2696838378906,
-      "learning_rate": 4.622e-05,
-      "loss": 112.657,
-      "step": 23110
-    },
-    {
-      "epoch": 0.0934077255299636,
-      "grad_norm": 680.420166015625,
-      "learning_rate": 4.624e-05,
-      "loss": 81.0697,
-      "step": 23120
-    },
-    {
-      "epoch": 0.09344812679533122,
-      "grad_norm": 1640.8831787109375,
-      "learning_rate": 4.626e-05,
-      "loss": 128.7615,
-      "step": 23130
-    },
-    {
-      "epoch": 0.09348852806069886,
-      "grad_norm": 666.759765625,
-      "learning_rate": 4.6280000000000004e-05,
-      "loss": 86.7895,
-      "step": 23140
-    },
-    {
-      "epoch": 0.09352892932606649,
-      "grad_norm": 2630.477783203125,
-      "learning_rate": 4.630000000000001e-05,
-      "loss": 174.0495,
-      "step": 23150
-    },
-    {
-      "epoch": 0.09356933059143413,
-      "grad_norm": 2171.40185546875,
-      "learning_rate": 4.6320000000000004e-05,
-      "loss": 126.4433,
-      "step": 23160
-    },
-    {
-      "epoch": 0.09360973185680176,
-      "grad_norm": 952.8081665039062,
-      "learning_rate": 4.634e-05,
-      "loss": 60.8379,
-      "step": 23170
-    },
-    {
-      "epoch": 0.09365013312216938,
-      "grad_norm": 1052.6181640625,
-      "learning_rate": 4.636e-05,
-      "loss": 155.2496,
-      "step": 23180
-    },
-    {
-      "epoch": 0.09369053438753702,
-      "grad_norm": 1012.393310546875,
-      "learning_rate": 4.638e-05,
-      "loss": 137.2208,
-      "step": 23190
-    },
-    {
-      "epoch": 0.09373093565290465,
-      "grad_norm": 1177.2757568359375,
-      "learning_rate": 4.64e-05,
-      "loss": 123.9954,
-      "step": 23200
-    },
-    {
-      "epoch": 0.09377133691827227,
-      "grad_norm": 1205.6624755859375,
-      "learning_rate": 4.642e-05,
-      "loss": 133.0356,
-      "step": 23210
-    },
-    {
-      "epoch": 0.09381173818363991,
-      "grad_norm": 790.0210571289062,
-      "learning_rate": 4.644e-05,
-      "loss": 134.6905,
-      "step": 23220
-    },
-    {
-      "epoch": 0.09385213944900754,
-      "grad_norm": 1255.626953125,
-      "learning_rate": 4.6460000000000006e-05,
-      "loss": 182.1293,
-      "step": 23230
-    },
-    {
-      "epoch": 0.09389254071437518,
-      "grad_norm": 705.6256103515625,
-      "learning_rate": 4.648e-05,
-      "loss": 80.2685,
-      "step": 23240
-    },
-    {
-      "epoch": 0.0939329419797428,
-      "grad_norm": 1400.9781494140625,
-      "learning_rate": 4.6500000000000005e-05,
-      "loss": 65.442,
-      "step": 23250
-    },
-    {
-      "epoch": 0.09397334324511043,
-      "grad_norm": 475.5975646972656,
-      "learning_rate": 4.652e-05,
-      "loss": 104.4814,
-      "step": 23260
-    },
-    {
-      "epoch": 0.09401374451047807,
-      "grad_norm": 656.1465454101562,
-      "learning_rate": 4.654e-05,
-      "loss": 64.6497,
-      "step": 23270
-    },
-    {
-      "epoch": 0.0940541457758457,
-      "grad_norm": 870.1063232421875,
-      "learning_rate": 4.656e-05,
-      "loss": 106.1441,
-      "step": 23280
-    },
-    {
-      "epoch": 0.09409454704121333,
-      "grad_norm": 461.961181640625,
-      "learning_rate": 4.6580000000000005e-05,
-      "loss": 77.1417,
-      "step": 23290
-    },
-    {
-      "epoch": 0.09413494830658097,
-      "grad_norm": 677.691162109375,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 99.122,
-      "step": 23300
-    },
-    {
-      "epoch": 0.09417534957194859,
-      "grad_norm": 651.6929931640625,
-      "learning_rate": 4.6620000000000004e-05,
-      "loss": 80.5311,
-      "step": 23310
-    },
-    {
-      "epoch": 0.09421575083731623,
-      "grad_norm": 4944.45703125,
-      "learning_rate": 4.664e-05,
-      "loss": 122.0651,
-      "step": 23320
-    },
-    {
-      "epoch": 0.09425615210268386,
-      "grad_norm": 585.9901733398438,
-      "learning_rate": 4.6660000000000004e-05,
-      "loss": 105.5626,
-      "step": 23330
-    },
-    {
-      "epoch": 0.09429655336805148,
-      "grad_norm": 869.3453979492188,
-      "learning_rate": 4.668e-05,
-      "loss": 107.5253,
-      "step": 23340
-    },
-    {
-      "epoch": 0.09433695463341912,
-      "grad_norm": 1403.1729736328125,
-      "learning_rate": 4.6700000000000003e-05,
-      "loss": 82.5326,
-      "step": 23350
-    },
-    {
-      "epoch": 0.09437735589878675,
-      "grad_norm": 1053.485595703125,
-      "learning_rate": 4.672e-05,
-      "loss": 110.7292,
-      "step": 23360
-    },
-    {
-      "epoch": 0.09441775716415438,
-      "grad_norm": 495.44793701171875,
-      "learning_rate": 4.674e-05,
-      "loss": 85.2537,
-      "step": 23370
-    },
-    {
-      "epoch": 0.09445815842952202,
-      "grad_norm": 1976.839599609375,
-      "learning_rate": 4.6760000000000006e-05,
-      "loss": 122.28,
-      "step": 23380
-    },
-    {
-      "epoch": 0.09449855969488964,
-      "grad_norm": 840.1267700195312,
-      "learning_rate": 4.678e-05,
-      "loss": 61.5691,
-      "step": 23390
-    },
-    {
-      "epoch": 0.09453896096025728,
-      "grad_norm": 1114.9774169921875,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 179.376,
-      "step": 23400
-    },
-    {
-      "epoch": 0.09457936222562491,
-      "grad_norm": 2026.411376953125,
-      "learning_rate": 4.682e-05,
-      "loss": 117.7227,
-      "step": 23410
-    },
-    {
-      "epoch": 0.09461976349099253,
-      "grad_norm": 1031.648193359375,
-      "learning_rate": 4.684e-05,
-      "loss": 95.3115,
-      "step": 23420
-    },
-    {
-      "epoch": 0.09466016475636017,
-      "grad_norm": 774.4593505859375,
-      "learning_rate": 4.686e-05,
-      "loss": 74.8853,
-      "step": 23430
-    },
-    {
-      "epoch": 0.0947005660217278,
-      "grad_norm": 799.5897216796875,
-      "learning_rate": 4.688e-05,
-      "loss": 106.8804,
-      "step": 23440
-    },
-    {
-      "epoch": 0.09474096728709543,
-      "grad_norm": 801.3352661132812,
-      "learning_rate": 4.69e-05,
-      "loss": 117.8924,
-      "step": 23450
-    },
-    {
-      "epoch": 0.09478136855246307,
-      "grad_norm": 562.31494140625,
-      "learning_rate": 4.6920000000000005e-05,
-      "loss": 70.9862,
-      "step": 23460
-    },
-    {
-      "epoch": 0.09482176981783069,
-      "grad_norm": 1088.81103515625,
-      "learning_rate": 4.694e-05,
-      "loss": 105.8252,
-      "step": 23470
-    },
-    {
-      "epoch": 0.09486217108319833,
-      "grad_norm": 773.8255004882812,
-      "learning_rate": 4.6960000000000004e-05,
-      "loss": 74.9372,
-      "step": 23480
-    },
-    {
-      "epoch": 0.09490257234856596,
-      "grad_norm": 2185.90771484375,
-      "learning_rate": 4.698e-05,
-      "loss": 137.0428,
-      "step": 23490
-    },
-    {
-      "epoch": 0.09494297361393358,
-      "grad_norm": 1338.7786865234375,
-      "learning_rate": 4.7e-05,
-      "loss": 138.3532,
-      "step": 23500
-    },
-    {
-      "epoch": 0.09498337487930122,
-      "grad_norm": 828.383544921875,
-      "learning_rate": 4.702e-05,
-      "loss": 103.3043,
-      "step": 23510
-    },
-    {
-      "epoch": 0.09502377614466885,
-      "grad_norm": 1320.6363525390625,
-      "learning_rate": 4.7040000000000004e-05,
-      "loss": 112.5485,
-      "step": 23520
-    },
-    {
-      "epoch": 0.09506417741003648,
-      "grad_norm": 1088.3404541015625,
-      "learning_rate": 4.706000000000001e-05,
-      "loss": 107.0843,
-      "step": 23530
-    },
-    {
-      "epoch": 0.09510457867540412,
-      "grad_norm": 697.0968627929688,
-      "learning_rate": 4.708e-05,
-      "loss": 86.7037,
-      "step": 23540
-    },
-    {
-      "epoch": 0.09514497994077174,
-      "grad_norm": 1503.7210693359375,
-      "learning_rate": 4.71e-05,
-      "loss": 113.7888,
-      "step": 23550
-    },
-    {
-      "epoch": 0.09518538120613938,
-      "grad_norm": 840.5352783203125,
-      "learning_rate": 4.712e-05,
-      "loss": 107.902,
-      "step": 23560
-    },
-    {
-      "epoch": 0.09522578247150701,
-      "grad_norm": 1638.9781494140625,
-      "learning_rate": 4.714e-05,
-      "loss": 102.9792,
-      "step": 23570
-    },
-    {
-      "epoch": 0.09526618373687464,
-      "grad_norm": 518.2499389648438,
-      "learning_rate": 4.716e-05,
-      "loss": 78.3873,
-      "step": 23580
-    },
-    {
-      "epoch": 0.09530658500224228,
-      "grad_norm": 837.2974853515625,
-      "learning_rate": 4.718e-05,
-      "loss": 71.646,
-      "step": 23590
-    },
-    {
-      "epoch": 0.0953469862676099,
-      "grad_norm": 560.1998291015625,
-      "learning_rate": 4.72e-05,
-      "loss": 131.5562,
-      "step": 23600
-    },
-    {
-      "epoch": 0.09538738753297753,
-      "grad_norm": 789.7793579101562,
-      "learning_rate": 4.7220000000000005e-05,
-      "loss": 95.4459,
-      "step": 23610
-    },
-    {
-      "epoch": 0.09542778879834517,
-      "grad_norm": 836.2322998046875,
-      "learning_rate": 4.724e-05,
-      "loss": 118.2009,
-      "step": 23620
-    },
-    {
-      "epoch": 0.0954681900637128,
-      "grad_norm": 1441.4818115234375,
-      "learning_rate": 4.7260000000000005e-05,
-      "loss": 164.6346,
-      "step": 23630
-    },
-    {
-      "epoch": 0.09550859132908043,
-      "grad_norm": 965.8220825195312,
-      "learning_rate": 4.728e-05,
-      "loss": 109.1539,
-      "step": 23640
-    },
-    {
-      "epoch": 0.09554899259444806,
-      "grad_norm": 1859.1063232421875,
-      "learning_rate": 4.73e-05,
-      "loss": 180.398,
-      "step": 23650
-    },
-    {
-      "epoch": 0.09558939385981569,
-      "grad_norm": 823.5313110351562,
-      "learning_rate": 4.732e-05,
-      "loss": 95.1897,
-      "step": 23660
-    },
-    {
-      "epoch": 0.09562979512518333,
-      "grad_norm": 880.113037109375,
-      "learning_rate": 4.7340000000000004e-05,
-      "loss": 107.1814,
-      "step": 23670
-    },
-    {
-      "epoch": 0.09567019639055095,
-      "grad_norm": 1045.581298828125,
-      "learning_rate": 4.736000000000001e-05,
-      "loss": 151.5835,
-      "step": 23680
-    },
-    {
-      "epoch": 0.09571059765591858,
-      "grad_norm": 1099.427978515625,
-      "learning_rate": 4.7380000000000004e-05,
-      "loss": 93.5232,
-      "step": 23690
-    },
-    {
-      "epoch": 0.09575099892128622,
-      "grad_norm": 640.9761962890625,
-      "learning_rate": 4.74e-05,
-      "loss": 117.26,
-      "step": 23700
-    },
-    {
-      "epoch": 0.09579140018665384,
-      "grad_norm": 337.58795166015625,
-      "learning_rate": 4.742e-05,
-      "loss": 114.4822,
-      "step": 23710
-    },
-    {
-      "epoch": 0.09583180145202148,
-      "grad_norm": 1089.921875,
-      "learning_rate": 4.744e-05,
-      "loss": 104.9418,
-      "step": 23720
-    },
-    {
-      "epoch": 0.09587220271738911,
-      "grad_norm": 2395.615234375,
-      "learning_rate": 4.746e-05,
-      "loss": 99.8505,
-      "step": 23730
-    },
-    {
-      "epoch": 0.09591260398275674,
-      "grad_norm": 1333.4490966796875,
-      "learning_rate": 4.748e-05,
-      "loss": 136.7725,
-      "step": 23740
-    },
-    {
-      "epoch": 0.09595300524812438,
-      "grad_norm": 1200.8509521484375,
-      "learning_rate": 4.75e-05,
-      "loss": 108.3065,
-      "step": 23750
-    },
-    {
-      "epoch": 0.095993406513492,
-      "grad_norm": 674.5408325195312,
-      "learning_rate": 4.7520000000000006e-05,
-      "loss": 104.4726,
-      "step": 23760
-    },
-    {
-      "epoch": 0.09603380777885963,
-      "grad_norm": 789.4915161132812,
-      "learning_rate": 4.754e-05,
-      "loss": 145.3513,
-      "step": 23770
-    },
-    {
-      "epoch": 0.09607420904422727,
-      "grad_norm": 797.5672607421875,
-      "learning_rate": 4.7560000000000005e-05,
-      "loss": 99.0135,
-      "step": 23780
-    },
-    {
-      "epoch": 0.0961146103095949,
-      "grad_norm": 1080.6739501953125,
-      "learning_rate": 4.758e-05,
-      "loss": 128.2187,
-      "step": 23790
-    },
-    {
-      "epoch": 0.09615501157496253,
-      "grad_norm": 1180.974365234375,
-      "learning_rate": 4.76e-05,
-      "loss": 112.952,
-      "step": 23800
-    },
-    {
-      "epoch": 0.09619541284033016,
-      "grad_norm": 782.664794921875,
-      "learning_rate": 4.762e-05,
-      "loss": 88.9206,
-      "step": 23810
-    },
-    {
-      "epoch": 0.09623581410569779,
-      "grad_norm": 731.3271484375,
-      "learning_rate": 4.7640000000000005e-05,
-      "loss": 109.8621,
-      "step": 23820
-    },
-    {
-      "epoch": 0.09627621537106543,
-      "grad_norm": 1363.64501953125,
-      "learning_rate": 4.766000000000001e-05,
-      "loss": 151.2704,
-      "step": 23830
-    },
-    {
-      "epoch": 0.09631661663643305,
-      "grad_norm": 802.5873413085938,
-      "learning_rate": 4.7680000000000004e-05,
-      "loss": 126.6234,
-      "step": 23840
-    },
-    {
-      "epoch": 0.09635701790180068,
-      "grad_norm": 1359.4644775390625,
-      "learning_rate": 4.77e-05,
-      "loss": 73.3318,
-      "step": 23850
-    },
-    {
-      "epoch": 0.09639741916716832,
-      "grad_norm": 733.4736328125,
-      "learning_rate": 4.7720000000000004e-05,
-      "loss": 115.1329,
-      "step": 23860
-    },
-    {
-      "epoch": 0.09643782043253595,
-      "grad_norm": 620.9826049804688,
-      "learning_rate": 4.774e-05,
-      "loss": 86.756,
-      "step": 23870
-    },
-    {
-      "epoch": 0.09647822169790359,
-      "grad_norm": 508.6841125488281,
-      "learning_rate": 4.7760000000000004e-05,
-      "loss": 90.5046,
-      "step": 23880
-    },
-    {
-      "epoch": 0.09651862296327121,
-      "grad_norm": 756.0390014648438,
-      "learning_rate": 4.778e-05,
-      "loss": 125.9883,
-      "step": 23890
-    },
-    {
-      "epoch": 0.09655902422863884,
-      "grad_norm": 531.0929565429688,
-      "learning_rate": 4.78e-05,
-      "loss": 152.4438,
-      "step": 23900
-    },
-    {
-      "epoch": 0.09659942549400648,
-      "grad_norm": 2243.71630859375,
-      "learning_rate": 4.7820000000000006e-05,
-      "loss": 155.722,
-      "step": 23910
-    },
-    {
-      "epoch": 0.0966398267593741,
-      "grad_norm": 1023.2913818359375,
-      "learning_rate": 4.784e-05,
-      "loss": 76.3717,
-      "step": 23920
-    },
-    {
-      "epoch": 0.09668022802474173,
-      "grad_norm": 896.178955078125,
-      "learning_rate": 4.7860000000000006e-05,
-      "loss": 114.2663,
-      "step": 23930
-    },
-    {
-      "epoch": 0.09672062929010937,
-      "grad_norm": 656.4266967773438,
-      "learning_rate": 4.788e-05,
-      "loss": 122.8116,
-      "step": 23940
-    },
-    {
-      "epoch": 0.096761030555477,
-      "grad_norm": 1333.90234375,
-      "learning_rate": 4.79e-05,
-      "loss": 119.1918,
-      "step": 23950
-    },
-    {
-      "epoch": 0.09680143182084464,
-      "grad_norm": 929.0885620117188,
-      "learning_rate": 4.792e-05,
-      "loss": 101.0765,
-      "step": 23960
-    },
-    {
-      "epoch": 0.09684183308621226,
-      "grad_norm": 783.88134765625,
-      "learning_rate": 4.794e-05,
-      "loss": 123.106,
-      "step": 23970
-    },
-    {
-      "epoch": 0.09688223435157989,
-      "grad_norm": 495.6070861816406,
-      "learning_rate": 4.796e-05,
-      "loss": 77.067,
-      "step": 23980
-    },
-    {
-      "epoch": 0.09692263561694753,
-      "grad_norm": 757.9783325195312,
-      "learning_rate": 4.7980000000000005e-05,
-      "loss": 108.6014,
-      "step": 23990
-    },
-    {
-      "epoch": 0.09696303688231515,
-      "grad_norm": 1557.5858154296875,
-      "learning_rate": 4.8e-05,
-      "loss": 132.5803,
-      "step": 24000
-    },
-    {
-      "epoch": 0.09700343814768278,
-      "grad_norm": 1372.1710205078125,
-      "learning_rate": 4.8020000000000004e-05,
-      "loss": 140.0741,
-      "step": 24010
-    },
-    {
-      "epoch": 0.09704383941305042,
-      "grad_norm": 708.5230712890625,
-      "learning_rate": 4.804e-05,
-      "loss": 136.9571,
-      "step": 24020
-    },
-    {
-      "epoch": 0.09708424067841805,
-      "grad_norm": 949.0543823242188,
-      "learning_rate": 4.8060000000000004e-05,
-      "loss": 119.4232,
-      "step": 24030
-    },
-    {
-      "epoch": 0.09712464194378569,
-      "grad_norm": 779.013916015625,
-      "learning_rate": 4.808e-05,
-      "loss": 132.5883,
-      "step": 24040
-    },
-    {
-      "epoch": 0.09716504320915331,
-      "grad_norm": 1494.10595703125,
-      "learning_rate": 4.8100000000000004e-05,
-      "loss": 101.6414,
-      "step": 24050
-    },
-    {
-      "epoch": 0.09720544447452094,
-      "grad_norm": 628.3251342773438,
-      "learning_rate": 4.812000000000001e-05,
-      "loss": 100.0648,
-      "step": 24060
-    },
-    {
-      "epoch": 0.09724584573988858,
-      "grad_norm": 518.7921752929688,
-      "learning_rate": 4.814e-05,
-      "loss": 55.6692,
-      "step": 24070
-    },
-    {
-      "epoch": 0.0972862470052562,
-      "grad_norm": 1588.790771484375,
-      "learning_rate": 4.816e-05,
-      "loss": 133.6212,
-      "step": 24080
-    },
-    {
-      "epoch": 0.09732664827062383,
-      "grad_norm": 843.473388671875,
-      "learning_rate": 4.818e-05,
-      "loss": 87.3461,
-      "step": 24090
-    },
-    {
-      "epoch": 0.09736704953599147,
-      "grad_norm": 659.4796752929688,
-      "learning_rate": 4.82e-05,
-      "loss": 117.1845,
-      "step": 24100
-    },
-    {
-      "epoch": 0.0974074508013591,
-      "grad_norm": 584.1058959960938,
-      "learning_rate": 4.822e-05,
-      "loss": 103.56,
-      "step": 24110
-    },
-    {
-      "epoch": 0.09744785206672674,
-      "grad_norm": 1724.538330078125,
-      "learning_rate": 4.824e-05,
-      "loss": 89.6239,
-      "step": 24120
-    },
-    {
-      "epoch": 0.09748825333209436,
-      "grad_norm": 693.462646484375,
-      "learning_rate": 4.826e-05,
-      "loss": 93.3504,
-      "step": 24130
-    },
-    {
-      "epoch": 0.09752865459746199,
-      "grad_norm": 1428.2559814453125,
-      "learning_rate": 4.8280000000000005e-05,
-      "loss": 112.4596,
-      "step": 24140
-    },
-    {
-      "epoch": 0.09756905586282963,
-      "grad_norm": 1738.0125732421875,
-      "learning_rate": 4.83e-05,
-      "loss": 107.2036,
-      "step": 24150
-    },
-    {
-      "epoch": 0.09760945712819726,
-      "grad_norm": 590.5738525390625,
-      "learning_rate": 4.8320000000000005e-05,
-      "loss": 95.526,
-      "step": 24160
-    },
-    {
-      "epoch": 0.09764985839356488,
-      "grad_norm": 1145.53515625,
-      "learning_rate": 4.834e-05,
-      "loss": 82.7261,
-      "step": 24170
-    },
-    {
-      "epoch": 0.09769025965893252,
-      "grad_norm": 1240.5450439453125,
-      "learning_rate": 4.836e-05,
-      "loss": 138.7817,
-      "step": 24180
-    },
-    {
-      "epoch": 0.09773066092430015,
-      "grad_norm": 1606.5479736328125,
-      "learning_rate": 4.838e-05,
-      "loss": 101.9048,
-      "step": 24190
-    },
-    {
-      "epoch": 0.09777106218966779,
-      "grad_norm": 1542.466796875,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 106.2202,
-      "step": 24200
-    },
-    {
-      "epoch": 0.09781146345503541,
-      "grad_norm": 889.3588256835938,
-      "learning_rate": 4.842000000000001e-05,
-      "loss": 144.4749,
-      "step": 24210
-    },
-    {
-      "epoch": 0.09785186472040304,
-      "grad_norm": 639.1673583984375,
-      "learning_rate": 4.8440000000000004e-05,
-      "loss": 103.4213,
-      "step": 24220
-    },
-    {
-      "epoch": 0.09789226598577068,
-      "grad_norm": 3171.987060546875,
-      "learning_rate": 4.846e-05,
-      "loss": 156.6668,
-      "step": 24230
-    },
-    {
-      "epoch": 0.0979326672511383,
-      "grad_norm": 709.1254272460938,
-      "learning_rate": 4.8480000000000003e-05,
-      "loss": 80.9064,
-      "step": 24240
-    },
-    {
-      "epoch": 0.09797306851650593,
-      "grad_norm": 561.7057495117188,
-      "learning_rate": 4.85e-05,
-      "loss": 171.3163,
-      "step": 24250
-    },
-    {
-      "epoch": 0.09801346978187357,
-      "grad_norm": 929.6841430664062,
-      "learning_rate": 4.852e-05,
-      "loss": 157.7511,
-      "step": 24260
-    },
-    {
-      "epoch": 0.0980538710472412,
-      "grad_norm": 1814.5821533203125,
-      "learning_rate": 4.854e-05,
-      "loss": 91.1438,
-      "step": 24270
-    },
-    {
-      "epoch": 0.09809427231260884,
-      "grad_norm": 4481.90673828125,
-      "learning_rate": 4.856e-05,
-      "loss": 116.3378,
-      "step": 24280
-    },
-    {
-      "epoch": 0.09813467357797646,
-      "grad_norm": 1093.2393798828125,
-      "learning_rate": 4.8580000000000006e-05,
-      "loss": 95.4637,
-      "step": 24290
-    },
-    {
-      "epoch": 0.09817507484334409,
-      "grad_norm": 559.2593383789062,
-      "learning_rate": 4.86e-05,
-      "loss": 80.5582,
-      "step": 24300
-    },
-    {
-      "epoch": 0.09821547610871173,
-      "grad_norm": 538.036865234375,
-      "learning_rate": 4.8620000000000005e-05,
-      "loss": 91.296,
-      "step": 24310
-    },
-    {
-      "epoch": 0.09825587737407936,
-      "grad_norm": 1021.3819580078125,
-      "learning_rate": 4.864e-05,
-      "loss": 96.4481,
-      "step": 24320
-    },
-    {
-      "epoch": 0.09829627863944698,
-      "grad_norm": 905.1778564453125,
-      "learning_rate": 4.866e-05,
-      "loss": 93.2619,
-      "step": 24330
-    },
-    {
-      "epoch": 0.09833667990481462,
-      "grad_norm": 604.1958618164062,
-      "learning_rate": 4.868e-05,
-      "loss": 62.2251,
-      "step": 24340
-    },
-    {
-      "epoch": 0.09837708117018225,
-      "grad_norm": 707.2637939453125,
-      "learning_rate": 4.87e-05,
-      "loss": 81.4563,
-      "step": 24350
-    },
-    {
-      "epoch": 0.09841748243554987,
-      "grad_norm": 1935.6922607421875,
-      "learning_rate": 4.872000000000001e-05,
-      "loss": 150.9556,
-      "step": 24360
-    },
-    {
-      "epoch": 0.09845788370091751,
-      "grad_norm": 529.8268432617188,
-      "learning_rate": 4.8740000000000004e-05,
-      "loss": 95.0914,
-      "step": 24370
-    },
-    {
-      "epoch": 0.09849828496628514,
-      "grad_norm": 1129.89599609375,
-      "learning_rate": 4.876e-05,
-      "loss": 128.5505,
-      "step": 24380
-    },
-    {
-      "epoch": 0.09853868623165278,
-      "grad_norm": 2489.7001953125,
-      "learning_rate": 4.8780000000000004e-05,
-      "loss": 138.6563,
-      "step": 24390
-    },
-    {
-      "epoch": 0.0985790874970204,
-      "grad_norm": 1026.3582763671875,
-      "learning_rate": 4.88e-05,
-      "loss": 126.4132,
-      "step": 24400
-    },
-    {
-      "epoch": 0.09861948876238803,
-      "grad_norm": 2506.63916015625,
-      "learning_rate": 4.8820000000000004e-05,
-      "loss": 151.2192,
-      "step": 24410
-    },
-    {
-      "epoch": 0.09865989002775567,
-      "grad_norm": 1200.955810546875,
-      "learning_rate": 4.884e-05,
-      "loss": 131.9323,
-      "step": 24420
-    },
-    {
-      "epoch": 0.0987002912931233,
-      "grad_norm": 888.474365234375,
-      "learning_rate": 4.886e-05,
-      "loss": 80.7957,
-      "step": 24430
-    },
-    {
-      "epoch": 0.09874069255849093,
-      "grad_norm": 853.1484375,
-      "learning_rate": 4.8880000000000006e-05,
-      "loss": 152.3066,
-      "step": 24440
-    },
-    {
-      "epoch": 0.09878109382385857,
-      "grad_norm": 931.940673828125,
-      "learning_rate": 4.89e-05,
-      "loss": 124.5411,
-      "step": 24450
-    },
-    {
-      "epoch": 0.09882149508922619,
-      "grad_norm": 755.48828125,
-      "learning_rate": 4.8920000000000006e-05,
-      "loss": 113.8801,
-      "step": 24460
-    },
-    {
-      "epoch": 0.09886189635459383,
-      "grad_norm": 1995.7860107421875,
-      "learning_rate": 4.894e-05,
-      "loss": 101.5088,
-      "step": 24470
-    },
-    {
-      "epoch": 0.09890229761996146,
-      "grad_norm": 645.9288940429688,
-      "learning_rate": 4.896e-05,
-      "loss": 75.8798,
-      "step": 24480
-    },
-    {
-      "epoch": 0.09894269888532908,
-      "grad_norm": 1073.67626953125,
-      "learning_rate": 4.898e-05,
-      "loss": 161.3777,
-      "step": 24490
-    },
-    {
-      "epoch": 0.09898310015069672,
-      "grad_norm": 1047.078125,
-      "learning_rate": 4.9e-05,
-      "loss": 136.8098,
-      "step": 24500
-    },
-    {
-      "epoch": 0.09902350141606435,
-      "grad_norm": 1193.5416259765625,
-      "learning_rate": 4.902e-05,
-      "loss": 112.7242,
-      "step": 24510
-    },
-    {
-      "epoch": 0.09906390268143198,
-      "grad_norm": 2082.607421875,
-      "learning_rate": 4.9040000000000005e-05,
-      "loss": 104.1718,
-      "step": 24520
-    },
-    {
-      "epoch": 0.09910430394679962,
-      "grad_norm": 657.3272705078125,
-      "learning_rate": 4.906e-05,
-      "loss": 94.1763,
-      "step": 24530
-    },
-    {
-      "epoch": 0.09914470521216724,
-      "grad_norm": 822.24462890625,
-      "learning_rate": 4.9080000000000004e-05,
-      "loss": 120.4415,
-      "step": 24540
-    },
-    {
-      "epoch": 0.09918510647753488,
-      "grad_norm": 1214.71142578125,
-      "learning_rate": 4.91e-05,
-      "loss": 95.7002,
-      "step": 24550
-    },
-    {
-      "epoch": 0.09922550774290251,
-      "grad_norm": 974.8594360351562,
-      "learning_rate": 4.9120000000000004e-05,
-      "loss": 122.8078,
-      "step": 24560
-    },
-    {
-      "epoch": 0.09926590900827013,
-      "grad_norm": 2204.4921875,
-      "learning_rate": 4.914e-05,
-      "loss": 181.9215,
-      "step": 24570
-    },
-    {
-      "epoch": 0.09930631027363777,
-      "grad_norm": 667.4011840820312,
-      "learning_rate": 4.9160000000000004e-05,
-      "loss": 80.917,
-      "step": 24580
-    },
-    {
-      "epoch": 0.0993467115390054,
-      "grad_norm": 896.2610473632812,
-      "learning_rate": 4.918000000000001e-05,
-      "loss": 144.0085,
-      "step": 24590
-    },
-    {
-      "epoch": 0.09938711280437303,
-      "grad_norm": 0.0,
-      "learning_rate": 4.92e-05,
-      "loss": 132.3478,
-      "step": 24600
-    },
-    {
-      "epoch": 0.09942751406974067,
-      "grad_norm": 941.837646484375,
-      "learning_rate": 4.9220000000000006e-05,
-      "loss": 98.3554,
-      "step": 24610
-    },
-    {
-      "epoch": 0.09946791533510829,
-      "grad_norm": 975.3572387695312,
-      "learning_rate": 4.924e-05,
-      "loss": 71.2832,
-      "step": 24620
-    },
-    {
-      "epoch": 0.09950831660047593,
-      "grad_norm": 467.11871337890625,
-      "learning_rate": 4.926e-05,
-      "loss": 84.4188,
-      "step": 24630
-    },
-    {
-      "epoch": 0.09954871786584356,
-      "grad_norm": 1425.18603515625,
-      "learning_rate": 4.928e-05,
-      "loss": 117.7631,
-      "step": 24640
-    },
-    {
-      "epoch": 0.09958911913121118,
-      "grad_norm": 992.929443359375,
-      "learning_rate": 4.93e-05,
-      "loss": 74.3671,
-      "step": 24650
-    },
-    {
-      "epoch": 0.09962952039657882,
-      "grad_norm": 1589.87646484375,
-      "learning_rate": 4.932e-05,
-      "loss": 127.9026,
-      "step": 24660
-    },
-    {
-      "epoch": 0.09966992166194645,
-      "grad_norm": 726.9409790039062,
-      "learning_rate": 4.9340000000000005e-05,
-      "loss": 127.6771,
-      "step": 24670
-    },
-    {
-      "epoch": 0.09971032292731408,
-      "grad_norm": 1687.093017578125,
-      "learning_rate": 4.936e-05,
-      "loss": 112.6813,
-      "step": 24680
-    },
-    {
-      "epoch": 0.09975072419268172,
-      "grad_norm": 891.7119140625,
-      "learning_rate": 4.9380000000000005e-05,
-      "loss": 131.5728,
-      "step": 24690
-    },
-    {
-      "epoch": 0.09979112545804934,
-      "grad_norm": 601.4475708007812,
-      "learning_rate": 4.94e-05,
-      "loss": 80.0845,
-      "step": 24700
-    },
-    {
-      "epoch": 0.09983152672341698,
-      "grad_norm": 1217.7177734375,
-      "learning_rate": 4.942e-05,
-      "loss": 108.1615,
-      "step": 24710
-    },
-    {
-      "epoch": 0.09987192798878461,
-      "grad_norm": 1115.55517578125,
-      "learning_rate": 4.944e-05,
-      "loss": 86.1,
-      "step": 24720
-    },
-    {
-      "epoch": 0.09991232925415224,
-      "grad_norm": 623.4889526367188,
-      "learning_rate": 4.946e-05,
-      "loss": 99.014,
-      "step": 24730
-    },
-    {
-      "epoch": 0.09995273051951988,
-      "grad_norm": 3404.5966796875,
-      "learning_rate": 4.948000000000001e-05,
-      "loss": 111.524,
-      "step": 24740
-    },
-    {
-      "epoch": 0.0999931317848875,
-      "grad_norm": 1838.208251953125,
-      "learning_rate": 4.9500000000000004e-05,
-      "loss": 162.1484,
-      "step": 24750
-    },
-    {
-      "epoch": 0.10003353305025513,
-      "grad_norm": 371.3464660644531,
-      "learning_rate": 4.952e-05,
-      "loss": 85.2928,
-      "step": 24760
-    },
-    {
-      "epoch": 0.10007393431562277,
-      "grad_norm": 718.6541748046875,
-      "learning_rate": 4.9540000000000003e-05,
-      "loss": 86.9755,
-      "step": 24770
-    },
-    {
-      "epoch": 0.1001143355809904,
-      "grad_norm": 1240.49169921875,
-      "learning_rate": 4.956e-05,
-      "loss": 72.236,
-      "step": 24780
-    },
-    {
-      "epoch": 0.10015473684635803,
-      "grad_norm": 2078.4365234375,
-      "learning_rate": 4.958e-05,
-      "loss": 130.2423,
-      "step": 24790
-    },
-    {
-      "epoch": 0.10019513811172566,
-      "grad_norm": 1719.6787109375,
-      "learning_rate": 4.96e-05,
-      "loss": 147.8617,
-      "step": 24800
-    },
-    {
-      "epoch": 0.10023553937709329,
-      "grad_norm": 1234.83203125,
-      "learning_rate": 4.962e-05,
-      "loss": 116.7895,
-      "step": 24810
-    },
-    {
-      "epoch": 0.10027594064246093,
-      "grad_norm": 1845.6007080078125,
-      "learning_rate": 4.9640000000000006e-05,
-      "loss": 123.7702,
-      "step": 24820
-    },
-    {
-      "epoch": 0.10031634190782855,
-      "grad_norm": 1477.3031005859375,
-      "learning_rate": 4.966e-05,
-      "loss": 175.7533,
-      "step": 24830
-    },
-    {
-      "epoch": 0.10035674317319618,
-      "grad_norm": 753.8565063476562,
-      "learning_rate": 4.9680000000000005e-05,
-      "loss": 137.9543,
-      "step": 24840
-    },
-    {
-      "epoch": 0.10039714443856382,
-      "grad_norm": 458.0389709472656,
-      "learning_rate": 4.97e-05,
-      "loss": 99.559,
-      "step": 24850
-    },
-    {
-      "epoch": 0.10043754570393144,
-      "grad_norm": 1369.2882080078125,
-      "learning_rate": 4.972e-05,
-      "loss": 151.6343,
-      "step": 24860
-    },
-    {
-      "epoch": 0.10047794696929908,
-      "grad_norm": 2486.626220703125,
-      "learning_rate": 4.974e-05,
-      "loss": 118.1052,
-      "step": 24870
-    },
-    {
-      "epoch": 0.10051834823466671,
-      "grad_norm": 1189.0513916015625,
-      "learning_rate": 4.976e-05,
-      "loss": 115.5461,
-      "step": 24880
-    },
-    {
-      "epoch": 0.10055874950003434,
-      "grad_norm": 3800.11083984375,
-      "learning_rate": 4.978e-05,
-      "loss": 113.4657,
-      "step": 24890
-    },
-    {
-      "epoch": 0.10059915076540198,
-      "grad_norm": 588.5828857421875,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 116.0871,
-      "step": 24900
-    },
-    {
-      "epoch": 0.1006395520307696,
-      "grad_norm": 748.7584228515625,
-      "learning_rate": 4.982e-05,
-      "loss": 96.397,
-      "step": 24910
-    },
-    {
-      "epoch": 0.10067995329613723,
-      "grad_norm": 883.9682006835938,
-      "learning_rate": 4.9840000000000004e-05,
-      "loss": 73.1233,
-      "step": 24920
-    },
-    {
-      "epoch": 0.10072035456150487,
-      "grad_norm": 747.5608520507812,
-      "learning_rate": 4.986e-05,
-      "loss": 101.5735,
-      "step": 24930
-    },
-    {
-      "epoch": 0.1007607558268725,
-      "grad_norm": 1152.760009765625,
-      "learning_rate": 4.9880000000000004e-05,
-      "loss": 106.2813,
-      "step": 24940
-    },
-    {
-      "epoch": 0.10080115709224013,
-      "grad_norm": 1556.455322265625,
-      "learning_rate": 4.99e-05,
-      "loss": 92.2315,
-      "step": 24950
-    },
-    {
-      "epoch": 0.10084155835760776,
-      "grad_norm": 564.4570922851562,
-      "learning_rate": 4.992e-05,
-      "loss": 105.7234,
-      "step": 24960
-    },
-    {
-      "epoch": 0.10088195962297539,
-      "grad_norm": 582.0642700195312,
-      "learning_rate": 4.9940000000000006e-05,
-      "loss": 85.4253,
-      "step": 24970
-    },
-    {
-      "epoch": 0.10092236088834303,
-      "grad_norm": 2132.204833984375,
-      "learning_rate": 4.996e-05,
-      "loss": 145.5456,
-      "step": 24980
-    },
-    {
-      "epoch": 0.10096276215371065,
-      "grad_norm": 1085.25048828125,
-      "learning_rate": 4.9980000000000006e-05,
-      "loss": 105.3973,
-      "step": 24990
-    },
-    {
-      "epoch": 0.10100316341907828,
-      "grad_norm": 482.8312072753906,
-      "learning_rate": 5e-05,
-      "loss": 107.6187,
-      "step": 25000
-    },
-    {
-      "epoch": 0.10104356468444592,
-      "grad_norm": 12273.1533203125,
-      "learning_rate": 4.999999975630607e-05,
-      "loss": 166.6157,
-      "step": 25010
-    },
-    {
-      "epoch": 0.10108396594981355,
-      "grad_norm": 1100.8514404296875,
-      "learning_rate": 4.999999902522426e-05,
-      "loss": 170.2824,
-      "step": 25020
-    },
-    {
-      "epoch": 0.10112436721518119,
-      "grad_norm": 1492.0712890625,
-      "learning_rate": 4.9999997806754614e-05,
-      "loss": 105.2635,
-      "step": 25030
-    },
-    {
-      "epoch": 0.10116476848054881,
-      "grad_norm": 2506.78173828125,
-      "learning_rate": 4.9999996100897126e-05,
-      "loss": 76.4485,
-      "step": 25040
-    },
-    {
-      "epoch": 0.10120516974591644,
-      "grad_norm": 2876.141845703125,
-      "learning_rate": 4.999999390765185e-05,
-      "loss": 115.6973,
-      "step": 25050
-    },
-    {
-      "epoch": 0.10124557101128408,
-      "grad_norm": 1318.9537353515625,
-      "learning_rate": 4.999999122701883e-05,
-      "loss": 120.0407,
-      "step": 25060
-    },
-    {
-      "epoch": 0.1012859722766517,
-      "grad_norm": 598.3362426757812,
-      "learning_rate": 4.99999880589981e-05,
-      "loss": 99.9749,
-      "step": 25070
-    },
-    {
-      "epoch": 0.10132637354201933,
-      "grad_norm": 680.303466796875,
-      "learning_rate": 4.999998440358973e-05,
-      "loss": 86.1489,
-      "step": 25080
-    },
-    {
-      "epoch": 0.10136677480738697,
-      "grad_norm": 1665.12548828125,
-      "learning_rate": 4.99999802607938e-05,
-      "loss": 125.203,
-      "step": 25090
-    },
-    {
-      "epoch": 0.1014071760727546,
-      "grad_norm": 4897.3798828125,
-      "learning_rate": 4.999997563061038e-05,
-      "loss": 99.004,
-      "step": 25100
-    },
-    {
-      "epoch": 0.10144757733812224,
-      "grad_norm": 561.159423828125,
-      "learning_rate": 4.999997051303956e-05,
-      "loss": 128.6623,
-      "step": 25110
-    },
-    {
-      "epoch": 0.10148797860348986,
-      "grad_norm": 407.36114501953125,
-      "learning_rate": 4.9999964908081455e-05,
-      "loss": 107.5606,
-      "step": 25120
-    },
-    {
-      "epoch": 0.10152837986885749,
-      "grad_norm": 602.0693359375,
-      "learning_rate": 4.999995881573616e-05,
-      "loss": 99.1402,
-      "step": 25130
-    },
-    {
-      "epoch": 0.10156878113422513,
-      "grad_norm": 428.0013427734375,
-      "learning_rate": 4.999995223600379e-05,
-      "loss": 176.0661,
-      "step": 25140
-    },
-    {
-      "epoch": 0.10160918239959275,
-      "grad_norm": 1730.8555908203125,
-      "learning_rate": 4.999994516888449e-05,
-      "loss": 143.4497,
-      "step": 25150
-    },
-    {
-      "epoch": 0.10164958366496038,
-      "grad_norm": 1302.3773193359375,
-      "learning_rate": 4.999993761437838e-05,
-      "loss": 137.5996,
-      "step": 25160
-    },
-    {
-      "epoch": 0.10168998493032802,
-      "grad_norm": 3554.90771484375,
-      "learning_rate": 4.9999929572485616e-05,
-      "loss": 139.0566,
-      "step": 25170
-    },
-    {
-      "epoch": 0.10173038619569565,
-      "grad_norm": 1487.5889892578125,
-      "learning_rate": 4.999992104320636e-05,
-      "loss": 108.4844,
-      "step": 25180
-    },
-    {
-      "epoch": 0.10177078746106329,
-      "grad_norm": 1060.8238525390625,
-      "learning_rate": 4.999991202654076e-05,
-      "loss": 110.4729,
-      "step": 25190
-    },
-    {
-      "epoch": 0.10181118872643091,
-      "grad_norm": 899.1284790039062,
-      "learning_rate": 4.9999902522489015e-05,
-      "loss": 117.7502,
-      "step": 25200
-    },
-    {
-      "epoch": 0.10185158999179854,
-      "grad_norm": 824.8314819335938,
-      "learning_rate": 4.99998925310513e-05,
-      "loss": 96.4199,
-      "step": 25210
-    },
-    {
-      "epoch": 0.10189199125716618,
-      "grad_norm": 729.9413452148438,
-      "learning_rate": 4.999988205222781e-05,
-      "loss": 58.8153,
-      "step": 25220
-    },
-    {
-      "epoch": 0.1019323925225338,
-      "grad_norm": 1475.010498046875,
-      "learning_rate": 4.999987108601874e-05,
-      "loss": 118.004,
-      "step": 25230
-    },
-    {
-      "epoch": 0.10197279378790143,
-      "grad_norm": 3924.0517578125,
-      "learning_rate": 4.999985963242432e-05,
-      "loss": 161.4323,
-      "step": 25240
-    },
-    {
-      "epoch": 0.10201319505326907,
-      "grad_norm": 684.8850708007812,
-      "learning_rate": 4.999984769144476e-05,
-      "loss": 93.1785,
-      "step": 25250
-    },
-    {
-      "epoch": 0.1020535963186367,
-      "grad_norm": 617.2882690429688,
-      "learning_rate": 4.99998352630803e-05,
-      "loss": 108.2454,
-      "step": 25260
-    },
-    {
-      "epoch": 0.10209399758400434,
-      "grad_norm": 713.0025634765625,
-      "learning_rate": 4.999982234733118e-05,
-      "loss": 128.2708,
-      "step": 25270
-    },
-    {
-      "epoch": 0.10213439884937196,
-      "grad_norm": 830.2299194335938,
-      "learning_rate": 4.9999808944197666e-05,
-      "loss": 93.2193,
-      "step": 25280
-    },
-    {
-      "epoch": 0.10217480011473959,
-      "grad_norm": 635.0062255859375,
-      "learning_rate": 4.999979505367999e-05,
-      "loss": 95.8657,
-      "step": 25290
-    },
-    {
-      "epoch": 0.10221520138010723,
-      "grad_norm": 681.8064575195312,
-      "learning_rate": 4.999978067577844e-05,
-      "loss": 81.1595,
-      "step": 25300
-    },
-    {
-      "epoch": 0.10225560264547486,
-      "grad_norm": 1729.1885986328125,
-      "learning_rate": 4.999976581049331e-05,
-      "loss": 104.3257,
-      "step": 25310
-    },
-    {
-      "epoch": 0.10229600391084248,
-      "grad_norm": 499.2225341796875,
-      "learning_rate": 4.999975045782486e-05,
-      "loss": 150.2334,
-      "step": 25320
-    },
-    {
-      "epoch": 0.10233640517621012,
-      "grad_norm": 1919.6650390625,
-      "learning_rate": 4.9999734617773405e-05,
-      "loss": 93.5073,
-      "step": 25330
-    },
-    {
-      "epoch": 0.10237680644157775,
-      "grad_norm": 628.6290283203125,
-      "learning_rate": 4.9999718290339256e-05,
-      "loss": 138.5961,
-      "step": 25340
-    },
-    {
-      "epoch": 0.10241720770694539,
-      "grad_norm": 877.5479125976562,
-      "learning_rate": 4.999970147552272e-05,
-      "loss": 106.7546,
-      "step": 25350
-    },
-    {
-      "epoch": 0.10245760897231301,
-      "grad_norm": 426.84027099609375,
-      "learning_rate": 4.999968417332415e-05,
-      "loss": 102.9967,
-      "step": 25360
-    },
-    {
-      "epoch": 0.10249801023768064,
-      "grad_norm": 1005.803466796875,
-      "learning_rate": 4.9999666383743854e-05,
-      "loss": 133.2389,
-      "step": 25370
-    },
-    {
-      "epoch": 0.10253841150304828,
-      "grad_norm": 973.8045654296875,
-      "learning_rate": 4.999964810678219e-05,
-      "loss": 127.7309,
-      "step": 25380
-    },
-    {
-      "epoch": 0.1025788127684159,
-      "grad_norm": 1207.312255859375,
-      "learning_rate": 4.9999629342439524e-05,
-      "loss": 91.1231,
-      "step": 25390
-    },
-    {
-      "epoch": 0.10261921403378353,
-      "grad_norm": 2072.83837890625,
-      "learning_rate": 4.999961009071621e-05,
-      "loss": 154.6031,
-      "step": 25400
-    },
-    {
-      "epoch": 0.10265961529915117,
-      "grad_norm": 547.6760864257812,
-      "learning_rate": 4.999959035161263e-05,
-      "loss": 112.7586,
-      "step": 25410
-    },
-    {
-      "epoch": 0.1027000165645188,
-      "grad_norm": 784.879150390625,
-      "learning_rate": 4.999957012512916e-05,
-      "loss": 104.8253,
-      "step": 25420
-    },
-    {
-      "epoch": 0.10274041782988644,
-      "grad_norm": 625.8154296875,
-      "learning_rate": 4.99995494112662e-05,
-      "loss": 80.9909,
-      "step": 25430
-    },
-    {
-      "epoch": 0.10278081909525406,
-      "grad_norm": 976.3950805664062,
-      "learning_rate": 4.999952821002415e-05,
-      "loss": 121.5834,
-      "step": 25440
-    },
-    {
-      "epoch": 0.10282122036062169,
-      "grad_norm": 658.44140625,
-      "learning_rate": 4.999950652140343e-05,
-      "loss": 152.7022,
-      "step": 25450
-    },
-    {
-      "epoch": 0.10286162162598933,
-      "grad_norm": 948.6815795898438,
-      "learning_rate": 4.999948434540446e-05,
-      "loss": 119.2311,
-      "step": 25460
-    },
-    {
-      "epoch": 0.10290202289135696,
-      "grad_norm": 2785.1484375,
-      "learning_rate": 4.999946168202767e-05,
-      "loss": 123.5097,
-      "step": 25470
-    },
-    {
-      "epoch": 0.10294242415672458,
-      "grad_norm": 2336.2509765625,
-      "learning_rate": 4.999943853127351e-05,
-      "loss": 92.8932,
-      "step": 25480
-    },
-    {
-      "epoch": 0.10298282542209222,
-      "grad_norm": 324.7106018066406,
-      "learning_rate": 4.9999414893142425e-05,
-      "loss": 84.0734,
-      "step": 25490
-    },
-    {
-      "epoch": 0.10302322668745985,
-      "grad_norm": 1529.5069580078125,
-      "learning_rate": 4.999939076763487e-05,
-      "loss": 151.1412,
-      "step": 25500
-    },
-    {
-      "epoch": 0.10306362795282749,
-      "grad_norm": 3243.043701171875,
-      "learning_rate": 4.999936615475133e-05,
-      "loss": 98.3063,
-      "step": 25510
-    },
-    {
-      "epoch": 0.10310402921819511,
-      "grad_norm": 656.1243286132812,
-      "learning_rate": 4.9999341054492265e-05,
-      "loss": 132.48,
-      "step": 25520
-    },
-    {
-      "epoch": 0.10314443048356274,
-      "grad_norm": 358.87744140625,
-      "learning_rate": 4.999931546685819e-05,
-      "loss": 160.6029,
-      "step": 25530
-    },
-    {
-      "epoch": 0.10318483174893038,
-      "grad_norm": 1219.5975341796875,
-      "learning_rate": 4.999928939184958e-05,
-      "loss": 122.454,
-      "step": 25540
-    },
-    {
-      "epoch": 0.103225233014298,
-      "grad_norm": 1476.14208984375,
-      "learning_rate": 4.999926282946695e-05,
-      "loss": 121.1642,
-      "step": 25550
-    },
-    {
-      "epoch": 0.10326563427966563,
-      "grad_norm": 737.3023071289062,
-      "learning_rate": 4.9999235779710826e-05,
-      "loss": 68.9884,
-      "step": 25560
-    },
-    {
-      "epoch": 0.10330603554503327,
-      "grad_norm": 0.0,
-      "learning_rate": 4.999920824258173e-05,
-      "loss": 70.6594,
-      "step": 25570
-    },
-    {
-      "epoch": 0.1033464368104009,
-      "grad_norm": 585.1388549804688,
-      "learning_rate": 4.999918021808019e-05,
-      "loss": 82.4569,
-      "step": 25580
-    },
-    {
-      "epoch": 0.10338683807576854,
-      "grad_norm": 1436.7415771484375,
-      "learning_rate": 4.999915170620677e-05,
-      "loss": 135.1959,
-      "step": 25590
-    },
-    {
-      "epoch": 0.10342723934113617,
-      "grad_norm": 1880.6636962890625,
-      "learning_rate": 4.999912270696202e-05,
-      "loss": 110.9196,
-      "step": 25600
-    },
-    {
-      "epoch": 0.10346764060650379,
-      "grad_norm": 435.46234130859375,
-      "learning_rate": 4.9999093220346495e-05,
-      "loss": 101.5219,
-      "step": 25610
-    },
-    {
-      "epoch": 0.10350804187187143,
-      "grad_norm": 1327.9521484375,
-      "learning_rate": 4.9999063246360786e-05,
-      "loss": 92.5202,
-      "step": 25620
-    },
-    {
-      "epoch": 0.10354844313723906,
-      "grad_norm": 1072.19775390625,
-      "learning_rate": 4.9999032785005464e-05,
-      "loss": 104.1919,
-      "step": 25630
-    },
-    {
-      "epoch": 0.10358884440260668,
-      "grad_norm": 1694.939697265625,
-      "learning_rate": 4.999900183628112e-05,
-      "loss": 165.591,
-      "step": 25640
-    },
-    {
-      "epoch": 0.10362924566797432,
-      "grad_norm": 773.9560546875,
-      "learning_rate": 4.999897040018837e-05,
-      "loss": 104.817,
-      "step": 25650
-    },
-    {
-      "epoch": 0.10366964693334195,
-      "grad_norm": 1010.9199829101562,
-      "learning_rate": 4.9998938476727826e-05,
-      "loss": 114.004,
-      "step": 25660
-    },
-    {
-      "epoch": 0.10371004819870959,
-      "grad_norm": 2661.3837890625,
-      "learning_rate": 4.99989060659001e-05,
-      "loss": 99.0087,
-      "step": 25670
-    },
-    {
-      "epoch": 0.10375044946407722,
-      "grad_norm": 553.9918823242188,
-      "learning_rate": 4.999887316770584e-05,
-      "loss": 72.9581,
-      "step": 25680
-    },
-    {
-      "epoch": 0.10379085072944484,
-      "grad_norm": 580.9917602539062,
-      "learning_rate": 4.999883978214567e-05,
-      "loss": 103.5152,
-      "step": 25690
-    },
-    {
-      "epoch": 0.10383125199481248,
-      "grad_norm": 4983.98974609375,
-      "learning_rate": 4.999880590922025e-05,
-      "loss": 135.9269,
-      "step": 25700
-    },
-    {
-      "epoch": 0.10387165326018011,
-      "grad_norm": 1303.5921630859375,
-      "learning_rate": 4.999877154893023e-05,
-      "loss": 125.694,
-      "step": 25710
-    },
-    {
-      "epoch": 0.10391205452554773,
-      "grad_norm": 416.0110778808594,
-      "learning_rate": 4.9998736701276295e-05,
-      "loss": 61.3954,
-      "step": 25720
-    },
-    {
-      "epoch": 0.10395245579091537,
-      "grad_norm": 1457.8369140625,
-      "learning_rate": 4.999870136625912e-05,
-      "loss": 148.0761,
-      "step": 25730
-    },
-    {
-      "epoch": 0.103992857056283,
-      "grad_norm": 758.3458251953125,
-      "learning_rate": 4.999866554387939e-05,
-      "loss": 94.0594,
-      "step": 25740
-    },
-    {
-      "epoch": 0.10403325832165064,
-      "grad_norm": 1217.02978515625,
-      "learning_rate": 4.999862923413781e-05,
-      "loss": 101.6771,
-      "step": 25750
-    },
-    {
-      "epoch": 0.10407365958701827,
-      "grad_norm": 1469.3179931640625,
-      "learning_rate": 4.9998592437035076e-05,
-      "loss": 119.159,
-      "step": 25760
-    },
-    {
-      "epoch": 0.10411406085238589,
-      "grad_norm": 2354.025390625,
-      "learning_rate": 4.9998555152571914e-05,
-      "loss": 115.2768,
-      "step": 25770
-    },
-    {
-      "epoch": 0.10415446211775353,
-      "grad_norm": 789.3027954101562,
-      "learning_rate": 4.999851738074904e-05,
-      "loss": 73.4068,
-      "step": 25780
-    },
-    {
-      "epoch": 0.10419486338312116,
-      "grad_norm": 841.7200317382812,
-      "learning_rate": 4.9998479121567214e-05,
-      "loss": 134.448,
-      "step": 25790
-    },
-    {
-      "epoch": 0.10423526464848878,
-      "grad_norm": 431.4704895019531,
-      "learning_rate": 4.9998440375027166e-05,
-      "loss": 150.1897,
-      "step": 25800
-    },
-    {
-      "epoch": 0.10427566591385642,
-      "grad_norm": 508.6786804199219,
-      "learning_rate": 4.999840114112965e-05,
-      "loss": 82.7051,
-      "step": 25810
-    },
-    {
-      "epoch": 0.10431606717922405,
-      "grad_norm": 818.539794921875,
-      "learning_rate": 4.999836141987543e-05,
-      "loss": 138.6107,
-      "step": 25820
-    },
-    {
-      "epoch": 0.10435646844459169,
-      "grad_norm": 503.4788513183594,
-      "learning_rate": 4.999832121126529e-05,
-      "loss": 119.86,
-      "step": 25830
-    },
-    {
-      "epoch": 0.10439686970995932,
-      "grad_norm": 838.92041015625,
-      "learning_rate": 4.99982805153e-05,
-      "loss": 121.616,
-      "step": 25840
-    },
-    {
-      "epoch": 0.10443727097532694,
-      "grad_norm": 631.1499633789062,
-      "learning_rate": 4.9998239331980366e-05,
-      "loss": 85.7237,
-      "step": 25850
-    },
-    {
-      "epoch": 0.10447767224069458,
-      "grad_norm": 566.1675415039062,
-      "learning_rate": 4.999819766130719e-05,
-      "loss": 91.9205,
-      "step": 25860
-    },
-    {
-      "epoch": 0.10451807350606221,
-      "grad_norm": 0.0,
-      "learning_rate": 4.999815550328128e-05,
-      "loss": 88.1384,
-      "step": 25870
-    },
-    {
-      "epoch": 0.10455847477142984,
-      "grad_norm": 735.2154541015625,
-      "learning_rate": 4.9998112857903454e-05,
-      "loss": 114.0636,
-      "step": 25880
-    },
-    {
-      "epoch": 0.10459887603679748,
-      "grad_norm": 1965.4488525390625,
-      "learning_rate": 4.9998069725174546e-05,
-      "loss": 117.463,
-      "step": 25890
-    },
-    {
-      "epoch": 0.1046392773021651,
-      "grad_norm": 954.0660400390625,
-      "learning_rate": 4.9998026105095405e-05,
-      "loss": 110.6166,
-      "step": 25900
-    },
-    {
-      "epoch": 0.10467967856753274,
-      "grad_norm": 2468.442626953125,
-      "learning_rate": 4.9997981997666874e-05,
-      "loss": 118.2137,
-      "step": 25910
-    },
-    {
-      "epoch": 0.10472007983290037,
-      "grad_norm": 1254.0142822265625,
-      "learning_rate": 4.999793740288982e-05,
-      "loss": 74.752,
-      "step": 25920
-    },
-    {
-      "epoch": 0.104760481098268,
-      "grad_norm": 625.595458984375,
-      "learning_rate": 4.999789232076509e-05,
-      "loss": 89.1264,
-      "step": 25930
-    },
-    {
-      "epoch": 0.10480088236363563,
-      "grad_norm": 579.2570190429688,
-      "learning_rate": 4.999784675129359e-05,
-      "loss": 128.8163,
-      "step": 25940
-    },
-    {
-      "epoch": 0.10484128362900326,
-      "grad_norm": 4238.0439453125,
-      "learning_rate": 4.999780069447619e-05,
-      "loss": 135.1213,
-      "step": 25950
-    },
-    {
-      "epoch": 0.10488168489437089,
-      "grad_norm": 2267.60693359375,
-      "learning_rate": 4.9997754150313815e-05,
-      "loss": 106.0803,
-      "step": 25960
-    },
-    {
-      "epoch": 0.10492208615973853,
-      "grad_norm": 1275.7972412109375,
-      "learning_rate": 4.999770711880734e-05,
-      "loss": 114.9827,
-      "step": 25970
-    },
-    {
-      "epoch": 0.10496248742510615,
-      "grad_norm": 1259.9520263671875,
-      "learning_rate": 4.999765959995769e-05,
-      "loss": 122.4034,
-      "step": 25980
-    },
-    {
-      "epoch": 0.10500288869047379,
-      "grad_norm": 1249.4486083984375,
-      "learning_rate": 4.99976115937658e-05,
-      "loss": 113.4047,
-      "step": 25990
-    },
-    {
-      "epoch": 0.10504328995584142,
-      "grad_norm": 829.5191650390625,
-      "learning_rate": 4.999756310023261e-05,
-      "loss": 100.0737,
-      "step": 26000
-    },
-    {
-      "epoch": 0.10508369122120904,
-      "grad_norm": 1288.53515625,
-      "learning_rate": 4.999751411935905e-05,
-      "loss": 86.3596,
-      "step": 26010
-    },
-    {
-      "epoch": 0.10512409248657668,
-      "grad_norm": 1140.1302490234375,
-      "learning_rate": 4.999746465114609e-05,
-      "loss": 62.8034,
-      "step": 26020
-    },
-    {
-      "epoch": 0.10516449375194431,
-      "grad_norm": 694.5878295898438,
-      "learning_rate": 4.999741469559468e-05,
-      "loss": 92.5841,
-      "step": 26030
-    },
-    {
-      "epoch": 0.10520489501731194,
-      "grad_norm": 1500.5892333984375,
-      "learning_rate": 4.99973642527058e-05,
-      "loss": 110.6927,
-      "step": 26040
-    },
-    {
-      "epoch": 0.10524529628267958,
-      "grad_norm": 874.5784301757812,
-      "learning_rate": 4.999731332248044e-05,
-      "loss": 63.3448,
-      "step": 26050
-    },
-    {
-      "epoch": 0.1052856975480472,
-      "grad_norm": 932.64501953125,
-      "learning_rate": 4.999726190491958e-05,
-      "loss": 107.8189,
-      "step": 26060
-    },
-    {
-      "epoch": 0.10532609881341484,
-      "grad_norm": 736.0410766601562,
-      "learning_rate": 4.9997210000024236e-05,
-      "loss": 99.0618,
-      "step": 26070
-    },
-    {
-      "epoch": 0.10536650007878247,
-      "grad_norm": 1410.8321533203125,
-      "learning_rate": 4.999715760779541e-05,
-      "loss": 126.3873,
-      "step": 26080
-    },
-    {
-      "epoch": 0.1054069013441501,
-      "grad_norm": 1724.4503173828125,
-      "learning_rate": 4.999710472823414e-05,
-      "loss": 104.2871,
-      "step": 26090
-    },
-    {
-      "epoch": 0.10544730260951773,
-      "grad_norm": 881.2642822265625,
-      "learning_rate": 4.9997051361341425e-05,
-      "loss": 67.8849,
-      "step": 26100
-    },
-    {
-      "epoch": 0.10548770387488536,
-      "grad_norm": 1165.669921875,
-      "learning_rate": 4.999699750711833e-05,
-      "loss": 99.132,
-      "step": 26110
-    },
-    {
-      "epoch": 0.10552810514025299,
-      "grad_norm": 994.9390869140625,
-      "learning_rate": 4.9996943165565905e-05,
-      "loss": 62.6341,
-      "step": 26120
-    },
-    {
-      "epoch": 0.10556850640562063,
-      "grad_norm": 1422.3941650390625,
-      "learning_rate": 4.99968883366852e-05,
-      "loss": 85.9586,
-      "step": 26130
-    },
-    {
-      "epoch": 0.10560890767098825,
-      "grad_norm": 791.3880615234375,
-      "learning_rate": 4.9996833020477285e-05,
-      "loss": 76.5841,
-      "step": 26140
-    },
-    {
-      "epoch": 0.10564930893635589,
-      "grad_norm": 1224.0389404296875,
-      "learning_rate": 4.999677721694325e-05,
-      "loss": 161.3394,
-      "step": 26150
-    },
-    {
-      "epoch": 0.10568971020172352,
-      "grad_norm": 562.48486328125,
-      "learning_rate": 4.9996720926084164e-05,
-      "loss": 117.8859,
-      "step": 26160
-    },
-    {
-      "epoch": 0.10573011146709115,
-      "grad_norm": 1088.6715087890625,
-      "learning_rate": 4.999666414790113e-05,
-      "loss": 119.7692,
-      "step": 26170
-    },
-    {
-      "epoch": 0.10577051273245879,
-      "grad_norm": 716.9673461914062,
-      "learning_rate": 4.999660688239527e-05,
-      "loss": 85.0265,
-      "step": 26180
-    },
-    {
-      "epoch": 0.10581091399782641,
-      "grad_norm": 469.45477294921875,
-      "learning_rate": 4.999654912956769e-05,
-      "loss": 75.7188,
-      "step": 26190
-    },
-    {
-      "epoch": 0.10585131526319404,
-      "grad_norm": 434.962890625,
-      "learning_rate": 4.9996490889419514e-05,
-      "loss": 92.6741,
-      "step": 26200
-    },
-    {
-      "epoch": 0.10589171652856168,
-      "grad_norm": 1787.8798828125,
-      "learning_rate": 4.9996432161951875e-05,
-      "loss": 101.0907,
-      "step": 26210
-    },
-    {
-      "epoch": 0.1059321177939293,
-      "grad_norm": 619.8673706054688,
-      "learning_rate": 4.999637294716593e-05,
-      "loss": 144.3787,
-      "step": 26220
-    },
-    {
-      "epoch": 0.10597251905929694,
-      "grad_norm": 1517.8831787109375,
-      "learning_rate": 4.9996313245062823e-05,
-      "loss": 106.3331,
-      "step": 26230
-    },
-    {
-      "epoch": 0.10601292032466457,
-      "grad_norm": 7921.58203125,
-      "learning_rate": 4.999625305564371e-05,
-      "loss": 83.1559,
-      "step": 26240
-    },
-    {
-      "epoch": 0.1060533215900322,
-      "grad_norm": 839.67822265625,
-      "learning_rate": 4.9996192378909786e-05,
-      "loss": 57.181,
-      "step": 26250
-    },
-    {
-      "epoch": 0.10609372285539984,
-      "grad_norm": 696.2799682617188,
-      "learning_rate": 4.999613121486222e-05,
-      "loss": 49.9272,
-      "step": 26260
-    },
-    {
-      "epoch": 0.10613412412076746,
-      "grad_norm": 1080.623291015625,
-      "learning_rate": 4.99960695635022e-05,
-      "loss": 120.3179,
-      "step": 26270
-    },
-    {
-      "epoch": 0.10617452538613509,
-      "grad_norm": 499.45166015625,
-      "learning_rate": 4.999600742483094e-05,
-      "loss": 85.3591,
-      "step": 26280
-    },
-    {
-      "epoch": 0.10621492665150273,
-      "grad_norm": 938.3174438476562,
-      "learning_rate": 4.999594479884965e-05,
-      "loss": 136.4883,
-      "step": 26290
-    },
-    {
-      "epoch": 0.10625532791687035,
-      "grad_norm": 1555.7998046875,
-      "learning_rate": 4.999588168555954e-05,
-      "loss": 122.8322,
-      "step": 26300
-    },
-    {
-      "epoch": 0.106295729182238,
-      "grad_norm": 4008.9033203125,
-      "learning_rate": 4.999581808496185e-05,
-      "loss": 91.5676,
-      "step": 26310
-    },
-    {
-      "epoch": 0.10633613044760562,
-      "grad_norm": 2325.766845703125,
-      "learning_rate": 4.999575399705783e-05,
-      "loss": 85.4827,
-      "step": 26320
-    },
-    {
-      "epoch": 0.10637653171297325,
-      "grad_norm": 954.0802612304688,
-      "learning_rate": 4.999568942184871e-05,
-      "loss": 161.0085,
-      "step": 26330
-    },
-    {
-      "epoch": 0.10641693297834089,
-      "grad_norm": 1845.4427490234375,
-      "learning_rate": 4.999562435933575e-05,
-      "loss": 106.469,
-      "step": 26340
-    },
-    {
-      "epoch": 0.10645733424370851,
-      "grad_norm": 1453.1446533203125,
-      "learning_rate": 4.999555880952023e-05,
-      "loss": 142.5144,
-      "step": 26350
-    },
-    {
-      "epoch": 0.10649773550907614,
-      "grad_norm": 824.1065673828125,
-      "learning_rate": 4.999549277240342e-05,
-      "loss": 84.602,
-      "step": 26360
-    },
-    {
-      "epoch": 0.10653813677444378,
-      "grad_norm": 3219.528564453125,
-      "learning_rate": 4.999542624798661e-05,
-      "loss": 79.4629,
-      "step": 26370
-    },
-    {
-      "epoch": 0.1065785380398114,
-      "grad_norm": 655.1124267578125,
-      "learning_rate": 4.999535923627109e-05,
-      "loss": 103.1578,
-      "step": 26380
-    },
-    {
-      "epoch": 0.10661893930517904,
-      "grad_norm": 840.5703125,
-      "learning_rate": 4.999529173725819e-05,
-      "loss": 146.742,
-      "step": 26390
-    },
-    {
-      "epoch": 0.10665934057054667,
-      "grad_norm": 1991.2275390625,
-      "learning_rate": 4.999522375094919e-05,
-      "loss": 88.6063,
-      "step": 26400
-    },
-    {
-      "epoch": 0.1066997418359143,
-      "grad_norm": 1580.2342529296875,
-      "learning_rate": 4.999515527734545e-05,
-      "loss": 113.3611,
-      "step": 26410
-    },
-    {
-      "epoch": 0.10674014310128194,
-      "grad_norm": 1060.3555908203125,
-      "learning_rate": 4.9995086316448284e-05,
-      "loss": 176.4278,
-      "step": 26420
-    },
-    {
-      "epoch": 0.10678054436664956,
-      "grad_norm": 2521.33447265625,
-      "learning_rate": 4.999501686825904e-05,
-      "loss": 104.2664,
-      "step": 26430
-    },
-    {
-      "epoch": 0.10682094563201719,
-      "grad_norm": 1635.3087158203125,
-      "learning_rate": 4.999494693277907e-05,
-      "loss": 129.5747,
-      "step": 26440
-    },
-    {
-      "epoch": 0.10686134689738483,
-      "grad_norm": 0.0,
-      "learning_rate": 4.999487651000975e-05,
-      "loss": 83.0078,
-      "step": 26450
-    },
-    {
-      "epoch": 0.10690174816275246,
-      "grad_norm": 1445.0679931640625,
-      "learning_rate": 4.9994805599952445e-05,
-      "loss": 115.5746,
-      "step": 26460
-    },
-    {
-      "epoch": 0.1069421494281201,
-      "grad_norm": 861.307861328125,
-      "learning_rate": 4.999473420260853e-05,
-      "loss": 63.1311,
-      "step": 26470
-    },
-    {
-      "epoch": 0.10698255069348772,
-      "grad_norm": 2009.19189453125,
-      "learning_rate": 4.999466231797941e-05,
-      "loss": 119.9433,
-      "step": 26480
-    },
-    {
-      "epoch": 0.10702295195885535,
-      "grad_norm": 1041.7818603515625,
-      "learning_rate": 4.9994589946066475e-05,
-      "loss": 134.1364,
-      "step": 26490
-    },
-    {
-      "epoch": 0.10706335322422299,
-      "grad_norm": 717.532470703125,
-      "learning_rate": 4.999451708687114e-05,
-      "loss": 99.0042,
-      "step": 26500
-    },
-    {
-      "epoch": 0.10710375448959061,
-      "grad_norm": 1365.530517578125,
-      "learning_rate": 4.999444374039483e-05,
-      "loss": 126.3179,
-      "step": 26510
-    },
-    {
-      "epoch": 0.10714415575495824,
-      "grad_norm": 570.315673828125,
-      "learning_rate": 4.999436990663897e-05,
-      "loss": 75.446,
-      "step": 26520
-    },
-    {
-      "epoch": 0.10718455702032588,
-      "grad_norm": 1172.0423583984375,
-      "learning_rate": 4.9994295585605e-05,
-      "loss": 117.2075,
-      "step": 26530
-    },
-    {
-      "epoch": 0.1072249582856935,
-      "grad_norm": 773.590087890625,
-      "learning_rate": 4.9994220777294364e-05,
-      "loss": 90.5786,
-      "step": 26540
-    },
-    {
-      "epoch": 0.10726535955106115,
-      "grad_norm": 983.9734497070312,
-      "learning_rate": 4.999414548170853e-05,
-      "loss": 89.9236,
-      "step": 26550
-    },
-    {
-      "epoch": 0.10730576081642877,
-      "grad_norm": 1194.6666259765625,
-      "learning_rate": 4.999406969884897e-05,
-      "loss": 64.2603,
-      "step": 26560
-    },
-    {
-      "epoch": 0.1073461620817964,
-      "grad_norm": 1597.0323486328125,
-      "learning_rate": 4.9993993428717144e-05,
-      "loss": 130.4407,
-      "step": 26570
-    },
-    {
-      "epoch": 0.10738656334716404,
-      "grad_norm": 1673.6080322265625,
-      "learning_rate": 4.999391667131455e-05,
-      "loss": 163.3066,
-      "step": 26580
-    },
-    {
-      "epoch": 0.10742696461253166,
-      "grad_norm": 663.078857421875,
-      "learning_rate": 4.9993839426642685e-05,
-      "loss": 130.6437,
-      "step": 26590
-    },
-    {
-      "epoch": 0.10746736587789929,
-      "grad_norm": 1640.7069091796875,
-      "learning_rate": 4.999376169470306e-05,
-      "loss": 131.2924,
-      "step": 26600
-    },
-    {
-      "epoch": 0.10750776714326693,
-      "grad_norm": 986.1610717773438,
-      "learning_rate": 4.9993683475497174e-05,
-      "loss": 115.7689,
-      "step": 26610
-    },
-    {
-      "epoch": 0.10754816840863456,
-      "grad_norm": 322.355224609375,
-      "learning_rate": 4.999360476902656e-05,
-      "loss": 75.3167,
-      "step": 26620
-    },
-    {
-      "epoch": 0.1075885696740022,
-      "grad_norm": 507.4998779296875,
-      "learning_rate": 4.999352557529275e-05,
-      "loss": 96.3537,
-      "step": 26630
-    },
-    {
-      "epoch": 0.10762897093936982,
-      "grad_norm": 1944.9493408203125,
-      "learning_rate": 4.99934458942973e-05,
-      "loss": 145.2764,
-      "step": 26640
-    },
-    {
-      "epoch": 0.10766937220473745,
-      "grad_norm": 2324.57470703125,
-      "learning_rate": 4.999336572604175e-05,
-      "loss": 90.6306,
-      "step": 26650
-    },
-    {
-      "epoch": 0.10770977347010509,
-      "grad_norm": 1668.3519287109375,
-      "learning_rate": 4.999328507052768e-05,
-      "loss": 104.2174,
-      "step": 26660
-    },
-    {
-      "epoch": 0.10775017473547271,
-      "grad_norm": 3942.664794921875,
-      "learning_rate": 4.999320392775663e-05,
-      "loss": 126.4141,
-      "step": 26670
-    },
-    {
-      "epoch": 0.10779057600084034,
-      "grad_norm": 1391.0018310546875,
-      "learning_rate": 4.999312229773022e-05,
-      "loss": 83.7744,
-      "step": 26680
-    },
-    {
-      "epoch": 0.10783097726620798,
-      "grad_norm": 1609.7559814453125,
-      "learning_rate": 4.999304018045001e-05,
-      "loss": 100.6681,
-      "step": 26690
-    },
-    {
-      "epoch": 0.1078713785315756,
-      "grad_norm": 1184.3446044921875,
-      "learning_rate": 4.999295757591762e-05,
-      "loss": 95.7704,
-      "step": 26700
-    },
-    {
-      "epoch": 0.10791177979694325,
-      "grad_norm": 2358.531494140625,
-      "learning_rate": 4.9992874484134653e-05,
-      "loss": 74.604,
-      "step": 26710
-    },
-    {
-      "epoch": 0.10795218106231087,
-      "grad_norm": 576.525146484375,
-      "learning_rate": 4.9992790905102734e-05,
-      "loss": 114.726,
-      "step": 26720
-    },
-    {
-      "epoch": 0.1079925823276785,
-      "grad_norm": 3263.18701171875,
-      "learning_rate": 4.999270683882349e-05,
-      "loss": 147.2892,
-      "step": 26730
-    },
-    {
-      "epoch": 0.10803298359304614,
-      "grad_norm": 882.455322265625,
-      "learning_rate": 4.999262228529855e-05,
-      "loss": 127.5431,
-      "step": 26740
-    },
-    {
-      "epoch": 0.10807338485841377,
-      "grad_norm": 1100.7166748046875,
-      "learning_rate": 4.999253724452958e-05,
-      "loss": 106.7672,
-      "step": 26750
-    },
-    {
-      "epoch": 0.10811378612378139,
-      "grad_norm": 687.6595458984375,
-      "learning_rate": 4.999245171651823e-05,
-      "loss": 89.8419,
-      "step": 26760
-    },
-    {
-      "epoch": 0.10815418738914903,
-      "grad_norm": 443.3103942871094,
-      "learning_rate": 4.999236570126616e-05,
-      "loss": 125.1742,
-      "step": 26770
-    },
-    {
-      "epoch": 0.10819458865451666,
-      "grad_norm": 1187.34912109375,
-      "learning_rate": 4.999227919877506e-05,
-      "loss": 127.4198,
-      "step": 26780
-    },
-    {
-      "epoch": 0.1082349899198843,
-      "grad_norm": 496.15240478515625,
-      "learning_rate": 4.9992192209046603e-05,
-      "loss": 117.4117,
-      "step": 26790
-    },
-    {
-      "epoch": 0.10827539118525192,
-      "grad_norm": 1453.9019775390625,
-      "learning_rate": 4.99921047320825e-05,
-      "loss": 93.6351,
-      "step": 26800
-    },
-    {
-      "epoch": 0.10831579245061955,
-      "grad_norm": 773.209228515625,
-      "learning_rate": 4.999201676788445e-05,
-      "loss": 105.4083,
-      "step": 26810
-    },
-    {
-      "epoch": 0.10835619371598719,
-      "grad_norm": 1077.73193359375,
-      "learning_rate": 4.999192831645416e-05,
-      "loss": 113.7033,
-      "step": 26820
-    },
-    {
-      "epoch": 0.10839659498135482,
-      "grad_norm": 3824.103515625,
-      "learning_rate": 4.999183937779336e-05,
-      "loss": 128.3056,
-      "step": 26830
-    },
-    {
-      "epoch": 0.10843699624672244,
-      "grad_norm": 1111.69189453125,
-      "learning_rate": 4.999174995190379e-05,
-      "loss": 69.169,
-      "step": 26840
-    },
-    {
-      "epoch": 0.10847739751209008,
-      "grad_norm": 355.9375,
-      "learning_rate": 4.999166003878718e-05,
-      "loss": 95.577,
-      "step": 26850
-    },
-    {
-      "epoch": 0.10851779877745771,
-      "grad_norm": 566.212158203125,
-      "learning_rate": 4.99915696384453e-05,
-      "loss": 92.3288,
-      "step": 26860
-    },
-    {
-      "epoch": 0.10855820004282535,
-      "grad_norm": 433.31170654296875,
-      "learning_rate": 4.99914787508799e-05,
-      "loss": 77.2623,
-      "step": 26870
-    },
-    {
-      "epoch": 0.10859860130819297,
-      "grad_norm": 1091.27734375,
-      "learning_rate": 4.999138737609276e-05,
-      "loss": 97.2431,
-      "step": 26880
-    },
-    {
-      "epoch": 0.1086390025735606,
-      "grad_norm": 779.114501953125,
-      "learning_rate": 4.9991295514085644e-05,
-      "loss": 90.7876,
-      "step": 26890
-    },
-    {
-      "epoch": 0.10867940383892824,
-      "grad_norm": 1826.2213134765625,
-      "learning_rate": 4.9991203164860365e-05,
-      "loss": 164.2726,
-      "step": 26900
-    },
-    {
-      "epoch": 0.10871980510429587,
-      "grad_norm": 1381.0306396484375,
-      "learning_rate": 4.999111032841871e-05,
-      "loss": 163.8954,
-      "step": 26910
-    },
-    {
-      "epoch": 0.10876020636966349,
-      "grad_norm": 535.5016479492188,
-      "learning_rate": 4.9991017004762496e-05,
-      "loss": 85.5764,
-      "step": 26920
-    },
-    {
-      "epoch": 0.10880060763503113,
-      "grad_norm": 822.9321899414062,
-      "learning_rate": 4.999092319389354e-05,
-      "loss": 117.8981,
-      "step": 26930
-    },
-    {
-      "epoch": 0.10884100890039876,
-      "grad_norm": 800.758544921875,
-      "learning_rate": 4.999082889581367e-05,
-      "loss": 76.8381,
-      "step": 26940
-    },
-    {
-      "epoch": 0.1088814101657664,
-      "grad_norm": 557.4238891601562,
-      "learning_rate": 4.999073411052472e-05,
-      "loss": 130.8072,
-      "step": 26950
-    },
-    {
-      "epoch": 0.10892181143113402,
-      "grad_norm": 1034.1680908203125,
-      "learning_rate": 4.9990638838028546e-05,
-      "loss": 75.0051,
-      "step": 26960
-    },
-    {
-      "epoch": 0.10896221269650165,
-      "grad_norm": 460.87469482421875,
-      "learning_rate": 4.9990543078327e-05,
-      "loss": 97.5543,
-      "step": 26970
-    },
-    {
-      "epoch": 0.10900261396186929,
-      "grad_norm": 726.0296630859375,
-      "learning_rate": 4.9990446831421955e-05,
-      "loss": 97.844,
-      "step": 26980
-    },
-    {
-      "epoch": 0.10904301522723692,
-      "grad_norm": 857.0488891601562,
-      "learning_rate": 4.9990350097315275e-05,
-      "loss": 110.277,
-      "step": 26990
-    },
-    {
-      "epoch": 0.10908341649260454,
-      "grad_norm": 494.9609375,
-      "learning_rate": 4.999025287600886e-05,
-      "loss": 78.5458,
-      "step": 27000
-    },
-    {
-      "epoch": 0.10912381775797218,
-      "grad_norm": 893.1571655273438,
-      "learning_rate": 4.99901551675046e-05,
-      "loss": 109.9289,
-      "step": 27010
-    },
-    {
-      "epoch": 0.10916421902333981,
-      "grad_norm": 970.8618774414062,
-      "learning_rate": 4.99900569718044e-05,
-      "loss": 116.1883,
-      "step": 27020
-    },
-    {
-      "epoch": 0.10920462028870745,
-      "grad_norm": 745.3575439453125,
-      "learning_rate": 4.9989958288910164e-05,
-      "loss": 161.3541,
-      "step": 27030
-    },
-    {
-      "epoch": 0.10924502155407508,
-      "grad_norm": 804.6139526367188,
-      "learning_rate": 4.998985911882384e-05,
-      "loss": 105.6759,
-      "step": 27040
-    },
-    {
-      "epoch": 0.1092854228194427,
-      "grad_norm": 988.0171508789062,
-      "learning_rate": 4.998975946154734e-05,
-      "loss": 167.7433,
-      "step": 27050
-    },
-    {
-      "epoch": 0.10932582408481034,
-      "grad_norm": 570.0489501953125,
-      "learning_rate": 4.998965931708261e-05,
-      "loss": 102.1875,
-      "step": 27060
-    },
-    {
-      "epoch": 0.10936622535017797,
-      "grad_norm": 677.1341552734375,
-      "learning_rate": 4.998955868543161e-05,
-      "loss": 128.7201,
-      "step": 27070
-    },
-    {
-      "epoch": 0.1094066266155456,
-      "grad_norm": 905.8904418945312,
-      "learning_rate": 4.99894575665963e-05,
-      "loss": 107.9805,
-      "step": 27080
-    },
-    {
-      "epoch": 0.10944702788091323,
-      "grad_norm": 737.1869506835938,
-      "learning_rate": 4.9989355960578645e-05,
-      "loss": 89.4114,
-      "step": 27090
-    },
-    {
-      "epoch": 0.10948742914628086,
-      "grad_norm": 670.5479125976562,
-      "learning_rate": 4.998925386738063e-05,
-      "loss": 175.9301,
-      "step": 27100
-    },
-    {
-      "epoch": 0.1095278304116485,
-      "grad_norm": 1188.797607421875,
-      "learning_rate": 4.9989151287004244e-05,
-      "loss": 90.327,
-      "step": 27110
-    },
-    {
-      "epoch": 0.10956823167701613,
-      "grad_norm": 1030.21533203125,
-      "learning_rate": 4.9989048219451495e-05,
-      "loss": 63.2848,
-      "step": 27120
-    },
-    {
-      "epoch": 0.10960863294238375,
-      "grad_norm": 539.4580688476562,
-      "learning_rate": 4.998894466472438e-05,
-      "loss": 104.4276,
-      "step": 27130
-    },
-    {
-      "epoch": 0.10964903420775139,
-      "grad_norm": 826.8709106445312,
-      "learning_rate": 4.998884062282492e-05,
-      "loss": 92.5154,
-      "step": 27140
-    },
-    {
-      "epoch": 0.10968943547311902,
-      "grad_norm": 916.9140014648438,
-      "learning_rate": 4.998873609375516e-05,
-      "loss": 104.0753,
-      "step": 27150
-    },
-    {
-      "epoch": 0.10972983673848664,
-      "grad_norm": 478.9193115234375,
-      "learning_rate": 4.998863107751711e-05,
-      "loss": 117.0698,
-      "step": 27160
-    },
-    {
-      "epoch": 0.10977023800385428,
-      "grad_norm": 944.5336303710938,
-      "learning_rate": 4.9988525574112846e-05,
-      "loss": 151.4209,
-      "step": 27170
-    },
-    {
-      "epoch": 0.10981063926922191,
-      "grad_norm": 552.6795654296875,
-      "learning_rate": 4.99884195835444e-05,
-      "loss": 88.4628,
-      "step": 27180
-    },
-    {
-      "epoch": 0.10985104053458955,
-      "grad_norm": 954.2455444335938,
-      "learning_rate": 4.9988313105813856e-05,
-      "loss": 112.3373,
-      "step": 27190
-    },
-    {
-      "epoch": 0.10989144179995718,
-      "grad_norm": 936.270751953125,
-      "learning_rate": 4.998820614092328e-05,
-      "loss": 75.6448,
-      "step": 27200
-    },
-    {
-      "epoch": 0.1099318430653248,
-      "grad_norm": 777.1414184570312,
-      "learning_rate": 4.9988098688874763e-05,
-      "loss": 96.6067,
-      "step": 27210
-    },
-    {
-      "epoch": 0.10997224433069244,
-      "grad_norm": 1599.469482421875,
-      "learning_rate": 4.9987990749670395e-05,
-      "loss": 126.3036,
-      "step": 27220
-    },
-    {
-      "epoch": 0.11001264559606007,
-      "grad_norm": 398.20196533203125,
-      "learning_rate": 4.9987882323312287e-05,
-      "loss": 86.396,
-      "step": 27230
-    },
-    {
-      "epoch": 0.1100530468614277,
-      "grad_norm": 1596.5191650390625,
-      "learning_rate": 4.998777340980254e-05,
-      "loss": 93.4679,
-      "step": 27240
-    },
-    {
-      "epoch": 0.11009344812679533,
-      "grad_norm": 807.2232055664062,
-      "learning_rate": 4.998766400914329e-05,
-      "loss": 75.585,
-      "step": 27250
-    },
-    {
-      "epoch": 0.11013384939216296,
-      "grad_norm": 1436.22216796875,
-      "learning_rate": 4.9987554121336666e-05,
-      "loss": 102.0856,
-      "step": 27260
-    },
-    {
-      "epoch": 0.1101742506575306,
-      "grad_norm": 1310.2745361328125,
-      "learning_rate": 4.998744374638481e-05,
-      "loss": 102.2359,
-      "step": 27270
-    },
-    {
-      "epoch": 0.11021465192289823,
-      "grad_norm": 1259.0927734375,
-      "learning_rate": 4.998733288428987e-05,
-      "loss": 82.0602,
-      "step": 27280
-    },
-    {
-      "epoch": 0.11025505318826585,
-      "grad_norm": 2409.748046875,
-      "learning_rate": 4.998722153505402e-05,
-      "loss": 95.3449,
-      "step": 27290
-    },
-    {
-      "epoch": 0.11029545445363349,
-      "grad_norm": 1046.286865234375,
-      "learning_rate": 4.998710969867942e-05,
-      "loss": 113.0733,
-      "step": 27300
-    },
-    {
-      "epoch": 0.11033585571900112,
-      "grad_norm": 630.7343139648438,
-      "learning_rate": 4.9986997375168246e-05,
-      "loss": 68.2742,
-      "step": 27310
-    },
-    {
-      "epoch": 0.11037625698436875,
-      "grad_norm": 553.9161987304688,
-      "learning_rate": 4.9986884564522696e-05,
-      "loss": 145.594,
-      "step": 27320
-    },
-    {
-      "epoch": 0.11041665824973639,
-      "grad_norm": 823.818603515625,
-      "learning_rate": 4.998677126674497e-05,
-      "loss": 106.8122,
-      "step": 27330
-    },
-    {
-      "epoch": 0.11045705951510401,
-      "grad_norm": 1561.4769287109375,
-      "learning_rate": 4.9986657481837277e-05,
-      "loss": 154.897,
-      "step": 27340
-    },
-    {
-      "epoch": 0.11049746078047164,
-      "grad_norm": 1459.7176513671875,
-      "learning_rate": 4.9986543209801825e-05,
-      "loss": 140.0131,
-      "step": 27350
-    },
-    {
-      "epoch": 0.11053786204583928,
-      "grad_norm": 797.2266235351562,
-      "learning_rate": 4.998642845064086e-05,
-      "loss": 117.8721,
-      "step": 27360
-    },
-    {
-      "epoch": 0.1105782633112069,
-      "grad_norm": 902.2604370117188,
-      "learning_rate": 4.9986313204356594e-05,
-      "loss": 169.9797,
-      "step": 27370
-    },
-    {
-      "epoch": 0.11061866457657454,
-      "grad_norm": 504.2903137207031,
-      "learning_rate": 4.998619747095129e-05,
-      "loss": 89.4714,
-      "step": 27380
-    },
-    {
-      "epoch": 0.11065906584194217,
-      "grad_norm": 713.7076416015625,
-      "learning_rate": 4.998608125042721e-05,
-      "loss": 97.6478,
-      "step": 27390
-    },
-    {
-      "epoch": 0.1106994671073098,
-      "grad_norm": 1151.92822265625,
-      "learning_rate": 4.9985964542786614e-05,
-      "loss": 115.0625,
-      "step": 27400
-    },
-    {
-      "epoch": 0.11073986837267744,
-      "grad_norm": 1444.137451171875,
-      "learning_rate": 4.9985847348031764e-05,
-      "loss": 119.7467,
-      "step": 27410
-    },
-    {
-      "epoch": 0.11078026963804506,
-      "grad_norm": 3067.8701171875,
-      "learning_rate": 4.998572966616496e-05,
-      "loss": 119.108,
-      "step": 27420
-    },
-    {
-      "epoch": 0.11082067090341269,
-      "grad_norm": 759.7105712890625,
-      "learning_rate": 4.99856114971885e-05,
-      "loss": 99.5433,
-      "step": 27430
-    },
-    {
-      "epoch": 0.11086107216878033,
-      "grad_norm": 1036.834716796875,
-      "learning_rate": 4.998549284110468e-05,
-      "loss": 85.0693,
-      "step": 27440
-    },
-    {
-      "epoch": 0.11090147343414795,
-      "grad_norm": 1072.163818359375,
-      "learning_rate": 4.998537369791581e-05,
-      "loss": 122.0357,
-      "step": 27450
-    },
-    {
-      "epoch": 0.1109418746995156,
-      "grad_norm": 1213.7109375,
-      "learning_rate": 4.9985254067624215e-05,
-      "loss": 81.71,
-      "step": 27460
-    },
-    {
-      "epoch": 0.11098227596488322,
-      "grad_norm": 641.5134887695312,
-      "learning_rate": 4.998513395023223e-05,
-      "loss": 111.3763,
-      "step": 27470
-    },
-    {
-      "epoch": 0.11102267723025085,
-      "grad_norm": 1474.4072265625,
-      "learning_rate": 4.99850133457422e-05,
-      "loss": 103.8446,
-      "step": 27480
-    },
-    {
-      "epoch": 0.11106307849561849,
-      "grad_norm": 837.8353271484375,
-      "learning_rate": 4.9984892254156465e-05,
-      "loss": 85.3186,
-      "step": 27490
-    },
-    {
-      "epoch": 0.11110347976098611,
-      "grad_norm": 714.3938598632812,
-      "learning_rate": 4.99847706754774e-05,
-      "loss": 116.3188,
-      "step": 27500
-    },
-    {
-      "epoch": 0.11114388102635374,
-      "grad_norm": 1514.875,
-      "learning_rate": 4.998464860970736e-05,
-      "loss": 61.2618,
-      "step": 27510
-    },
-    {
-      "epoch": 0.11118428229172138,
-      "grad_norm": 122.57615661621094,
-      "learning_rate": 4.998452605684874e-05,
-      "loss": 103.8598,
-      "step": 27520
-    },
-    {
-      "epoch": 0.111224683557089,
-      "grad_norm": 700.09521484375,
-      "learning_rate": 4.9984403016903915e-05,
-      "loss": 96.256,
-      "step": 27530
-    },
-    {
-      "epoch": 0.11126508482245664,
-      "grad_norm": 2113.217041015625,
-      "learning_rate": 4.998427948987528e-05,
-      "loss": 94.1343,
-      "step": 27540
-    },
-    {
-      "epoch": 0.11130548608782427,
-      "grad_norm": 710.7130126953125,
-      "learning_rate": 4.998415547576527e-05,
-      "loss": 134.276,
-      "step": 27550
-    },
-    {
-      "epoch": 0.1113458873531919,
-      "grad_norm": 974.3263549804688,
-      "learning_rate": 4.9984030974576285e-05,
-      "loss": 70.3129,
-      "step": 27560
-    },
-    {
-      "epoch": 0.11138628861855954,
-      "grad_norm": 1219.5396728515625,
-      "learning_rate": 4.998390598631075e-05,
-      "loss": 119.3183,
-      "step": 27570
-    },
-    {
-      "epoch": 0.11142668988392716,
-      "grad_norm": 2306.385498046875,
-      "learning_rate": 4.998378051097111e-05,
-      "loss": 96.3586,
-      "step": 27580
-    },
-    {
-      "epoch": 0.11146709114929479,
-      "grad_norm": 676.72998046875,
-      "learning_rate": 4.99836545485598e-05,
-      "loss": 87.2172,
-      "step": 27590
-    },
-    {
-      "epoch": 0.11150749241466243,
-      "grad_norm": 1688.67578125,
-      "learning_rate": 4.998352809907928e-05,
-      "loss": 158.3624,
-      "step": 27600
-    },
-    {
-      "epoch": 0.11154789368003006,
-      "grad_norm": 3076.226806640625,
-      "learning_rate": 4.9983401162532025e-05,
-      "loss": 105.7579,
-      "step": 27610
-    },
-    {
-      "epoch": 0.1115882949453977,
-      "grad_norm": 789.8070678710938,
-      "learning_rate": 4.99832737389205e-05,
-      "loss": 114.1961,
-      "step": 27620
-    },
-    {
-      "epoch": 0.11162869621076532,
-      "grad_norm": 1299.20166015625,
-      "learning_rate": 4.998314582824719e-05,
-      "loss": 85.1794,
-      "step": 27630
-    },
-    {
-      "epoch": 0.11166909747613295,
-      "grad_norm": 611.26123046875,
-      "learning_rate": 4.998301743051459e-05,
-      "loss": 63.9051,
-      "step": 27640
-    },
-    {
-      "epoch": 0.11170949874150059,
-      "grad_norm": 967.321044921875,
-      "learning_rate": 4.99828885457252e-05,
-      "loss": 127.1219,
-      "step": 27650
-    },
-    {
-      "epoch": 0.11174990000686821,
-      "grad_norm": 1000.7772216796875,
-      "learning_rate": 4.998275917388154e-05,
-      "loss": 103.4082,
-      "step": 27660
-    },
-    {
-      "epoch": 0.11179030127223584,
-      "grad_norm": 627.480224609375,
-      "learning_rate": 4.9982629314986126e-05,
-      "loss": 96.7553,
-      "step": 27670
-    },
-    {
-      "epoch": 0.11183070253760348,
-      "grad_norm": 1001.6890869140625,
-      "learning_rate": 4.998249896904149e-05,
-      "loss": 92.7391,
-      "step": 27680
-    },
-    {
-      "epoch": 0.1118711038029711,
-      "grad_norm": 781.9347534179688,
-      "learning_rate": 4.998236813605017e-05,
-      "loss": 86.0648,
-      "step": 27690
-    },
-    {
-      "epoch": 0.11191150506833875,
-      "grad_norm": 2122.2900390625,
-      "learning_rate": 4.998223681601473e-05,
-      "loss": 106.2797,
-      "step": 27700
-    },
-    {
-      "epoch": 0.11195190633370637,
-      "grad_norm": 2083.68798828125,
-      "learning_rate": 4.9982105008937726e-05,
-      "loss": 91.295,
-      "step": 27710
-    },
-    {
-      "epoch": 0.111992307599074,
-      "grad_norm": 1193.2694091796875,
-      "learning_rate": 4.998197271482171e-05,
-      "loss": 69.5858,
-      "step": 27720
-    },
-    {
-      "epoch": 0.11203270886444164,
-      "grad_norm": 2355.148193359375,
-      "learning_rate": 4.998183993366928e-05,
-      "loss": 133.6512,
-      "step": 27730
-    },
-    {
-      "epoch": 0.11207311012980926,
-      "grad_norm": 745.9656372070312,
-      "learning_rate": 4.998170666548302e-05,
-      "loss": 99.9143,
-      "step": 27740
-    },
-    {
-      "epoch": 0.11211351139517689,
-      "grad_norm": 559.490478515625,
-      "learning_rate": 4.998157291026553e-05,
-      "loss": 116.1021,
-      "step": 27750
-    },
-    {
-      "epoch": 0.11215391266054453,
-      "grad_norm": 613.9931030273438,
-      "learning_rate": 4.998143866801942e-05,
-      "loss": 79.9714,
-      "step": 27760
-    },
-    {
-      "epoch": 0.11219431392591216,
-      "grad_norm": 838.7278442382812,
-      "learning_rate": 4.9981303938747286e-05,
-      "loss": 74.1118,
-      "step": 27770
-    },
-    {
-      "epoch": 0.1122347151912798,
-      "grad_norm": 731.7330322265625,
-      "learning_rate": 4.9981168722451776e-05,
-      "loss": 85.4063,
-      "step": 27780
-    },
-    {
-      "epoch": 0.11227511645664742,
-      "grad_norm": 826.3333129882812,
-      "learning_rate": 4.998103301913552e-05,
-      "loss": 83.0917,
-      "step": 27790
-    },
-    {
-      "epoch": 0.11231551772201505,
-      "grad_norm": 1546.168701171875,
-      "learning_rate": 4.998089682880117e-05,
-      "loss": 128.3176,
-      "step": 27800
-    },
-    {
-      "epoch": 0.11235591898738269,
-      "grad_norm": 1162.286865234375,
-      "learning_rate": 4.998076015145138e-05,
-      "loss": 122.4255,
-      "step": 27810
-    },
-    {
-      "epoch": 0.11239632025275031,
-      "grad_norm": 1132.9892578125,
-      "learning_rate": 4.9980622987088795e-05,
-      "loss": 98.7629,
-      "step": 27820
-    },
-    {
-      "epoch": 0.11243672151811794,
-      "grad_norm": 1079.5322265625,
-      "learning_rate": 4.9980485335716114e-05,
-      "loss": 119.6111,
-      "step": 27830
-    },
-    {
-      "epoch": 0.11247712278348558,
-      "grad_norm": 1292.2762451171875,
-      "learning_rate": 4.9980347197336005e-05,
-      "loss": 72.3516,
-      "step": 27840
-    },
-    {
-      "epoch": 0.1125175240488532,
-      "grad_norm": 767.2022094726562,
-      "learning_rate": 4.998020857195117e-05,
-      "loss": 144.7648,
-      "step": 27850
-    },
-    {
-      "epoch": 0.11255792531422085,
-      "grad_norm": 679.2119750976562,
-      "learning_rate": 4.998006945956431e-05,
-      "loss": 117.3672,
-      "step": 27860
-    },
-    {
-      "epoch": 0.11259832657958847,
-      "grad_norm": 1033.4124755859375,
-      "learning_rate": 4.997992986017813e-05,
-      "loss": 91.8273,
-      "step": 27870
-    },
-    {
-      "epoch": 0.1126387278449561,
-      "grad_norm": 1781.468505859375,
-      "learning_rate": 4.997978977379536e-05,
-      "loss": 108.8957,
-      "step": 27880
-    },
-    {
-      "epoch": 0.11267912911032374,
-      "grad_norm": 875.4254150390625,
-      "learning_rate": 4.9979649200418735e-05,
-      "loss": 65.4817,
-      "step": 27890
-    },
-    {
-      "epoch": 0.11271953037569137,
-      "grad_norm": 760.1170654296875,
-      "learning_rate": 4.997950814005098e-05,
-      "loss": 105.1915,
-      "step": 27900
-    },
-    {
-      "epoch": 0.11275993164105899,
-      "grad_norm": 816.7872924804688,
-      "learning_rate": 4.997936659269486e-05,
-      "loss": 72.9768,
-      "step": 27910
-    },
-    {
-      "epoch": 0.11280033290642663,
-      "grad_norm": 1037.8052978515625,
-      "learning_rate": 4.997922455835311e-05,
-      "loss": 84.1713,
-      "step": 27920
-    },
-    {
-      "epoch": 0.11284073417179426,
-      "grad_norm": 599.27197265625,
-      "learning_rate": 4.9979082037028535e-05,
-      "loss": 89.3494,
-      "step": 27930
-    },
-    {
-      "epoch": 0.1128811354371619,
-      "grad_norm": 2486.75830078125,
-      "learning_rate": 4.9978939028723894e-05,
-      "loss": 96.0437,
-      "step": 27940
-    },
-    {
-      "epoch": 0.11292153670252952,
-      "grad_norm": 933.3982543945312,
-      "learning_rate": 4.9978795533441966e-05,
-      "loss": 159.8028,
-      "step": 27950
-    },
-    {
-      "epoch": 0.11296193796789715,
-      "grad_norm": 783.9051513671875,
-      "learning_rate": 4.997865155118557e-05,
-      "loss": 80.7255,
-      "step": 27960
-    },
-    {
-      "epoch": 0.11300233923326479,
-      "grad_norm": 1356.3843994140625,
-      "learning_rate": 4.9978507081957494e-05,
-      "loss": 135.2147,
-      "step": 27970
-    },
-    {
-      "epoch": 0.11304274049863242,
-      "grad_norm": 2129.216796875,
-      "learning_rate": 4.997836212576057e-05,
-      "loss": 73.888,
-      "step": 27980
-    },
-    {
-      "epoch": 0.11308314176400004,
-      "grad_norm": 959.6722412109375,
-      "learning_rate": 4.9978216682597614e-05,
-      "loss": 159.4311,
-      "step": 27990
-    },
-    {
-      "epoch": 0.11312354302936768,
-      "grad_norm": 2303.733642578125,
-      "learning_rate": 4.997807075247146e-05,
-      "loss": 98.653,
-      "step": 28000
-    },
-    {
-      "epoch": 0.11316394429473531,
-      "grad_norm": 906.4110717773438,
-      "learning_rate": 4.997792433538496e-05,
-      "loss": 97.6867,
-      "step": 28010
-    },
-    {
-      "epoch": 0.11320434556010295,
-      "grad_norm": 814.7039184570312,
-      "learning_rate": 4.997777743134097e-05,
-      "loss": 91.2434,
-      "step": 28020
-    },
-    {
-      "epoch": 0.11324474682547057,
-      "grad_norm": 496.1783752441406,
-      "learning_rate": 4.9977630040342346e-05,
-      "loss": 109.5629,
-      "step": 28030
-    },
-    {
-      "epoch": 0.1132851480908382,
-      "grad_norm": 1374.5599365234375,
-      "learning_rate": 4.997748216239196e-05,
-      "loss": 115.9625,
-      "step": 28040
-    },
-    {
-      "epoch": 0.11332554935620584,
-      "grad_norm": 1610.67529296875,
-      "learning_rate": 4.997733379749271e-05,
-      "loss": 138.7809,
-      "step": 28050
-    },
-    {
-      "epoch": 0.11336595062157347,
-      "grad_norm": 655.5691528320312,
-      "learning_rate": 4.9977184945647473e-05,
-      "loss": 72.344,
-      "step": 28060
-    },
-    {
-      "epoch": 0.11340635188694109,
-      "grad_norm": 1008.762939453125,
-      "learning_rate": 4.9977035606859156e-05,
-      "loss": 170.6503,
-      "step": 28070
-    },
-    {
-      "epoch": 0.11344675315230873,
-      "grad_norm": 692.5789184570312,
-      "learning_rate": 4.9976885781130665e-05,
-      "loss": 79.409,
-      "step": 28080
-    },
-    {
-      "epoch": 0.11348715441767636,
-      "grad_norm": 2432.585693359375,
-      "learning_rate": 4.9976735468464935e-05,
-      "loss": 181.1055,
-      "step": 28090
-    },
-    {
-      "epoch": 0.113527555683044,
-      "grad_norm": 1090.251220703125,
-      "learning_rate": 4.997658466886489e-05,
-      "loss": 114.0751,
-      "step": 28100
-    },
-    {
-      "epoch": 0.11356795694841162,
-      "grad_norm": 861.4591674804688,
-      "learning_rate": 4.997643338233346e-05,
-      "loss": 97.7165,
-      "step": 28110
-    },
-    {
-      "epoch": 0.11360835821377925,
-      "grad_norm": 1879.217529296875,
-      "learning_rate": 4.997628160887361e-05,
-      "loss": 99.4693,
-      "step": 28120
-    },
-    {
-      "epoch": 0.11364875947914689,
-      "grad_norm": 617.5592651367188,
-      "learning_rate": 4.997612934848829e-05,
-      "loss": 98.8888,
-      "step": 28130
-    },
-    {
-      "epoch": 0.11368916074451452,
-      "grad_norm": 671.50341796875,
-      "learning_rate": 4.997597660118046e-05,
-      "loss": 96.949,
-      "step": 28140
-    },
-    {
-      "epoch": 0.11372956200988214,
-      "grad_norm": 800.4517211914062,
-      "learning_rate": 4.9975823366953124e-05,
-      "loss": 71.7921,
-      "step": 28150
-    },
-    {
-      "epoch": 0.11376996327524978,
-      "grad_norm": 889.1356201171875,
-      "learning_rate": 4.9975669645809244e-05,
-      "loss": 124.3523,
-      "step": 28160
-    },
-    {
-      "epoch": 0.11381036454061741,
-      "grad_norm": 1013.8363037109375,
-      "learning_rate": 4.997551543775182e-05,
-      "loss": 111.264,
-      "step": 28170
-    },
-    {
-      "epoch": 0.11385076580598505,
-      "grad_norm": 1395.8243408203125,
-      "learning_rate": 4.997536074278387e-05,
-      "loss": 112.4192,
-      "step": 28180
-    },
-    {
-      "epoch": 0.11389116707135268,
-      "grad_norm": 1430.334228515625,
-      "learning_rate": 4.997520556090841e-05,
-      "loss": 121.1184,
-      "step": 28190
-    },
-    {
-      "epoch": 0.1139315683367203,
-      "grad_norm": 1214.85400390625,
-      "learning_rate": 4.9975049892128455e-05,
-      "loss": 87.6032,
-      "step": 28200
-    },
-    {
-      "epoch": 0.11397196960208794,
-      "grad_norm": 867.681396484375,
-      "learning_rate": 4.9974893736447045e-05,
-      "loss": 90.5741,
-      "step": 28210
-    },
-    {
-      "epoch": 0.11401237086745557,
-      "grad_norm": 759.8204345703125,
-      "learning_rate": 4.997473709386722e-05,
-      "loss": 79.1436,
-      "step": 28220
-    },
-    {
-      "epoch": 0.1140527721328232,
-      "grad_norm": 1119.670654296875,
-      "learning_rate": 4.997457996439204e-05,
-      "loss": 89.3735,
-      "step": 28230
-    },
-    {
-      "epoch": 0.11409317339819083,
-      "grad_norm": 843.7247924804688,
-      "learning_rate": 4.997442234802456e-05,
-      "loss": 111.084,
-      "step": 28240
-    },
-    {
-      "epoch": 0.11413357466355846,
-      "grad_norm": 775.5316772460938,
-      "learning_rate": 4.997426424476787e-05,
-      "loss": 96.1073,
-      "step": 28250
-    },
-    {
-      "epoch": 0.1141739759289261,
-      "grad_norm": 1531.5697021484375,
-      "learning_rate": 4.9974105654625036e-05,
-      "loss": 82.1885,
-      "step": 28260
-    },
-    {
-      "epoch": 0.11421437719429373,
-      "grad_norm": 693.6162719726562,
-      "learning_rate": 4.997394657759915e-05,
-      "loss": 119.7303,
-      "step": 28270
-    },
-    {
-      "epoch": 0.11425477845966135,
-      "grad_norm": 4390.083984375,
-      "learning_rate": 4.997378701369332e-05,
-      "loss": 156.4863,
-      "step": 28280
-    },
-    {
-      "epoch": 0.11429517972502899,
-      "grad_norm": 781.9625854492188,
-      "learning_rate": 4.9973626962910656e-05,
-      "loss": 90.4073,
-      "step": 28290
-    },
-    {
-      "epoch": 0.11433558099039662,
-      "grad_norm": 1085.052978515625,
-      "learning_rate": 4.9973466425254286e-05,
-      "loss": 74.8492,
-      "step": 28300
-    },
-    {
-      "epoch": 0.11437598225576424,
-      "grad_norm": 975.8035278320312,
-      "learning_rate": 4.997330540072732e-05,
-      "loss": 108.0216,
-      "step": 28310
-    },
-    {
-      "epoch": 0.11441638352113188,
-      "grad_norm": 664.2997436523438,
-      "learning_rate": 4.997314388933291e-05,
-      "loss": 96.9389,
-      "step": 28320
-    },
-    {
-      "epoch": 0.11445678478649951,
-      "grad_norm": 395.10052490234375,
-      "learning_rate": 4.997298189107421e-05,
-      "loss": 114.4039,
-      "step": 28330
-    },
-    {
-      "epoch": 0.11449718605186715,
-      "grad_norm": 894.0714721679688,
-      "learning_rate": 4.9972819405954366e-05,
-      "loss": 66.7973,
-      "step": 28340
-    },
-    {
-      "epoch": 0.11453758731723478,
-      "grad_norm": 3636.150390625,
-      "learning_rate": 4.997265643397654e-05,
-      "loss": 137.2116,
-      "step": 28350
-    },
-    {
-      "epoch": 0.1145779885826024,
-      "grad_norm": 543.2451782226562,
-      "learning_rate": 4.997249297514394e-05,
-      "loss": 72.1923,
-      "step": 28360
-    },
-    {
-      "epoch": 0.11461838984797004,
-      "grad_norm": 1202.521484375,
-      "learning_rate": 4.997232902945971e-05,
-      "loss": 101.5618,
-      "step": 28370
-    },
-    {
-      "epoch": 0.11465879111333767,
-      "grad_norm": 1286.06494140625,
-      "learning_rate": 4.997216459692709e-05,
-      "loss": 99.6231,
-      "step": 28380
-    },
-    {
-      "epoch": 0.1146991923787053,
-      "grad_norm": 747.28271484375,
-      "learning_rate": 4.997199967754925e-05,
-      "loss": 101.1043,
-      "step": 28390
-    },
-    {
-      "epoch": 0.11473959364407293,
-      "grad_norm": 397.4785461425781,
-      "learning_rate": 4.997183427132943e-05,
-      "loss": 99.9764,
-      "step": 28400
-    },
-    {
-      "epoch": 0.11477999490944056,
-      "grad_norm": 1810.4814453125,
-      "learning_rate": 4.997166837827084e-05,
-      "loss": 107.5269,
-      "step": 28410
-    },
-    {
-      "epoch": 0.1148203961748082,
-      "grad_norm": 1796.510009765625,
-      "learning_rate": 4.997150199837671e-05,
-      "loss": 98.7808,
-      "step": 28420
-    },
-    {
-      "epoch": 0.11486079744017583,
-      "grad_norm": 997.3446044921875,
-      "learning_rate": 4.997133513165031e-05,
-      "loss": 101.5585,
-      "step": 28430
-    },
-    {
-      "epoch": 0.11490119870554345,
-      "grad_norm": 575.0126342773438,
-      "learning_rate": 4.9971167778094863e-05,
-      "loss": 115.6277,
-      "step": 28440
-    },
-    {
-      "epoch": 0.11494159997091109,
-      "grad_norm": 1271.26953125,
-      "learning_rate": 4.997099993771365e-05,
-      "loss": 60.656,
-      "step": 28450
-    },
-    {
-      "epoch": 0.11498200123627872,
-      "grad_norm": 1329.1561279296875,
-      "learning_rate": 4.997083161050994e-05,
-      "loss": 83.6766,
-      "step": 28460
-    },
-    {
-      "epoch": 0.11502240250164635,
-      "grad_norm": 527.1378784179688,
-      "learning_rate": 4.9970662796487e-05,
-      "loss": 87.6013,
-      "step": 28470
-    },
-    {
-      "epoch": 0.11506280376701399,
-      "grad_norm": 888.7230224609375,
-      "learning_rate": 4.997049349564814e-05,
-      "loss": 86.2665,
-      "step": 28480
-    },
-    {
-      "epoch": 0.11510320503238161,
-      "grad_norm": 338.1375732421875,
-      "learning_rate": 4.997032370799666e-05,
-      "loss": 130.7292,
-      "step": 28490
-    },
-    {
-      "epoch": 0.11514360629774925,
-      "grad_norm": 2081.712158203125,
-      "learning_rate": 4.997015343353585e-05,
-      "loss": 128.4502,
-      "step": 28500
-    },
-    {
-      "epoch": 0.11518400756311688,
-      "grad_norm": 1699.38623046875,
-      "learning_rate": 4.996998267226905e-05,
-      "loss": 115.8591,
-      "step": 28510
-    },
-    {
-      "epoch": 0.1152244088284845,
-      "grad_norm": 467.2063293457031,
-      "learning_rate": 4.996981142419959e-05,
-      "loss": 73.2754,
-      "step": 28520
-    },
-    {
-      "epoch": 0.11526481009385214,
-      "grad_norm": 804.551025390625,
-      "learning_rate": 4.996963968933079e-05,
-      "loss": 67.7496,
-      "step": 28530
-    },
-    {
-      "epoch": 0.11530521135921977,
-      "grad_norm": 1599.342529296875,
-      "learning_rate": 4.996946746766601e-05,
-      "loss": 91.9091,
-      "step": 28540
-    },
-    {
-      "epoch": 0.1153456126245874,
-      "grad_norm": 1292.8143310546875,
-      "learning_rate": 4.996929475920862e-05,
-      "loss": 113.546,
-      "step": 28550
-    },
-    {
-      "epoch": 0.11538601388995504,
-      "grad_norm": 1679.7252197265625,
-      "learning_rate": 4.9969121563961956e-05,
-      "loss": 83.8834,
-      "step": 28560
-    },
-    {
-      "epoch": 0.11542641515532266,
-      "grad_norm": 1275.5130615234375,
-      "learning_rate": 4.9968947881929414e-05,
-      "loss": 115.3006,
-      "step": 28570
-    },
-    {
-      "epoch": 0.1154668164206903,
-      "grad_norm": 959.7946166992188,
-      "learning_rate": 4.996877371311439e-05,
-      "loss": 87.6582,
-      "step": 28580
-    },
-    {
-      "epoch": 0.11550721768605793,
-      "grad_norm": 1284.703369140625,
-      "learning_rate": 4.996859905752026e-05,
-      "loss": 132.2889,
-      "step": 28590
-    },
-    {
-      "epoch": 0.11554761895142555,
-      "grad_norm": 0.0,
-      "learning_rate": 4.996842391515044e-05,
-      "loss": 91.88,
-      "step": 28600
-    },
-    {
-      "epoch": 0.1155880202167932,
-      "grad_norm": 1002.761962890625,
-      "learning_rate": 4.996824828600834e-05,
-      "loss": 125.265,
-      "step": 28610
-    },
-    {
-      "epoch": 0.11562842148216082,
-      "grad_norm": 665.4733276367188,
-      "learning_rate": 4.996807217009738e-05,
-      "loss": 99.0348,
-      "step": 28620
-    },
-    {
-      "epoch": 0.11566882274752845,
-      "grad_norm": 1002.7938232421875,
-      "learning_rate": 4.996789556742101e-05,
-      "loss": 75.1195,
-      "step": 28630
-    },
-    {
-      "epoch": 0.11570922401289609,
-      "grad_norm": 858.1141357421875,
-      "learning_rate": 4.996771847798265e-05,
-      "loss": 73.2454,
-      "step": 28640
-    },
-    {
-      "epoch": 0.11574962527826371,
-      "grad_norm": 1226.3555908203125,
-      "learning_rate": 4.996754090178577e-05,
-      "loss": 146.9862,
-      "step": 28650
-    },
-    {
-      "epoch": 0.11579002654363135,
-      "grad_norm": 1391.651123046875,
-      "learning_rate": 4.996736283883382e-05,
-      "loss": 142.1579,
-      "step": 28660
-    },
-    {
-      "epoch": 0.11583042780899898,
-      "grad_norm": 1489.2674560546875,
-      "learning_rate": 4.9967184289130286e-05,
-      "loss": 115.9638,
-      "step": 28670
-    },
-    {
-      "epoch": 0.1158708290743666,
-      "grad_norm": 1078.4761962890625,
-      "learning_rate": 4.9967005252678634e-05,
-      "loss": 118.9085,
-      "step": 28680
-    },
-    {
-      "epoch": 0.11591123033973424,
-      "grad_norm": 2236.838623046875,
-      "learning_rate": 4.9966825729482364e-05,
-      "loss": 130.9481,
-      "step": 28690
-    },
-    {
-      "epoch": 0.11595163160510187,
-      "grad_norm": 862.959716796875,
-      "learning_rate": 4.996664571954497e-05,
-      "loss": 111.0184,
-      "step": 28700
-    },
-    {
-      "epoch": 0.1159920328704695,
-      "grad_norm": 1136.712646484375,
-      "learning_rate": 4.996646522286997e-05,
-      "loss": 103.9295,
-      "step": 28710
-    },
-    {
-      "epoch": 0.11603243413583714,
-      "grad_norm": 1675.60791015625,
-      "learning_rate": 4.996628423946087e-05,
-      "loss": 109.6175,
-      "step": 28720
-    },
-    {
-      "epoch": 0.11607283540120476,
-      "grad_norm": 1874.5843505859375,
-      "learning_rate": 4.996610276932121e-05,
-      "loss": 85.4252,
-      "step": 28730
-    },
-    {
-      "epoch": 0.1161132366665724,
-      "grad_norm": 1308.438232421875,
-      "learning_rate": 4.996592081245451e-05,
-      "loss": 115.684,
-      "step": 28740
-    },
-    {
-      "epoch": 0.11615363793194003,
-      "grad_norm": 428.2576904296875,
-      "learning_rate": 4.996573836886435e-05,
-      "loss": 57.5053,
-      "step": 28750
-    },
-    {
-      "epoch": 0.11619403919730766,
-      "grad_norm": 1469.609619140625,
-      "learning_rate": 4.9965555438554254e-05,
-      "loss": 77.9164,
-      "step": 28760
-    },
-    {
-      "epoch": 0.1162344404626753,
-      "grad_norm": 2238.658447265625,
-      "learning_rate": 4.9965372021527814e-05,
-      "loss": 129.5649,
-      "step": 28770
-    },
-    {
-      "epoch": 0.11627484172804292,
-      "grad_norm": 1731.619140625,
-      "learning_rate": 4.996518811778858e-05,
-      "loss": 92.2746,
-      "step": 28780
-    },
-    {
-      "epoch": 0.11631524299341055,
-      "grad_norm": 2262.8583984375,
-      "learning_rate": 4.996500372734015e-05,
-      "loss": 126.2437,
-      "step": 28790
-    },
-    {
-      "epoch": 0.11635564425877819,
-      "grad_norm": 1330.55908203125,
-      "learning_rate": 4.9964818850186135e-05,
-      "loss": 118.3177,
-      "step": 28800
-    },
-    {
-      "epoch": 0.11639604552414581,
-      "grad_norm": 1293.2772216796875,
-      "learning_rate": 4.9964633486330116e-05,
-      "loss": 131.9247,
-      "step": 28810
-    },
-    {
-      "epoch": 0.11643644678951345,
-      "grad_norm": 626.7706298828125,
-      "learning_rate": 4.9964447635775714e-05,
-      "loss": 80.6821,
-      "step": 28820
-    },
-    {
-      "epoch": 0.11647684805488108,
-      "grad_norm": 680.4353637695312,
-      "learning_rate": 4.996426129852655e-05,
-      "loss": 79.4926,
-      "step": 28830
-    },
-    {
-      "epoch": 0.1165172493202487,
-      "grad_norm": 1769.197998046875,
-      "learning_rate": 4.996407447458626e-05,
-      "loss": 105.4986,
-      "step": 28840
-    },
-    {
-      "epoch": 0.11655765058561635,
-      "grad_norm": 1792.2196044921875,
-      "learning_rate": 4.996388716395848e-05,
-      "loss": 108.3764,
-      "step": 28850
-    },
-    {
-      "epoch": 0.11659805185098397,
-      "grad_norm": 582.6719970703125,
-      "learning_rate": 4.996369936664688e-05,
-      "loss": 79.2789,
-      "step": 28860
-    },
-    {
-      "epoch": 0.1166384531163516,
-      "grad_norm": 1305.8157958984375,
-      "learning_rate": 4.99635110826551e-05,
-      "loss": 112.6347,
-      "step": 28870
-    },
-    {
-      "epoch": 0.11667885438171924,
-      "grad_norm": 1331.4168701171875,
-      "learning_rate": 4.996332231198683e-05,
-      "loss": 76.5,
-      "step": 28880
-    },
-    {
-      "epoch": 0.11671925564708686,
-      "grad_norm": 1037.36767578125,
-      "learning_rate": 4.996313305464573e-05,
-      "loss": 159.2613,
-      "step": 28890
-    },
-    {
-      "epoch": 0.1167596569124545,
-      "grad_norm": 1378.345458984375,
-      "learning_rate": 4.99629433106355e-05,
-      "loss": 157.8604,
-      "step": 28900
-    },
-    {
-      "epoch": 0.11680005817782213,
-      "grad_norm": 846.559814453125,
-      "learning_rate": 4.9962753079959836e-05,
-      "loss": 112.2688,
-      "step": 28910
-    },
-    {
-      "epoch": 0.11684045944318976,
-      "grad_norm": 1050.1434326171875,
-      "learning_rate": 4.996256236262245e-05,
-      "loss": 88.0946,
-      "step": 28920
-    },
-    {
-      "epoch": 0.1168808607085574,
-      "grad_norm": 410.0029296875,
-      "learning_rate": 4.996237115862706e-05,
-      "loss": 59.4956,
-      "step": 28930
-    },
-    {
-      "epoch": 0.11692126197392502,
-      "grad_norm": 573.1680908203125,
-      "learning_rate": 4.99621794679774e-05,
-      "loss": 111.6136,
-      "step": 28940
-    },
-    {
-      "epoch": 0.11696166323929265,
-      "grad_norm": 869.5213012695312,
-      "learning_rate": 4.996198729067719e-05,
-      "loss": 146.8108,
-      "step": 28950
-    },
-    {
-      "epoch": 0.11700206450466029,
-      "grad_norm": 634.8200073242188,
-      "learning_rate": 4.99617946267302e-05,
-      "loss": 119.8773,
-      "step": 28960
-    },
-    {
-      "epoch": 0.11704246577002791,
-      "grad_norm": 1196.7987060546875,
-      "learning_rate": 4.996160147614016e-05,
-      "loss": 71.1213,
-      "step": 28970
-    },
-    {
-      "epoch": 0.11708286703539555,
-      "grad_norm": 1217.557861328125,
-      "learning_rate": 4.996140783891085e-05,
-      "loss": 121.3136,
-      "step": 28980
-    },
-    {
-      "epoch": 0.11712326830076318,
-      "grad_norm": 1255.81396484375,
-      "learning_rate": 4.9961213715046045e-05,
-      "loss": 116.6792,
-      "step": 28990
-    },
-    {
-      "epoch": 0.11716366956613081,
-      "grad_norm": 627.504150390625,
-      "learning_rate": 4.996101910454953e-05,
-      "loss": 75.7639,
-      "step": 29000
-    },
-    {
-      "epoch": 0.11720407083149845,
-      "grad_norm": 529.0521240234375,
-      "learning_rate": 4.996082400742509e-05,
-      "loss": 139.7255,
-      "step": 29010
-    },
-    {
-      "epoch": 0.11724447209686607,
-      "grad_norm": 1349.41357421875,
-      "learning_rate": 4.996062842367654e-05,
-      "loss": 138.5896,
-      "step": 29020
-    },
-    {
-      "epoch": 0.1172848733622337,
-      "grad_norm": 1360.37255859375,
-      "learning_rate": 4.9960432353307686e-05,
-      "loss": 80.0865,
-      "step": 29030
-    },
-    {
-      "epoch": 0.11732527462760134,
-      "grad_norm": 1388.444091796875,
-      "learning_rate": 4.996023579632236e-05,
-      "loss": 112.3245,
-      "step": 29040
-    },
-    {
-      "epoch": 0.11736567589296897,
-      "grad_norm": 1930.4471435546875,
-      "learning_rate": 4.996003875272438e-05,
-      "loss": 114.5048,
-      "step": 29050
-    },
-    {
-      "epoch": 0.1174060771583366,
-      "grad_norm": 569.8203125,
-      "learning_rate": 4.9959841222517596e-05,
-      "loss": 62.5382,
-      "step": 29060
-    },
-    {
-      "epoch": 0.11744647842370423,
-      "grad_norm": 455.6988830566406,
-      "learning_rate": 4.9959643205705854e-05,
-      "loss": 100.6463,
-      "step": 29070
-    },
-    {
-      "epoch": 0.11748687968907186,
-      "grad_norm": 908.3660888671875,
-      "learning_rate": 4.995944470229302e-05,
-      "loss": 114.3609,
-      "step": 29080
-    },
-    {
-      "epoch": 0.1175272809544395,
-      "grad_norm": 755.6206665039062,
-      "learning_rate": 4.9959245712282966e-05,
-      "loss": 95.5638,
-      "step": 29090
-    },
-    {
-      "epoch": 0.11756768221980712,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9959046235679565e-05,
-      "loss": 64.4007,
-      "step": 29100
-    },
-    {
-      "epoch": 0.11760808348517475,
-      "grad_norm": 808.4530639648438,
-      "learning_rate": 4.9958846272486704e-05,
-      "loss": 112.1131,
-      "step": 29110
-    },
-    {
-      "epoch": 0.11764848475054239,
-      "grad_norm": 1141.5194091796875,
-      "learning_rate": 4.9958645822708285e-05,
-      "loss": 87.7617,
-      "step": 29120
-    },
-    {
-      "epoch": 0.11768888601591002,
-      "grad_norm": 653.6248779296875,
-      "learning_rate": 4.995844488634822e-05,
-      "loss": 70.7442,
-      "step": 29130
-    },
-    {
-      "epoch": 0.11772928728127766,
-      "grad_norm": 1141.1375732421875,
-      "learning_rate": 4.9958243463410414e-05,
-      "loss": 100.08,
-      "step": 29140
-    },
-    {
-      "epoch": 0.11776968854664528,
-      "grad_norm": 675.2451171875,
-      "learning_rate": 4.995804155389881e-05,
-      "loss": 113.5737,
-      "step": 29150
-    },
-    {
-      "epoch": 0.11781008981201291,
-      "grad_norm": 822.73486328125,
-      "learning_rate": 4.995783915781734e-05,
-      "loss": 78.6838,
-      "step": 29160
-    },
-    {
-      "epoch": 0.11785049107738055,
-      "grad_norm": 1152.279541015625,
-      "learning_rate": 4.995763627516994e-05,
-      "loss": 124.2494,
-      "step": 29170
-    },
-    {
-      "epoch": 0.11789089234274817,
-      "grad_norm": 696.2450561523438,
-      "learning_rate": 4.995743290596057e-05,
-      "loss": 85.5477,
-      "step": 29180
-    },
-    {
-      "epoch": 0.1179312936081158,
-      "grad_norm": 669.2838745117188,
-      "learning_rate": 4.9957229050193197e-05,
-      "loss": 158.3468,
-      "step": 29190
-    },
-    {
-      "epoch": 0.11797169487348344,
-      "grad_norm": 629.5454711914062,
-      "learning_rate": 4.9957024707871806e-05,
-      "loss": 72.3909,
-      "step": 29200
-    },
-    {
-      "epoch": 0.11801209613885107,
-      "grad_norm": 2218.03564453125,
-      "learning_rate": 4.995681987900036e-05,
-      "loss": 131.4396,
-      "step": 29210
-    },
-    {
-      "epoch": 0.1180524974042187,
-      "grad_norm": 1187.0064697265625,
-      "learning_rate": 4.995661456358286e-05,
-      "loss": 91.9417,
-      "step": 29220
-    },
-    {
-      "epoch": 0.11809289866958633,
-      "grad_norm": 539.1826171875,
-      "learning_rate": 4.995640876162332e-05,
-      "loss": 119.5583,
-      "step": 29230
-    },
-    {
-      "epoch": 0.11813329993495396,
-      "grad_norm": 1318.2589111328125,
-      "learning_rate": 4.9956202473125736e-05,
-      "loss": 115.7666,
-      "step": 29240
-    },
-    {
-      "epoch": 0.1181737012003216,
-      "grad_norm": 466.1606750488281,
-      "learning_rate": 4.995599569809414e-05,
-      "loss": 118.0604,
-      "step": 29250
-    },
-    {
-      "epoch": 0.11821410246568922,
-      "grad_norm": 797.7738647460938,
-      "learning_rate": 4.9955788436532545e-05,
-      "loss": 72.8938,
-      "step": 29260
-    },
-    {
-      "epoch": 0.11825450373105685,
-      "grad_norm": 1671.8890380859375,
-      "learning_rate": 4.995558068844503e-05,
-      "loss": 96.0493,
-      "step": 29270
-    },
-    {
-      "epoch": 0.11829490499642449,
-      "grad_norm": 934.203125,
-      "learning_rate": 4.99553724538356e-05,
-      "loss": 97.1457,
-      "step": 29280
-    },
-    {
-      "epoch": 0.11833530626179212,
-      "grad_norm": 1040.1793212890625,
-      "learning_rate": 4.9955163732708346e-05,
-      "loss": 138.6697,
-      "step": 29290
-    },
-    {
-      "epoch": 0.11837570752715976,
-      "grad_norm": 2738.3046875,
-      "learning_rate": 4.9954954525067334e-05,
-      "loss": 120.2477,
-      "step": 29300
-    },
-    {
-      "epoch": 0.11841610879252738,
-      "grad_norm": 543.4052124023438,
-      "learning_rate": 4.995474483091662e-05,
-      "loss": 129.9302,
-      "step": 29310
-    },
-    {
-      "epoch": 0.11845651005789501,
-      "grad_norm": 534.146240234375,
-      "learning_rate": 4.995453465026032e-05,
-      "loss": 67.6277,
-      "step": 29320
-    },
-    {
-      "epoch": 0.11849691132326265,
-      "grad_norm": 1281.582275390625,
-      "learning_rate": 4.995432398310252e-05,
-      "loss": 117.349,
-      "step": 29330
-    },
-    {
-      "epoch": 0.11853731258863028,
-      "grad_norm": 3242.07568359375,
-      "learning_rate": 4.995411282944732e-05,
-      "loss": 136.7699,
-      "step": 29340
-    },
-    {
-      "epoch": 0.1185777138539979,
-      "grad_norm": 1517.0467529296875,
-      "learning_rate": 4.9953901189298845e-05,
-      "loss": 127.2586,
-      "step": 29350
-    },
-    {
-      "epoch": 0.11861811511936554,
-      "grad_norm": 689.41015625,
-      "learning_rate": 4.9953689062661226e-05,
-      "loss": 108.8368,
-      "step": 29360
-    },
-    {
-      "epoch": 0.11865851638473317,
-      "grad_norm": 663.7586669921875,
-      "learning_rate": 4.995347644953858e-05,
-      "loss": 86.3188,
-      "step": 29370
-    },
-    {
-      "epoch": 0.11869891765010081,
-      "grad_norm": 1621.5384521484375,
-      "learning_rate": 4.9953263349935074e-05,
-      "loss": 100.3996,
-      "step": 29380
-    },
-    {
-      "epoch": 0.11873931891546843,
-      "grad_norm": 1026.15576171875,
-      "learning_rate": 4.995304976385484e-05,
-      "loss": 124.3401,
-      "step": 29390
-    },
-    {
-      "epoch": 0.11877972018083606,
-      "grad_norm": 840.7376098632812,
-      "learning_rate": 4.995283569130207e-05,
-      "loss": 99.3957,
-      "step": 29400
-    },
-    {
-      "epoch": 0.1188201214462037,
-      "grad_norm": 1305.7696533203125,
-      "learning_rate": 4.995262113228091e-05,
-      "loss": 87.9966,
-      "step": 29410
-    },
-    {
-      "epoch": 0.11886052271157133,
-      "grad_norm": 1312.4971923828125,
-      "learning_rate": 4.9952406086795564e-05,
-      "loss": 114.1831,
-      "step": 29420
-    },
-    {
-      "epoch": 0.11890092397693895,
-      "grad_norm": 1050.6826171875,
-      "learning_rate": 4.995219055485021e-05,
-      "loss": 117.2113,
-      "step": 29430
-    },
-    {
-      "epoch": 0.11894132524230659,
-      "grad_norm": 733.8092041015625,
-      "learning_rate": 4.995197453644905e-05,
-      "loss": 120.497,
-      "step": 29440
-    },
-    {
-      "epoch": 0.11898172650767422,
-      "grad_norm": 980.0513305664062,
-      "learning_rate": 4.9951758031596304e-05,
-      "loss": 97.3482,
-      "step": 29450
-    },
-    {
-      "epoch": 0.11902212777304186,
-      "grad_norm": 646.7477416992188,
-      "learning_rate": 4.995154104029619e-05,
-      "loss": 103.4561,
-      "step": 29460
-    },
-    {
-      "epoch": 0.11906252903840948,
-      "grad_norm": 2361.02978515625,
-      "learning_rate": 4.9951323562552934e-05,
-      "loss": 104.5779,
-      "step": 29470
-    },
-    {
-      "epoch": 0.11910293030377711,
-      "grad_norm": 844.0553588867188,
-      "learning_rate": 4.995110559837078e-05,
-      "loss": 91.3419,
-      "step": 29480
-    },
-    {
-      "epoch": 0.11914333156914475,
-      "grad_norm": 2193.67431640625,
-      "learning_rate": 4.995088714775398e-05,
-      "loss": 127.1109,
-      "step": 29490
-    },
-    {
-      "epoch": 0.11918373283451238,
-      "grad_norm": 732.5401611328125,
-      "learning_rate": 4.995066821070679e-05,
-      "loss": 120.7417,
-      "step": 29500
-    },
-    {
-      "epoch": 0.11922413409988,
-      "grad_norm": 1001.8763427734375,
-      "learning_rate": 4.995044878723348e-05,
-      "loss": 114.4594,
-      "step": 29510
-    },
-    {
-      "epoch": 0.11926453536524764,
-      "grad_norm": 928.3828735351562,
-      "learning_rate": 4.995022887733832e-05,
-      "loss": 83.6775,
-      "step": 29520
-    },
-    {
-      "epoch": 0.11930493663061527,
-      "grad_norm": 858.733154296875,
-      "learning_rate": 4.99500084810256e-05,
-      "loss": 144.456,
-      "step": 29530
-    },
-    {
-      "epoch": 0.11934533789598291,
-      "grad_norm": 2473.096435546875,
-      "learning_rate": 4.994978759829963e-05,
-      "loss": 117.7391,
-      "step": 29540
-    },
-    {
-      "epoch": 0.11938573916135053,
-      "grad_norm": 1232.745361328125,
-      "learning_rate": 4.9949566229164704e-05,
-      "loss": 87.8614,
-      "step": 29550
-    },
-    {
-      "epoch": 0.11942614042671816,
-      "grad_norm": 730.4893188476562,
-      "learning_rate": 4.994934437362513e-05,
-      "loss": 94.9959,
-      "step": 29560
-    },
-    {
-      "epoch": 0.1194665416920858,
-      "grad_norm": 1123.4429931640625,
-      "learning_rate": 4.9949122031685245e-05,
-      "loss": 116.839,
-      "step": 29570
-    },
-    {
-      "epoch": 0.11950694295745343,
-      "grad_norm": 723.6360473632812,
-      "learning_rate": 4.9948899203349384e-05,
-      "loss": 86.0155,
-      "step": 29580
-    },
-    {
-      "epoch": 0.11954734422282105,
-      "grad_norm": 480.5324401855469,
-      "learning_rate": 4.994867588862189e-05,
-      "loss": 94.0604,
-      "step": 29590
-    },
-    {
-      "epoch": 0.1195877454881887,
-      "grad_norm": 1117.021484375,
-      "learning_rate": 4.9948452087507116e-05,
-      "loss": 69.5577,
-      "step": 29600
-    },
-    {
-      "epoch": 0.11962814675355632,
-      "grad_norm": 1889.9290771484375,
-      "learning_rate": 4.9948227800009416e-05,
-      "loss": 130.7532,
-      "step": 29610
-    },
-    {
-      "epoch": 0.11966854801892396,
-      "grad_norm": 811.6279907226562,
-      "learning_rate": 4.994800302613318e-05,
-      "loss": 93.5169,
-      "step": 29620
-    },
-    {
-      "epoch": 0.11970894928429159,
-      "grad_norm": 339.779052734375,
-      "learning_rate": 4.994777776588278e-05,
-      "loss": 112.6328,
-      "step": 29630
-    },
-    {
-      "epoch": 0.11974935054965921,
-      "grad_norm": 794.0516967773438,
-      "learning_rate": 4.9947552019262605e-05,
-      "loss": 61.7838,
-      "step": 29640
-    },
-    {
-      "epoch": 0.11978975181502685,
-      "grad_norm": 1287.187255859375,
-      "learning_rate": 4.994732578627706e-05,
-      "loss": 156.2145,
-      "step": 29650
-    },
-    {
-      "epoch": 0.11983015308039448,
-      "grad_norm": 921.3040161132812,
-      "learning_rate": 4.994709906693056e-05,
-      "loss": 77.835,
-      "step": 29660
-    },
-    {
-      "epoch": 0.1198705543457621,
-      "grad_norm": 985.0899047851562,
-      "learning_rate": 4.9946871861227514e-05,
-      "loss": 101.6619,
-      "step": 29670
-    },
-    {
-      "epoch": 0.11991095561112974,
-      "grad_norm": 1570.173583984375,
-      "learning_rate": 4.9946644169172355e-05,
-      "loss": 132.0515,
-      "step": 29680
-    },
-    {
-      "epoch": 0.11995135687649737,
-      "grad_norm": 1566.288818359375,
-      "learning_rate": 4.9946415990769534e-05,
-      "loss": 109.4148,
-      "step": 29690
-    },
-    {
-      "epoch": 0.11999175814186501,
-      "grad_norm": 654.2642822265625,
-      "learning_rate": 4.994618732602349e-05,
-      "loss": 92.2775,
-      "step": 29700
-    },
-    {
-      "epoch": 0.12003215940723264,
-      "grad_norm": 5245.0458984375,
-      "learning_rate": 4.994595817493867e-05,
-      "loss": 144.156,
-      "step": 29710
-    },
-    {
-      "epoch": 0.12007256067260026,
-      "grad_norm": 4157.51708984375,
-      "learning_rate": 4.9945728537519555e-05,
-      "loss": 124.7138,
-      "step": 29720
-    },
-    {
-      "epoch": 0.1201129619379679,
-      "grad_norm": 1033.809814453125,
-      "learning_rate": 4.994549841377063e-05,
-      "loss": 108.5572,
-      "step": 29730
-    },
-    {
-      "epoch": 0.12015336320333553,
-      "grad_norm": 681.5181884765625,
-      "learning_rate": 4.9945267803696364e-05,
-      "loss": 99.638,
-      "step": 29740
-    },
-    {
-      "epoch": 0.12019376446870315,
-      "grad_norm": 2804.89501953125,
-      "learning_rate": 4.994503670730125e-05,
-      "loss": 99.5536,
-      "step": 29750
-    },
-    {
-      "epoch": 0.1202341657340708,
-      "grad_norm": 346.3814392089844,
-      "learning_rate": 4.994480512458981e-05,
-      "loss": 108.1357,
-      "step": 29760
-    },
-    {
-      "epoch": 0.12027456699943842,
-      "grad_norm": 1421.3365478515625,
-      "learning_rate": 4.9944573055566556e-05,
-      "loss": 103.8683,
-      "step": 29770
-    },
-    {
-      "epoch": 0.12031496826480606,
-      "grad_norm": 1197.5367431640625,
-      "learning_rate": 4.994434050023601e-05,
-      "loss": 107.7595,
-      "step": 29780
-    },
-    {
-      "epoch": 0.12035536953017369,
-      "grad_norm": 256.7560729980469,
-      "learning_rate": 4.9944107458602693e-05,
-      "loss": 81.2983,
-      "step": 29790
-    },
-    {
-      "epoch": 0.12039577079554131,
-      "grad_norm": 1710.7000732421875,
-      "learning_rate": 4.994387393067117e-05,
-      "loss": 98.1409,
-      "step": 29800
-    },
-    {
-      "epoch": 0.12043617206090895,
-      "grad_norm": 583.5015258789062,
-      "learning_rate": 4.994363991644597e-05,
-      "loss": 67.0093,
-      "step": 29810
-    },
-    {
-      "epoch": 0.12047657332627658,
-      "grad_norm": 1057.0147705078125,
-      "learning_rate": 4.9943405415931674e-05,
-      "loss": 116.4649,
-      "step": 29820
-    },
-    {
-      "epoch": 0.1205169745916442,
-      "grad_norm": 671.4649658203125,
-      "learning_rate": 4.9943170429132855e-05,
-      "loss": 95.8829,
-      "step": 29830
-    },
-    {
-      "epoch": 0.12055737585701184,
-      "grad_norm": 1301.7279052734375,
-      "learning_rate": 4.9942934956054076e-05,
-      "loss": 90.3948,
-      "step": 29840
-    },
-    {
-      "epoch": 0.12059777712237947,
-      "grad_norm": 1199.511474609375,
-      "learning_rate": 4.994269899669994e-05,
-      "loss": 85.8095,
-      "step": 29850
-    },
-    {
-      "epoch": 0.12063817838774711,
-      "grad_norm": 648.86376953125,
-      "learning_rate": 4.9942462551075056e-05,
-      "loss": 77.0479,
-      "step": 29860
-    },
-    {
-      "epoch": 0.12067857965311474,
-      "grad_norm": 788.9052124023438,
-      "learning_rate": 4.994222561918401e-05,
-      "loss": 81.1479,
-      "step": 29870
-    },
-    {
-      "epoch": 0.12071898091848236,
-      "grad_norm": 931.2744750976562,
-      "learning_rate": 4.994198820103145e-05,
-      "loss": 85.522,
-      "step": 29880
-    },
-    {
-      "epoch": 0.12075938218385,
-      "grad_norm": 1237.2655029296875,
-      "learning_rate": 4.994175029662198e-05,
-      "loss": 95.5039,
-      "step": 29890
-    },
-    {
-      "epoch": 0.12079978344921763,
-      "grad_norm": 704.9238891601562,
-      "learning_rate": 4.994151190596025e-05,
-      "loss": 85.7873,
-      "step": 29900
-    },
-    {
-      "epoch": 0.12084018471458526,
-      "grad_norm": 826.0241088867188,
-      "learning_rate": 4.9941273029050894e-05,
-      "loss": 129.6459,
-      "step": 29910
-    },
-    {
-      "epoch": 0.1208805859799529,
-      "grad_norm": 479.6162109375,
-      "learning_rate": 4.994103366589859e-05,
-      "loss": 117.3531,
-      "step": 29920
-    },
-    {
-      "epoch": 0.12092098724532052,
-      "grad_norm": 1103.497314453125,
-      "learning_rate": 4.994079381650799e-05,
-      "loss": 183.9052,
-      "step": 29930
-    },
-    {
-      "epoch": 0.12096138851068816,
-      "grad_norm": 710.986572265625,
-      "learning_rate": 4.994055348088378e-05,
-      "loss": 105.431,
-      "step": 29940
-    },
-    {
-      "epoch": 0.12100178977605579,
-      "grad_norm": 1297.952880859375,
-      "learning_rate": 4.994031265903063e-05,
-      "loss": 128.99,
-      "step": 29950
-    },
-    {
-      "epoch": 0.12104219104142341,
-      "grad_norm": 1707.3641357421875,
-      "learning_rate": 4.9940071350953255e-05,
-      "loss": 127.0656,
-      "step": 29960
-    },
-    {
-      "epoch": 0.12108259230679105,
-      "grad_norm": 787.633544921875,
-      "learning_rate": 4.993982955665634e-05,
-      "loss": 106.8927,
-      "step": 29970
-    },
-    {
-      "epoch": 0.12112299357215868,
-      "grad_norm": 1233.0062255859375,
-      "learning_rate": 4.9939587276144616e-05,
-      "loss": 104.3319,
-      "step": 29980
-    },
-    {
-      "epoch": 0.1211633948375263,
-      "grad_norm": 1296.74658203125,
-      "learning_rate": 4.993934450942279e-05,
-      "loss": 77.4372,
-      "step": 29990
-    },
-    {
-      "epoch": 0.12120379610289395,
-      "grad_norm": 735.8043212890625,
-      "learning_rate": 4.993910125649561e-05,
-      "loss": 147.1198,
-      "step": 30000
-    },
-    {
-      "epoch": 0.12124419736826157,
-      "grad_norm": 364.9338684082031,
-      "learning_rate": 4.993885751736781e-05,
-      "loss": 84.0477,
-      "step": 30010
-    },
-    {
-      "epoch": 0.12128459863362921,
-      "grad_norm": 719.6060180664062,
-      "learning_rate": 4.993861329204414e-05,
-      "loss": 115.0903,
-      "step": 30020
-    },
-    {
-      "epoch": 0.12132499989899684,
-      "grad_norm": 1294.3851318359375,
-      "learning_rate": 4.993836858052937e-05,
-      "loss": 151.7286,
-      "step": 30030
-    },
-    {
-      "epoch": 0.12136540116436446,
-      "grad_norm": 711.005126953125,
-      "learning_rate": 4.993812338282826e-05,
-      "loss": 86.3682,
-      "step": 30040
-    },
-    {
-      "epoch": 0.1214058024297321,
-      "grad_norm": 723.4884643554688,
-      "learning_rate": 4.993787769894559e-05,
-      "loss": 120.8179,
-      "step": 30050
-    },
-    {
-      "epoch": 0.12144620369509973,
-      "grad_norm": 1060.3338623046875,
-      "learning_rate": 4.993763152888617e-05,
-      "loss": 134.1443,
-      "step": 30060
-    },
-    {
-      "epoch": 0.12148660496046736,
-      "grad_norm": 829.1416625976562,
-      "learning_rate": 4.993738487265478e-05,
-      "loss": 119.1408,
-      "step": 30070
-    },
-    {
-      "epoch": 0.121527006225835,
-      "grad_norm": 267.49755859375,
-      "learning_rate": 4.993713773025623e-05,
-      "loss": 112.0577,
-      "step": 30080
-    },
-    {
-      "epoch": 0.12156740749120262,
-      "grad_norm": 2977.7763671875,
-      "learning_rate": 4.993689010169534e-05,
-      "loss": 115.676,
-      "step": 30090
-    },
-    {
-      "epoch": 0.12160780875657026,
-      "grad_norm": 927.7879638671875,
-      "learning_rate": 4.993664198697694e-05,
-      "loss": 80.4086,
-      "step": 30100
-    },
-    {
-      "epoch": 0.12164821002193789,
-      "grad_norm": 1482.4476318359375,
-      "learning_rate": 4.993639338610587e-05,
-      "loss": 90.1057,
-      "step": 30110
-    },
-    {
-      "epoch": 0.12168861128730551,
-      "grad_norm": 1078.0257568359375,
-      "learning_rate": 4.993614429908697e-05,
-      "loss": 112.5462,
-      "step": 30120
-    },
-    {
-      "epoch": 0.12172901255267315,
-      "grad_norm": 1431.45947265625,
-      "learning_rate": 4.99358947259251e-05,
-      "loss": 83.491,
-      "step": 30130
-    },
-    {
-      "epoch": 0.12176941381804078,
-      "grad_norm": 792.9324340820312,
-      "learning_rate": 4.9935644666625125e-05,
-      "loss": 82.8931,
-      "step": 30140
-    },
-    {
-      "epoch": 0.12180981508340841,
-      "grad_norm": 1509.551025390625,
-      "learning_rate": 4.9935394121191915e-05,
-      "loss": 103.0119,
-      "step": 30150
-    },
-    {
-      "epoch": 0.12185021634877605,
-      "grad_norm": 1039.8441162109375,
-      "learning_rate": 4.993514308963036e-05,
-      "loss": 119.8,
-      "step": 30160
-    },
-    {
-      "epoch": 0.12189061761414367,
-      "grad_norm": 399.6064758300781,
-      "learning_rate": 4.993489157194536e-05,
-      "loss": 115.0279,
-      "step": 30170
-    },
-    {
-      "epoch": 0.12193101887951131,
-      "grad_norm": 794.4744873046875,
-      "learning_rate": 4.993463956814181e-05,
-      "loss": 101.3931,
-      "step": 30180
-    },
-    {
-      "epoch": 0.12197142014487894,
-      "grad_norm": 2071.046142578125,
-      "learning_rate": 4.993438707822462e-05,
-      "loss": 114.3679,
-      "step": 30190
-    },
-    {
-      "epoch": 0.12201182141024657,
-      "grad_norm": 996.6055297851562,
-      "learning_rate": 4.993413410219871e-05,
-      "loss": 84.8786,
-      "step": 30200
-    },
-    {
-      "epoch": 0.1220522226756142,
-      "grad_norm": 1139.685302734375,
-      "learning_rate": 4.993388064006903e-05,
-      "loss": 78.988,
-      "step": 30210
-    },
-    {
-      "epoch": 0.12209262394098183,
-      "grad_norm": 622.7205200195312,
-      "learning_rate": 4.993362669184051e-05,
-      "loss": 88.1808,
-      "step": 30220
-    },
-    {
-      "epoch": 0.12213302520634946,
-      "grad_norm": 7026.68359375,
-      "learning_rate": 4.9933372257518096e-05,
-      "loss": 126.5632,
-      "step": 30230
-    },
-    {
-      "epoch": 0.1221734264717171,
-      "grad_norm": 945.2219848632812,
-      "learning_rate": 4.993311733710676e-05,
-      "loss": 78.7985,
-      "step": 30240
-    },
-    {
-      "epoch": 0.12221382773708472,
-      "grad_norm": 826.4041137695312,
-      "learning_rate": 4.9932861930611454e-05,
-      "loss": 72.6881,
-      "step": 30250
-    },
-    {
-      "epoch": 0.12225422900245236,
-      "grad_norm": 714.5296630859375,
-      "learning_rate": 4.993260603803718e-05,
-      "loss": 94.0831,
-      "step": 30260
-    },
-    {
-      "epoch": 0.12229463026781999,
-      "grad_norm": 692.105224609375,
-      "learning_rate": 4.99323496593889e-05,
-      "loss": 60.1789,
-      "step": 30270
-    },
-    {
-      "epoch": 0.12233503153318762,
-      "grad_norm": 534.4913330078125,
-      "learning_rate": 4.993209279467164e-05,
-      "loss": 69.5706,
-      "step": 30280
-    },
-    {
-      "epoch": 0.12237543279855526,
-      "grad_norm": 261.96044921875,
-      "learning_rate": 4.99318354438904e-05,
-      "loss": 77.032,
-      "step": 30290
-    },
-    {
-      "epoch": 0.12241583406392288,
-      "grad_norm": 1099.048095703125,
-      "learning_rate": 4.9931577607050175e-05,
-      "loss": 92.2519,
-      "step": 30300
-    },
-    {
-      "epoch": 0.12245623532929051,
-      "grad_norm": 1755.97705078125,
-      "learning_rate": 4.993131928415602e-05,
-      "loss": 159.9417,
-      "step": 30310
-    },
-    {
-      "epoch": 0.12249663659465815,
-      "grad_norm": 1634.8353271484375,
-      "learning_rate": 4.993106047521296e-05,
-      "loss": 91.0221,
-      "step": 30320
-    },
-    {
-      "epoch": 0.12253703786002577,
-      "grad_norm": 980.1278686523438,
-      "learning_rate": 4.993080118022604e-05,
-      "loss": 97.5692,
-      "step": 30330
-    },
-    {
-      "epoch": 0.12257743912539341,
-      "grad_norm": 2488.35009765625,
-      "learning_rate": 4.993054139920032e-05,
-      "loss": 114.5058,
-      "step": 30340
-    },
-    {
-      "epoch": 0.12261784039076104,
-      "grad_norm": 1487.991455078125,
-      "learning_rate": 4.9930281132140846e-05,
-      "loss": 82.5714,
-      "step": 30350
-    },
-    {
-      "epoch": 0.12265824165612867,
-      "grad_norm": 513.70654296875,
-      "learning_rate": 4.993002037905272e-05,
-      "loss": 127.0465,
-      "step": 30360
-    },
-    {
-      "epoch": 0.1226986429214963,
-      "grad_norm": 2360.14208984375,
-      "learning_rate": 4.9929759139941e-05,
-      "loss": 128.6292,
-      "step": 30370
-    },
-    {
-      "epoch": 0.12273904418686393,
-      "grad_norm": 537.433349609375,
-      "learning_rate": 4.99294974148108e-05,
-      "loss": 84.4188,
-      "step": 30380
-    },
-    {
-      "epoch": 0.12277944545223156,
-      "grad_norm": 339.3077697753906,
-      "learning_rate": 4.9929235203667214e-05,
-      "loss": 84.647,
-      "step": 30390
-    },
-    {
-      "epoch": 0.1228198467175992,
-      "grad_norm": 697.3407592773438,
-      "learning_rate": 4.992897250651535e-05,
-      "loss": 96.315,
-      "step": 30400
-    },
-    {
-      "epoch": 0.12286024798296682,
-      "grad_norm": 637.2626342773438,
-      "learning_rate": 4.9928709323360337e-05,
-      "loss": 87.7648,
-      "step": 30410
-    },
-    {
-      "epoch": 0.12290064924833445,
-      "grad_norm": 591.4133911132812,
-      "learning_rate": 4.99284456542073e-05,
-      "loss": 57.4863,
-      "step": 30420
-    },
-    {
-      "epoch": 0.12294105051370209,
-      "grad_norm": 1242.639404296875,
-      "learning_rate": 4.992818149906138e-05,
-      "loss": 105.4673,
-      "step": 30430
-    },
-    {
-      "epoch": 0.12298145177906972,
-      "grad_norm": 946.1543579101562,
-      "learning_rate": 4.992791685792772e-05,
-      "loss": 101.4292,
-      "step": 30440
-    },
-    {
-      "epoch": 0.12302185304443736,
-      "grad_norm": 554.2642211914062,
-      "learning_rate": 4.9927651730811495e-05,
-      "loss": 55.5201,
-      "step": 30450
-    },
-    {
-      "epoch": 0.12306225430980498,
-      "grad_norm": 1658.3990478515625,
-      "learning_rate": 4.992738611771787e-05,
-      "loss": 90.6332,
-      "step": 30460
-    },
-    {
-      "epoch": 0.12310265557517261,
-      "grad_norm": 1110.182373046875,
-      "learning_rate": 4.9927120018652004e-05,
-      "loss": 112.0799,
-      "step": 30470
-    },
-    {
-      "epoch": 0.12314305684054025,
-      "grad_norm": 2135.968505859375,
-      "learning_rate": 4.992685343361911e-05,
-      "loss": 124.2809,
-      "step": 30480
-    },
-    {
-      "epoch": 0.12318345810590788,
-      "grad_norm": 0.0,
-      "learning_rate": 4.992658636262438e-05,
-      "loss": 130.0038,
-      "step": 30490
-    },
-    {
-      "epoch": 0.1232238593712755,
-      "grad_norm": 897.2399291992188,
-      "learning_rate": 4.992631880567301e-05,
-      "loss": 110.8611,
-      "step": 30500
-    },
-    {
-      "epoch": 0.12326426063664314,
-      "grad_norm": 1721.460693359375,
-      "learning_rate": 4.9926050762770224e-05,
-      "loss": 110.5094,
-      "step": 30510
-    },
-    {
-      "epoch": 0.12330466190201077,
-      "grad_norm": 994.7217407226562,
-      "learning_rate": 4.992578223392124e-05,
-      "loss": 95.7169,
-      "step": 30520
-    },
-    {
-      "epoch": 0.12334506316737841,
-      "grad_norm": 1169.87939453125,
-      "learning_rate": 4.9925513219131303e-05,
-      "loss": 82.8276,
-      "step": 30530
-    },
-    {
-      "epoch": 0.12338546443274603,
-      "grad_norm": 677.4993286132812,
-      "learning_rate": 4.992524371840566e-05,
-      "loss": 105.6593,
-      "step": 30540
-    },
-    {
-      "epoch": 0.12342586569811366,
-      "grad_norm": 885.8995361328125,
-      "learning_rate": 4.992497373174955e-05,
-      "loss": 114.9273,
-      "step": 30550
-    },
-    {
-      "epoch": 0.1234662669634813,
-      "grad_norm": 1765.774658203125,
-      "learning_rate": 4.9924703259168244e-05,
-      "loss": 73.1132,
-      "step": 30560
-    },
-    {
-      "epoch": 0.12350666822884893,
-      "grad_norm": 1306.00244140625,
-      "learning_rate": 4.992443230066701e-05,
-      "loss": 96.6844,
-      "step": 30570
-    },
-    {
-      "epoch": 0.12354706949421655,
-      "grad_norm": 2164.46044921875,
-      "learning_rate": 4.992416085625115e-05,
-      "loss": 101.1072,
-      "step": 30580
-    },
-    {
-      "epoch": 0.12358747075958419,
-      "grad_norm": 1472.5419921875,
-      "learning_rate": 4.992388892592594e-05,
-      "loss": 93.9604,
-      "step": 30590
-    },
-    {
-      "epoch": 0.12362787202495182,
-      "grad_norm": 376.5589294433594,
-      "learning_rate": 4.9923616509696683e-05,
-      "loss": 89.4311,
-      "step": 30600
-    },
-    {
-      "epoch": 0.12366827329031946,
-      "grad_norm": 809.4692993164062,
-      "learning_rate": 4.9923343607568684e-05,
-      "loss": 90.6958,
-      "step": 30610
-    },
-    {
-      "epoch": 0.12370867455568708,
-      "grad_norm": 1123.7208251953125,
-      "learning_rate": 4.9923070219547275e-05,
-      "loss": 113.3677,
-      "step": 30620
-    },
-    {
-      "epoch": 0.12374907582105471,
-      "grad_norm": 1361.58056640625,
-      "learning_rate": 4.9922796345637776e-05,
-      "loss": 109.5225,
-      "step": 30630
-    },
-    {
-      "epoch": 0.12378947708642235,
-      "grad_norm": 1546.5401611328125,
-      "learning_rate": 4.992252198584554e-05,
-      "loss": 156.2372,
-      "step": 30640
-    },
-    {
-      "epoch": 0.12382987835178998,
-      "grad_norm": 1252.5113525390625,
-      "learning_rate": 4.99222471401759e-05,
-      "loss": 98.325,
-      "step": 30650
-    },
-    {
-      "epoch": 0.1238702796171576,
-      "grad_norm": 587.108642578125,
-      "learning_rate": 4.992197180863422e-05,
-      "loss": 83.3731,
-      "step": 30660
-    },
-    {
-      "epoch": 0.12391068088252524,
-      "grad_norm": 546.9779663085938,
-      "learning_rate": 4.992169599122587e-05,
-      "loss": 69.5896,
-      "step": 30670
-    },
-    {
-      "epoch": 0.12395108214789287,
-      "grad_norm": 533.7169189453125,
-      "learning_rate": 4.992141968795623e-05,
-      "loss": 59.3518,
-      "step": 30680
-    },
-    {
-      "epoch": 0.12399148341326051,
-      "grad_norm": 609.4190673828125,
-      "learning_rate": 4.992114289883068e-05,
-      "loss": 132.897,
-      "step": 30690
-    },
-    {
-      "epoch": 0.12403188467862813,
-      "grad_norm": 1489.8643798828125,
-      "learning_rate": 4.9920865623854615e-05,
-      "loss": 84.5661,
-      "step": 30700
-    },
-    {
-      "epoch": 0.12407228594399576,
-      "grad_norm": 1392.722900390625,
-      "learning_rate": 4.992058786303345e-05,
-      "loss": 80.1483,
-      "step": 30710
-    },
-    {
-      "epoch": 0.1241126872093634,
-      "grad_norm": 1709.162841796875,
-      "learning_rate": 4.9920309616372596e-05,
-      "loss": 78.6077,
-      "step": 30720
-    },
-    {
-      "epoch": 0.12415308847473103,
-      "grad_norm": 319.7249450683594,
-      "learning_rate": 4.9920030883877476e-05,
-      "loss": 49.0658,
-      "step": 30730
-    },
-    {
-      "epoch": 0.12419348974009865,
-      "grad_norm": 2077.73095703125,
-      "learning_rate": 4.9919751665553525e-05,
-      "loss": 119.2495,
-      "step": 30740
-    },
-    {
-      "epoch": 0.1242338910054663,
-      "grad_norm": 1575.6136474609375,
-      "learning_rate": 4.991947196140618e-05,
-      "loss": 65.2858,
-      "step": 30750
-    },
-    {
-      "epoch": 0.12427429227083392,
-      "grad_norm": 891.6358032226562,
-      "learning_rate": 4.9919191771440905e-05,
-      "loss": 76.9454,
-      "step": 30760
-    },
-    {
-      "epoch": 0.12431469353620156,
-      "grad_norm": 836.199951171875,
-      "learning_rate": 4.991891109566316e-05,
-      "loss": 86.9683,
-      "step": 30770
-    },
-    {
-      "epoch": 0.12435509480156919,
-      "grad_norm": 1238.918701171875,
-      "learning_rate": 4.99186299340784e-05,
-      "loss": 95.3734,
-      "step": 30780
-    },
-    {
-      "epoch": 0.12439549606693681,
-      "grad_norm": 1337.448974609375,
-      "learning_rate": 4.991834828669213e-05,
-      "loss": 103.7555,
-      "step": 30790
-    },
-    {
-      "epoch": 0.12443589733230445,
-      "grad_norm": 2682.815673828125,
-      "learning_rate": 4.9918066153509834e-05,
-      "loss": 157.3261,
-      "step": 30800
-    },
-    {
-      "epoch": 0.12447629859767208,
-      "grad_norm": 2427.59033203125,
-      "learning_rate": 4.9917783534537006e-05,
-      "loss": 79.6278,
-      "step": 30810
-    },
-    {
-      "epoch": 0.1245166998630397,
-      "grad_norm": 668.7774658203125,
-      "learning_rate": 4.991750042977916e-05,
-      "loss": 77.9551,
-      "step": 30820
-    },
-    {
-      "epoch": 0.12455710112840734,
-      "grad_norm": 1772.7213134765625,
-      "learning_rate": 4.991721683924182e-05,
-      "loss": 109.0242,
-      "step": 30830
-    },
-    {
-      "epoch": 0.12459750239377497,
-      "grad_norm": 896.2188110351562,
-      "learning_rate": 4.99169327629305e-05,
-      "loss": 103.7983,
-      "step": 30840
-    },
-    {
-      "epoch": 0.12463790365914261,
-      "grad_norm": 1777.919677734375,
-      "learning_rate": 4.991664820085074e-05,
-      "loss": 69.7507,
-      "step": 30850
-    },
-    {
-      "epoch": 0.12467830492451024,
-      "grad_norm": 414.7098693847656,
-      "learning_rate": 4.9916363153008114e-05,
-      "loss": 137.439,
-      "step": 30860
-    },
-    {
-      "epoch": 0.12471870618987786,
-      "grad_norm": 598.8345947265625,
-      "learning_rate": 4.9916077619408155e-05,
-      "loss": 120.2765,
-      "step": 30870
-    },
-    {
-      "epoch": 0.1247591074552455,
-      "grad_norm": 1064.3018798828125,
-      "learning_rate": 4.991579160005644e-05,
-      "loss": 84.4768,
-      "step": 30880
-    },
-    {
-      "epoch": 0.12479950872061313,
-      "grad_norm": 666.4903564453125,
-      "learning_rate": 4.9915505094958526e-05,
-      "loss": 136.433,
-      "step": 30890
-    },
-    {
-      "epoch": 0.12483990998598075,
-      "grad_norm": 1181.445556640625,
-      "learning_rate": 4.991521810412002e-05,
-      "loss": 63.556,
-      "step": 30900
-    },
-    {
-      "epoch": 0.1248803112513484,
-      "grad_norm": 605.0240478515625,
-      "learning_rate": 4.991493062754651e-05,
-      "loss": 94.3815,
-      "step": 30910
-    },
-    {
-      "epoch": 0.12492071251671602,
-      "grad_norm": 794.223388671875,
-      "learning_rate": 4.99146426652436e-05,
-      "loss": 76.5061,
-      "step": 30920
-    },
-    {
-      "epoch": 0.12496111378208366,
-      "grad_norm": 1127.827392578125,
-      "learning_rate": 4.991435421721691e-05,
-      "loss": 80.6668,
-      "step": 30930
-    },
-    {
-      "epoch": 0.12500151504745127,
-      "grad_norm": 494.918212890625,
-      "learning_rate": 4.991406528347206e-05,
-      "loss": 137.6075,
-      "step": 30940
-    },
-    {
-      "epoch": 0.1250419163128189,
-      "grad_norm": 688.9254760742188,
-      "learning_rate": 4.9913775864014665e-05,
-      "loss": 77.563,
-      "step": 30950
-    },
-    {
-      "epoch": 0.12508231757818655,
-      "grad_norm": 1798.6043701171875,
-      "learning_rate": 4.991348595885039e-05,
-      "loss": 172.5805,
-      "step": 30960
-    },
-    {
-      "epoch": 0.1251227188435542,
-      "grad_norm": 629.1405639648438,
-      "learning_rate": 4.991319556798488e-05,
-      "loss": 94.1099,
-      "step": 30970
-    },
-    {
-      "epoch": 0.1251631201089218,
-      "grad_norm": 831.2258911132812,
-      "learning_rate": 4.99129046914238e-05,
-      "loss": 82.6184,
-      "step": 30980
-    },
-    {
-      "epoch": 0.12520352137428944,
-      "grad_norm": 899.6831665039062,
-      "learning_rate": 4.991261332917282e-05,
-      "loss": 89.8317,
-      "step": 30990
-    },
-    {
-      "epoch": 0.12524392263965708,
-      "grad_norm": 1243.12255859375,
-      "learning_rate": 4.991232148123761e-05,
-      "loss": 106.8137,
-      "step": 31000
-    },
-    {
-      "epoch": 0.1252843239050247,
-      "grad_norm": 708.5358276367188,
-      "learning_rate": 4.9912029147623875e-05,
-      "loss": 71.765,
-      "step": 31010
-    },
-    {
-      "epoch": 0.12532472517039234,
-      "grad_norm": 1078.65185546875,
-      "learning_rate": 4.9911736328337296e-05,
-      "loss": 98.1934,
-      "step": 31020
-    },
-    {
-      "epoch": 0.12536512643575998,
-      "grad_norm": 863.23193359375,
-      "learning_rate": 4.99114430233836e-05,
-      "loss": 110.7916,
-      "step": 31030
-    },
-    {
-      "epoch": 0.1254055277011276,
-      "grad_norm": 948.326416015625,
-      "learning_rate": 4.991114923276849e-05,
-      "loss": 128.7791,
-      "step": 31040
-    },
-    {
-      "epoch": 0.12544592896649523,
-      "grad_norm": 1923.48681640625,
-      "learning_rate": 4.9910854956497696e-05,
-      "loss": 81.3787,
-      "step": 31050
-    },
-    {
-      "epoch": 0.12548633023186287,
-      "grad_norm": 1134.7197265625,
-      "learning_rate": 4.991056019457697e-05,
-      "loss": 97.8454,
-      "step": 31060
-    },
-    {
-      "epoch": 0.12552673149723048,
-      "grad_norm": 1076.4554443359375,
-      "learning_rate": 4.991026494701205e-05,
-      "loss": 93.0418,
-      "step": 31070
-    },
-    {
-      "epoch": 0.12556713276259812,
-      "grad_norm": 1976.7174072265625,
-      "learning_rate": 4.9909969213808683e-05,
-      "loss": 87.8301,
-      "step": 31080
-    },
-    {
-      "epoch": 0.12560753402796576,
-      "grad_norm": 1298.41162109375,
-      "learning_rate": 4.990967299497264e-05,
-      "loss": 125.1484,
-      "step": 31090
-    },
-    {
-      "epoch": 0.12564793529333337,
-      "grad_norm": 532.9257202148438,
-      "learning_rate": 4.990937629050971e-05,
-      "loss": 101.9732,
-      "step": 31100
-    },
-    {
-      "epoch": 0.125688336558701,
-      "grad_norm": 1158.4649658203125,
-      "learning_rate": 4.990907910042566e-05,
-      "loss": 88.4633,
-      "step": 31110
-    },
-    {
-      "epoch": 0.12572873782406865,
-      "grad_norm": 1732.9954833984375,
-      "learning_rate": 4.990878142472628e-05,
-      "loss": 80.0504,
-      "step": 31120
-    },
-    {
-      "epoch": 0.1257691390894363,
-      "grad_norm": 432.7400207519531,
-      "learning_rate": 4.990848326341739e-05,
-      "loss": 92.5247,
-      "step": 31130
-    },
-    {
-      "epoch": 0.1258095403548039,
-      "grad_norm": 985.176025390625,
-      "learning_rate": 4.990818461650479e-05,
-      "loss": 100.5391,
-      "step": 31140
-    },
-    {
-      "epoch": 0.12584994162017155,
-      "grad_norm": 789.0088500976562,
-      "learning_rate": 4.990788548399432e-05,
-      "loss": 88.8649,
-      "step": 31150
-    },
-    {
-      "epoch": 0.12589034288553919,
-      "grad_norm": 1790.52978515625,
-      "learning_rate": 4.990758586589178e-05,
-      "loss": 108.7384,
-      "step": 31160
-    },
-    {
-      "epoch": 0.1259307441509068,
-      "grad_norm": 1069.7830810546875,
-      "learning_rate": 4.9907285762203046e-05,
-      "loss": 104.5124,
-      "step": 31170
-    },
-    {
-      "epoch": 0.12597114541627444,
-      "grad_norm": 766.6657104492188,
-      "learning_rate": 4.990698517293395e-05,
-      "loss": 59.5578,
-      "step": 31180
-    },
-    {
-      "epoch": 0.12601154668164208,
-      "grad_norm": 1233.1490478515625,
-      "learning_rate": 4.990668409809034e-05,
-      "loss": 63.7087,
-      "step": 31190
-    },
-    {
-      "epoch": 0.1260519479470097,
-      "grad_norm": 1039.3233642578125,
-      "learning_rate": 4.990638253767812e-05,
-      "loss": 110.4396,
-      "step": 31200
-    },
-    {
-      "epoch": 0.12609234921237733,
-      "grad_norm": 1719.74267578125,
-      "learning_rate": 4.9906080491703146e-05,
-      "loss": 84.7901,
-      "step": 31210
-    },
-    {
-      "epoch": 0.12613275047774497,
-      "grad_norm": 2833.946533203125,
-      "learning_rate": 4.9905777960171304e-05,
-      "loss": 146.1511,
-      "step": 31220
-    },
-    {
-      "epoch": 0.12617315174311258,
-      "grad_norm": 820.8349609375,
-      "learning_rate": 4.99054749430885e-05,
-      "loss": 120.1305,
-      "step": 31230
-    },
-    {
-      "epoch": 0.12621355300848022,
-      "grad_norm": 782.0814208984375,
-      "learning_rate": 4.990517144046064e-05,
-      "loss": 68.8798,
-      "step": 31240
-    },
-    {
-      "epoch": 0.12625395427384786,
-      "grad_norm": 1526.1461181640625,
-      "learning_rate": 4.990486745229364e-05,
-      "loss": 109.5388,
-      "step": 31250
-    },
-    {
-      "epoch": 0.12629435553921547,
-      "grad_norm": 2700.19677734375,
-      "learning_rate": 4.9904562978593426e-05,
-      "loss": 141.1139,
-      "step": 31260
-    },
-    {
-      "epoch": 0.12633475680458311,
-      "grad_norm": 578.1785278320312,
-      "learning_rate": 4.990425801936594e-05,
-      "loss": 63.9474,
-      "step": 31270
-    },
-    {
-      "epoch": 0.12637515806995075,
-      "grad_norm": 1284.421630859375,
-      "learning_rate": 4.990395257461712e-05,
-      "loss": 64.6959,
-      "step": 31280
-    },
-    {
-      "epoch": 0.1264155593353184,
-      "grad_norm": 722.0364379882812,
-      "learning_rate": 4.9903646644352925e-05,
-      "loss": 136.6368,
-      "step": 31290
-    },
-    {
-      "epoch": 0.126455960600686,
-      "grad_norm": 639.962646484375,
-      "learning_rate": 4.990334022857932e-05,
-      "loss": 85.277,
-      "step": 31300
-    },
-    {
-      "epoch": 0.12649636186605365,
-      "grad_norm": 1369.47265625,
-      "learning_rate": 4.990303332730226e-05,
-      "loss": 95.2965,
-      "step": 31310
-    },
-    {
-      "epoch": 0.1265367631314213,
-      "grad_norm": 710.7479858398438,
-      "learning_rate": 4.990272594052776e-05,
-      "loss": 103.4021,
-      "step": 31320
-    },
-    {
-      "epoch": 0.1265771643967889,
-      "grad_norm": 967.9599609375,
-      "learning_rate": 4.990241806826179e-05,
-      "loss": 55.237,
-      "step": 31330
-    },
-    {
-      "epoch": 0.12661756566215654,
-      "grad_norm": 1026.2791748046875,
-      "learning_rate": 4.990210971051037e-05,
-      "loss": 76.5986,
-      "step": 31340
-    },
-    {
-      "epoch": 0.12665796692752418,
-      "grad_norm": 1351.9713134765625,
-      "learning_rate": 4.990180086727949e-05,
-      "loss": 78.9581,
-      "step": 31350
-    },
-    {
-      "epoch": 0.1266983681928918,
-      "grad_norm": 1424.3463134765625,
-      "learning_rate": 4.9901491538575185e-05,
-      "loss": 72.768,
-      "step": 31360
-    },
-    {
-      "epoch": 0.12673876945825943,
-      "grad_norm": 251.47360229492188,
-      "learning_rate": 4.990118172440348e-05,
-      "loss": 97.3149,
-      "step": 31370
-    },
-    {
-      "epoch": 0.12677917072362707,
-      "grad_norm": 1101.8797607421875,
-      "learning_rate": 4.9900871424770424e-05,
-      "loss": 113.3635,
-      "step": 31380
-    },
-    {
-      "epoch": 0.12681957198899468,
-      "grad_norm": 3568.4501953125,
-      "learning_rate": 4.9900560639682045e-05,
-      "loss": 123.9855,
-      "step": 31390
-    },
-    {
-      "epoch": 0.12685997325436232,
-      "grad_norm": 1042.5748291015625,
-      "learning_rate": 4.9900249369144434e-05,
-      "loss": 74.6795,
-      "step": 31400
-    },
-    {
-      "epoch": 0.12690037451972996,
-      "grad_norm": 2803.200439453125,
-      "learning_rate": 4.9899937613163635e-05,
-      "loss": 122.4246,
-      "step": 31410
-    },
-    {
-      "epoch": 0.12694077578509758,
-      "grad_norm": 999.986083984375,
-      "learning_rate": 4.9899625371745726e-05,
-      "loss": 103.829,
-      "step": 31420
-    },
-    {
-      "epoch": 0.12698117705046522,
-      "grad_norm": 420.03607177734375,
-      "learning_rate": 4.989931264489681e-05,
-      "loss": 109.0738,
-      "step": 31430
-    },
-    {
-      "epoch": 0.12702157831583286,
-      "grad_norm": 679.0191040039062,
-      "learning_rate": 4.9898999432622974e-05,
-      "loss": 63.1434,
-      "step": 31440
-    },
-    {
-      "epoch": 0.1270619795812005,
-      "grad_norm": 1536.908203125,
-      "learning_rate": 4.989868573493032e-05,
-      "loss": 114.7305,
-      "step": 31450
-    },
-    {
-      "epoch": 0.1271023808465681,
-      "grad_norm": 2571.055419921875,
-      "learning_rate": 4.9898371551824974e-05,
-      "loss": 91.3618,
-      "step": 31460
-    },
-    {
-      "epoch": 0.12714278211193575,
-      "grad_norm": 1286.792724609375,
-      "learning_rate": 4.989805688331306e-05,
-      "loss": 96.6507,
-      "step": 31470
-    },
-    {
-      "epoch": 0.1271831833773034,
-      "grad_norm": 1014.9740600585938,
-      "learning_rate": 4.9897741729400705e-05,
-      "loss": 82.7385,
-      "step": 31480
-    },
-    {
-      "epoch": 0.127223584642671,
-      "grad_norm": 3557.023681640625,
-      "learning_rate": 4.989742609009405e-05,
-      "loss": 105.2578,
-      "step": 31490
-    },
-    {
-      "epoch": 0.12726398590803864,
-      "grad_norm": 1261.7489013671875,
-      "learning_rate": 4.989710996539926e-05,
-      "loss": 110.2257,
-      "step": 31500
-    },
-    {
-      "epoch": 0.12730438717340628,
-      "grad_norm": 620.3264770507812,
-      "learning_rate": 4.9896793355322495e-05,
-      "loss": 130.9313,
-      "step": 31510
-    },
-    {
-      "epoch": 0.1273447884387739,
-      "grad_norm": 1063.13623046875,
-      "learning_rate": 4.989647625986993e-05,
-      "loss": 84.6815,
-      "step": 31520
-    },
-    {
-      "epoch": 0.12738518970414153,
-      "grad_norm": 1258.741943359375,
-      "learning_rate": 4.989615867904773e-05,
-      "loss": 101.0329,
-      "step": 31530
-    },
-    {
-      "epoch": 0.12742559096950917,
-      "grad_norm": 483.7554931640625,
-      "learning_rate": 4.989584061286211e-05,
-      "loss": 115.0386,
-      "step": 31540
-    },
-    {
-      "epoch": 0.12746599223487678,
-      "grad_norm": 1390.3809814453125,
-      "learning_rate": 4.989552206131925e-05,
-      "loss": 95.0824,
-      "step": 31550
-    },
-    {
-      "epoch": 0.12750639350024442,
-      "grad_norm": 474.9691162109375,
-      "learning_rate": 4.9895203024425385e-05,
-      "loss": 123.0268,
-      "step": 31560
-    },
-    {
-      "epoch": 0.12754679476561206,
-      "grad_norm": 953.542236328125,
-      "learning_rate": 4.989488350218671e-05,
-      "loss": 132.4043,
-      "step": 31570
-    },
-    {
-      "epoch": 0.12758719603097968,
-      "grad_norm": 602.45751953125,
-      "learning_rate": 4.989456349460947e-05,
-      "loss": 81.3055,
-      "step": 31580
-    },
-    {
-      "epoch": 0.12762759729634732,
-      "grad_norm": 1298.2872314453125,
-      "learning_rate": 4.989424300169989e-05,
-      "loss": 85.7651,
-      "step": 31590
-    },
-    {
-      "epoch": 0.12766799856171496,
-      "grad_norm": 539.2799072265625,
-      "learning_rate": 4.9893922023464236e-05,
-      "loss": 73.1652,
-      "step": 31600
-    },
-    {
-      "epoch": 0.1277083998270826,
-      "grad_norm": 506.7164001464844,
-      "learning_rate": 4.989360055990875e-05,
-      "loss": 85.522,
-      "step": 31610
-    },
-    {
-      "epoch": 0.1277488010924502,
-      "grad_norm": 1420.2366943359375,
-      "learning_rate": 4.98932786110397e-05,
-      "loss": 101.5808,
-      "step": 31620
-    },
-    {
-      "epoch": 0.12778920235781785,
-      "grad_norm": 931.4535522460938,
-      "learning_rate": 4.989295617686337e-05,
-      "loss": 104.2542,
-      "step": 31630
-    },
-    {
-      "epoch": 0.1278296036231855,
-      "grad_norm": 1208.561767578125,
-      "learning_rate": 4.989263325738605e-05,
-      "loss": 118.4887,
-      "step": 31640
-    },
-    {
-      "epoch": 0.1278700048885531,
-      "grad_norm": 1164.8492431640625,
-      "learning_rate": 4.989230985261403e-05,
-      "loss": 133.7977,
-      "step": 31650
-    },
-    {
-      "epoch": 0.12791040615392074,
-      "grad_norm": 2086.97900390625,
-      "learning_rate": 4.9891985962553606e-05,
-      "loss": 138.2947,
-      "step": 31660
-    },
-    {
-      "epoch": 0.12795080741928838,
-      "grad_norm": 896.4519653320312,
-      "learning_rate": 4.98916615872111e-05,
-      "loss": 108.6655,
-      "step": 31670
-    },
-    {
-      "epoch": 0.127991208684656,
-      "grad_norm": 921.4371337890625,
-      "learning_rate": 4.9891336726592844e-05,
-      "loss": 72.2636,
-      "step": 31680
-    },
-    {
-      "epoch": 0.12803160995002363,
-      "grad_norm": 2175.284423828125,
-      "learning_rate": 4.989101138070516e-05,
-      "loss": 108.8024,
-      "step": 31690
-    },
-    {
-      "epoch": 0.12807201121539127,
-      "grad_norm": 456.76068115234375,
-      "learning_rate": 4.989068554955439e-05,
-      "loss": 124.3368,
-      "step": 31700
-    },
-    {
-      "epoch": 0.12811241248075889,
-      "grad_norm": 2323.61962890625,
-      "learning_rate": 4.9890359233146897e-05,
-      "loss": 83.3551,
-      "step": 31710
-    },
-    {
-      "epoch": 0.12815281374612653,
-      "grad_norm": 630.8578491210938,
-      "learning_rate": 4.989003243148904e-05,
-      "loss": 112.8626,
-      "step": 31720
-    },
-    {
-      "epoch": 0.12819321501149417,
-      "grad_norm": 1055.7919921875,
-      "learning_rate": 4.988970514458718e-05,
-      "loss": 85.2178,
-      "step": 31730
-    },
-    {
-      "epoch": 0.12823361627686178,
-      "grad_norm": 4026.397705078125,
-      "learning_rate": 4.9889377372447706e-05,
-      "loss": 136.7049,
-      "step": 31740
-    },
-    {
-      "epoch": 0.12827401754222942,
-      "grad_norm": 1399.34130859375,
-      "learning_rate": 4.9889049115077005e-05,
-      "loss": 90.4039,
-      "step": 31750
-    },
-    {
-      "epoch": 0.12831441880759706,
-      "grad_norm": 718.3490600585938,
-      "learning_rate": 4.988872037248148e-05,
-      "loss": 122.6609,
-      "step": 31760
-    },
-    {
-      "epoch": 0.1283548200729647,
-      "grad_norm": 1984.33740234375,
-      "learning_rate": 4.988839114466753e-05,
-      "loss": 118.7678,
-      "step": 31770
-    },
-    {
-      "epoch": 0.1283952213383323,
-      "grad_norm": 732.53662109375,
-      "learning_rate": 4.988806143164159e-05,
-      "loss": 59.4425,
-      "step": 31780
-    },
-    {
-      "epoch": 0.12843562260369995,
-      "grad_norm": 559.675537109375,
-      "learning_rate": 4.988773123341007e-05,
-      "loss": 67.2548,
-      "step": 31790
-    },
-    {
-      "epoch": 0.1284760238690676,
-      "grad_norm": 1089.072998046875,
-      "learning_rate": 4.988740054997943e-05,
-      "loss": 96.6428,
-      "step": 31800
-    },
-    {
-      "epoch": 0.1285164251344352,
-      "grad_norm": 2197.211181640625,
-      "learning_rate": 4.9887069381356094e-05,
-      "loss": 116.156,
-      "step": 31810
-    },
-    {
-      "epoch": 0.12855682639980284,
-      "grad_norm": 459.337158203125,
-      "learning_rate": 4.988673772754653e-05,
-      "loss": 105.2984,
-      "step": 31820
-    },
-    {
-      "epoch": 0.12859722766517048,
-      "grad_norm": 1142.6385498046875,
-      "learning_rate": 4.98864055885572e-05,
-      "loss": 97.5586,
-      "step": 31830
-    },
-    {
-      "epoch": 0.1286376289305381,
-      "grad_norm": 952.4442749023438,
-      "learning_rate": 4.988607296439458e-05,
-      "loss": 112.1516,
-      "step": 31840
-    },
-    {
-      "epoch": 0.12867803019590573,
-      "grad_norm": 633.4367065429688,
-      "learning_rate": 4.988573985506516e-05,
-      "loss": 82.8759,
-      "step": 31850
-    },
-    {
-      "epoch": 0.12871843146127337,
-      "grad_norm": 341.5099792480469,
-      "learning_rate": 4.988540626057543e-05,
-      "loss": 79.2266,
-      "step": 31860
-    },
-    {
-      "epoch": 0.128758832726641,
-      "grad_norm": 521.6610717773438,
-      "learning_rate": 4.988507218093189e-05,
-      "loss": 74.551,
-      "step": 31870
-    },
-    {
-      "epoch": 0.12879923399200863,
-      "grad_norm": 848.8890991210938,
-      "learning_rate": 4.988473761614105e-05,
-      "loss": 68.6997,
-      "step": 31880
-    },
-    {
-      "epoch": 0.12883963525737627,
-      "grad_norm": 1142.2823486328125,
-      "learning_rate": 4.9884402566209445e-05,
-      "loss": 84.3612,
-      "step": 31890
-    },
-    {
-      "epoch": 0.12888003652274388,
-      "grad_norm": 1104.7357177734375,
-      "learning_rate": 4.98840670311436e-05,
-      "loss": 114.3409,
-      "step": 31900
-    },
-    {
-      "epoch": 0.12892043778811152,
-      "grad_norm": 1430.170654296875,
-      "learning_rate": 4.988373101095006e-05,
-      "loss": 66.6711,
-      "step": 31910
-    },
-    {
-      "epoch": 0.12896083905347916,
-      "grad_norm": 962.3909301757812,
-      "learning_rate": 4.9883394505635364e-05,
-      "loss": 103.3977,
-      "step": 31920
-    },
-    {
-      "epoch": 0.1290012403188468,
-      "grad_norm": 1590.525634765625,
-      "learning_rate": 4.988305751520609e-05,
-      "loss": 107.3967,
-      "step": 31930
-    },
-    {
-      "epoch": 0.1290416415842144,
-      "grad_norm": 373.5509033203125,
-      "learning_rate": 4.988272003966879e-05,
-      "loss": 61.0093,
-      "step": 31940
-    },
-    {
-      "epoch": 0.12908204284958205,
-      "grad_norm": 845.4537353515625,
-      "learning_rate": 4.9882382079030064e-05,
-      "loss": 87.1438,
-      "step": 31950
-    },
-    {
-      "epoch": 0.1291224441149497,
-      "grad_norm": 945.7670288085938,
-      "learning_rate": 4.988204363329648e-05,
-      "loss": 67.9447,
-      "step": 31960
-    },
-    {
-      "epoch": 0.1291628453803173,
-      "grad_norm": 398.878173828125,
-      "learning_rate": 4.988170470247465e-05,
-      "loss": 81.1569,
-      "step": 31970
-    },
-    {
-      "epoch": 0.12920324664568494,
-      "grad_norm": 329.99859619140625,
-      "learning_rate": 4.988136528657118e-05,
-      "loss": 59.9264,
-      "step": 31980
-    },
-    {
-      "epoch": 0.12924364791105258,
-      "grad_norm": 875.9232177734375,
-      "learning_rate": 4.988102538559268e-05,
-      "loss": 118.8401,
-      "step": 31990
-    },
-    {
-      "epoch": 0.1292840491764202,
-      "grad_norm": 1466.2398681640625,
-      "learning_rate": 4.988068499954578e-05,
-      "loss": 106.1032,
-      "step": 32000
-    },
-    {
-      "epoch": 0.12932445044178784,
-      "grad_norm": 1423.832763671875,
-      "learning_rate": 4.9880344128437115e-05,
-      "loss": 138.5585,
-      "step": 32010
-    },
-    {
-      "epoch": 0.12936485170715548,
-      "grad_norm": 2598.307373046875,
-      "learning_rate": 4.988000277227334e-05,
-      "loss": 115.3402,
-      "step": 32020
-    },
-    {
-      "epoch": 0.1294052529725231,
-      "grad_norm": 1134.399169921875,
-      "learning_rate": 4.987966093106109e-05,
-      "loss": 90.1487,
-      "step": 32030
-    },
-    {
-      "epoch": 0.12944565423789073,
-      "grad_norm": 765.7633056640625,
-      "learning_rate": 4.987931860480705e-05,
-      "loss": 100.8343,
-      "step": 32040
-    },
-    {
-      "epoch": 0.12948605550325837,
-      "grad_norm": 1252.3284912109375,
-      "learning_rate": 4.987897579351788e-05,
-      "loss": 78.8595,
-      "step": 32050
-    },
-    {
-      "epoch": 0.12952645676862598,
-      "grad_norm": 4336.3466796875,
-      "learning_rate": 4.987863249720027e-05,
-      "loss": 95.2753,
-      "step": 32060
-    },
-    {
-      "epoch": 0.12956685803399362,
-      "grad_norm": 1328.06396484375,
-      "learning_rate": 4.987828871586091e-05,
-      "loss": 101.3385,
-      "step": 32070
-    },
-    {
-      "epoch": 0.12960725929936126,
-      "grad_norm": 520.3338012695312,
-      "learning_rate": 4.987794444950651e-05,
-      "loss": 153.1254,
-      "step": 32080
-    },
-    {
-      "epoch": 0.1296476605647289,
-      "grad_norm": 694.16015625,
-      "learning_rate": 4.987759969814377e-05,
-      "loss": 93.5572,
-      "step": 32090
-    },
-    {
-      "epoch": 0.1296880618300965,
-      "grad_norm": 1401.1295166015625,
-      "learning_rate": 4.987725446177941e-05,
-      "loss": 93.8664,
-      "step": 32100
-    },
-    {
-      "epoch": 0.12972846309546415,
-      "grad_norm": 928.5399780273438,
-      "learning_rate": 4.9876908740420175e-05,
-      "loss": 74.3067,
-      "step": 32110
-    },
-    {
-      "epoch": 0.1297688643608318,
-      "grad_norm": 1556.6954345703125,
-      "learning_rate": 4.98765625340728e-05,
-      "loss": 85.9339,
-      "step": 32120
-    },
-    {
-      "epoch": 0.1298092656261994,
-      "grad_norm": 783.3317260742188,
-      "learning_rate": 4.987621584274402e-05,
-      "loss": 85.9349,
-      "step": 32130
-    },
-    {
-      "epoch": 0.12984966689156704,
-      "grad_norm": 423.84014892578125,
-      "learning_rate": 4.9875868666440604e-05,
-      "loss": 49.8796,
-      "step": 32140
-    },
-    {
-      "epoch": 0.12989006815693468,
-      "grad_norm": 806.3920288085938,
-      "learning_rate": 4.987552100516933e-05,
-      "loss": 66.9737,
-      "step": 32150
-    },
-    {
-      "epoch": 0.1299304694223023,
-      "grad_norm": 974.2014770507812,
-      "learning_rate": 4.987517285893697e-05,
-      "loss": 131.7362,
-      "step": 32160
-    },
-    {
-      "epoch": 0.12997087068766994,
-      "grad_norm": 3982.545654296875,
-      "learning_rate": 4.9874824227750305e-05,
-      "loss": 130.7934,
-      "step": 32170
-    },
-    {
-      "epoch": 0.13001127195303758,
-      "grad_norm": 936.3251953125,
-      "learning_rate": 4.987447511161612e-05,
-      "loss": 86.1245,
-      "step": 32180
-    },
-    {
-      "epoch": 0.1300516732184052,
-      "grad_norm": 1825.6453857421875,
-      "learning_rate": 4.987412551054126e-05,
-      "loss": 95.4473,
-      "step": 32190
-    },
-    {
-      "epoch": 0.13009207448377283,
-      "grad_norm": 1493.9964599609375,
-      "learning_rate": 4.987377542453251e-05,
-      "loss": 78.5434,
-      "step": 32200
-    },
-    {
-      "epoch": 0.13013247574914047,
-      "grad_norm": 748.4346923828125,
-      "learning_rate": 4.9873424853596695e-05,
-      "loss": 87.2075,
-      "step": 32210
-    },
-    {
-      "epoch": 0.13017287701450808,
-      "grad_norm": 1414.2127685546875,
-      "learning_rate": 4.987307379774066e-05,
-      "loss": 95.8589,
-      "step": 32220
-    },
-    {
-      "epoch": 0.13021327827987572,
-      "grad_norm": 799.97705078125,
-      "learning_rate": 4.987272225697125e-05,
-      "loss": 94.7114,
-      "step": 32230
-    },
-    {
-      "epoch": 0.13025367954524336,
-      "grad_norm": 2368.305419921875,
-      "learning_rate": 4.987237023129531e-05,
-      "loss": 93.2615,
-      "step": 32240
-    },
-    {
-      "epoch": 0.130294080810611,
-      "grad_norm": 525.53759765625,
-      "learning_rate": 4.987201772071971e-05,
-      "loss": 100.8188,
-      "step": 32250
-    },
-    {
-      "epoch": 0.1303344820759786,
-      "grad_norm": 1076.550048828125,
-      "learning_rate": 4.9871664725251314e-05,
-      "loss": 59.2903,
-      "step": 32260
-    },
-    {
-      "epoch": 0.13037488334134625,
-      "grad_norm": 1305.9718017578125,
-      "learning_rate": 4.987131124489701e-05,
-      "loss": 62.7737,
-      "step": 32270
-    },
-    {
-      "epoch": 0.1304152846067139,
-      "grad_norm": 1377.45458984375,
-      "learning_rate": 4.98709572796637e-05,
-      "loss": 98.2218,
-      "step": 32280
-    },
-    {
-      "epoch": 0.1304556858720815,
-      "grad_norm": 1250.3414306640625,
-      "learning_rate": 4.987060282955826e-05,
-      "loss": 141.3949,
-      "step": 32290
-    },
-    {
-      "epoch": 0.13049608713744915,
-      "grad_norm": 362.1710205078125,
-      "learning_rate": 4.987024789458762e-05,
-      "loss": 56.7362,
-      "step": 32300
-    },
-    {
-      "epoch": 0.13053648840281679,
-      "grad_norm": 1179.3179931640625,
-      "learning_rate": 4.9869892474758694e-05,
-      "loss": 110.4925,
-      "step": 32310
-    },
-    {
-      "epoch": 0.1305768896681844,
-      "grad_norm": 885.8596801757812,
-      "learning_rate": 4.986953657007841e-05,
-      "loss": 103.9362,
-      "step": 32320
-    },
-    {
-      "epoch": 0.13061729093355204,
-      "grad_norm": 2053.674072265625,
-      "learning_rate": 4.98691801805537e-05,
-      "loss": 108.4903,
-      "step": 32330
-    },
-    {
-      "epoch": 0.13065769219891968,
-      "grad_norm": 628.6939086914062,
-      "learning_rate": 4.986882330619152e-05,
-      "loss": 66.5057,
-      "step": 32340
-    },
-    {
-      "epoch": 0.1306980934642873,
-      "grad_norm": 9528.09765625,
-      "learning_rate": 4.986846594699883e-05,
-      "loss": 103.5715,
-      "step": 32350
-    },
-    {
-      "epoch": 0.13073849472965493,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9868108102982604e-05,
-      "loss": 121.9924,
-      "step": 32360
-    },
-    {
-      "epoch": 0.13077889599502257,
-      "grad_norm": 710.1858520507812,
-      "learning_rate": 4.986774977414979e-05,
-      "loss": 65.1207,
-      "step": 32370
-    },
-    {
-      "epoch": 0.13081929726039018,
-      "grad_norm": 474.6985778808594,
-      "learning_rate": 4.98673909605074e-05,
-      "loss": 102.3303,
-      "step": 32380
-    },
-    {
-      "epoch": 0.13085969852575782,
-      "grad_norm": 0.0,
-      "learning_rate": 4.986703166206242e-05,
-      "loss": 75.2703,
-      "step": 32390
-    },
-    {
-      "epoch": 0.13090009979112546,
-      "grad_norm": 781.9612426757812,
-      "learning_rate": 4.986667187882186e-05,
-      "loss": 115.7585,
-      "step": 32400
-    },
-    {
-      "epoch": 0.1309405010564931,
-      "grad_norm": 1456.1153564453125,
-      "learning_rate": 4.986631161079272e-05,
-      "loss": 119.8207,
-      "step": 32410
-    },
-    {
-      "epoch": 0.13098090232186071,
-      "grad_norm": 767.140625,
-      "learning_rate": 4.986595085798204e-05,
-      "loss": 79.3004,
-      "step": 32420
-    },
-    {
-      "epoch": 0.13102130358722835,
-      "grad_norm": 1090.2476806640625,
-      "learning_rate": 4.9865589620396837e-05,
-      "loss": 104.849,
-      "step": 32430
-    },
-    {
-      "epoch": 0.131061704852596,
-      "grad_norm": 618.4625854492188,
-      "learning_rate": 4.986522789804417e-05,
-      "loss": 88.606,
-      "step": 32440
-    },
-    {
-      "epoch": 0.1311021061179636,
-      "grad_norm": 1400.4354248046875,
-      "learning_rate": 4.9864865690931086e-05,
-      "loss": 112.4108,
-      "step": 32450
-    },
-    {
-      "epoch": 0.13114250738333125,
-      "grad_norm": 1226.9744873046875,
-      "learning_rate": 4.986450299906464e-05,
-      "loss": 110.0555,
-      "step": 32460
-    },
-    {
-      "epoch": 0.1311829086486989,
-      "grad_norm": 1217.69970703125,
-      "learning_rate": 4.9864139822451905e-05,
-      "loss": 105.7761,
-      "step": 32470
-    },
-    {
-      "epoch": 0.1312233099140665,
-      "grad_norm": 944.0647583007812,
-      "learning_rate": 4.9863776161099964e-05,
-      "loss": 68.9442,
-      "step": 32480
-    },
-    {
-      "epoch": 0.13126371117943414,
-      "grad_norm": 1241.845458984375,
-      "learning_rate": 4.986341201501591e-05,
-      "loss": 95.3951,
-      "step": 32490
-    },
-    {
-      "epoch": 0.13130411244480178,
-      "grad_norm": 1810.86572265625,
-      "learning_rate": 4.9863047384206835e-05,
-      "loss": 132.1521,
-      "step": 32500
-    },
-    {
-      "epoch": 0.1313445137101694,
-      "grad_norm": 1479.045166015625,
-      "learning_rate": 4.986268226867985e-05,
-      "loss": 109.1159,
-      "step": 32510
-    },
-    {
-      "epoch": 0.13138491497553703,
-      "grad_norm": 958.012451171875,
-      "learning_rate": 4.986231666844208e-05,
-      "loss": 84.0408,
-      "step": 32520
-    },
-    {
-      "epoch": 0.13142531624090467,
-      "grad_norm": 1029.1376953125,
-      "learning_rate": 4.9861950583500636e-05,
-      "loss": 161.0161,
-      "step": 32530
-    },
-    {
-      "epoch": 0.13146571750627228,
-      "grad_norm": 2178.518798828125,
-      "learning_rate": 4.986158401386268e-05,
-      "loss": 125.5231,
-      "step": 32540
-    },
-    {
-      "epoch": 0.13150611877163992,
-      "grad_norm": 873.2677612304688,
-      "learning_rate": 4.9861216959535335e-05,
-      "loss": 82.9297,
-      "step": 32550
-    },
-    {
-      "epoch": 0.13154652003700756,
-      "grad_norm": 3118.9599609375,
-      "learning_rate": 4.9860849420525766e-05,
-      "loss": 133.3087,
-      "step": 32560
-    },
-    {
-      "epoch": 0.1315869213023752,
-      "grad_norm": 342.8658752441406,
-      "learning_rate": 4.986048139684114e-05,
-      "loss": 77.4298,
-      "step": 32570
-    },
-    {
-      "epoch": 0.13162732256774282,
-      "grad_norm": 1096.45458984375,
-      "learning_rate": 4.986011288848863e-05,
-      "loss": 64.3582,
-      "step": 32580
-    },
-    {
-      "epoch": 0.13166772383311046,
-      "grad_norm": 802.4067993164062,
-      "learning_rate": 4.9859743895475416e-05,
-      "loss": 106.3808,
-      "step": 32590
-    },
-    {
-      "epoch": 0.1317081250984781,
-      "grad_norm": 1518.53173828125,
-      "learning_rate": 4.98593744178087e-05,
-      "loss": 86.2783,
-      "step": 32600
-    },
-    {
-      "epoch": 0.1317485263638457,
-      "grad_norm": 662.360107421875,
-      "learning_rate": 4.985900445549568e-05,
-      "loss": 127.7222,
-      "step": 32610
-    },
-    {
-      "epoch": 0.13178892762921335,
-      "grad_norm": 1433.412109375,
-      "learning_rate": 4.985863400854358e-05,
-      "loss": 109.9828,
-      "step": 32620
-    },
-    {
-      "epoch": 0.131829328894581,
-      "grad_norm": 524.9939575195312,
-      "learning_rate": 4.98582630769596e-05,
-      "loss": 114.5297,
-      "step": 32630
-    },
-    {
-      "epoch": 0.1318697301599486,
-      "grad_norm": 432.99395751953125,
-      "learning_rate": 4.9857891660750986e-05,
-      "loss": 74.3833,
-      "step": 32640
-    },
-    {
-      "epoch": 0.13191013142531624,
-      "grad_norm": 1059.305908203125,
-      "learning_rate": 4.9857519759924974e-05,
-      "loss": 103.7896,
-      "step": 32650
-    },
-    {
-      "epoch": 0.13195053269068388,
-      "grad_norm": 1521.5408935546875,
-      "learning_rate": 4.985714737448882e-05,
-      "loss": 110.8204,
-      "step": 32660
-    },
-    {
-      "epoch": 0.1319909339560515,
-      "grad_norm": 712.7562255859375,
-      "learning_rate": 4.9856774504449776e-05,
-      "loss": 77.0656,
-      "step": 32670
-    },
-    {
-      "epoch": 0.13203133522141913,
-      "grad_norm": 738.4374389648438,
-      "learning_rate": 4.9856401149815126e-05,
-      "loss": 76.575,
-      "step": 32680
-    },
-    {
-      "epoch": 0.13207173648678677,
-      "grad_norm": 521.0355834960938,
-      "learning_rate": 4.9856027310592134e-05,
-      "loss": 79.7847,
-      "step": 32690
-    },
-    {
-      "epoch": 0.13211213775215438,
-      "grad_norm": 738.5291137695312,
-      "learning_rate": 4.985565298678809e-05,
-      "loss": 74.041,
-      "step": 32700
-    },
-    {
-      "epoch": 0.13215253901752202,
-      "grad_norm": 1271.053955078125,
-      "learning_rate": 4.985527817841029e-05,
-      "loss": 87.5718,
-      "step": 32710
-    },
-    {
-      "epoch": 0.13219294028288966,
-      "grad_norm": 1135.2099609375,
-      "learning_rate": 4.985490288546606e-05,
-      "loss": 121.2673,
-      "step": 32720
-    },
-    {
-      "epoch": 0.1322333415482573,
-      "grad_norm": 866.3362426757812,
-      "learning_rate": 4.9854527107962686e-05,
-      "loss": 107.6865,
-      "step": 32730
-    },
-    {
-      "epoch": 0.13227374281362492,
-      "grad_norm": 1969.1063232421875,
-      "learning_rate": 4.985415084590752e-05,
-      "loss": 79.8334,
-      "step": 32740
-    },
-    {
-      "epoch": 0.13231414407899256,
-      "grad_norm": 1149.1275634765625,
-      "learning_rate": 4.985377409930789e-05,
-      "loss": 109.8118,
-      "step": 32750
-    },
-    {
-      "epoch": 0.1323545453443602,
-      "grad_norm": 756.9971313476562,
-      "learning_rate": 4.985339686817113e-05,
-      "loss": 118.7679,
-      "step": 32760
-    },
-    {
-      "epoch": 0.1323949466097278,
-      "grad_norm": 795.5819702148438,
-      "learning_rate": 4.9853019152504607e-05,
-      "loss": 91.6352,
-      "step": 32770
-    },
-    {
-      "epoch": 0.13243534787509545,
-      "grad_norm": 1227.4793701171875,
-      "learning_rate": 4.9852640952315674e-05,
-      "loss": 60.6576,
-      "step": 32780
-    },
-    {
-      "epoch": 0.1324757491404631,
-      "grad_norm": 1278.319091796875,
-      "learning_rate": 4.985226226761172e-05,
-      "loss": 113.7857,
-      "step": 32790
-    },
-    {
-      "epoch": 0.1325161504058307,
-      "grad_norm": 610.24560546875,
-      "learning_rate": 4.985188309840012e-05,
-      "loss": 78.6529,
-      "step": 32800
-    },
-    {
-      "epoch": 0.13255655167119834,
-      "grad_norm": 1682.79736328125,
-      "learning_rate": 4.9851503444688255e-05,
-      "loss": 107.8541,
-      "step": 32810
-    },
-    {
-      "epoch": 0.13259695293656598,
-      "grad_norm": 493.7384033203125,
-      "learning_rate": 4.985112330648354e-05,
-      "loss": 79.7233,
-      "step": 32820
-    },
-    {
-      "epoch": 0.1326373542019336,
-      "grad_norm": 790.5695190429688,
-      "learning_rate": 4.985074268379338e-05,
-      "loss": 91.9399,
-      "step": 32830
-    },
-    {
-      "epoch": 0.13267775546730123,
-      "grad_norm": 1110.0252685546875,
-      "learning_rate": 4.985036157662521e-05,
-      "loss": 160.5353,
-      "step": 32840
-    },
-    {
-      "epoch": 0.13271815673266887,
-      "grad_norm": 1280.8482666015625,
-      "learning_rate": 4.9849979984986426e-05,
-      "loss": 139.881,
-      "step": 32850
-    },
-    {
-      "epoch": 0.13275855799803649,
-      "grad_norm": 973.560302734375,
-      "learning_rate": 4.98495979088845e-05,
-      "loss": 124.0516,
-      "step": 32860
-    },
-    {
-      "epoch": 0.13279895926340413,
-      "grad_norm": 496.84893798828125,
-      "learning_rate": 4.9849215348326875e-05,
-      "loss": 90.3565,
-      "step": 32870
-    },
-    {
-      "epoch": 0.13283936052877177,
-      "grad_norm": 1748.6497802734375,
-      "learning_rate": 4.984883230332099e-05,
-      "loss": 84.9742,
-      "step": 32880
-    },
-    {
-      "epoch": 0.1328797617941394,
-      "grad_norm": 1350.8636474609375,
-      "learning_rate": 4.984844877387433e-05,
-      "loss": 115.242,
-      "step": 32890
-    },
-    {
-      "epoch": 0.13292016305950702,
-      "grad_norm": 776.8285522460938,
-      "learning_rate": 4.984806475999437e-05,
-      "loss": 77.2372,
-      "step": 32900
-    },
-    {
-      "epoch": 0.13296056432487466,
-      "grad_norm": 863.73583984375,
-      "learning_rate": 4.98476802616886e-05,
-      "loss": 74.2184,
-      "step": 32910
-    },
-    {
-      "epoch": 0.1330009655902423,
-      "grad_norm": 793.2598876953125,
-      "learning_rate": 4.9847295278964514e-05,
-      "loss": 67.6055,
-      "step": 32920
-    },
-    {
-      "epoch": 0.1330413668556099,
-      "grad_norm": 1486.9317626953125,
-      "learning_rate": 4.9846909811829604e-05,
-      "loss": 127.2083,
-      "step": 32930
-    },
-    {
-      "epoch": 0.13308176812097755,
-      "grad_norm": 773.4334716796875,
-      "learning_rate": 4.984652386029139e-05,
-      "loss": 116.8202,
-      "step": 32940
-    },
-    {
-      "epoch": 0.1331221693863452,
-      "grad_norm": 1226.421630859375,
-      "learning_rate": 4.984613742435742e-05,
-      "loss": 112.299,
-      "step": 32950
-    },
-    {
-      "epoch": 0.1331625706517128,
-      "grad_norm": 445.5863342285156,
-      "learning_rate": 4.9845750504035195e-05,
-      "loss": 87.6496,
-      "step": 32960
-    },
-    {
-      "epoch": 0.13320297191708044,
-      "grad_norm": 894.6140747070312,
-      "learning_rate": 4.984536309933227e-05,
-      "loss": 95.4575,
-      "step": 32970
-    },
-    {
-      "epoch": 0.13324337318244808,
-      "grad_norm": 823.8974609375,
-      "learning_rate": 4.9844975210256217e-05,
-      "loss": 93.921,
-      "step": 32980
-    },
-    {
-      "epoch": 0.1332837744478157,
-      "grad_norm": 755.107177734375,
-      "learning_rate": 4.984458683681457e-05,
-      "loss": 84.3962,
-      "step": 32990
-    },
-    {
-      "epoch": 0.13332417571318333,
-      "grad_norm": 885.6873779296875,
-      "learning_rate": 4.984419797901491e-05,
-      "loss": 105.6254,
-      "step": 33000
-    },
-    {
-      "epoch": 0.13336457697855097,
-      "grad_norm": 411.86053466796875,
-      "learning_rate": 4.984380863686482e-05,
-      "loss": 97.5172,
-      "step": 33010
-    },
-    {
-      "epoch": 0.1334049782439186,
-      "grad_norm": 563.2962646484375,
-      "learning_rate": 4.98434188103719e-05,
-      "loss": 80.259,
-      "step": 33020
-    },
-    {
-      "epoch": 0.13344537950928623,
-      "grad_norm": 1379.732666015625,
-      "learning_rate": 4.984302849954373e-05,
-      "loss": 73.214,
-      "step": 33030
-    },
-    {
-      "epoch": 0.13348578077465387,
-      "grad_norm": 750.8110961914062,
-      "learning_rate": 4.984263770438793e-05,
-      "loss": 106.9937,
-      "step": 33040
-    },
-    {
-      "epoch": 0.1335261820400215,
-      "grad_norm": 1013.02392578125,
-      "learning_rate": 4.984224642491212e-05,
-      "loss": 75.5326,
-      "step": 33050
-    },
-    {
-      "epoch": 0.13356658330538912,
-      "grad_norm": 1022.1004028320312,
-      "learning_rate": 4.9841854661123936e-05,
-      "loss": 108.553,
-      "step": 33060
-    },
-    {
-      "epoch": 0.13360698457075676,
-      "grad_norm": 809.585205078125,
-      "learning_rate": 4.9841462413030995e-05,
-      "loss": 84.0129,
-      "step": 33070
-    },
-    {
-      "epoch": 0.1336473858361244,
-      "grad_norm": 842.181884765625,
-      "learning_rate": 4.984106968064095e-05,
-      "loss": 112.834,
-      "step": 33080
-    },
-    {
-      "epoch": 0.133687787101492,
-      "grad_norm": 467.4346618652344,
-      "learning_rate": 4.984067646396147e-05,
-      "loss": 100.528,
-      "step": 33090
-    },
-    {
-      "epoch": 0.13372818836685965,
-      "grad_norm": 714.903076171875,
-      "learning_rate": 4.984028276300021e-05,
-      "loss": 129.6565,
-      "step": 33100
-    },
-    {
-      "epoch": 0.1337685896322273,
-      "grad_norm": 709.3721923828125,
-      "learning_rate": 4.983988857776486e-05,
-      "loss": 101.3018,
-      "step": 33110
-    },
-    {
-      "epoch": 0.1338089908975949,
-      "grad_norm": 988.5148315429688,
-      "learning_rate": 4.983949390826308e-05,
-      "loss": 86.796,
-      "step": 33120
-    },
-    {
-      "epoch": 0.13384939216296254,
-      "grad_norm": 1472.410400390625,
-      "learning_rate": 4.983909875450258e-05,
-      "loss": 94.508,
-      "step": 33130
-    },
-    {
-      "epoch": 0.13388979342833018,
-      "grad_norm": 900.3340454101562,
-      "learning_rate": 4.983870311649107e-05,
-      "loss": 95.947,
-      "step": 33140
-    },
-    {
-      "epoch": 0.1339301946936978,
-      "grad_norm": 1466.88134765625,
-      "learning_rate": 4.983830699423625e-05,
-      "loss": 74.9536,
-      "step": 33150
-    },
-    {
-      "epoch": 0.13397059595906544,
-      "grad_norm": 576.2639770507812,
-      "learning_rate": 4.9837910387745845e-05,
-      "loss": 115.5023,
-      "step": 33160
-    },
-    {
-      "epoch": 0.13401099722443308,
-      "grad_norm": 1318.8853759765625,
-      "learning_rate": 4.9837513297027594e-05,
-      "loss": 84.5853,
-      "step": 33170
-    },
-    {
-      "epoch": 0.1340513984898007,
-      "grad_norm": 934.3829345703125,
-      "learning_rate": 4.983711572208924e-05,
-      "loss": 75.3381,
-      "step": 33180
-    },
-    {
-      "epoch": 0.13409179975516833,
-      "grad_norm": 1154.966796875,
-      "learning_rate": 4.983671766293851e-05,
-      "loss": 83.0238,
-      "step": 33190
-    },
-    {
-      "epoch": 0.13413220102053597,
-      "grad_norm": 1403.0806884765625,
-      "learning_rate": 4.983631911958319e-05,
-      "loss": 82.2895,
-      "step": 33200
-    },
-    {
-      "epoch": 0.1341726022859036,
-      "grad_norm": 1025.5306396484375,
-      "learning_rate": 4.983592009203105e-05,
-      "loss": 64.9099,
-      "step": 33210
-    },
-    {
-      "epoch": 0.13421300355127122,
-      "grad_norm": 570.6565551757812,
-      "learning_rate": 4.9835520580289854e-05,
-      "loss": 60.6472,
-      "step": 33220
-    },
-    {
-      "epoch": 0.13425340481663886,
-      "grad_norm": 241.88272094726562,
-      "learning_rate": 4.98351205843674e-05,
-      "loss": 101.3504,
-      "step": 33230
-    },
-    {
-      "epoch": 0.1342938060820065,
-      "grad_norm": 786.84912109375,
-      "learning_rate": 4.9834720104271484e-05,
-      "loss": 80.2996,
-      "step": 33240
-    },
-    {
-      "epoch": 0.1343342073473741,
-      "grad_norm": 1382.0052490234375,
-      "learning_rate": 4.983431914000991e-05,
-      "loss": 90.2512,
-      "step": 33250
-    },
-    {
-      "epoch": 0.13437460861274175,
-      "grad_norm": 1485.1060791015625,
-      "learning_rate": 4.9833917691590506e-05,
-      "loss": 80.0044,
-      "step": 33260
-    },
-    {
-      "epoch": 0.1344150098781094,
-      "grad_norm": 698.7577514648438,
-      "learning_rate": 4.9833515759021085e-05,
-      "loss": 102.9575,
-      "step": 33270
-    },
-    {
-      "epoch": 0.134455411143477,
-      "grad_norm": 693.4671020507812,
-      "learning_rate": 4.98331133423095e-05,
-      "loss": 94.7776,
-      "step": 33280
-    },
-    {
-      "epoch": 0.13449581240884464,
-      "grad_norm": 664.5126342773438,
-      "learning_rate": 4.983271044146357e-05,
-      "loss": 85.3929,
-      "step": 33290
-    },
-    {
-      "epoch": 0.13453621367421228,
-      "grad_norm": 841.61962890625,
-      "learning_rate": 4.983230705649118e-05,
-      "loss": 92.7831,
-      "step": 33300
-    },
-    {
-      "epoch": 0.1345766149395799,
-      "grad_norm": 1524.328125,
-      "learning_rate": 4.9831903187400166e-05,
-      "loss": 85.1477,
-      "step": 33310
-    },
-    {
-      "epoch": 0.13461701620494754,
-      "grad_norm": 551.5435180664062,
-      "learning_rate": 4.983149883419842e-05,
-      "loss": 128.3945,
-      "step": 33320
-    },
-    {
-      "epoch": 0.13465741747031518,
-      "grad_norm": 387.754638671875,
-      "learning_rate": 4.983109399689382e-05,
-      "loss": 75.982,
-      "step": 33330
-    },
-    {
-      "epoch": 0.1346978187356828,
-      "grad_norm": 2052.11669921875,
-      "learning_rate": 4.9830688675494265e-05,
-      "loss": 166.9485,
-      "step": 33340
-    },
-    {
-      "epoch": 0.13473822000105043,
-      "grad_norm": 656.3280639648438,
-      "learning_rate": 4.9830282870007646e-05,
-      "loss": 98.0935,
-      "step": 33350
-    },
-    {
-      "epoch": 0.13477862126641807,
-      "grad_norm": 960.6358032226562,
-      "learning_rate": 4.982987658044188e-05,
-      "loss": 88.8502,
-      "step": 33360
-    },
-    {
-      "epoch": 0.1348190225317857,
-      "grad_norm": 303.2107238769531,
-      "learning_rate": 4.982946980680488e-05,
-      "loss": 82.7864,
-      "step": 33370
-    },
-    {
-      "epoch": 0.13485942379715332,
-      "grad_norm": 786.0392456054688,
-      "learning_rate": 4.982906254910459e-05,
-      "loss": 75.8299,
-      "step": 33380
-    },
-    {
-      "epoch": 0.13489982506252096,
-      "grad_norm": 554.2486572265625,
-      "learning_rate": 4.982865480734894e-05,
-      "loss": 71.6516,
-      "step": 33390
-    },
-    {
-      "epoch": 0.1349402263278886,
-      "grad_norm": 901.5972900390625,
-      "learning_rate": 4.982824658154589e-05,
-      "loss": 106.4935,
-      "step": 33400
-    },
-    {
-      "epoch": 0.1349806275932562,
-      "grad_norm": 363.12115478515625,
-      "learning_rate": 4.982783787170338e-05,
-      "loss": 76.4278,
-      "step": 33410
-    },
-    {
-      "epoch": 0.13502102885862385,
-      "grad_norm": 3618.558349609375,
-      "learning_rate": 4.982742867782939e-05,
-      "loss": 130.8146,
-      "step": 33420
-    },
-    {
-      "epoch": 0.1350614301239915,
-      "grad_norm": 810.9340209960938,
-      "learning_rate": 4.982701899993189e-05,
-      "loss": 109.7431,
-      "step": 33430
-    },
-    {
-      "epoch": 0.1351018313893591,
-      "grad_norm": 1115.0909423828125,
-      "learning_rate": 4.982660883801889e-05,
-      "loss": 97.727,
-      "step": 33440
-    },
-    {
-      "epoch": 0.13514223265472675,
-      "grad_norm": 726.3258056640625,
-      "learning_rate": 4.982619819209836e-05,
-      "loss": 129.2327,
-      "step": 33450
-    },
-    {
-      "epoch": 0.13518263392009439,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9825787062178315e-05,
-      "loss": 62.2287,
-      "step": 33460
-    },
-    {
-      "epoch": 0.135223035185462,
-      "grad_norm": 2895.0556640625,
-      "learning_rate": 4.982537544826677e-05,
-      "loss": 99.4745,
-      "step": 33470
-    },
-    {
-      "epoch": 0.13526343645082964,
-      "grad_norm": 496.2646789550781,
-      "learning_rate": 4.982496335037175e-05,
-      "loss": 76.2467,
-      "step": 33480
-    },
-    {
-      "epoch": 0.13530383771619728,
-      "grad_norm": 1723.678955078125,
-      "learning_rate": 4.982455076850129e-05,
-      "loss": 87.5003,
-      "step": 33490
-    },
-    {
-      "epoch": 0.1353442389815649,
-      "grad_norm": 622.0546875,
-      "learning_rate": 4.982413770266342e-05,
-      "loss": 60.5188,
-      "step": 33500
-    },
-    {
-      "epoch": 0.13538464024693253,
-      "grad_norm": 579.060302734375,
-      "learning_rate": 4.9823724152866226e-05,
-      "loss": 121.2273,
-      "step": 33510
-    },
-    {
-      "epoch": 0.13542504151230017,
-      "grad_norm": 638.8704833984375,
-      "learning_rate": 4.982331011911774e-05,
-      "loss": 139.4159,
-      "step": 33520
-    },
-    {
-      "epoch": 0.13546544277766778,
-      "grad_norm": 684.1176147460938,
-      "learning_rate": 4.9822895601426034e-05,
-      "loss": 107.1239,
-      "step": 33530
-    },
-    {
-      "epoch": 0.13550584404303542,
-      "grad_norm": 626.8060302734375,
-      "learning_rate": 4.982248059979921e-05,
-      "loss": 86.9168,
-      "step": 33540
-    },
-    {
-      "epoch": 0.13554624530840306,
-      "grad_norm": 539.0343017578125,
-      "learning_rate": 4.982206511424534e-05,
-      "loss": 69.0915,
-      "step": 33550
-    },
-    {
-      "epoch": 0.1355866465737707,
-      "grad_norm": 2398.862548828125,
-      "learning_rate": 4.9821649144772545e-05,
-      "loss": 70.2374,
-      "step": 33560
-    },
-    {
-      "epoch": 0.13562704783913831,
-      "grad_norm": 586.1361694335938,
-      "learning_rate": 4.9821232691388906e-05,
-      "loss": 70.2269,
-      "step": 33570
-    },
-    {
-      "epoch": 0.13566744910450595,
-      "grad_norm": 850.7914428710938,
-      "learning_rate": 4.982081575410256e-05,
-      "loss": 113.3011,
-      "step": 33580
-    },
-    {
-      "epoch": 0.1357078503698736,
-      "grad_norm": 642.4421997070312,
-      "learning_rate": 4.9820398332921634e-05,
-      "loss": 95.5486,
-      "step": 33590
-    },
-    {
-      "epoch": 0.1357482516352412,
-      "grad_norm": 706.2947998046875,
-      "learning_rate": 4.981998042785427e-05,
-      "loss": 97.3085,
-      "step": 33600
-    },
-    {
-      "epoch": 0.13578865290060885,
-      "grad_norm": 1023.1542358398438,
-      "learning_rate": 4.98195620389086e-05,
-      "loss": 117.8821,
-      "step": 33610
-    },
-    {
-      "epoch": 0.1358290541659765,
-      "grad_norm": 480.221923828125,
-      "learning_rate": 4.9819143166092796e-05,
-      "loss": 93.8683,
-      "step": 33620
-    },
-    {
-      "epoch": 0.1358694554313441,
-      "grad_norm": 520.1914672851562,
-      "learning_rate": 4.9818723809415016e-05,
-      "loss": 83.5975,
-      "step": 33630
-    },
-    {
-      "epoch": 0.13590985669671174,
-      "grad_norm": 2167.59619140625,
-      "learning_rate": 4.981830396888344e-05,
-      "loss": 103.9532,
-      "step": 33640
-    },
-    {
-      "epoch": 0.13595025796207938,
-      "grad_norm": 801.8353271484375,
-      "learning_rate": 4.981788364450625e-05,
-      "loss": 146.1724,
-      "step": 33650
-    },
-    {
-      "epoch": 0.135990659227447,
-      "grad_norm": 862.556884765625,
-      "learning_rate": 4.981746283629164e-05,
-      "loss": 92.8243,
-      "step": 33660
-    },
-    {
-      "epoch": 0.13603106049281463,
-      "grad_norm": 1226.8758544921875,
-      "learning_rate": 4.981704154424781e-05,
-      "loss": 107.6284,
-      "step": 33670
-    },
-    {
-      "epoch": 0.13607146175818227,
-      "grad_norm": 1550.146484375,
-      "learning_rate": 4.981661976838299e-05,
-      "loss": 109.7946,
-      "step": 33680
-    },
-    {
-      "epoch": 0.13611186302354988,
-      "grad_norm": 619.8020629882812,
-      "learning_rate": 4.981619750870537e-05,
-      "loss": 108.0941,
-      "step": 33690
-    },
-    {
-      "epoch": 0.13615226428891752,
-      "grad_norm": 551.2567749023438,
-      "learning_rate": 4.9815774765223226e-05,
-      "loss": 75.5076,
-      "step": 33700
-    },
-    {
-      "epoch": 0.13619266555428516,
-      "grad_norm": 1045.827880859375,
-      "learning_rate": 4.9815351537944774e-05,
-      "loss": 64.202,
-      "step": 33710
-    },
-    {
-      "epoch": 0.1362330668196528,
-      "grad_norm": 435.85296630859375,
-      "learning_rate": 4.9814927826878256e-05,
-      "loss": 123.6437,
-      "step": 33720
-    },
-    {
-      "epoch": 0.13627346808502042,
-      "grad_norm": 1007.7003784179688,
-      "learning_rate": 4.9814503632031954e-05,
-      "loss": 133.8719,
-      "step": 33730
-    },
-    {
-      "epoch": 0.13631386935038806,
-      "grad_norm": 1092.2225341796875,
-      "learning_rate": 4.981407895341412e-05,
-      "loss": 91.8184,
-      "step": 33740
-    },
-    {
-      "epoch": 0.1363542706157557,
-      "grad_norm": 656.14990234375,
-      "learning_rate": 4.9813653791033057e-05,
-      "loss": 68.1897,
-      "step": 33750
-    },
-    {
-      "epoch": 0.1363946718811233,
-      "grad_norm": 1275.6177978515625,
-      "learning_rate": 4.981322814489703e-05,
-      "loss": 91.9896,
-      "step": 33760
-    },
-    {
-      "epoch": 0.13643507314649095,
-      "grad_norm": 900.5950927734375,
-      "learning_rate": 4.9812802015014334e-05,
-      "loss": 90.7882,
-      "step": 33770
-    },
-    {
-      "epoch": 0.1364754744118586,
-      "grad_norm": 0.0,
-      "learning_rate": 4.981237540139331e-05,
-      "loss": 62.0134,
-      "step": 33780
-    },
-    {
-      "epoch": 0.1365158756772262,
-      "grad_norm": 1441.5318603515625,
-      "learning_rate": 4.9811948304042234e-05,
-      "loss": 75.5356,
-      "step": 33790
-    },
-    {
-      "epoch": 0.13655627694259384,
-      "grad_norm": 783.706787109375,
-      "learning_rate": 4.9811520722969465e-05,
-      "loss": 109.1127,
-      "step": 33800
-    },
-    {
-      "epoch": 0.13659667820796148,
-      "grad_norm": 1493.0281982421875,
-      "learning_rate": 4.981109265818332e-05,
-      "loss": 150.1899,
-      "step": 33810
-    },
-    {
-      "epoch": 0.1366370794733291,
-      "grad_norm": 728.7564086914062,
-      "learning_rate": 4.981066410969215e-05,
-      "loss": 64.3308,
-      "step": 33820
-    },
-    {
-      "epoch": 0.13667748073869673,
-      "grad_norm": 1312.0137939453125,
-      "learning_rate": 4.981023507750431e-05,
-      "loss": 119.9241,
-      "step": 33830
-    },
-    {
-      "epoch": 0.13671788200406437,
-      "grad_norm": 805.4586181640625,
-      "learning_rate": 4.980980556162816e-05,
-      "loss": 107.8721,
-      "step": 33840
-    },
-    {
-      "epoch": 0.13675828326943198,
-      "grad_norm": 609.840087890625,
-      "learning_rate": 4.980937556207208e-05,
-      "loss": 89.4216,
-      "step": 33850
-    },
-    {
-      "epoch": 0.13679868453479962,
-      "grad_norm": 973.7212524414062,
-      "learning_rate": 4.9808945078844456e-05,
-      "loss": 81.5434,
-      "step": 33860
-    },
-    {
-      "epoch": 0.13683908580016726,
-      "grad_norm": 982.0657348632812,
-      "learning_rate": 4.9808514111953674e-05,
-      "loss": 67.6982,
-      "step": 33870
-    },
-    {
-      "epoch": 0.1368794870655349,
-      "grad_norm": 524.6974487304688,
-      "learning_rate": 4.980808266140813e-05,
-      "loss": 150.7908,
-      "step": 33880
-    },
-    {
-      "epoch": 0.13691988833090252,
-      "grad_norm": 600.6156005859375,
-      "learning_rate": 4.980765072721625e-05,
-      "loss": 107.8641,
-      "step": 33890
-    },
-    {
-      "epoch": 0.13696028959627016,
-      "grad_norm": 464.95550537109375,
-      "learning_rate": 4.9807218309386444e-05,
-      "loss": 81.4719,
-      "step": 33900
-    },
-    {
-      "epoch": 0.1370006908616378,
-      "grad_norm": 810.7647094726562,
-      "learning_rate": 4.980678540792715e-05,
-      "loss": 91.2513,
-      "step": 33910
-    },
-    {
-      "epoch": 0.1370410921270054,
-      "grad_norm": 904.2017211914062,
-      "learning_rate": 4.980635202284679e-05,
-      "loss": 128.9432,
-      "step": 33920
-    },
-    {
-      "epoch": 0.13708149339237305,
-      "grad_norm": 751.0636596679688,
-      "learning_rate": 4.980591815415384e-05,
-      "loss": 67.3943,
-      "step": 33930
-    },
-    {
-      "epoch": 0.1371218946577407,
-      "grad_norm": 999.1458740234375,
-      "learning_rate": 4.980548380185674e-05,
-      "loss": 122.1365,
-      "step": 33940
-    },
-    {
-      "epoch": 0.1371622959231083,
-      "grad_norm": 1589.1033935546875,
-      "learning_rate": 4.980504896596396e-05,
-      "loss": 77.5396,
-      "step": 33950
-    },
-    {
-      "epoch": 0.13720269718847594,
-      "grad_norm": 1546.7073974609375,
-      "learning_rate": 4.980461364648398e-05,
-      "loss": 117.3774,
-      "step": 33960
-    },
-    {
-      "epoch": 0.13724309845384358,
-      "grad_norm": 920.711669921875,
-      "learning_rate": 4.9804177843425295e-05,
-      "loss": 112.9565,
-      "step": 33970
-    },
-    {
-      "epoch": 0.1372834997192112,
-      "grad_norm": 1206.650146484375,
-      "learning_rate": 4.980374155679639e-05,
-      "loss": 111.5966,
-      "step": 33980
-    },
-    {
-      "epoch": 0.13732390098457883,
-      "grad_norm": 1469.6834716796875,
-      "learning_rate": 4.980330478660576e-05,
-      "loss": 107.332,
-      "step": 33990
-    },
-    {
-      "epoch": 0.13736430224994647,
-      "grad_norm": 1957.3115234375,
-      "learning_rate": 4.980286753286195e-05,
-      "loss": 128.1132,
-      "step": 34000
-    },
-    {
-      "epoch": 0.13740470351531409,
-      "grad_norm": 337.71295166015625,
-      "learning_rate": 4.9802429795573455e-05,
-      "loss": 70.1459,
-      "step": 34010
-    },
-    {
-      "epoch": 0.13744510478068173,
-      "grad_norm": 714.4182739257812,
-      "learning_rate": 4.980199157474884e-05,
-      "loss": 116.3985,
-      "step": 34020
-    },
-    {
-      "epoch": 0.13748550604604937,
-      "grad_norm": 691.6640014648438,
-      "learning_rate": 4.980155287039662e-05,
-      "loss": 54.4459,
-      "step": 34030
-    },
-    {
-      "epoch": 0.137525907311417,
-      "grad_norm": 1381.4683837890625,
-      "learning_rate": 4.980111368252535e-05,
-      "loss": 59.3446,
-      "step": 34040
-    },
-    {
-      "epoch": 0.13756630857678462,
-      "grad_norm": 1003.1581420898438,
-      "learning_rate": 4.9800674011143614e-05,
-      "loss": 79.539,
-      "step": 34050
-    },
-    {
-      "epoch": 0.13760670984215226,
-      "grad_norm": 705.7440795898438,
-      "learning_rate": 4.980023385625996e-05,
-      "loss": 104.7091,
-      "step": 34060
-    },
-    {
-      "epoch": 0.1376471111075199,
-      "grad_norm": 655.6250610351562,
-      "learning_rate": 4.979979321788298e-05,
-      "loss": 65.6317,
-      "step": 34070
-    },
-    {
-      "epoch": 0.1376875123728875,
-      "grad_norm": 741.8604125976562,
-      "learning_rate": 4.9799352096021266e-05,
-      "loss": 70.4658,
-      "step": 34080
-    },
-    {
-      "epoch": 0.13772791363825515,
-      "grad_norm": 2940.106689453125,
-      "learning_rate": 4.979891049068342e-05,
-      "loss": 108.38,
-      "step": 34090
-    },
-    {
-      "epoch": 0.1377683149036228,
-      "grad_norm": 924.1039428710938,
-      "learning_rate": 4.979846840187804e-05,
-      "loss": 69.7543,
-      "step": 34100
-    },
-    {
-      "epoch": 0.1378087161689904,
-      "grad_norm": 937.28515625,
-      "learning_rate": 4.979802582961375e-05,
-      "loss": 97.4457,
-      "step": 34110
-    },
-    {
-      "epoch": 0.13784911743435804,
-      "grad_norm": 1431.635498046875,
-      "learning_rate": 4.979758277389919e-05,
-      "loss": 91.3253,
-      "step": 34120
-    },
-    {
-      "epoch": 0.13788951869972568,
-      "grad_norm": 953.5215454101562,
-      "learning_rate": 4.9797139234742975e-05,
-      "loss": 81.4589,
-      "step": 34130
-    },
-    {
-      "epoch": 0.1379299199650933,
-      "grad_norm": 4141.5361328125,
-      "learning_rate": 4.9796695212153764e-05,
-      "loss": 154.3838,
-      "step": 34140
-    },
-    {
-      "epoch": 0.13797032123046093,
-      "grad_norm": 779.7581176757812,
-      "learning_rate": 4.9796250706140224e-05,
-      "loss": 69.3488,
-      "step": 34150
-    },
-    {
-      "epoch": 0.13801072249582857,
-      "grad_norm": 778.5997314453125,
-      "learning_rate": 4.9795805716711e-05,
-      "loss": 135.2858,
-      "step": 34160
-    },
-    {
-      "epoch": 0.1380511237611962,
-      "grad_norm": 990.5801391601562,
-      "learning_rate": 4.979536024387479e-05,
-      "loss": 108.3737,
-      "step": 34170
-    },
-    {
-      "epoch": 0.13809152502656383,
-      "grad_norm": 715.4221801757812,
-      "learning_rate": 4.979491428764026e-05,
-      "loss": 95.1732,
-      "step": 34180
-    },
-    {
-      "epoch": 0.13813192629193147,
-      "grad_norm": 2391.443115234375,
-      "learning_rate": 4.979446784801611e-05,
-      "loss": 151.664,
-      "step": 34190
-    },
-    {
-      "epoch": 0.1381723275572991,
-      "grad_norm": 1992.6207275390625,
-      "learning_rate": 4.9794020925011044e-05,
-      "loss": 82.8294,
-      "step": 34200
-    },
-    {
-      "epoch": 0.13821272882266672,
-      "grad_norm": 1085.1295166015625,
-      "learning_rate": 4.979357351863377e-05,
-      "loss": 116.4222,
-      "step": 34210
-    },
-    {
-      "epoch": 0.13825313008803436,
-      "grad_norm": 607.43603515625,
-      "learning_rate": 4.979312562889302e-05,
-      "loss": 71.7245,
-      "step": 34220
-    },
-    {
-      "epoch": 0.138293531353402,
-      "grad_norm": 782.8214111328125,
-      "learning_rate": 4.9792677255797525e-05,
-      "loss": 93.9537,
-      "step": 34230
-    },
-    {
-      "epoch": 0.1383339326187696,
-      "grad_norm": 735.2789306640625,
-      "learning_rate": 4.979222839935602e-05,
-      "loss": 71.7033,
-      "step": 34240
-    },
-    {
-      "epoch": 0.13837433388413725,
-      "grad_norm": 1233.6981201171875,
-      "learning_rate": 4.979177905957726e-05,
-      "loss": 90.7622,
-      "step": 34250
-    },
-    {
-      "epoch": 0.1384147351495049,
-      "grad_norm": 567.0103149414062,
-      "learning_rate": 4.979132923647001e-05,
-      "loss": 135.1103,
-      "step": 34260
-    },
-    {
-      "epoch": 0.1384551364148725,
-      "grad_norm": 594.86669921875,
-      "learning_rate": 4.979087893004302e-05,
-      "loss": 82.0568,
-      "step": 34270
-    },
-    {
-      "epoch": 0.13849553768024014,
-      "grad_norm": 900.0689086914062,
-      "learning_rate": 4.979042814030509e-05,
-      "loss": 91.8436,
-      "step": 34280
-    },
-    {
-      "epoch": 0.13853593894560778,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9789976867265e-05,
-      "loss": 67.3411,
-      "step": 34290
-    },
-    {
-      "epoch": 0.1385763402109754,
-      "grad_norm": 1077.3468017578125,
-      "learning_rate": 4.9789525110931545e-05,
-      "loss": 119.6317,
-      "step": 34300
-    },
-    {
-      "epoch": 0.13861674147634304,
-      "grad_norm": 2667.85400390625,
-      "learning_rate": 4.978907287131354e-05,
-      "loss": 95.113,
-      "step": 34310
-    },
-    {
-      "epoch": 0.13865714274171068,
-      "grad_norm": 1062.1884765625,
-      "learning_rate": 4.978862014841979e-05,
-      "loss": 77.8131,
-      "step": 34320
-    },
-    {
-      "epoch": 0.1386975440070783,
-      "grad_norm": 3490.418701171875,
-      "learning_rate": 4.9788166942259135e-05,
-      "loss": 124.19,
-      "step": 34330
-    },
-    {
-      "epoch": 0.13873794527244593,
-      "grad_norm": 351.10784912109375,
-      "learning_rate": 4.97877132528404e-05,
-      "loss": 100.0667,
-      "step": 34340
-    },
-    {
-      "epoch": 0.13877834653781357,
-      "grad_norm": 901.481201171875,
-      "learning_rate": 4.978725908017243e-05,
-      "loss": 80.7147,
-      "step": 34350
-    },
-    {
-      "epoch": 0.1388187478031812,
-      "grad_norm": 575.747802734375,
-      "learning_rate": 4.9786804424264085e-05,
-      "loss": 107.6961,
-      "step": 34360
-    },
-    {
-      "epoch": 0.13885914906854882,
-      "grad_norm": 816.9763793945312,
-      "learning_rate": 4.9786349285124235e-05,
-      "loss": 104.7212,
-      "step": 34370
-    },
-    {
-      "epoch": 0.13889955033391646,
-      "grad_norm": 672.7227783203125,
-      "learning_rate": 4.978589366276174e-05,
-      "loss": 95.8187,
-      "step": 34380
-    },
-    {
-      "epoch": 0.1389399515992841,
-      "grad_norm": 1123.2491455078125,
-      "learning_rate": 4.978543755718549e-05,
-      "loss": 97.5948,
-      "step": 34390
-    },
-    {
-      "epoch": 0.1389803528646517,
-      "grad_norm": 741.401611328125,
-      "learning_rate": 4.978498096840436e-05,
-      "loss": 77.3318,
-      "step": 34400
-    },
-    {
-      "epoch": 0.13902075413001935,
-      "grad_norm": 2102.46923828125,
-      "learning_rate": 4.978452389642728e-05,
-      "loss": 80.4057,
-      "step": 34410
-    },
-    {
-      "epoch": 0.139061155395387,
-      "grad_norm": 1185.681396484375,
-      "learning_rate": 4.978406634126315e-05,
-      "loss": 105.0425,
-      "step": 34420
-    },
-    {
-      "epoch": 0.1391015566607546,
-      "grad_norm": 1234.7724609375,
-      "learning_rate": 4.9783608302920873e-05,
-      "loss": 94.0603,
-      "step": 34430
-    },
-    {
-      "epoch": 0.13914195792612225,
-      "grad_norm": 1568.1796875,
-      "learning_rate": 4.9783149781409404e-05,
-      "loss": 123.3275,
-      "step": 34440
-    },
-    {
-      "epoch": 0.13918235919148988,
-      "grad_norm": 1667.6737060546875,
-      "learning_rate": 4.978269077673767e-05,
-      "loss": 62.5816,
-      "step": 34450
-    },
-    {
-      "epoch": 0.1392227604568575,
-      "grad_norm": 880.3914794921875,
-      "learning_rate": 4.9782231288914614e-05,
-      "loss": 80.6274,
-      "step": 34460
-    },
-    {
-      "epoch": 0.13926316172222514,
-      "grad_norm": 473.5460510253906,
-      "learning_rate": 4.97817713179492e-05,
-      "loss": 136.4202,
-      "step": 34470
-    },
-    {
-      "epoch": 0.13930356298759278,
-      "grad_norm": 1051.316162109375,
-      "learning_rate": 4.9781310863850405e-05,
-      "loss": 79.2705,
-      "step": 34480
-    },
-    {
-      "epoch": 0.1393439642529604,
-      "grad_norm": 942.3121948242188,
-      "learning_rate": 4.978084992662719e-05,
-      "loss": 87.1078,
-      "step": 34490
-    },
-    {
-      "epoch": 0.13938436551832803,
-      "grad_norm": 663.4690551757812,
-      "learning_rate": 4.978038850628854e-05,
-      "loss": 87.4095,
-      "step": 34500
-    },
-    {
-      "epoch": 0.13942476678369567,
-      "grad_norm": 469.06475830078125,
-      "learning_rate": 4.977992660284347e-05,
-      "loss": 92.2786,
-      "step": 34510
-    },
-    {
-      "epoch": 0.1394651680490633,
-      "grad_norm": 452.9181213378906,
-      "learning_rate": 4.977946421630098e-05,
-      "loss": 109.0747,
-      "step": 34520
-    },
-    {
-      "epoch": 0.13950556931443092,
-      "grad_norm": 3418.622314453125,
-      "learning_rate": 4.977900134667006e-05,
-      "loss": 92.9968,
-      "step": 34530
-    },
-    {
-      "epoch": 0.13954597057979856,
-      "grad_norm": 934.139892578125,
-      "learning_rate": 4.977853799395976e-05,
-      "loss": 86.7001,
-      "step": 34540
-    },
-    {
-      "epoch": 0.1395863718451662,
-      "grad_norm": 541.4644165039062,
-      "learning_rate": 4.97780741581791e-05,
-      "loss": 125.5777,
-      "step": 34550
-    },
-    {
-      "epoch": 0.13962677311053381,
-      "grad_norm": 1703.493896484375,
-      "learning_rate": 4.977760983933714e-05,
-      "loss": 113.32,
-      "step": 34560
-    },
-    {
-      "epoch": 0.13966717437590145,
-      "grad_norm": 469.1068115234375,
-      "learning_rate": 4.9777145037442906e-05,
-      "loss": 73.509,
-      "step": 34570
-    },
-    {
-      "epoch": 0.1397075756412691,
-      "grad_norm": 445.8856201171875,
-      "learning_rate": 4.9776679752505476e-05,
-      "loss": 72.2943,
-      "step": 34580
-    },
-    {
-      "epoch": 0.1397479769066367,
-      "grad_norm": 636.2007446289062,
-      "learning_rate": 4.977621398453393e-05,
-      "loss": 91.6618,
-      "step": 34590
-    },
-    {
-      "epoch": 0.13978837817200435,
-      "grad_norm": 727.9812622070312,
-      "learning_rate": 4.977574773353732e-05,
-      "loss": 101.3671,
-      "step": 34600
-    },
-    {
-      "epoch": 0.13982877943737199,
-      "grad_norm": 1291.48974609375,
-      "learning_rate": 4.9775280999524766e-05,
-      "loss": 75.3459,
-      "step": 34610
-    },
-    {
-      "epoch": 0.1398691807027396,
-      "grad_norm": 975.12109375,
-      "learning_rate": 4.9774813782505346e-05,
-      "loss": 108.9374,
-      "step": 34620
-    },
-    {
-      "epoch": 0.13990958196810724,
-      "grad_norm": 1339.5286865234375,
-      "learning_rate": 4.9774346082488176e-05,
-      "loss": 62.3226,
-      "step": 34630
-    },
-    {
-      "epoch": 0.13994998323347488,
-      "grad_norm": 1278.2696533203125,
-      "learning_rate": 4.977387789948238e-05,
-      "loss": 111.346,
-      "step": 34640
-    },
-    {
-      "epoch": 0.1399903844988425,
-      "grad_norm": 582.800537109375,
-      "learning_rate": 4.977340923349707e-05,
-      "loss": 59.1344,
-      "step": 34650
-    },
-    {
-      "epoch": 0.14003078576421013,
-      "grad_norm": 980.9688110351562,
-      "learning_rate": 4.9772940084541405e-05,
-      "loss": 116.2872,
-      "step": 34660
-    },
-    {
-      "epoch": 0.14007118702957777,
-      "grad_norm": 600.323486328125,
-      "learning_rate": 4.9772470452624506e-05,
-      "loss": 98.4954,
-      "step": 34670
-    },
-    {
-      "epoch": 0.1401115882949454,
-      "grad_norm": 1239.6783447265625,
-      "learning_rate": 4.977200033775555e-05,
-      "loss": 108.7115,
-      "step": 34680
-    },
-    {
-      "epoch": 0.14015198956031302,
-      "grad_norm": 1736.9547119140625,
-      "learning_rate": 4.97715297399437e-05,
-      "loss": 134.5554,
-      "step": 34690
-    },
-    {
-      "epoch": 0.14019239082568066,
-      "grad_norm": 2366.77490234375,
-      "learning_rate": 4.977105865919812e-05,
-      "loss": 127.1646,
-      "step": 34700
-    },
-    {
-      "epoch": 0.1402327920910483,
-      "grad_norm": 1988.361083984375,
-      "learning_rate": 4.9770587095527995e-05,
-      "loss": 107.3173,
-      "step": 34710
-    },
-    {
-      "epoch": 0.14027319335641592,
-      "grad_norm": 632.1370239257812,
-      "learning_rate": 4.977011504894252e-05,
-      "loss": 84.2821,
-      "step": 34720
-    },
-    {
-      "epoch": 0.14031359462178356,
-      "grad_norm": 1778.7666015625,
-      "learning_rate": 4.9769642519450904e-05,
-      "loss": 83.9897,
-      "step": 34730
-    },
-    {
-      "epoch": 0.1403539958871512,
-      "grad_norm": 652.8768310546875,
-      "learning_rate": 4.9769169507062355e-05,
-      "loss": 102.6689,
-      "step": 34740
-    },
-    {
-      "epoch": 0.1403943971525188,
-      "grad_norm": 1475.0391845703125,
-      "learning_rate": 4.976869601178609e-05,
-      "loss": 117.3591,
-      "step": 34750
-    },
-    {
-      "epoch": 0.14043479841788645,
-      "grad_norm": 1975.065673828125,
-      "learning_rate": 4.976822203363135e-05,
-      "loss": 98.2487,
-      "step": 34760
-    },
-    {
-      "epoch": 0.1404751996832541,
-      "grad_norm": 1050.9573974609375,
-      "learning_rate": 4.976774757260737e-05,
-      "loss": 94.4713,
-      "step": 34770
-    },
-    {
-      "epoch": 0.1405156009486217,
-      "grad_norm": 1176.8922119140625,
-      "learning_rate": 4.9767272628723396e-05,
-      "loss": 99.8589,
-      "step": 34780
-    },
-    {
-      "epoch": 0.14055600221398934,
-      "grad_norm": 1319.9659423828125,
-      "learning_rate": 4.976679720198869e-05,
-      "loss": 100.7659,
-      "step": 34790
-    },
-    {
-      "epoch": 0.14059640347935698,
-      "grad_norm": 515.001220703125,
-      "learning_rate": 4.976632129241252e-05,
-      "loss": 77.551,
-      "step": 34800
-    },
-    {
-      "epoch": 0.1406368047447246,
-      "grad_norm": 822.0165405273438,
-      "learning_rate": 4.9765844900004176e-05,
-      "loss": 82.4817,
-      "step": 34810
-    },
-    {
-      "epoch": 0.14067720601009223,
-      "grad_norm": 1176.460693359375,
-      "learning_rate": 4.976536802477293e-05,
-      "loss": 116.3207,
-      "step": 34820
-    },
-    {
-      "epoch": 0.14071760727545987,
-      "grad_norm": 1214.6893310546875,
-      "learning_rate": 4.976489066672808e-05,
-      "loss": 90.3449,
-      "step": 34830
-    },
-    {
-      "epoch": 0.1407580085408275,
-      "grad_norm": 610.4821166992188,
-      "learning_rate": 4.9764412825878943e-05,
-      "loss": 104.2971,
-      "step": 34840
-    },
-    {
-      "epoch": 0.14079840980619512,
-      "grad_norm": 1629.2425537109375,
-      "learning_rate": 4.976393450223482e-05,
-      "loss": 100.4276,
-      "step": 34850
-    },
-    {
-      "epoch": 0.14083881107156276,
-      "grad_norm": 1354.5982666015625,
-      "learning_rate": 4.9763455695805056e-05,
-      "loss": 87.7344,
-      "step": 34860
-    },
-    {
-      "epoch": 0.1408792123369304,
-      "grad_norm": 569.3548583984375,
-      "learning_rate": 4.976297640659897e-05,
-      "loss": 72.0556,
-      "step": 34870
-    },
-    {
-      "epoch": 0.14091961360229802,
-      "grad_norm": 1153.3983154296875,
-      "learning_rate": 4.97624966346259e-05,
-      "loss": 76.03,
-      "step": 34880
-    },
-    {
-      "epoch": 0.14096001486766566,
-      "grad_norm": 550.9409790039062,
-      "learning_rate": 4.9762016379895225e-05,
-      "loss": 106.2172,
-      "step": 34890
-    },
-    {
-      "epoch": 0.1410004161330333,
-      "grad_norm": 621.2599487304688,
-      "learning_rate": 4.976153564241628e-05,
-      "loss": 85.7488,
-      "step": 34900
-    },
-    {
-      "epoch": 0.1410408173984009,
-      "grad_norm": 1340.081787109375,
-      "learning_rate": 4.976105442219846e-05,
-      "loss": 108.7937,
-      "step": 34910
-    },
-    {
-      "epoch": 0.14108121866376855,
-      "grad_norm": 2728.06201171875,
-      "learning_rate": 4.976057271925113e-05,
-      "loss": 80.4239,
-      "step": 34920
-    },
-    {
-      "epoch": 0.1411216199291362,
-      "grad_norm": 571.938720703125,
-      "learning_rate": 4.9760090533583686e-05,
-      "loss": 61.1496,
-      "step": 34930
-    },
-    {
-      "epoch": 0.1411620211945038,
-      "grad_norm": 421.5497131347656,
-      "learning_rate": 4.9759607865205534e-05,
-      "loss": 151.0666,
-      "step": 34940
-    },
-    {
-      "epoch": 0.14120242245987144,
-      "grad_norm": 1543.9168701171875,
-      "learning_rate": 4.975912471412607e-05,
-      "loss": 121.0911,
-      "step": 34950
-    },
-    {
-      "epoch": 0.14124282372523908,
-      "grad_norm": 1396.1658935546875,
-      "learning_rate": 4.975864108035474e-05,
-      "loss": 94.3398,
-      "step": 34960
-    },
-    {
-      "epoch": 0.1412832249906067,
-      "grad_norm": 785.8390502929688,
-      "learning_rate": 4.975815696390094e-05,
-      "loss": 89.429,
-      "step": 34970
-    },
-    {
-      "epoch": 0.14132362625597433,
-      "grad_norm": 719.430908203125,
-      "learning_rate": 4.975767236477413e-05,
-      "loss": 116.3824,
-      "step": 34980
-    },
-    {
-      "epoch": 0.14136402752134197,
-      "grad_norm": 662.4757080078125,
-      "learning_rate": 4.975718728298375e-05,
-      "loss": 87.5541,
-      "step": 34990
-    },
-    {
-      "epoch": 0.1414044287867096,
-      "grad_norm": 1017.3776245117188,
-      "learning_rate": 4.975670171853926e-05,
-      "loss": 230.2885,
-      "step": 35000
-    },
-    {
-      "epoch": 0.14144483005207723,
-      "grad_norm": 391.976806640625,
-      "learning_rate": 4.975621567145012e-05,
-      "loss": 81.4888,
-      "step": 35010
-    },
-    {
-      "epoch": 0.14148523131744487,
-      "grad_norm": 1794.4031982421875,
-      "learning_rate": 4.975572914172582e-05,
-      "loss": 102.6515,
-      "step": 35020
-    },
-    {
-      "epoch": 0.1415256325828125,
-      "grad_norm": 937.76123046875,
-      "learning_rate": 4.975524212937582e-05,
-      "loss": 108.7728,
-      "step": 35030
-    },
-    {
-      "epoch": 0.14156603384818012,
-      "grad_norm": 1814.7794189453125,
-      "learning_rate": 4.975475463440964e-05,
-      "loss": 93.3038,
-      "step": 35040
-    },
-    {
-      "epoch": 0.14160643511354776,
-      "grad_norm": 1418.080322265625,
-      "learning_rate": 4.975426665683678e-05,
-      "loss": 92.7187,
-      "step": 35050
-    },
-    {
-      "epoch": 0.1416468363789154,
-      "grad_norm": 1647.2652587890625,
-      "learning_rate": 4.9753778196666737e-05,
-      "loss": 99.5951,
-      "step": 35060
-    },
-    {
-      "epoch": 0.141687237644283,
-      "grad_norm": 728.3031005859375,
-      "learning_rate": 4.975328925390904e-05,
-      "loss": 78.1284,
-      "step": 35070
-    },
-    {
-      "epoch": 0.14172763890965065,
-      "grad_norm": 1265.145263671875,
-      "learning_rate": 4.975279982857324e-05,
-      "loss": 141.2944,
-      "step": 35080
-    },
-    {
-      "epoch": 0.1417680401750183,
-      "grad_norm": 1588.732666015625,
-      "learning_rate": 4.975230992066885e-05,
-      "loss": 80.3437,
-      "step": 35090
-    },
-    {
-      "epoch": 0.1418084414403859,
-      "grad_norm": 0.0,
-      "learning_rate": 4.975181953020544e-05,
-      "loss": 101.6614,
-      "step": 35100
-    },
-    {
-      "epoch": 0.14184884270575354,
-      "grad_norm": 2776.19140625,
-      "learning_rate": 4.9751328657192565e-05,
-      "loss": 95.6715,
-      "step": 35110
-    },
-    {
-      "epoch": 0.14188924397112118,
-      "grad_norm": 2080.83837890625,
-      "learning_rate": 4.9750837301639796e-05,
-      "loss": 153.1414,
-      "step": 35120
-    },
-    {
-      "epoch": 0.1419296452364888,
-      "grad_norm": 912.7608642578125,
-      "learning_rate": 4.975034546355671e-05,
-      "loss": 106.0554,
-      "step": 35130
-    },
-    {
-      "epoch": 0.14197004650185643,
-      "grad_norm": 358.8157043457031,
-      "learning_rate": 4.97498531429529e-05,
-      "loss": 95.7549,
-      "step": 35140
-    },
-    {
-      "epoch": 0.14201044776722407,
-      "grad_norm": 387.9512634277344,
-      "learning_rate": 4.974936033983795e-05,
-      "loss": 88.8157,
-      "step": 35150
-    },
-    {
-      "epoch": 0.14205084903259171,
-      "grad_norm": 1014.5821533203125,
-      "learning_rate": 4.974886705422149e-05,
-      "loss": 103.6284,
-      "step": 35160
-    },
-    {
-      "epoch": 0.14209125029795933,
-      "grad_norm": 1092.1754150390625,
-      "learning_rate": 4.974837328611312e-05,
-      "loss": 107.7396,
-      "step": 35170
-    },
-    {
-      "epoch": 0.14213165156332697,
-      "grad_norm": 926.5281372070312,
-      "learning_rate": 4.974787903552247e-05,
-      "loss": 78.8734,
-      "step": 35180
-    },
-    {
-      "epoch": 0.1421720528286946,
-      "grad_norm": 546.8348999023438,
-      "learning_rate": 4.974738430245918e-05,
-      "loss": 189.5167,
-      "step": 35190
-    },
-    {
-      "epoch": 0.14221245409406222,
-      "grad_norm": 581.217041015625,
-      "learning_rate": 4.9746889086932895e-05,
-      "loss": 82.9471,
-      "step": 35200
-    },
-    {
-      "epoch": 0.14225285535942986,
-      "grad_norm": 414.46807861328125,
-      "learning_rate": 4.974639338895326e-05,
-      "loss": 61.2459,
-      "step": 35210
-    },
-    {
-      "epoch": 0.1422932566247975,
-      "grad_norm": 1268.71826171875,
-      "learning_rate": 4.9745897208529956e-05,
-      "loss": 102.3453,
-      "step": 35220
-    },
-    {
-      "epoch": 0.1423336578901651,
-      "grad_norm": 1139.93115234375,
-      "learning_rate": 4.974540054567264e-05,
-      "loss": 82.09,
-      "step": 35230
-    },
-    {
-      "epoch": 0.14237405915553275,
-      "grad_norm": 1343.3614501953125,
-      "learning_rate": 4.9744903400391e-05,
-      "loss": 98.7603,
-      "step": 35240
-    },
-    {
-      "epoch": 0.1424144604209004,
-      "grad_norm": 1799.45263671875,
-      "learning_rate": 4.9744405772694725e-05,
-      "loss": 90.476,
-      "step": 35250
-    },
-    {
-      "epoch": 0.142454861686268,
-      "grad_norm": 740.581787109375,
-      "learning_rate": 4.9743907662593524e-05,
-      "loss": 80.5426,
-      "step": 35260
-    },
-    {
-      "epoch": 0.14249526295163564,
-      "grad_norm": 689.4889526367188,
-      "learning_rate": 4.97434090700971e-05,
-      "loss": 109.3305,
-      "step": 35270
-    },
-    {
-      "epoch": 0.14253566421700328,
-      "grad_norm": 1570.7303466796875,
-      "learning_rate": 4.974290999521519e-05,
-      "loss": 115.1295,
-      "step": 35280
-    },
-    {
-      "epoch": 0.1425760654823709,
-      "grad_norm": 1887.6724853515625,
-      "learning_rate": 4.97424104379575e-05,
-      "loss": 78.6015,
-      "step": 35290
-    },
-    {
-      "epoch": 0.14261646674773854,
-      "grad_norm": 906.8424072265625,
-      "learning_rate": 4.974191039833378e-05,
-      "loss": 62.9002,
-      "step": 35300
-    },
-    {
-      "epoch": 0.14265686801310618,
-      "grad_norm": 1138.534423828125,
-      "learning_rate": 4.974140987635378e-05,
-      "loss": 114.5066,
-      "step": 35310
-    },
-    {
-      "epoch": 0.14269726927847382,
-      "grad_norm": 5530.13232421875,
-      "learning_rate": 4.974090887202726e-05,
-      "loss": 101.5712,
-      "step": 35320
-    },
-    {
-      "epoch": 0.14273767054384143,
-      "grad_norm": 571.6856079101562,
-      "learning_rate": 4.9740407385363983e-05,
-      "loss": 87.484,
-      "step": 35330
-    },
-    {
-      "epoch": 0.14277807180920907,
-      "grad_norm": 742.740478515625,
-      "learning_rate": 4.973990541637373e-05,
-      "loss": 116.8295,
-      "step": 35340
-    },
-    {
-      "epoch": 0.1428184730745767,
-      "grad_norm": 1002.74072265625,
-      "learning_rate": 4.9739402965066276e-05,
-      "loss": 61.5279,
-      "step": 35350
-    },
-    {
-      "epoch": 0.14285887433994432,
-      "grad_norm": 782.5518188476562,
-      "learning_rate": 4.973890003145143e-05,
-      "loss": 104.612,
-      "step": 35360
-    },
-    {
-      "epoch": 0.14289927560531196,
-      "grad_norm": 3795.69970703125,
-      "learning_rate": 4.973839661553899e-05,
-      "loss": 90.4234,
-      "step": 35370
-    },
-    {
-      "epoch": 0.1429396768706796,
-      "grad_norm": 679.3167724609375,
-      "learning_rate": 4.9737892717338774e-05,
-      "loss": 101.094,
-      "step": 35380
-    },
-    {
-      "epoch": 0.1429800781360472,
-      "grad_norm": 1362.4091796875,
-      "learning_rate": 4.97373883368606e-05,
-      "loss": 121.7002,
-      "step": 35390
-    },
-    {
-      "epoch": 0.14302047940141485,
-      "grad_norm": 692.154296875,
-      "learning_rate": 4.973688347411431e-05,
-      "loss": 99.3686,
-      "step": 35400
-    },
-    {
-      "epoch": 0.1430608806667825,
-      "grad_norm": 1434.7564697265625,
-      "learning_rate": 4.973637812910973e-05,
-      "loss": 98.6788,
-      "step": 35410
-    },
-    {
-      "epoch": 0.1431012819321501,
-      "grad_norm": 993.9842529296875,
-      "learning_rate": 4.9735872301856734e-05,
-      "loss": 98.5837,
-      "step": 35420
-    },
-    {
-      "epoch": 0.14314168319751774,
-      "grad_norm": 772.5789794921875,
-      "learning_rate": 4.973536599236517e-05,
-      "loss": 114.3075,
-      "step": 35430
-    },
-    {
-      "epoch": 0.14318208446288538,
-      "grad_norm": 1945.123046875,
-      "learning_rate": 4.9734859200644905e-05,
-      "loss": 80.5528,
-      "step": 35440
-    },
-    {
-      "epoch": 0.143222485728253,
-      "grad_norm": 1043.6229248046875,
-      "learning_rate": 4.973435192670584e-05,
-      "loss": 104.7841,
-      "step": 35450
-    },
-    {
-      "epoch": 0.14326288699362064,
-      "grad_norm": 0.0,
-      "learning_rate": 4.973384417055784e-05,
-      "loss": 83.0118,
-      "step": 35460
-    },
-    {
-      "epoch": 0.14330328825898828,
-      "grad_norm": 1075.596435546875,
-      "learning_rate": 4.9733335932210814e-05,
-      "loss": 87.0017,
-      "step": 35470
-    },
-    {
-      "epoch": 0.14334368952435592,
-      "grad_norm": 1106.34765625,
-      "learning_rate": 4.973282721167467e-05,
-      "loss": 84.2105,
-      "step": 35480
-    },
-    {
-      "epoch": 0.14338409078972353,
-      "grad_norm": 680.8936157226562,
-      "learning_rate": 4.973231800895932e-05,
-      "loss": 102.21,
-      "step": 35490
-    },
-    {
-      "epoch": 0.14342449205509117,
-      "grad_norm": 971.0728759765625,
-      "learning_rate": 4.9731808324074717e-05,
-      "loss": 89.0123,
-      "step": 35500
-    },
-    {
-      "epoch": 0.1434648933204588,
-      "grad_norm": 1773.6971435546875,
-      "learning_rate": 4.973129815703076e-05,
-      "loss": 60.7568,
-      "step": 35510
-    },
-    {
-      "epoch": 0.14350529458582642,
-      "grad_norm": 1183.9600830078125,
-      "learning_rate": 4.973078750783742e-05,
-      "loss": 91.8321,
-      "step": 35520
-    },
-    {
-      "epoch": 0.14354569585119406,
-      "grad_norm": 540.8861694335938,
-      "learning_rate": 4.973027637650464e-05,
-      "loss": 82.3345,
-      "step": 35530
-    },
-    {
-      "epoch": 0.1435860971165617,
-      "grad_norm": 2457.4091796875,
-      "learning_rate": 4.9729764763042394e-05,
-      "loss": 98.1432,
-      "step": 35540
-    },
-    {
-      "epoch": 0.1436264983819293,
-      "grad_norm": 778.1422729492188,
-      "learning_rate": 4.9729252667460655e-05,
-      "loss": 85.1171,
-      "step": 35550
-    },
-    {
-      "epoch": 0.14366689964729695,
-      "grad_norm": 501.1910095214844,
-      "learning_rate": 4.97287400897694e-05,
-      "loss": 138.8153,
-      "step": 35560
-    },
-    {
-      "epoch": 0.1437073009126646,
-      "grad_norm": 1607.3275146484375,
-      "learning_rate": 4.972822702997863e-05,
-      "loss": 80.1325,
-      "step": 35570
-    },
-    {
-      "epoch": 0.1437477021780322,
-      "grad_norm": 796.763427734375,
-      "learning_rate": 4.9727713488098335e-05,
-      "loss": 79.0268,
-      "step": 35580
-    },
-    {
-      "epoch": 0.14378810344339985,
-      "grad_norm": 661.5986938476562,
-      "learning_rate": 4.972719946413854e-05,
-      "loss": 88.3111,
-      "step": 35590
-    },
-    {
-      "epoch": 0.14382850470876749,
-      "grad_norm": 1435.8564453125,
-      "learning_rate": 4.9726684958109266e-05,
-      "loss": 107.5516,
-      "step": 35600
-    },
-    {
-      "epoch": 0.1438689059741351,
-      "grad_norm": 679.6939697265625,
-      "learning_rate": 4.972616997002053e-05,
-      "loss": 75.3328,
-      "step": 35610
-    },
-    {
-      "epoch": 0.14390930723950274,
-      "grad_norm": 1368.540283203125,
-      "learning_rate": 4.972565449988239e-05,
-      "loss": 83.6638,
-      "step": 35620
-    },
-    {
-      "epoch": 0.14394970850487038,
-      "grad_norm": 503.572998046875,
-      "learning_rate": 4.972513854770487e-05,
-      "loss": 82.9908,
-      "step": 35630
-    },
-    {
-      "epoch": 0.14399010977023802,
-      "grad_norm": 1151.1427001953125,
-      "learning_rate": 4.972462211349806e-05,
-      "loss": 94.219,
-      "step": 35640
-    },
-    {
-      "epoch": 0.14403051103560563,
-      "grad_norm": 2658.9482421875,
-      "learning_rate": 4.972410519727201e-05,
-      "loss": 91.6265,
-      "step": 35650
-    },
-    {
-      "epoch": 0.14407091230097327,
-      "grad_norm": 953.5581665039062,
-      "learning_rate": 4.97235877990368e-05,
-      "loss": 96.4948,
-      "step": 35660
-    },
-    {
-      "epoch": 0.1441113135663409,
-      "grad_norm": 898.052490234375,
-      "learning_rate": 4.972306991880251e-05,
-      "loss": 144.139,
-      "step": 35670
-    },
-    {
-      "epoch": 0.14415171483170852,
-      "grad_norm": 338.9049072265625,
-      "learning_rate": 4.972255155657925e-05,
-      "loss": 81.2305,
-      "step": 35680
-    },
-    {
-      "epoch": 0.14419211609707616,
-      "grad_norm": 662.849365234375,
-      "learning_rate": 4.972203271237712e-05,
-      "loss": 82.9598,
-      "step": 35690
-    },
-    {
-      "epoch": 0.1442325173624438,
-      "grad_norm": 488.6361999511719,
-      "learning_rate": 4.972151338620623e-05,
-      "loss": 64.3388,
-      "step": 35700
-    },
-    {
-      "epoch": 0.14427291862781141,
-      "grad_norm": 1745.017578125,
-      "learning_rate": 4.972099357807671e-05,
-      "loss": 146.5352,
-      "step": 35710
-    },
-    {
-      "epoch": 0.14431331989317905,
-      "grad_norm": 717.8941650390625,
-      "learning_rate": 4.9720473287998695e-05,
-      "loss": 82.1567,
-      "step": 35720
-    },
-    {
-      "epoch": 0.1443537211585467,
-      "grad_norm": 3314.039794921875,
-      "learning_rate": 4.9719952515982324e-05,
-      "loss": 81.9564,
-      "step": 35730
-    },
-    {
-      "epoch": 0.1443941224239143,
-      "grad_norm": 716.9427490234375,
-      "learning_rate": 4.9719431262037755e-05,
-      "loss": 94.6004,
-      "step": 35740
-    },
-    {
-      "epoch": 0.14443452368928195,
-      "grad_norm": 591.388427734375,
-      "learning_rate": 4.971890952617515e-05,
-      "loss": 107.062,
-      "step": 35750
-    },
-    {
-      "epoch": 0.1444749249546496,
-      "grad_norm": 1554.075927734375,
-      "learning_rate": 4.9718387308404675e-05,
-      "loss": 119.4833,
-      "step": 35760
-    },
-    {
-      "epoch": 0.1445153262200172,
-      "grad_norm": 440.0912780761719,
-      "learning_rate": 4.9717864608736506e-05,
-      "loss": 104.854,
-      "step": 35770
-    },
-    {
-      "epoch": 0.14455572748538484,
-      "grad_norm": 692.7317504882812,
-      "learning_rate": 4.971734142718085e-05,
-      "loss": 116.6411,
-      "step": 35780
-    },
-    {
-      "epoch": 0.14459612875075248,
-      "grad_norm": 520.4461059570312,
-      "learning_rate": 4.971681776374789e-05,
-      "loss": 111.4582,
-      "step": 35790
-    },
-    {
-      "epoch": 0.14463653001612012,
-      "grad_norm": 762.310546875,
-      "learning_rate": 4.971629361844785e-05,
-      "loss": 102.7609,
-      "step": 35800
-    },
-    {
-      "epoch": 0.14467693128148773,
-      "grad_norm": 852.2344970703125,
-      "learning_rate": 4.971576899129094e-05,
-      "loss": 121.1159,
-      "step": 35810
-    },
-    {
-      "epoch": 0.14471733254685537,
-      "grad_norm": 445.6109924316406,
-      "learning_rate": 4.9715243882287386e-05,
-      "loss": 89.8152,
-      "step": 35820
-    },
-    {
-      "epoch": 0.144757733812223,
-      "grad_norm": 950.6737060546875,
-      "learning_rate": 4.971471829144743e-05,
-      "loss": 75.9464,
-      "step": 35830
-    },
-    {
-      "epoch": 0.14479813507759062,
-      "grad_norm": 726.3805541992188,
-      "learning_rate": 4.9714192218781316e-05,
-      "loss": 81.0347,
-      "step": 35840
-    },
-    {
-      "epoch": 0.14483853634295826,
-      "grad_norm": 1368.5889892578125,
-      "learning_rate": 4.97136656642993e-05,
-      "loss": 94.3249,
-      "step": 35850
-    },
-    {
-      "epoch": 0.1448789376083259,
-      "grad_norm": 2891.8837890625,
-      "learning_rate": 4.9713138628011654e-05,
-      "loss": 88.5745,
-      "step": 35860
-    },
-    {
-      "epoch": 0.14491933887369352,
-      "grad_norm": 883.2448120117188,
-      "learning_rate": 4.971261110992864e-05,
-      "loss": 86.1088,
-      "step": 35870
-    },
-    {
-      "epoch": 0.14495974013906116,
-      "grad_norm": 750.9356689453125,
-      "learning_rate": 4.9712083110060556e-05,
-      "loss": 157.1846,
-      "step": 35880
-    },
-    {
-      "epoch": 0.1450001414044288,
-      "grad_norm": 1293.0548095703125,
-      "learning_rate": 4.971155462841769e-05,
-      "loss": 97.4993,
-      "step": 35890
-    },
-    {
-      "epoch": 0.1450405426697964,
-      "grad_norm": 1317.3045654296875,
-      "learning_rate": 4.971102566501034e-05,
-      "loss": 77.7684,
-      "step": 35900
-    },
-    {
-      "epoch": 0.14508094393516405,
-      "grad_norm": 698.4030151367188,
-      "learning_rate": 4.971049621984882e-05,
-      "loss": 68.1086,
-      "step": 35910
-    },
-    {
-      "epoch": 0.1451213452005317,
-      "grad_norm": 491.6965637207031,
-      "learning_rate": 4.9709966292943455e-05,
-      "loss": 85.3335,
-      "step": 35920
-    },
-    {
-      "epoch": 0.1451617464658993,
-      "grad_norm": 3422.90380859375,
-      "learning_rate": 4.970943588430458e-05,
-      "loss": 122.6019,
-      "step": 35930
-    },
-    {
-      "epoch": 0.14520214773126694,
-      "grad_norm": 933.2683715820312,
-      "learning_rate": 4.970890499394253e-05,
-      "loss": 116.233,
-      "step": 35940
-    },
-    {
-      "epoch": 0.14524254899663458,
-      "grad_norm": 1370.2513427734375,
-      "learning_rate": 4.9708373621867656e-05,
-      "loss": 68.2386,
-      "step": 35950
-    },
-    {
-      "epoch": 0.14528295026200222,
-      "grad_norm": 2012.014892578125,
-      "learning_rate": 4.9707841768090314e-05,
-      "loss": 73.6452,
-      "step": 35960
-    },
-    {
-      "epoch": 0.14532335152736983,
-      "grad_norm": 1204.0938720703125,
-      "learning_rate": 4.9707309432620874e-05,
-      "loss": 82.8698,
-      "step": 35970
-    },
-    {
-      "epoch": 0.14536375279273747,
-      "grad_norm": 1037.1630859375,
-      "learning_rate": 4.9706776615469716e-05,
-      "loss": 80.2174,
-      "step": 35980
-    },
-    {
-      "epoch": 0.1454041540581051,
-      "grad_norm": 566.7116088867188,
-      "learning_rate": 4.970624331664724e-05,
-      "loss": 80.2169,
-      "step": 35990
-    },
-    {
-      "epoch": 0.14544455532347272,
-      "grad_norm": 2502.911865234375,
-      "learning_rate": 4.9705709536163824e-05,
-      "loss": 148.008,
-      "step": 36000
-    },
-    {
-      "epoch": 0.14548495658884036,
-      "grad_norm": 569.8108520507812,
-      "learning_rate": 4.970517527402988e-05,
-      "loss": 70.3217,
-      "step": 36010
-    },
-    {
-      "epoch": 0.145525357854208,
-      "grad_norm": 635.7393188476562,
-      "learning_rate": 4.9704640530255826e-05,
-      "loss": 84.3072,
-      "step": 36020
-    },
-    {
-      "epoch": 0.14556575911957562,
-      "grad_norm": 365.2387390136719,
-      "learning_rate": 4.970410530485209e-05,
-      "loss": 60.3312,
-      "step": 36030
-    },
-    {
-      "epoch": 0.14560616038494326,
-      "grad_norm": 529.880126953125,
-      "learning_rate": 4.970356959782909e-05,
-      "loss": 125.5716,
-      "step": 36040
-    },
-    {
-      "epoch": 0.1456465616503109,
-      "grad_norm": 928.6548461914062,
-      "learning_rate": 4.97030334091973e-05,
-      "loss": 98.0715,
-      "step": 36050
-    },
-    {
-      "epoch": 0.1456869629156785,
-      "grad_norm": 937.4443359375,
-      "learning_rate": 4.970249673896714e-05,
-      "loss": 123.7811,
-      "step": 36060
-    },
-    {
-      "epoch": 0.14572736418104615,
-      "grad_norm": 664.5932006835938,
-      "learning_rate": 4.970195958714909e-05,
-      "loss": 103.4508,
-      "step": 36070
-    },
-    {
-      "epoch": 0.1457677654464138,
-      "grad_norm": 600.3163452148438,
-      "learning_rate": 4.970142195375363e-05,
-      "loss": 72.4148,
-      "step": 36080
-    },
-    {
-      "epoch": 0.1458081667117814,
-      "grad_norm": 602.3806762695312,
-      "learning_rate": 4.970088383879123e-05,
-      "loss": 103.1301,
-      "step": 36090
-    },
-    {
-      "epoch": 0.14584856797714904,
-      "grad_norm": 737.54443359375,
-      "learning_rate": 4.970034524227238e-05,
-      "loss": 80.4645,
-      "step": 36100
-    },
-    {
-      "epoch": 0.14588896924251668,
-      "grad_norm": 1445.3460693359375,
-      "learning_rate": 4.969980616420759e-05,
-      "loss": 75.5717,
-      "step": 36110
-    },
-    {
-      "epoch": 0.14592937050788432,
-      "grad_norm": 385.9198913574219,
-      "learning_rate": 4.9699266604607355e-05,
-      "loss": 137.0707,
-      "step": 36120
-    },
-    {
-      "epoch": 0.14596977177325193,
-      "grad_norm": 0.0,
-      "learning_rate": 4.96987265634822e-05,
-      "loss": 88.1988,
-      "step": 36130
-    },
-    {
-      "epoch": 0.14601017303861957,
-      "grad_norm": 793.6890869140625,
-      "learning_rate": 4.9698186040842654e-05,
-      "loss": 93.8744,
-      "step": 36140
-    },
-    {
-      "epoch": 0.1460505743039872,
-      "grad_norm": 1040.031494140625,
-      "learning_rate": 4.969764503669926e-05,
-      "loss": 67.5155,
-      "step": 36150
-    },
-    {
-      "epoch": 0.14609097556935483,
-      "grad_norm": 1021.9496459960938,
-      "learning_rate": 4.9697103551062556e-05,
-      "loss": 69.8323,
-      "step": 36160
-    },
-    {
-      "epoch": 0.14613137683472247,
-      "grad_norm": 1474.1693115234375,
-      "learning_rate": 4.9696561583943106e-05,
-      "loss": 100.1957,
-      "step": 36170
-    },
-    {
-      "epoch": 0.1461717781000901,
-      "grad_norm": 1472.6767578125,
-      "learning_rate": 4.969601913535148e-05,
-      "loss": 91.6,
-      "step": 36180
-    },
-    {
-      "epoch": 0.14621217936545772,
-      "grad_norm": 689.4755249023438,
-      "learning_rate": 4.9695476205298235e-05,
-      "loss": 146.9316,
-      "step": 36190
-    },
-    {
-      "epoch": 0.14625258063082536,
-      "grad_norm": 1301.0394287109375,
-      "learning_rate": 4.969493279379398e-05,
-      "loss": 75.4633,
-      "step": 36200
-    },
-    {
-      "epoch": 0.146292981896193,
-      "grad_norm": 708.2933349609375,
-      "learning_rate": 4.9694388900849284e-05,
-      "loss": 73.6771,
-      "step": 36210
-    },
-    {
-      "epoch": 0.1463333831615606,
-      "grad_norm": 0.0,
-      "learning_rate": 4.969384452647477e-05,
-      "loss": 81.5898,
-      "step": 36220
-    },
-    {
-      "epoch": 0.14637378442692825,
-      "grad_norm": 1018.5614013671875,
-      "learning_rate": 4.969329967068104e-05,
-      "loss": 98.2319,
-      "step": 36230
-    },
-    {
-      "epoch": 0.1464141856922959,
-      "grad_norm": 1839.9981689453125,
-      "learning_rate": 4.969275433347872e-05,
-      "loss": 88.1999,
-      "step": 36240
-    },
-    {
-      "epoch": 0.1464545869576635,
-      "grad_norm": 549.9813842773438,
-      "learning_rate": 4.9692208514878444e-05,
-      "loss": 79.6468,
-      "step": 36250
-    },
-    {
-      "epoch": 0.14649498822303114,
-      "grad_norm": 615.52099609375,
-      "learning_rate": 4.9691662214890856e-05,
-      "loss": 93.557,
-      "step": 36260
-    },
-    {
-      "epoch": 0.14653538948839878,
-      "grad_norm": 0.0,
-      "learning_rate": 4.969111543352659e-05,
-      "loss": 159.444,
-      "step": 36270
-    },
-    {
-      "epoch": 0.14657579075376642,
-      "grad_norm": 1021.6466674804688,
-      "learning_rate": 4.969056817079633e-05,
-      "loss": 93.9579,
-      "step": 36280
-    },
-    {
-      "epoch": 0.14661619201913403,
-      "grad_norm": 1050.8099365234375,
-      "learning_rate": 4.969002042671072e-05,
-      "loss": 92.0323,
-      "step": 36290
-    },
-    {
-      "epoch": 0.14665659328450167,
-      "grad_norm": 933.8421020507812,
-      "learning_rate": 4.968947220128045e-05,
-      "loss": 74.9989,
-      "step": 36300
-    },
-    {
-      "epoch": 0.14669699454986931,
-      "grad_norm": 775.214111328125,
-      "learning_rate": 4.968892349451621e-05,
-      "loss": 109.3062,
-      "step": 36310
-    },
-    {
-      "epoch": 0.14673739581523693,
-      "grad_norm": 275.0602111816406,
-      "learning_rate": 4.9688374306428696e-05,
-      "loss": 79.9519,
-      "step": 36320
-    },
-    {
-      "epoch": 0.14677779708060457,
-      "grad_norm": 1294.7794189453125,
-      "learning_rate": 4.9687824637028625e-05,
-      "loss": 113.0867,
-      "step": 36330
-    },
-    {
-      "epoch": 0.1468181983459722,
-      "grad_norm": 745.861572265625,
-      "learning_rate": 4.968727448632669e-05,
-      "loss": 76.9137,
-      "step": 36340
-    },
-    {
-      "epoch": 0.14685859961133982,
-      "grad_norm": 776.559326171875,
-      "learning_rate": 4.968672385433364e-05,
-      "loss": 86.1525,
-      "step": 36350
-    },
-    {
-      "epoch": 0.14689900087670746,
-      "grad_norm": 1309.9783935546875,
-      "learning_rate": 4.968617274106019e-05,
-      "loss": 79.5693,
-      "step": 36360
-    },
-    {
-      "epoch": 0.1469394021420751,
-      "grad_norm": 1547.9395751953125,
-      "learning_rate": 4.968562114651709e-05,
-      "loss": 116.9439,
-      "step": 36370
-    },
-    {
-      "epoch": 0.1469798034074427,
-      "grad_norm": 453.9913024902344,
-      "learning_rate": 4.9685069070715106e-05,
-      "loss": 74.6536,
-      "step": 36380
-    },
-    {
-      "epoch": 0.14702020467281035,
-      "grad_norm": 346.8396911621094,
-      "learning_rate": 4.968451651366498e-05,
-      "loss": 95.8461,
-      "step": 36390
-    },
-    {
-      "epoch": 0.147060605938178,
-      "grad_norm": 537.4244995117188,
-      "learning_rate": 4.968396347537751e-05,
-      "loss": 74.5145,
-      "step": 36400
-    },
-    {
-      "epoch": 0.1471010072035456,
-      "grad_norm": 897.6705932617188,
-      "learning_rate": 4.968340995586346e-05,
-      "loss": 109.4366,
-      "step": 36410
-    },
-    {
-      "epoch": 0.14714140846891324,
-      "grad_norm": 712.1412963867188,
-      "learning_rate": 4.9682855955133625e-05,
-      "loss": 67.6998,
-      "step": 36420
-    },
-    {
-      "epoch": 0.14718180973428088,
-      "grad_norm": 0.0,
-      "learning_rate": 4.96823014731988e-05,
-      "loss": 102.4061,
-      "step": 36430
-    },
-    {
-      "epoch": 0.14722221099964852,
-      "grad_norm": 1074.514404296875,
-      "learning_rate": 4.9681746510069805e-05,
-      "loss": 79.5955,
-      "step": 36440
-    },
-    {
-      "epoch": 0.14726261226501614,
-      "grad_norm": 1605.676025390625,
-      "learning_rate": 4.9681191065757455e-05,
-      "loss": 89.505,
-      "step": 36450
-    },
-    {
-      "epoch": 0.14730301353038378,
-      "grad_norm": 981.2547607421875,
-      "learning_rate": 4.9680635140272575e-05,
-      "loss": 99.4992,
-      "step": 36460
-    },
-    {
-      "epoch": 0.14734341479575142,
-      "grad_norm": 937.9696655273438,
-      "learning_rate": 4.9680078733626015e-05,
-      "loss": 75.9712,
-      "step": 36470
-    },
-    {
-      "epoch": 0.14738381606111903,
-      "grad_norm": 501.3292236328125,
-      "learning_rate": 4.9679521845828604e-05,
-      "loss": 75.7125,
-      "step": 36480
-    },
-    {
-      "epoch": 0.14742421732648667,
-      "grad_norm": 1447.997802734375,
-      "learning_rate": 4.967896447689121e-05,
-      "loss": 83.492,
-      "step": 36490
-    },
-    {
-      "epoch": 0.1474646185918543,
-      "grad_norm": 555.2802734375,
-      "learning_rate": 4.96784066268247e-05,
-      "loss": 63.0523,
-      "step": 36500
-    },
-    {
-      "epoch": 0.14750501985722192,
-      "grad_norm": 784.3593139648438,
-      "learning_rate": 4.967784829563995e-05,
-      "loss": 93.0994,
-      "step": 36510
-    },
-    {
-      "epoch": 0.14754542112258956,
-      "grad_norm": 1216.3717041015625,
-      "learning_rate": 4.967728948334784e-05,
-      "loss": 99.9734,
-      "step": 36520
-    },
-    {
-      "epoch": 0.1475858223879572,
-      "grad_norm": 717.1402587890625,
-      "learning_rate": 4.967673018995926e-05,
-      "loss": 80.4648,
-      "step": 36530
-    },
-    {
-      "epoch": 0.1476262236533248,
-      "grad_norm": 468.6872253417969,
-      "learning_rate": 4.967617041548513e-05,
-      "loss": 96.3743,
-      "step": 36540
-    },
-    {
-      "epoch": 0.14766662491869245,
-      "grad_norm": 709.4956665039062,
-      "learning_rate": 4.967561015993635e-05,
-      "loss": 96.8744,
-      "step": 36550
-    },
-    {
-      "epoch": 0.1477070261840601,
-      "grad_norm": 691.0542602539062,
-      "learning_rate": 4.967504942332385e-05,
-      "loss": 86.1126,
-      "step": 36560
-    },
-    {
-      "epoch": 0.1477474274494277,
-      "grad_norm": 1028.8739013671875,
-      "learning_rate": 4.967448820565856e-05,
-      "loss": 125.4938,
-      "step": 36570
-    },
-    {
-      "epoch": 0.14778782871479534,
-      "grad_norm": 599.9513549804688,
-      "learning_rate": 4.9673926506951404e-05,
-      "loss": 117.7221,
-      "step": 36580
-    },
-    {
-      "epoch": 0.14782822998016298,
-      "grad_norm": 609.7288818359375,
-      "learning_rate": 4.967336432721337e-05,
-      "loss": 80.1608,
-      "step": 36590
-    },
-    {
-      "epoch": 0.1478686312455306,
-      "grad_norm": 1172.2431640625,
-      "learning_rate": 4.967280166645538e-05,
-      "loss": 97.0124,
-      "step": 36600
-    },
-    {
-      "epoch": 0.14790903251089824,
-      "grad_norm": 774.5439453125,
-      "learning_rate": 4.967223852468842e-05,
-      "loss": 106.1079,
-      "step": 36610
-    },
-    {
-      "epoch": 0.14794943377626588,
-      "grad_norm": 414.7299499511719,
-      "learning_rate": 4.967167490192347e-05,
-      "loss": 95.8307,
-      "step": 36620
-    },
-    {
-      "epoch": 0.14798983504163352,
-      "grad_norm": 1053.8829345703125,
-      "learning_rate": 4.967111079817151e-05,
-      "loss": 102.4389,
-      "step": 36630
-    },
-    {
-      "epoch": 0.14803023630700113,
-      "grad_norm": 730.1495971679688,
-      "learning_rate": 4.967054621344356e-05,
-      "loss": 87.2563,
-      "step": 36640
-    },
-    {
-      "epoch": 0.14807063757236877,
-      "grad_norm": 2739.23583984375,
-      "learning_rate": 4.96699811477506e-05,
-      "loss": 126.1713,
-      "step": 36650
-    },
-    {
-      "epoch": 0.1481110388377364,
-      "grad_norm": 3278.61767578125,
-      "learning_rate": 4.966941560110366e-05,
-      "loss": 115.4281,
-      "step": 36660
-    },
-    {
-      "epoch": 0.14815144010310402,
-      "grad_norm": 1437.7108154296875,
-      "learning_rate": 4.966884957351375e-05,
-      "loss": 98.0438,
-      "step": 36670
-    },
-    {
-      "epoch": 0.14819184136847166,
-      "grad_norm": 582.1204833984375,
-      "learning_rate": 4.966828306499193e-05,
-      "loss": 85.8503,
-      "step": 36680
-    },
-    {
-      "epoch": 0.1482322426338393,
-      "grad_norm": 790.8889770507812,
-      "learning_rate": 4.966771607554923e-05,
-      "loss": 75.0409,
-      "step": 36690
-    },
-    {
-      "epoch": 0.1482726438992069,
-      "grad_norm": 370.1599426269531,
-      "learning_rate": 4.96671486051967e-05,
-      "loss": 94.7309,
-      "step": 36700
-    },
-    {
-      "epoch": 0.14831304516457455,
-      "grad_norm": 2078.58837890625,
-      "learning_rate": 4.966658065394542e-05,
-      "loss": 133.5383,
-      "step": 36710
-    },
-    {
-      "epoch": 0.1483534464299422,
-      "grad_norm": 374.68475341796875,
-      "learning_rate": 4.9666012221806434e-05,
-      "loss": 102.0996,
-      "step": 36720
-    },
-    {
-      "epoch": 0.1483938476953098,
-      "grad_norm": 1200.5596923828125,
-      "learning_rate": 4.966544330879085e-05,
-      "loss": 89.5257,
-      "step": 36730
-    },
-    {
-      "epoch": 0.14843424896067745,
-      "grad_norm": 430.6545104980469,
-      "learning_rate": 4.9664873914909755e-05,
-      "loss": 73.1253,
-      "step": 36740
-    },
-    {
-      "epoch": 0.14847465022604509,
-      "grad_norm": 1570.5799560546875,
-      "learning_rate": 4.966430404017424e-05,
-      "loss": 72.1421,
-      "step": 36750
-    },
-    {
-      "epoch": 0.1485150514914127,
-      "grad_norm": 422.2970886230469,
-      "learning_rate": 4.966373368459541e-05,
-      "loss": 147.6407,
-      "step": 36760
-    },
-    {
-      "epoch": 0.14855545275678034,
-      "grad_norm": 567.98046875,
-      "learning_rate": 4.966316284818441e-05,
-      "loss": 78.6361,
-      "step": 36770
-    },
-    {
-      "epoch": 0.14859585402214798,
-      "grad_norm": 0.0,
-      "learning_rate": 4.966259153095235e-05,
-      "loss": 72.8352,
-      "step": 36780
-    },
-    {
-      "epoch": 0.14863625528751562,
-      "grad_norm": 871.2714233398438,
-      "learning_rate": 4.966201973291036e-05,
-      "loss": 104.7993,
-      "step": 36790
-    },
-    {
-      "epoch": 0.14867665655288323,
-      "grad_norm": 762.0390625,
-      "learning_rate": 4.966144745406961e-05,
-      "loss": 63.3974,
-      "step": 36800
-    },
-    {
-      "epoch": 0.14871705781825087,
-      "grad_norm": 1256.638427734375,
-      "learning_rate": 4.966087469444124e-05,
-      "loss": 82.0591,
-      "step": 36810
-    },
-    {
-      "epoch": 0.1487574590836185,
-      "grad_norm": 1347.9171142578125,
-      "learning_rate": 4.966030145403642e-05,
-      "loss": 97.7492,
-      "step": 36820
-    },
-    {
-      "epoch": 0.14879786034898612,
-      "grad_norm": 2305.42236328125,
-      "learning_rate": 4.965972773286633e-05,
-      "loss": 98.8773,
-      "step": 36830
-    },
-    {
-      "epoch": 0.14883826161435376,
-      "grad_norm": 759.54931640625,
-      "learning_rate": 4.965915353094215e-05,
-      "loss": 104.5514,
-      "step": 36840
-    },
-    {
-      "epoch": 0.1488786628797214,
-      "grad_norm": 409.0998840332031,
-      "learning_rate": 4.9658578848275076e-05,
-      "loss": 72.8223,
-      "step": 36850
-    },
-    {
-      "epoch": 0.14891906414508901,
-      "grad_norm": 464.468017578125,
-      "learning_rate": 4.965800368487632e-05,
-      "loss": 112.4646,
-      "step": 36860
-    },
-    {
-      "epoch": 0.14895946541045665,
-      "grad_norm": 923.630859375,
-      "learning_rate": 4.9657428040757084e-05,
-      "loss": 107.7895,
-      "step": 36870
-    },
-    {
-      "epoch": 0.1489998666758243,
-      "grad_norm": 1179.0242919921875,
-      "learning_rate": 4.965685191592859e-05,
-      "loss": 86.0383,
-      "step": 36880
-    },
-    {
-      "epoch": 0.1490402679411919,
-      "grad_norm": 636.7333374023438,
-      "learning_rate": 4.9656275310402074e-05,
-      "loss": 114.6134,
-      "step": 36890
-    },
-    {
-      "epoch": 0.14908066920655955,
-      "grad_norm": 2465.29296875,
-      "learning_rate": 4.965569822418877e-05,
-      "loss": 58.8899,
-      "step": 36900
-    },
-    {
-      "epoch": 0.1491210704719272,
-      "grad_norm": 757.9246215820312,
-      "learning_rate": 4.9655120657299945e-05,
-      "loss": 66.7647,
-      "step": 36910
-    },
-    {
-      "epoch": 0.1491614717372948,
-      "grad_norm": 671.7369995117188,
-      "learning_rate": 4.965454260974685e-05,
-      "loss": 87.831,
-      "step": 36920
-    },
-    {
-      "epoch": 0.14920187300266244,
-      "grad_norm": 409.745849609375,
-      "learning_rate": 4.9653964081540756e-05,
-      "loss": 107.4229,
-      "step": 36930
-    },
-    {
-      "epoch": 0.14924227426803008,
-      "grad_norm": 1466.3577880859375,
-      "learning_rate": 4.965338507269294e-05,
-      "loss": 93.4886,
-      "step": 36940
-    },
-    {
-      "epoch": 0.14928267553339772,
-      "grad_norm": 554.5855712890625,
-      "learning_rate": 4.965280558321468e-05,
-      "loss": 97.7594,
-      "step": 36950
-    },
-    {
-      "epoch": 0.14932307679876533,
-      "grad_norm": 1820.7589111328125,
-      "learning_rate": 4.9652225613117284e-05,
-      "loss": 122.0696,
-      "step": 36960
-    },
-    {
-      "epoch": 0.14936347806413297,
-      "grad_norm": 1617.1182861328125,
-      "learning_rate": 4.965164516241206e-05,
-      "loss": 123.7657,
-      "step": 36970
-    },
-    {
-      "epoch": 0.1494038793295006,
-      "grad_norm": 669.3226928710938,
-      "learning_rate": 4.965106423111033e-05,
-      "loss": 103.5812,
-      "step": 36980
-    },
-    {
-      "epoch": 0.14944428059486822,
-      "grad_norm": 2139.23876953125,
-      "learning_rate": 4.9650482819223405e-05,
-      "loss": 119.0735,
-      "step": 36990
-    },
-    {
-      "epoch": 0.14948468186023586,
-      "grad_norm": 532.9093017578125,
-      "learning_rate": 4.964990092676263e-05,
-      "loss": 82.2448,
-      "step": 37000
-    },
-    {
-      "epoch": 0.1495250831256035,
-      "grad_norm": 804.912109375,
-      "learning_rate": 4.964931855373934e-05,
-      "loss": 80.909,
-      "step": 37010
-    },
-    {
-      "epoch": 0.14956548439097112,
-      "grad_norm": 847.0060424804688,
-      "learning_rate": 4.9648735700164895e-05,
-      "loss": 92.2945,
-      "step": 37020
-    },
-    {
-      "epoch": 0.14960588565633876,
-      "grad_norm": 1183.10302734375,
-      "learning_rate": 4.964815236605066e-05,
-      "loss": 94.6774,
-      "step": 37030
-    },
-    {
-      "epoch": 0.1496462869217064,
-      "grad_norm": 885.68896484375,
-      "learning_rate": 4.964756855140801e-05,
-      "loss": 75.244,
-      "step": 37040
-    },
-    {
-      "epoch": 0.149686688187074,
-      "grad_norm": 843.5673828125,
-      "learning_rate": 4.964698425624831e-05,
-      "loss": 82.0223,
-      "step": 37050
-    },
-    {
-      "epoch": 0.14972708945244165,
-      "grad_norm": 817.964599609375,
-      "learning_rate": 4.964639948058297e-05,
-      "loss": 131.6297,
-      "step": 37060
-    },
-    {
-      "epoch": 0.1497674907178093,
-      "grad_norm": 413.62335205078125,
-      "learning_rate": 4.964581422442338e-05,
-      "loss": 93.1363,
-      "step": 37070
-    },
-    {
-      "epoch": 0.1498078919831769,
-      "grad_norm": 569.75146484375,
-      "learning_rate": 4.964522848778096e-05,
-      "loss": 70.2936,
-      "step": 37080
-    },
-    {
-      "epoch": 0.14984829324854454,
-      "grad_norm": 507.4297180175781,
-      "learning_rate": 4.964464227066712e-05,
-      "loss": 53.4872,
-      "step": 37090
-    },
-    {
-      "epoch": 0.14988869451391218,
-      "grad_norm": 1582.9383544921875,
-      "learning_rate": 4.964405557309328e-05,
-      "loss": 128.3349,
-      "step": 37100
-    },
-    {
-      "epoch": 0.14992909577927982,
-      "grad_norm": 871.0317993164062,
-      "learning_rate": 4.9643468395070904e-05,
-      "loss": 106.9411,
-      "step": 37110
-    },
-    {
-      "epoch": 0.14996949704464743,
-      "grad_norm": 4390.7822265625,
-      "learning_rate": 4.964288073661142e-05,
-      "loss": 98.8651,
-      "step": 37120
-    },
-    {
-      "epoch": 0.15000989831001507,
-      "grad_norm": 1010.7752075195312,
-      "learning_rate": 4.9642292597726284e-05,
-      "loss": 134.204,
-      "step": 37130
-    },
-    {
-      "epoch": 0.1500502995753827,
-      "grad_norm": 3052.423583984375,
-      "learning_rate": 4.964170397842697e-05,
-      "loss": 129.5846,
-      "step": 37140
-    },
-    {
-      "epoch": 0.15009070084075032,
-      "grad_norm": 363.93865966796875,
-      "learning_rate": 4.9641114878724956e-05,
-      "loss": 101.122,
-      "step": 37150
-    },
-    {
-      "epoch": 0.15013110210611796,
-      "grad_norm": 963.4545288085938,
-      "learning_rate": 4.964052529863171e-05,
-      "loss": 98.0533,
-      "step": 37160
-    },
-    {
-      "epoch": 0.1501715033714856,
-      "grad_norm": 939.9786987304688,
-      "learning_rate": 4.9639935238158744e-05,
-      "loss": 89.1619,
-      "step": 37170
-    },
-    {
-      "epoch": 0.15021190463685322,
-      "grad_norm": 1083.3187255859375,
-      "learning_rate": 4.963934469731756e-05,
-      "loss": 77.3688,
-      "step": 37180
-    },
-    {
-      "epoch": 0.15025230590222086,
-      "grad_norm": 818.65478515625,
-      "learning_rate": 4.963875367611966e-05,
-      "loss": 96.2921,
-      "step": 37190
-    },
-    {
-      "epoch": 0.1502927071675885,
-      "grad_norm": 723.4608154296875,
-      "learning_rate": 4.963816217457657e-05,
-      "loss": 128.7513,
-      "step": 37200
-    },
-    {
-      "epoch": 0.1503331084329561,
-      "grad_norm": 737.654541015625,
-      "learning_rate": 4.963757019269983e-05,
-      "loss": 107.9872,
-      "step": 37210
-    },
-    {
-      "epoch": 0.15037350969832375,
-      "grad_norm": 771.156005859375,
-      "learning_rate": 4.963697773050097e-05,
-      "loss": 109.9101,
-      "step": 37220
-    },
-    {
-      "epoch": 0.1504139109636914,
-      "grad_norm": 748.536376953125,
-      "learning_rate": 4.9636384787991547e-05,
-      "loss": 65.0804,
-      "step": 37230
-    },
-    {
-      "epoch": 0.150454312229059,
-      "grad_norm": 430.6962890625,
-      "learning_rate": 4.963579136518312e-05,
-      "loss": 59.6868,
-      "step": 37240
-    },
-    {
-      "epoch": 0.15049471349442664,
-      "grad_norm": 703.9127807617188,
-      "learning_rate": 4.963519746208726e-05,
-      "loss": 111.1007,
-      "step": 37250
-    },
-    {
-      "epoch": 0.15053511475979428,
-      "grad_norm": 813.0443115234375,
-      "learning_rate": 4.963460307871553e-05,
-      "loss": 102.864,
-      "step": 37260
-    },
-    {
-      "epoch": 0.15057551602516192,
-      "grad_norm": 1400.834716796875,
-      "learning_rate": 4.963400821507954e-05,
-      "loss": 116.5066,
-      "step": 37270
-    },
-    {
-      "epoch": 0.15061591729052953,
-      "grad_norm": 941.7633666992188,
-      "learning_rate": 4.9633412871190873e-05,
-      "loss": 97.3467,
-      "step": 37280
-    },
-    {
-      "epoch": 0.15065631855589717,
-      "grad_norm": 895.1846923828125,
-      "learning_rate": 4.963281704706115e-05,
-      "loss": 88.1651,
-      "step": 37290
-    },
-    {
-      "epoch": 0.1506967198212648,
-      "grad_norm": 1722.0013427734375,
-      "learning_rate": 4.9632220742701965e-05,
-      "loss": 64.2686,
-      "step": 37300
-    },
-    {
-      "epoch": 0.15073712108663243,
-      "grad_norm": 499.2967224121094,
-      "learning_rate": 4.963162395812496e-05,
-      "loss": 57.7127,
-      "step": 37310
-    },
-    {
-      "epoch": 0.15077752235200007,
-      "grad_norm": 642.2814331054688,
-      "learning_rate": 4.9631026693341764e-05,
-      "loss": 47.7118,
-      "step": 37320
-    },
-    {
-      "epoch": 0.1508179236173677,
-      "grad_norm": 985.0557250976562,
-      "learning_rate": 4.963042894836403e-05,
-      "loss": 113.7733,
-      "step": 37330
-    },
-    {
-      "epoch": 0.15085832488273532,
-      "grad_norm": 558.4752197265625,
-      "learning_rate": 4.9629830723203384e-05,
-      "loss": 55.6255,
-      "step": 37340
-    },
-    {
-      "epoch": 0.15089872614810296,
-      "grad_norm": 366.1856689453125,
-      "learning_rate": 4.9629232017871524e-05,
-      "loss": 98.4935,
-      "step": 37350
-    },
-    {
-      "epoch": 0.1509391274134706,
-      "grad_norm": 817.6262817382812,
-      "learning_rate": 4.96286328323801e-05,
-      "loss": 69.1073,
-      "step": 37360
-    },
-    {
-      "epoch": 0.1509795286788382,
-      "grad_norm": 724.3482055664062,
-      "learning_rate": 4.96280331667408e-05,
-      "loss": 101.978,
-      "step": 37370
-    },
-    {
-      "epoch": 0.15101992994420585,
-      "grad_norm": 2478.86962890625,
-      "learning_rate": 4.9627433020965314e-05,
-      "loss": 104.1645,
-      "step": 37380
-    },
-    {
-      "epoch": 0.1510603312095735,
-      "grad_norm": 663.1130981445312,
-      "learning_rate": 4.962683239506534e-05,
-      "loss": 80.2843,
-      "step": 37390
-    },
-    {
-      "epoch": 0.1511007324749411,
-      "grad_norm": 757.1906127929688,
-      "learning_rate": 4.9626231289052596e-05,
-      "loss": 87.6918,
-      "step": 37400
-    },
-    {
-      "epoch": 0.15114113374030874,
-      "grad_norm": 427.3059997558594,
-      "learning_rate": 4.962562970293879e-05,
-      "loss": 79.3416,
-      "step": 37410
-    },
-    {
-      "epoch": 0.15118153500567638,
-      "grad_norm": 821.1558837890625,
-      "learning_rate": 4.962502763673565e-05,
-      "loss": 116.5157,
-      "step": 37420
-    },
-    {
-      "epoch": 0.15122193627104402,
-      "grad_norm": 1046.214111328125,
-      "learning_rate": 4.962442509045493e-05,
-      "loss": 53.0469,
-      "step": 37430
-    },
-    {
-      "epoch": 0.15126233753641163,
-      "grad_norm": 1114.373291015625,
-      "learning_rate": 4.9623822064108364e-05,
-      "loss": 75.2076,
-      "step": 37440
-    },
-    {
-      "epoch": 0.15130273880177927,
-      "grad_norm": 1147.0633544921875,
-      "learning_rate": 4.9623218557707694e-05,
-      "loss": 80.5338,
-      "step": 37450
-    },
-    {
-      "epoch": 0.15134314006714691,
-      "grad_norm": 1153.3177490234375,
-      "learning_rate": 4.9622614571264715e-05,
-      "loss": 115.0096,
-      "step": 37460
-    },
-    {
-      "epoch": 0.15138354133251453,
-      "grad_norm": 560.9694213867188,
-      "learning_rate": 4.962201010479119e-05,
-      "loss": 86.4352,
-      "step": 37470
-    },
-    {
-      "epoch": 0.15142394259788217,
-      "grad_norm": 584.9893188476562,
-      "learning_rate": 4.96214051582989e-05,
-      "loss": 63.2892,
-      "step": 37480
-    },
-    {
-      "epoch": 0.1514643438632498,
-      "grad_norm": 573.4566650390625,
-      "learning_rate": 4.962079973179963e-05,
-      "loss": 71.3577,
-      "step": 37490
-    },
-    {
-      "epoch": 0.15150474512861742,
-      "grad_norm": 696.969482421875,
-      "learning_rate": 4.962019382530521e-05,
-      "loss": 111.7284,
-      "step": 37500
-    },
-    {
-      "epoch": 0.15154514639398506,
-      "grad_norm": 674.3920288085938,
-      "learning_rate": 4.961958743882742e-05,
-      "loss": 82.3515,
-      "step": 37510
-    },
-    {
-      "epoch": 0.1515855476593527,
-      "grad_norm": 835.2359008789062,
-      "learning_rate": 4.96189805723781e-05,
-      "loss": 81.7163,
-      "step": 37520
-    },
-    {
-      "epoch": 0.1516259489247203,
-      "grad_norm": 1307.296142578125,
-      "learning_rate": 4.96183732259691e-05,
-      "loss": 92.0704,
-      "step": 37530
-    },
-    {
-      "epoch": 0.15166635019008795,
-      "grad_norm": 808.1261596679688,
-      "learning_rate": 4.961776539961222e-05,
-      "loss": 82.1407,
-      "step": 37540
-    },
-    {
-      "epoch": 0.1517067514554556,
-      "grad_norm": 1993.1292724609375,
-      "learning_rate": 4.9617157093319326e-05,
-      "loss": 85.8768,
-      "step": 37550
-    },
-    {
-      "epoch": 0.1517471527208232,
-      "grad_norm": 784.2838134765625,
-      "learning_rate": 4.961654830710229e-05,
-      "loss": 92.2036,
-      "step": 37560
-    },
-    {
-      "epoch": 0.15178755398619084,
-      "grad_norm": 412.1575622558594,
-      "learning_rate": 4.961593904097297e-05,
-      "loss": 82.7349,
-      "step": 37570
-    },
-    {
-      "epoch": 0.15182795525155848,
-      "grad_norm": 726.8958129882812,
-      "learning_rate": 4.961532929494325e-05,
-      "loss": 71.485,
-      "step": 37580
-    },
-    {
-      "epoch": 0.15186835651692612,
-      "grad_norm": 1616.0050048828125,
-      "learning_rate": 4.9614719069025e-05,
-      "loss": 65.669,
-      "step": 37590
-    },
-    {
-      "epoch": 0.15190875778229374,
-      "grad_norm": 1195.6494140625,
-      "learning_rate": 4.9614108363230135e-05,
-      "loss": 120.262,
-      "step": 37600
-    },
-    {
-      "epoch": 0.15194915904766138,
-      "grad_norm": 785.0847778320312,
-      "learning_rate": 4.961349717757056e-05,
-      "loss": 55.0058,
-      "step": 37610
-    },
-    {
-      "epoch": 0.15198956031302902,
-      "grad_norm": 367.303466796875,
-      "learning_rate": 4.961288551205818e-05,
-      "loss": 85.8025,
-      "step": 37620
-    },
-    {
-      "epoch": 0.15202996157839663,
-      "grad_norm": 1064.3392333984375,
-      "learning_rate": 4.961227336670493e-05,
-      "loss": 120.8695,
-      "step": 37630
-    },
-    {
-      "epoch": 0.15207036284376427,
-      "grad_norm": 963.6973876953125,
-      "learning_rate": 4.961166074152274e-05,
-      "loss": 119.0628,
-      "step": 37640
-    },
-    {
-      "epoch": 0.1521107641091319,
-      "grad_norm": 1672.767333984375,
-      "learning_rate": 4.961104763652355e-05,
-      "loss": 100.2525,
-      "step": 37650
-    },
-    {
-      "epoch": 0.15215116537449952,
-      "grad_norm": 850.4930419921875,
-      "learning_rate": 4.961043405171931e-05,
-      "loss": 63.3998,
-      "step": 37660
-    },
-    {
-      "epoch": 0.15219156663986716,
-      "grad_norm": 1047.08935546875,
-      "learning_rate": 4.9609819987122e-05,
-      "loss": 78.7605,
-      "step": 37670
-    },
-    {
-      "epoch": 0.1522319679052348,
-      "grad_norm": 689.37744140625,
-      "learning_rate": 4.9609205442743566e-05,
-      "loss": 100.5812,
-      "step": 37680
-    },
-    {
-      "epoch": 0.1522723691706024,
-      "grad_norm": 686.3281860351562,
-      "learning_rate": 4.9608590418596016e-05,
-      "loss": 58.7595,
-      "step": 37690
-    },
-    {
-      "epoch": 0.15231277043597005,
-      "grad_norm": 992.6712036132812,
-      "learning_rate": 4.9607974914691316e-05,
-      "loss": 117.9191,
-      "step": 37700
-    },
-    {
-      "epoch": 0.1523531717013377,
-      "grad_norm": 518.0152587890625,
-      "learning_rate": 4.960735893104148e-05,
-      "loss": 74.092,
-      "step": 37710
-    },
-    {
-      "epoch": 0.1523935729667053,
-      "grad_norm": 0.0,
-      "learning_rate": 4.960674246765851e-05,
-      "loss": 74.5977,
-      "step": 37720
-    },
-    {
-      "epoch": 0.15243397423207294,
-      "grad_norm": 524.2943115234375,
-      "learning_rate": 4.9606125524554434e-05,
-      "loss": 47.5876,
-      "step": 37730
-    },
-    {
-      "epoch": 0.15247437549744058,
-      "grad_norm": 1206.4256591796875,
-      "learning_rate": 4.960550810174126e-05,
-      "loss": 84.7391,
-      "step": 37740
-    },
-    {
-      "epoch": 0.15251477676280822,
-      "grad_norm": 762.7168579101562,
-      "learning_rate": 4.960489019923105e-05,
-      "loss": 67.8448,
-      "step": 37750
-    },
-    {
-      "epoch": 0.15255517802817584,
-      "grad_norm": 1341.15869140625,
-      "learning_rate": 4.9604271817035834e-05,
-      "loss": 112.8694,
-      "step": 37760
-    },
-    {
-      "epoch": 0.15259557929354348,
-      "grad_norm": 603.6417236328125,
-      "learning_rate": 4.960365295516767e-05,
-      "loss": 123.1844,
-      "step": 37770
-    },
-    {
-      "epoch": 0.15263598055891112,
-      "grad_norm": 868.97998046875,
-      "learning_rate": 4.9603033613638626e-05,
-      "loss": 75.7884,
-      "step": 37780
-    },
-    {
-      "epoch": 0.15267638182427873,
-      "grad_norm": 696.1629638671875,
-      "learning_rate": 4.9602413792460776e-05,
-      "loss": 82.7375,
-      "step": 37790
-    },
-    {
-      "epoch": 0.15271678308964637,
-      "grad_norm": 713.6837768554688,
-      "learning_rate": 4.960179349164621e-05,
-      "loss": 60.8608,
-      "step": 37800
-    },
-    {
-      "epoch": 0.152757184355014,
-      "grad_norm": 903.1224365234375,
-      "learning_rate": 4.9601172711207005e-05,
-      "loss": 111.0102,
-      "step": 37810
-    },
-    {
-      "epoch": 0.15279758562038162,
-      "grad_norm": 1082.6353759765625,
-      "learning_rate": 4.9600551451155274e-05,
-      "loss": 102.1816,
-      "step": 37820
-    },
-    {
-      "epoch": 0.15283798688574926,
-      "grad_norm": 1227.940673828125,
-      "learning_rate": 4.959992971150313e-05,
-      "loss": 98.8093,
-      "step": 37830
-    },
-    {
-      "epoch": 0.1528783881511169,
-      "grad_norm": 1349.7529296875,
-      "learning_rate": 4.959930749226269e-05,
-      "loss": 99.6163,
-      "step": 37840
-    },
-    {
-      "epoch": 0.1529187894164845,
-      "grad_norm": 1902.737548828125,
-      "learning_rate": 4.9598684793446085e-05,
-      "loss": 108.2246,
-      "step": 37850
-    },
-    {
-      "epoch": 0.15295919068185215,
-      "grad_norm": 888.9658203125,
-      "learning_rate": 4.959806161506545e-05,
-      "loss": 79.167,
-      "step": 37860
-    },
-    {
-      "epoch": 0.1529995919472198,
-      "grad_norm": 984.05322265625,
-      "learning_rate": 4.9597437957132955e-05,
-      "loss": 76.7845,
-      "step": 37870
-    },
-    {
-      "epoch": 0.1530399932125874,
-      "grad_norm": 849.2679443359375,
-      "learning_rate": 4.959681381966073e-05,
-      "loss": 122.3138,
-      "step": 37880
-    },
-    {
-      "epoch": 0.15308039447795505,
-      "grad_norm": 725.888916015625,
-      "learning_rate": 4.959618920266096e-05,
-      "loss": 73.6324,
-      "step": 37890
-    },
-    {
-      "epoch": 0.15312079574332269,
-      "grad_norm": 792.6710205078125,
-      "learning_rate": 4.959556410614582e-05,
-      "loss": 54.1202,
-      "step": 37900
-    },
-    {
-      "epoch": 0.15316119700869033,
-      "grad_norm": 583.5054931640625,
-      "learning_rate": 4.959493853012749e-05,
-      "loss": 81.214,
-      "step": 37910
-    },
-    {
-      "epoch": 0.15320159827405794,
-      "grad_norm": 793.6455688476562,
-      "learning_rate": 4.9594312474618175e-05,
-      "loss": 70.9016,
-      "step": 37920
-    },
-    {
-      "epoch": 0.15324199953942558,
-      "grad_norm": 1003.0465087890625,
-      "learning_rate": 4.959368593963007e-05,
-      "loss": 117.324,
-      "step": 37930
-    },
-    {
-      "epoch": 0.15328240080479322,
-      "grad_norm": 450.6336364746094,
-      "learning_rate": 4.9593058925175406e-05,
-      "loss": 95.6128,
-      "step": 37940
-    },
-    {
-      "epoch": 0.15332280207016083,
-      "grad_norm": 674.2539672851562,
-      "learning_rate": 4.959243143126639e-05,
-      "loss": 86.3896,
-      "step": 37950
-    },
-    {
-      "epoch": 0.15336320333552847,
-      "grad_norm": 396.0630798339844,
-      "learning_rate": 4.959180345791528e-05,
-      "loss": 80.5643,
-      "step": 37960
-    },
-    {
-      "epoch": 0.1534036046008961,
-      "grad_norm": 601.1489868164062,
-      "learning_rate": 4.9591175005134286e-05,
-      "loss": 105.7729,
-      "step": 37970
-    },
-    {
-      "epoch": 0.15344400586626372,
-      "grad_norm": 884.35693359375,
-      "learning_rate": 4.959054607293567e-05,
-      "loss": 100.5772,
-      "step": 37980
-    },
-    {
-      "epoch": 0.15348440713163136,
-      "grad_norm": 670.9600830078125,
-      "learning_rate": 4.95899166613317e-05,
-      "loss": 75.9151,
-      "step": 37990
-    },
-    {
-      "epoch": 0.153524808396999,
-      "grad_norm": 1574.9869384765625,
-      "learning_rate": 4.9589286770334654e-05,
-      "loss": 74.2453,
-      "step": 38000
-    },
-    {
-      "epoch": 0.15356520966236661,
-      "grad_norm": 1199.295654296875,
-      "learning_rate": 4.958865639995679e-05,
-      "loss": 89.8758,
-      "step": 38010
-    },
-    {
-      "epoch": 0.15360561092773425,
-      "grad_norm": 605.9471435546875,
-      "learning_rate": 4.958802555021042e-05,
-      "loss": 96.6403,
-      "step": 38020
-    },
-    {
-      "epoch": 0.1536460121931019,
-      "grad_norm": 1894.856201171875,
-      "learning_rate": 4.958739422110783e-05,
-      "loss": 87.7068,
-      "step": 38030
-    },
-    {
-      "epoch": 0.1536864134584695,
-      "grad_norm": 1081.8231201171875,
-      "learning_rate": 4.9586762412661333e-05,
-      "loss": 88.5522,
-      "step": 38040
-    },
-    {
-      "epoch": 0.15372681472383715,
-      "grad_norm": 452.8377685546875,
-      "learning_rate": 4.958613012488324e-05,
-      "loss": 75.0825,
-      "step": 38050
-    },
-    {
-      "epoch": 0.1537672159892048,
-      "grad_norm": 855.8710327148438,
-      "learning_rate": 4.958549735778589e-05,
-      "loss": 106.2082,
-      "step": 38060
-    },
-    {
-      "epoch": 0.15380761725457243,
-      "grad_norm": 0.0,
-      "learning_rate": 4.958486411138161e-05,
-      "loss": 50.9362,
-      "step": 38070
-    },
-    {
-      "epoch": 0.15384801851994004,
-      "grad_norm": 677.1991577148438,
-      "learning_rate": 4.958423038568274e-05,
-      "loss": 95.3129,
-      "step": 38080
-    },
-    {
-      "epoch": 0.15388841978530768,
-      "grad_norm": 470.16778564453125,
-      "learning_rate": 4.958359618070165e-05,
-      "loss": 92.6209,
-      "step": 38090
-    },
-    {
-      "epoch": 0.15392882105067532,
-      "grad_norm": 839.3685302734375,
-      "learning_rate": 4.958296149645069e-05,
-      "loss": 96.7531,
-      "step": 38100
-    },
-    {
-      "epoch": 0.15396922231604293,
-      "grad_norm": 1347.764892578125,
-      "learning_rate": 4.9582326332942244e-05,
-      "loss": 85.9113,
-      "step": 38110
-    },
-    {
-      "epoch": 0.15400962358141057,
-      "grad_norm": 452.386962890625,
-      "learning_rate": 4.958169069018869e-05,
-      "loss": 104.578,
-      "step": 38120
-    },
-    {
-      "epoch": 0.1540500248467782,
-      "grad_norm": 1746.528076171875,
-      "learning_rate": 4.958105456820242e-05,
-      "loss": 81.6607,
-      "step": 38130
-    },
-    {
-      "epoch": 0.15409042611214582,
-      "grad_norm": 862.51708984375,
-      "learning_rate": 4.958041796699583e-05,
-      "loss": 92.2541,
-      "step": 38140
-    },
-    {
-      "epoch": 0.15413082737751346,
-      "grad_norm": 1134.7611083984375,
-      "learning_rate": 4.957978088658134e-05,
-      "loss": 85.8542,
-      "step": 38150
-    },
-    {
-      "epoch": 0.1541712286428811,
-      "grad_norm": 737.6195068359375,
-      "learning_rate": 4.957914332697137e-05,
-      "loss": 84.3832,
-      "step": 38160
-    },
-    {
-      "epoch": 0.15421162990824872,
-      "grad_norm": 614.6725463867188,
-      "learning_rate": 4.957850528817834e-05,
-      "loss": 83.8153,
-      "step": 38170
-    },
-    {
-      "epoch": 0.15425203117361636,
-      "grad_norm": 848.0565795898438,
-      "learning_rate": 4.957786677021471e-05,
-      "loss": 72.8551,
-      "step": 38180
-    },
-    {
-      "epoch": 0.154292432438984,
-      "grad_norm": 1637.4676513671875,
-      "learning_rate": 4.9577227773092904e-05,
-      "loss": 69.5406,
-      "step": 38190
-    },
-    {
-      "epoch": 0.1543328337043516,
-      "grad_norm": 1447.67431640625,
-      "learning_rate": 4.9576588296825386e-05,
-      "loss": 60.8636,
-      "step": 38200
-    },
-    {
-      "epoch": 0.15437323496971925,
-      "grad_norm": 815.7072143554688,
-      "learning_rate": 4.9575948341424634e-05,
-      "loss": 86.5249,
-      "step": 38210
-    },
-    {
-      "epoch": 0.1544136362350869,
-      "grad_norm": 455.055419921875,
-      "learning_rate": 4.957530790690311e-05,
-      "loss": 87.8417,
-      "step": 38220
-    },
-    {
-      "epoch": 0.15445403750045453,
-      "grad_norm": 1008.4957275390625,
-      "learning_rate": 4.957466699327331e-05,
-      "loss": 111.3395,
-      "step": 38230
-    },
-    {
-      "epoch": 0.15449443876582214,
-      "grad_norm": 0.0,
-      "learning_rate": 4.957402560054773e-05,
-      "loss": 127.6305,
-      "step": 38240
-    },
-    {
-      "epoch": 0.15453484003118978,
-      "grad_norm": 2538.266357421875,
-      "learning_rate": 4.957338372873886e-05,
-      "loss": 126.8811,
-      "step": 38250
-    },
-    {
-      "epoch": 0.15457524129655742,
-      "grad_norm": 765.6140747070312,
-      "learning_rate": 4.957274137785922e-05,
-      "loss": 98.6367,
-      "step": 38260
-    },
-    {
-      "epoch": 0.15461564256192503,
-      "grad_norm": 1128.1064453125,
-      "learning_rate": 4.957209854792135e-05,
-      "loss": 115.5567,
-      "step": 38270
-    },
-    {
-      "epoch": 0.15465604382729267,
-      "grad_norm": 1656.940673828125,
-      "learning_rate": 4.957145523893776e-05,
-      "loss": 96.0133,
-      "step": 38280
-    },
-    {
-      "epoch": 0.1546964450926603,
-      "grad_norm": 776.5398559570312,
-      "learning_rate": 4.9570811450921e-05,
-      "loss": 96.1438,
-      "step": 38290
-    },
-    {
-      "epoch": 0.15473684635802792,
-      "grad_norm": 541.0315551757812,
-      "learning_rate": 4.957016718388362e-05,
-      "loss": 94.2679,
-      "step": 38300
-    },
-    {
-      "epoch": 0.15477724762339556,
-      "grad_norm": 1425.260498046875,
-      "learning_rate": 4.956952243783818e-05,
-      "loss": 85.9477,
-      "step": 38310
-    },
-    {
-      "epoch": 0.1548176488887632,
-      "grad_norm": 495.0509948730469,
-      "learning_rate": 4.956887721279726e-05,
-      "loss": 128.2712,
-      "step": 38320
-    },
-    {
-      "epoch": 0.15485805015413082,
-      "grad_norm": 1667.0662841796875,
-      "learning_rate": 4.956823150877342e-05,
-      "loss": 81.26,
-      "step": 38330
-    },
-    {
-      "epoch": 0.15489845141949846,
-      "grad_norm": 483.2498474121094,
-      "learning_rate": 4.956758532577926e-05,
-      "loss": 76.7887,
-      "step": 38340
-    },
-    {
-      "epoch": 0.1549388526848661,
-      "grad_norm": 940.1998901367188,
-      "learning_rate": 4.9566938663827377e-05,
-      "loss": 62.8369,
-      "step": 38350
-    },
-    {
-      "epoch": 0.1549792539502337,
-      "grad_norm": 826.6288452148438,
-      "learning_rate": 4.9566291522930375e-05,
-      "loss": 82.23,
-      "step": 38360
-    },
-    {
-      "epoch": 0.15501965521560135,
-      "grad_norm": 596.3945922851562,
-      "learning_rate": 4.956564390310088e-05,
-      "loss": 68.8034,
-      "step": 38370
-    },
-    {
-      "epoch": 0.155060056480969,
-      "grad_norm": 843.1058349609375,
-      "learning_rate": 4.95649958043515e-05,
-      "loss": 92.2614,
-      "step": 38380
-    },
-    {
-      "epoch": 0.15510045774633663,
-      "grad_norm": 368.0083923339844,
-      "learning_rate": 4.956434722669489e-05,
-      "loss": 80.4238,
-      "step": 38390
-    },
-    {
-      "epoch": 0.15514085901170424,
-      "grad_norm": 527.663330078125,
-      "learning_rate": 4.9563698170143666e-05,
-      "loss": 98.6083,
-      "step": 38400
-    },
-    {
-      "epoch": 0.15518126027707188,
-      "grad_norm": 847.8034057617188,
-      "learning_rate": 4.9563048634710516e-05,
-      "loss": 154.9312,
-      "step": 38410
-    },
-    {
-      "epoch": 0.15522166154243952,
-      "grad_norm": 872.8273315429688,
-      "learning_rate": 4.956239862040808e-05,
-      "loss": 89.4222,
-      "step": 38420
-    },
-    {
-      "epoch": 0.15526206280780713,
-      "grad_norm": 2756.822509765625,
-      "learning_rate": 4.956174812724904e-05,
-      "loss": 77.4832,
-      "step": 38430
-    },
-    {
-      "epoch": 0.15530246407317477,
-      "grad_norm": 405.55841064453125,
-      "learning_rate": 4.956109715524608e-05,
-      "loss": 104.6995,
-      "step": 38440
-    },
-    {
-      "epoch": 0.1553428653385424,
-      "grad_norm": 514.676513671875,
-      "learning_rate": 4.956044570441188e-05,
-      "loss": 112.402,
-      "step": 38450
-    },
-    {
-      "epoch": 0.15538326660391003,
-      "grad_norm": 867.9771118164062,
-      "learning_rate": 4.955979377475915e-05,
-      "loss": 72.6036,
-      "step": 38460
-    },
-    {
-      "epoch": 0.15542366786927767,
-      "grad_norm": 407.26898193359375,
-      "learning_rate": 4.9559141366300594e-05,
-      "loss": 89.936,
-      "step": 38470
-    },
-    {
-      "epoch": 0.1554640691346453,
-      "grad_norm": 906.7838745117188,
-      "learning_rate": 4.955848847904894e-05,
-      "loss": 111.0247,
-      "step": 38480
-    },
-    {
-      "epoch": 0.15550447040001292,
-      "grad_norm": 907.695068359375,
-      "learning_rate": 4.955783511301689e-05,
-      "loss": 87.2485,
-      "step": 38490
-    },
-    {
-      "epoch": 0.15554487166538056,
-      "grad_norm": 1016.2330932617188,
-      "learning_rate": 4.9557181268217227e-05,
-      "loss": 130.9673,
-      "step": 38500
-    },
-    {
-      "epoch": 0.1555852729307482,
-      "grad_norm": 1297.313232421875,
-      "learning_rate": 4.955652694466265e-05,
-      "loss": 104.8052,
-      "step": 38510
-    },
-    {
-      "epoch": 0.1556256741961158,
-      "grad_norm": 704.80126953125,
-      "learning_rate": 4.9555872142365945e-05,
-      "loss": 97.9365,
-      "step": 38520
-    },
-    {
-      "epoch": 0.15566607546148345,
-      "grad_norm": 1771.6796875,
-      "learning_rate": 4.9555216861339876e-05,
-      "loss": 74.8663,
-      "step": 38530
-    },
-    {
-      "epoch": 0.1557064767268511,
-      "grad_norm": 527.6610717773438,
-      "learning_rate": 4.9554561101597206e-05,
-      "loss": 99.9333,
-      "step": 38540
-    },
-    {
-      "epoch": 0.15574687799221873,
-      "grad_norm": 380.7573547363281,
-      "learning_rate": 4.955390486315073e-05,
-      "loss": 72.2091,
-      "step": 38550
-    },
-    {
-      "epoch": 0.15578727925758634,
-      "grad_norm": 1313.071044921875,
-      "learning_rate": 4.955324814601324e-05,
-      "loss": 84.0544,
-      "step": 38560
-    },
-    {
-      "epoch": 0.15582768052295398,
-      "grad_norm": 879.2240600585938,
-      "learning_rate": 4.955259095019753e-05,
-      "loss": 100.0556,
-      "step": 38570
-    },
-    {
-      "epoch": 0.15586808178832162,
-      "grad_norm": 1069.8026123046875,
-      "learning_rate": 4.955193327571642e-05,
-      "loss": 60.2459,
-      "step": 38580
-    },
-    {
-      "epoch": 0.15590848305368923,
-      "grad_norm": 1302.9310302734375,
-      "learning_rate": 4.955127512258273e-05,
-      "loss": 92.8039,
-      "step": 38590
-    },
-    {
-      "epoch": 0.15594888431905687,
-      "grad_norm": 1001.041259765625,
-      "learning_rate": 4.95506164908093e-05,
-      "loss": 110.6155,
-      "step": 38600
-    },
-    {
-      "epoch": 0.15598928558442451,
-      "grad_norm": 895.0310668945312,
-      "learning_rate": 4.954995738040895e-05,
-      "loss": 107.8179,
-      "step": 38610
-    },
-    {
-      "epoch": 0.15602968684979213,
-      "grad_norm": 823.3916625976562,
-      "learning_rate": 4.954929779139455e-05,
-      "loss": 78.2662,
-      "step": 38620
-    },
-    {
-      "epoch": 0.15607008811515977,
-      "grad_norm": 409.09527587890625,
-      "learning_rate": 4.954863772377894e-05,
-      "loss": 123.7662,
-      "step": 38630
-    },
-    {
-      "epoch": 0.1561104893805274,
-      "grad_norm": 569.7661743164062,
-      "learning_rate": 4.9547977177575014e-05,
-      "loss": 89.6015,
-      "step": 38640
-    },
-    {
-      "epoch": 0.15615089064589502,
-      "grad_norm": 616.7923583984375,
-      "learning_rate": 4.954731615279563e-05,
-      "loss": 94.4896,
-      "step": 38650
-    },
-    {
-      "epoch": 0.15619129191126266,
-      "grad_norm": 479.69647216796875,
-      "learning_rate": 4.9546654649453675e-05,
-      "loss": 66.053,
-      "step": 38660
-    },
-    {
-      "epoch": 0.1562316931766303,
-      "grad_norm": 2162.659912109375,
-      "learning_rate": 4.954599266756205e-05,
-      "loss": 138.2849,
-      "step": 38670
-    },
-    {
-      "epoch": 0.1562720944419979,
-      "grad_norm": 729.6651000976562,
-      "learning_rate": 4.9545330207133664e-05,
-      "loss": 107.9969,
-      "step": 38680
-    },
-    {
-      "epoch": 0.15631249570736555,
-      "grad_norm": 1174.257568359375,
-      "learning_rate": 4.9544667268181436e-05,
-      "loss": 94.6718,
-      "step": 38690
-    },
-    {
-      "epoch": 0.1563528969727332,
-      "grad_norm": 638.7692260742188,
-      "learning_rate": 4.9544003850718266e-05,
-      "loss": 100.2124,
-      "step": 38700
-    },
-    {
-      "epoch": 0.15639329823810083,
-      "grad_norm": 991.5917358398438,
-      "learning_rate": 4.954333995475712e-05,
-      "loss": 58.1896,
-      "step": 38710
-    },
-    {
-      "epoch": 0.15643369950346844,
-      "grad_norm": 2954.35302734375,
-      "learning_rate": 4.954267558031092e-05,
-      "loss": 65.0267,
-      "step": 38720
-    },
-    {
-      "epoch": 0.15647410076883608,
-      "grad_norm": 1842.9337158203125,
-      "learning_rate": 4.954201072739262e-05,
-      "loss": 92.5574,
-      "step": 38730
-    },
-    {
-      "epoch": 0.15651450203420372,
-      "grad_norm": 1125.166748046875,
-      "learning_rate": 4.9541345396015193e-05,
-      "loss": 89.0116,
-      "step": 38740
-    },
-    {
-      "epoch": 0.15655490329957134,
-      "grad_norm": 420.5506896972656,
-      "learning_rate": 4.9540679586191605e-05,
-      "loss": 91.9055,
-      "step": 38750
-    },
-    {
-      "epoch": 0.15659530456493898,
-      "grad_norm": 1417.18115234375,
-      "learning_rate": 4.9540013297934826e-05,
-      "loss": 72.1004,
-      "step": 38760
-    },
-    {
-      "epoch": 0.15663570583030662,
-      "grad_norm": 711.8256225585938,
-      "learning_rate": 4.953934653125786e-05,
-      "loss": 110.1404,
-      "step": 38770
-    },
-    {
-      "epoch": 0.15667610709567423,
-      "grad_norm": 936.8480834960938,
-      "learning_rate": 4.9538679286173696e-05,
-      "loss": 63.1972,
-      "step": 38780
-    },
-    {
-      "epoch": 0.15671650836104187,
-      "grad_norm": 965.4445190429688,
-      "learning_rate": 4.953801156269534e-05,
-      "loss": 96.635,
-      "step": 38790
-    },
-    {
-      "epoch": 0.1567569096264095,
-      "grad_norm": 388.2055969238281,
-      "learning_rate": 4.953734336083583e-05,
-      "loss": 69.8237,
-      "step": 38800
-    },
-    {
-      "epoch": 0.15679731089177712,
-      "grad_norm": 1032.505126953125,
-      "learning_rate": 4.953667468060816e-05,
-      "loss": 80.3915,
-      "step": 38810
-    },
-    {
-      "epoch": 0.15683771215714476,
-      "grad_norm": 2164.00341796875,
-      "learning_rate": 4.95360055220254e-05,
-      "loss": 117.7873,
-      "step": 38820
-    },
-    {
-      "epoch": 0.1568781134225124,
-      "grad_norm": 820.93310546875,
-      "learning_rate": 4.9535335885100575e-05,
-      "loss": 106.5048,
-      "step": 38830
-    },
-    {
-      "epoch": 0.15691851468788,
-      "grad_norm": 513.0703735351562,
-      "learning_rate": 4.953466576984675e-05,
-      "loss": 67.942,
-      "step": 38840
-    },
-    {
-      "epoch": 0.15695891595324765,
-      "grad_norm": 479.7278137207031,
-      "learning_rate": 4.953399517627698e-05,
-      "loss": 92.5827,
-      "step": 38850
-    },
-    {
-      "epoch": 0.1569993172186153,
-      "grad_norm": 3824.49951171875,
-      "learning_rate": 4.953332410440435e-05,
-      "loss": 76.3334,
-      "step": 38860
-    },
-    {
-      "epoch": 0.15703971848398293,
-      "grad_norm": 492.0223693847656,
-      "learning_rate": 4.953265255424192e-05,
-      "loss": 90.9314,
-      "step": 38870
-    },
-    {
-      "epoch": 0.15708011974935054,
-      "grad_norm": 487.1667175292969,
-      "learning_rate": 4.953198052580281e-05,
-      "loss": 83.6485,
-      "step": 38880
-    },
-    {
-      "epoch": 0.15712052101471818,
-      "grad_norm": 821.9052734375,
-      "learning_rate": 4.953130801910011e-05,
-      "loss": 66.1471,
-      "step": 38890
-    },
-    {
-      "epoch": 0.15716092228008582,
-      "grad_norm": 501.9080810546875,
-      "learning_rate": 4.953063503414692e-05,
-      "loss": 59.9543,
-      "step": 38900
-    },
-    {
-      "epoch": 0.15720132354545344,
-      "grad_norm": 1479.7249755859375,
-      "learning_rate": 4.9529961570956383e-05,
-      "loss": 83.3378,
-      "step": 38910
-    },
-    {
-      "epoch": 0.15724172481082108,
-      "grad_norm": 636.5172119140625,
-      "learning_rate": 4.952928762954161e-05,
-      "loss": 94.2658,
-      "step": 38920
-    },
-    {
-      "epoch": 0.15728212607618872,
-      "grad_norm": 1404.88134765625,
-      "learning_rate": 4.952861320991575e-05,
-      "loss": 86.6847,
-      "step": 38930
-    },
-    {
-      "epoch": 0.15732252734155633,
-      "grad_norm": 1033.673828125,
-      "learning_rate": 4.952793831209195e-05,
-      "loss": 109.079,
-      "step": 38940
-    },
-    {
-      "epoch": 0.15736292860692397,
-      "grad_norm": 990.5618896484375,
-      "learning_rate": 4.952726293608335e-05,
-      "loss": 80.7069,
-      "step": 38950
-    },
-    {
-      "epoch": 0.1574033298722916,
-      "grad_norm": 675.2631225585938,
-      "learning_rate": 4.9526587081903145e-05,
-      "loss": 81.1813,
-      "step": 38960
-    },
-    {
-      "epoch": 0.15744373113765922,
-      "grad_norm": 911.697509765625,
-      "learning_rate": 4.9525910749564494e-05,
-      "loss": 107.4708,
-      "step": 38970
-    },
-    {
-      "epoch": 0.15748413240302686,
-      "grad_norm": 506.453369140625,
-      "learning_rate": 4.952523393908059e-05,
-      "loss": 94.5604,
-      "step": 38980
-    },
-    {
-      "epoch": 0.1575245336683945,
-      "grad_norm": 1586.1180419921875,
-      "learning_rate": 4.9524556650464616e-05,
-      "loss": 97.9683,
-      "step": 38990
-    },
-    {
-      "epoch": 0.1575649349337621,
-      "grad_norm": 525.472900390625,
-      "learning_rate": 4.952387888372979e-05,
-      "loss": 126.3809,
-      "step": 39000
-    },
-    {
-      "epoch": 0.15760533619912975,
-      "grad_norm": 2915.437255859375,
-      "learning_rate": 4.952320063888932e-05,
-      "loss": 100.4383,
-      "step": 39010
-    },
-    {
-      "epoch": 0.1576457374644974,
-      "grad_norm": 656.8854370117188,
-      "learning_rate": 4.952252191595643e-05,
-      "loss": 41.7866,
-      "step": 39020
-    },
-    {
-      "epoch": 0.15768613872986503,
-      "grad_norm": 1504.2880859375,
-      "learning_rate": 4.9521842714944345e-05,
-      "loss": 95.3066,
-      "step": 39030
-    },
-    {
-      "epoch": 0.15772653999523265,
-      "grad_norm": 425.61871337890625,
-      "learning_rate": 4.952116303586631e-05,
-      "loss": 89.983,
-      "step": 39040
-    },
-    {
-      "epoch": 0.15776694126060029,
-      "grad_norm": 1021.040771484375,
-      "learning_rate": 4.952048287873558e-05,
-      "loss": 103.6552,
-      "step": 39050
-    },
-    {
-      "epoch": 0.15780734252596793,
-      "grad_norm": 770.2539672851562,
-      "learning_rate": 4.9519802243565414e-05,
-      "loss": 88.245,
-      "step": 39060
-    },
-    {
-      "epoch": 0.15784774379133554,
-      "grad_norm": 509.4902038574219,
-      "learning_rate": 4.951912113036908e-05,
-      "loss": 125.9709,
-      "step": 39070
-    },
-    {
-      "epoch": 0.15788814505670318,
-      "grad_norm": 577.9376831054688,
-      "learning_rate": 4.951843953915985e-05,
-      "loss": 56.5803,
-      "step": 39080
-    },
-    {
-      "epoch": 0.15792854632207082,
-      "grad_norm": 1800.318115234375,
-      "learning_rate": 4.951775746995102e-05,
-      "loss": 126.1482,
-      "step": 39090
-    },
-    {
-      "epoch": 0.15796894758743843,
-      "grad_norm": 644.34521484375,
-      "learning_rate": 4.951707492275589e-05,
-      "loss": 62.9948,
-      "step": 39100
-    },
-    {
-      "epoch": 0.15800934885280607,
-      "grad_norm": 2003.1749267578125,
-      "learning_rate": 4.9516391897587764e-05,
-      "loss": 83.1591,
-      "step": 39110
-    },
-    {
-      "epoch": 0.1580497501181737,
-      "grad_norm": 663.0642700195312,
-      "learning_rate": 4.951570839445995e-05,
-      "loss": 86.2755,
-      "step": 39120
-    },
-    {
-      "epoch": 0.15809015138354132,
-      "grad_norm": 2192.492919921875,
-      "learning_rate": 4.951502441338578e-05,
-      "loss": 69.6609,
-      "step": 39130
-    },
-    {
-      "epoch": 0.15813055264890896,
-      "grad_norm": 437.15411376953125,
-      "learning_rate": 4.951433995437859e-05,
-      "loss": 98.7007,
-      "step": 39140
-    },
-    {
-      "epoch": 0.1581709539142766,
-      "grad_norm": 643.024658203125,
-      "learning_rate": 4.951365501745172e-05,
-      "loss": 69.265,
-      "step": 39150
-    },
-    {
-      "epoch": 0.15821135517964421,
-      "grad_norm": 818.115234375,
-      "learning_rate": 4.951296960261853e-05,
-      "loss": 82.4836,
-      "step": 39160
-    },
-    {
-      "epoch": 0.15825175644501185,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9512283709892374e-05,
-      "loss": 64.5836,
-      "step": 39170
-    },
-    {
-      "epoch": 0.1582921577103795,
-      "grad_norm": 476.3262634277344,
-      "learning_rate": 4.951159733928663e-05,
-      "loss": 113.6856,
-      "step": 39180
-    },
-    {
-      "epoch": 0.15833255897574713,
-      "grad_norm": 617.4471435546875,
-      "learning_rate": 4.9510910490814666e-05,
-      "loss": 53.7248,
-      "step": 39190
-    },
-    {
-      "epoch": 0.15837296024111475,
-      "grad_norm": 1832.762939453125,
-      "learning_rate": 4.95102231644899e-05,
-      "loss": 80.7921,
-      "step": 39200
-    },
-    {
-      "epoch": 0.1584133615064824,
-      "grad_norm": 736.6915283203125,
-      "learning_rate": 4.95095353603257e-05,
-      "loss": 81.6178,
-      "step": 39210
-    },
-    {
-      "epoch": 0.15845376277185003,
-      "grad_norm": 881.178466796875,
-      "learning_rate": 4.9508847078335495e-05,
-      "loss": 78.0822,
-      "step": 39220
-    },
-    {
-      "epoch": 0.15849416403721764,
-      "grad_norm": 482.23944091796875,
-      "learning_rate": 4.9508158318532696e-05,
-      "loss": 83.8018,
-      "step": 39230
-    },
-    {
-      "epoch": 0.15853456530258528,
-      "grad_norm": 717.8798217773438,
-      "learning_rate": 4.9507469080930734e-05,
-      "loss": 69.0619,
-      "step": 39240
-    },
-    {
-      "epoch": 0.15857496656795292,
-      "grad_norm": 502.1875305175781,
-      "learning_rate": 4.9506779365543046e-05,
-      "loss": 53.7278,
-      "step": 39250
-    },
-    {
-      "epoch": 0.15861536783332053,
-      "grad_norm": 437.0827331542969,
-      "learning_rate": 4.950608917238308e-05,
-      "loss": 79.8096,
-      "step": 39260
-    },
-    {
-      "epoch": 0.15865576909868817,
-      "grad_norm": 1342.6405029296875,
-      "learning_rate": 4.9505398501464284e-05,
-      "loss": 78.4081,
-      "step": 39270
-    },
-    {
-      "epoch": 0.1586961703640558,
-      "grad_norm": 849.6737670898438,
-      "learning_rate": 4.9504707352800125e-05,
-      "loss": 79.9384,
-      "step": 39280
-    },
-    {
-      "epoch": 0.15873657162942342,
-      "grad_norm": 925.02392578125,
-      "learning_rate": 4.95040157264041e-05,
-      "loss": 59.0614,
-      "step": 39290
-    },
-    {
-      "epoch": 0.15877697289479106,
-      "grad_norm": 1974.9837646484375,
-      "learning_rate": 4.9503323622289655e-05,
-      "loss": 91.3175,
-      "step": 39300
-    },
-    {
-      "epoch": 0.1588173741601587,
-      "grad_norm": 1192.585693359375,
-      "learning_rate": 4.950263104047031e-05,
-      "loss": 62.4365,
-      "step": 39310
-    },
-    {
-      "epoch": 0.15885777542552632,
-      "grad_norm": 1465.087158203125,
-      "learning_rate": 4.9501937980959545e-05,
-      "loss": 75.6416,
-      "step": 39320
-    },
-    {
-      "epoch": 0.15889817669089396,
-      "grad_norm": 669.2102661132812,
-      "learning_rate": 4.950124444377089e-05,
-      "loss": 101.0025,
-      "step": 39330
-    },
-    {
-      "epoch": 0.1589385779562616,
-      "grad_norm": 1937.0089111328125,
-      "learning_rate": 4.950055042891786e-05,
-      "loss": 82.6265,
-      "step": 39340
-    },
-    {
-      "epoch": 0.15897897922162924,
-      "grad_norm": 1510.9599609375,
-      "learning_rate": 4.949985593641399e-05,
-      "loss": 117.1577,
-      "step": 39350
-    },
-    {
-      "epoch": 0.15901938048699685,
-      "grad_norm": 1171.20703125,
-      "learning_rate": 4.949916096627282e-05,
-      "loss": 106.9967,
-      "step": 39360
-    },
-    {
-      "epoch": 0.1590597817523645,
-      "grad_norm": 1075.4632568359375,
-      "learning_rate": 4.949846551850788e-05,
-      "loss": 110.2698,
-      "step": 39370
-    },
-    {
-      "epoch": 0.15910018301773213,
-      "grad_norm": 1426.9444580078125,
-      "learning_rate": 4.949776959313275e-05,
-      "loss": 68.2341,
-      "step": 39380
-    },
-    {
-      "epoch": 0.15914058428309974,
-      "grad_norm": 734.63525390625,
-      "learning_rate": 4.9497073190160994e-05,
-      "loss": 97.0607,
-      "step": 39390
-    },
-    {
-      "epoch": 0.15918098554846738,
-      "grad_norm": 682.6970825195312,
-      "learning_rate": 4.949637630960617e-05,
-      "loss": 83.067,
-      "step": 39400
-    },
-    {
-      "epoch": 0.15922138681383502,
-      "grad_norm": 239.862060546875,
-      "learning_rate": 4.9495678951481896e-05,
-      "loss": 93.8866,
-      "step": 39410
-    },
-    {
-      "epoch": 0.15926178807920263,
-      "grad_norm": 1805.31591796875,
-      "learning_rate": 4.949498111580174e-05,
-      "loss": 81.097,
-      "step": 39420
-    },
-    {
-      "epoch": 0.15930218934457027,
-      "grad_norm": 1094.3526611328125,
-      "learning_rate": 4.949428280257932e-05,
-      "loss": 122.2046,
-      "step": 39430
-    },
-    {
-      "epoch": 0.1593425906099379,
-      "grad_norm": 395.03936767578125,
-      "learning_rate": 4.949358401182824e-05,
-      "loss": 80.3976,
-      "step": 39440
-    },
-    {
-      "epoch": 0.15938299187530552,
-      "grad_norm": 896.0783081054688,
-      "learning_rate": 4.949288474356213e-05,
-      "loss": 100.0894,
-      "step": 39450
-    },
-    {
-      "epoch": 0.15942339314067316,
-      "grad_norm": 1049.211669921875,
-      "learning_rate": 4.9492184997794624e-05,
-      "loss": 115.2995,
-      "step": 39460
-    },
-    {
-      "epoch": 0.1594637944060408,
-      "grad_norm": 1161.3402099609375,
-      "learning_rate": 4.949148477453936e-05,
-      "loss": 81.019,
-      "step": 39470
-    },
-    {
-      "epoch": 0.15950419567140842,
-      "grad_norm": 2834.7724609375,
-      "learning_rate": 4.949078407381e-05,
-      "loss": 96.1574,
-      "step": 39480
-    },
-    {
-      "epoch": 0.15954459693677606,
-      "grad_norm": 808.9170532226562,
-      "learning_rate": 4.949008289562019e-05,
-      "loss": 79.7376,
-      "step": 39490
-    },
-    {
-      "epoch": 0.1595849982021437,
-      "grad_norm": 530.290771484375,
-      "learning_rate": 4.94893812399836e-05,
-      "loss": 69.7185,
-      "step": 39500
-    },
-    {
-      "epoch": 0.1596253994675113,
-      "grad_norm": 1201.916015625,
-      "learning_rate": 4.9488679106913924e-05,
-      "loss": 76.6368,
-      "step": 39510
-    },
-    {
-      "epoch": 0.15966580073287895,
-      "grad_norm": 1412.6275634765625,
-      "learning_rate": 4.948797649642484e-05,
-      "loss": 64.3421,
-      "step": 39520
-    },
-    {
-      "epoch": 0.1597062019982466,
-      "grad_norm": 927.9067993164062,
-      "learning_rate": 4.9487273408530044e-05,
-      "loss": 103.6948,
-      "step": 39530
-    },
-    {
-      "epoch": 0.15974660326361423,
-      "grad_norm": 722.8312377929688,
-      "learning_rate": 4.9486569843243244e-05,
-      "loss": 76.603,
-      "step": 39540
-    },
-    {
-      "epoch": 0.15978700452898184,
-      "grad_norm": 606.1805419921875,
-      "learning_rate": 4.948586580057816e-05,
-      "loss": 96.2999,
-      "step": 39550
-    },
-    {
-      "epoch": 0.15982740579434948,
-      "grad_norm": 787.3512573242188,
-      "learning_rate": 4.948516128054852e-05,
-      "loss": 91.6384,
-      "step": 39560
-    },
-    {
-      "epoch": 0.15986780705971712,
-      "grad_norm": 477.5809631347656,
-      "learning_rate": 4.948445628316805e-05,
-      "loss": 84.977,
-      "step": 39570
-    },
-    {
-      "epoch": 0.15990820832508473,
-      "grad_norm": 2748.46630859375,
-      "learning_rate": 4.94837508084505e-05,
-      "loss": 113.5827,
-      "step": 39580
-    },
-    {
-      "epoch": 0.15994860959045237,
-      "grad_norm": 1089.883544921875,
-      "learning_rate": 4.948304485640963e-05,
-      "loss": 110.2332,
-      "step": 39590
-    },
-    {
-      "epoch": 0.15998901085582,
-      "grad_norm": 2496.34326171875,
-      "learning_rate": 4.948233842705919e-05,
-      "loss": 62.8154,
-      "step": 39600
-    },
-    {
-      "epoch": 0.16002941212118763,
-      "grad_norm": 1067.39501953125,
-      "learning_rate": 4.948163152041295e-05,
-      "loss": 92.6578,
-      "step": 39610
-    },
-    {
-      "epoch": 0.16006981338655527,
-      "grad_norm": 401.0006408691406,
-      "learning_rate": 4.948092413648471e-05,
-      "loss": 94.2235,
-      "step": 39620
-    },
-    {
-      "epoch": 0.1601102146519229,
-      "grad_norm": 599.7701416015625,
-      "learning_rate": 4.948021627528825e-05,
-      "loss": 119.0192,
-      "step": 39630
-    },
-    {
-      "epoch": 0.16015061591729052,
-      "grad_norm": 5756.70068359375,
-      "learning_rate": 4.9479507936837364e-05,
-      "loss": 117.7867,
-      "step": 39640
-    },
-    {
-      "epoch": 0.16019101718265816,
-      "grad_norm": 1125.0499267578125,
-      "learning_rate": 4.947879912114588e-05,
-      "loss": 90.0595,
-      "step": 39650
-    },
-    {
-      "epoch": 0.1602314184480258,
-      "grad_norm": 5096.72216796875,
-      "learning_rate": 4.947808982822759e-05,
-      "loss": 109.1604,
-      "step": 39660
-    },
-    {
-      "epoch": 0.1602718197133934,
-      "grad_norm": 750.5372314453125,
-      "learning_rate": 4.9477380058096343e-05,
-      "loss": 75.6724,
-      "step": 39670
-    },
-    {
-      "epoch": 0.16031222097876105,
-      "grad_norm": 721.302734375,
-      "learning_rate": 4.947666981076597e-05,
-      "loss": 76.214,
-      "step": 39680
-    },
-    {
-      "epoch": 0.1603526222441287,
-      "grad_norm": 430.94671630859375,
-      "learning_rate": 4.947595908625032e-05,
-      "loss": 78.1926,
-      "step": 39690
-    },
-    {
-      "epoch": 0.16039302350949633,
-      "grad_norm": 871.6290893554688,
-      "learning_rate": 4.947524788456325e-05,
-      "loss": 56.1409,
-      "step": 39700
-    },
-    {
-      "epoch": 0.16043342477486394,
-      "grad_norm": 1097.3226318359375,
-      "learning_rate": 4.9474536205718615e-05,
-      "loss": 70.7656,
-      "step": 39710
-    },
-    {
-      "epoch": 0.16047382604023158,
-      "grad_norm": 821.6055908203125,
-      "learning_rate": 4.94738240497303e-05,
-      "loss": 116.6263,
-      "step": 39720
-    },
-    {
-      "epoch": 0.16051422730559922,
-      "grad_norm": 988.2919311523438,
-      "learning_rate": 4.947311141661218e-05,
-      "loss": 75.0355,
-      "step": 39730
-    },
-    {
-      "epoch": 0.16055462857096683,
-      "grad_norm": 1874.0450439453125,
-      "learning_rate": 4.947239830637815e-05,
-      "loss": 93.3599,
-      "step": 39740
-    },
-    {
-      "epoch": 0.16059502983633447,
-      "grad_norm": 1665.5904541015625,
-      "learning_rate": 4.947168471904213e-05,
-      "loss": 110.6867,
-      "step": 39750
-    },
-    {
-      "epoch": 0.16063543110170211,
-      "grad_norm": 768.0989990234375,
-      "learning_rate": 4.947097065461801e-05,
-      "loss": 74.4268,
-      "step": 39760
-    },
-    {
-      "epoch": 0.16067583236706973,
-      "grad_norm": 1051.240234375,
-      "learning_rate": 4.947025611311972e-05,
-      "loss": 114.7893,
-      "step": 39770
-    },
-    {
-      "epoch": 0.16071623363243737,
-      "grad_norm": 4345.51953125,
-      "learning_rate": 4.946954109456118e-05,
-      "loss": 108.7796,
-      "step": 39780
-    },
-    {
-      "epoch": 0.160756634897805,
-      "grad_norm": 1246.1304931640625,
-      "learning_rate": 4.946882559895635e-05,
-      "loss": 58.0561,
-      "step": 39790
-    },
-    {
-      "epoch": 0.16079703616317262,
-      "grad_norm": 3426.0712890625,
-      "learning_rate": 4.946810962631916e-05,
-      "loss": 86.2152,
-      "step": 39800
-    },
-    {
-      "epoch": 0.16083743742854026,
-      "grad_norm": 971.04052734375,
-      "learning_rate": 4.9467393176663576e-05,
-      "loss": 80.0207,
-      "step": 39810
-    },
-    {
-      "epoch": 0.1608778386939079,
-      "grad_norm": 791.5706787109375,
-      "learning_rate": 4.9466676250003576e-05,
-      "loss": 83.3073,
-      "step": 39820
-    },
-    {
-      "epoch": 0.1609182399592755,
-      "grad_norm": 1009.4277954101562,
-      "learning_rate": 4.9465958846353114e-05,
-      "loss": 81.5609,
-      "step": 39830
-    },
-    {
-      "epoch": 0.16095864122464315,
-      "grad_norm": 645.2051391601562,
-      "learning_rate": 4.9465240965726195e-05,
-      "loss": 110.7899,
-      "step": 39840
-    },
-    {
-      "epoch": 0.1609990424900108,
-      "grad_norm": 2333.77294921875,
-      "learning_rate": 4.9464522608136805e-05,
-      "loss": 67.2439,
-      "step": 39850
-    },
-    {
-      "epoch": 0.16103944375537843,
-      "grad_norm": 657.8132934570312,
-      "learning_rate": 4.946380377359895e-05,
-      "loss": 60.2396,
-      "step": 39860
-    },
-    {
-      "epoch": 0.16107984502074604,
-      "grad_norm": 1080.5504150390625,
-      "learning_rate": 4.9463084462126655e-05,
-      "loss": 101.5745,
-      "step": 39870
-    },
-    {
-      "epoch": 0.16112024628611368,
-      "grad_norm": 550.1509399414062,
-      "learning_rate": 4.946236467373392e-05,
-      "loss": 76.9606,
-      "step": 39880
-    },
-    {
-      "epoch": 0.16116064755148132,
-      "grad_norm": 686.2342529296875,
-      "learning_rate": 4.94616444084348e-05,
-      "loss": 87.1771,
-      "step": 39890
-    },
-    {
-      "epoch": 0.16120104881684894,
-      "grad_norm": 563.68017578125,
-      "learning_rate": 4.946092366624333e-05,
-      "loss": 90.5833,
-      "step": 39900
-    },
-    {
-      "epoch": 0.16124145008221658,
-      "grad_norm": 508.9067687988281,
-      "learning_rate": 4.946020244717355e-05,
-      "loss": 110.8761,
-      "step": 39910
-    },
-    {
-      "epoch": 0.16128185134758422,
-      "grad_norm": 447.7995910644531,
-      "learning_rate": 4.945948075123954e-05,
-      "loss": 84.4401,
-      "step": 39920
-    },
-    {
-      "epoch": 0.16132225261295183,
-      "grad_norm": 4074.44482421875,
-      "learning_rate": 4.9458758578455354e-05,
-      "loss": 130.2105,
-      "step": 39930
-    },
-    {
-      "epoch": 0.16136265387831947,
-      "grad_norm": 434.82470703125,
-      "learning_rate": 4.945803592883509e-05,
-      "loss": 76.4253,
-      "step": 39940
-    },
-    {
-      "epoch": 0.1614030551436871,
-      "grad_norm": 761.01123046875,
-      "learning_rate": 4.945731280239281e-05,
-      "loss": 106.6753,
-      "step": 39950
-    },
-    {
-      "epoch": 0.16144345640905472,
-      "grad_norm": 1182.587646484375,
-      "learning_rate": 4.9456589199142637e-05,
-      "loss": 101.2267,
-      "step": 39960
-    },
-    {
-      "epoch": 0.16148385767442236,
-      "grad_norm": 1377.45947265625,
-      "learning_rate": 4.945586511909865e-05,
-      "loss": 135.843,
-      "step": 39970
-    },
-    {
-      "epoch": 0.16152425893979,
-      "grad_norm": 1164.2666015625,
-      "learning_rate": 4.9455140562274995e-05,
-      "loss": 104.891,
-      "step": 39980
-    },
-    {
-      "epoch": 0.1615646602051576,
-      "grad_norm": 602.760498046875,
-      "learning_rate": 4.9454415528685785e-05,
-      "loss": 117.2372,
-      "step": 39990
-    },
-    {
-      "epoch": 0.16160506147052525,
-      "grad_norm": 1094.737548828125,
-      "learning_rate": 4.9453690018345144e-05,
-      "loss": 85.9165,
-      "step": 40000
-    },
-    {
-      "epoch": 0.1616454627358929,
-      "grad_norm": 599.16015625,
-      "learning_rate": 4.9452964031267236e-05,
-      "loss": 69.1269,
-      "step": 40010
-    },
-    {
-      "epoch": 0.16168586400126053,
-      "grad_norm": 613.2879028320312,
-      "learning_rate": 4.9452237567466194e-05,
-      "loss": 88.8299,
-      "step": 40020
-    },
-    {
-      "epoch": 0.16172626526662814,
-      "grad_norm": 735.5689086914062,
-      "learning_rate": 4.9451510626956196e-05,
-      "loss": 76.922,
-      "step": 40030
-    },
-    {
-      "epoch": 0.16176666653199578,
-      "grad_norm": 696.985595703125,
-      "learning_rate": 4.945078320975142e-05,
-      "loss": 79.6425,
-      "step": 40040
-    },
-    {
-      "epoch": 0.16180706779736342,
-      "grad_norm": 1030.8795166015625,
-      "learning_rate": 4.9450055315866026e-05,
-      "loss": 104.0844,
-      "step": 40050
-    },
-    {
-      "epoch": 0.16184746906273104,
-      "grad_norm": 3068.77294921875,
-      "learning_rate": 4.944932694531422e-05,
-      "loss": 85.7902,
-      "step": 40060
-    },
-    {
-      "epoch": 0.16188787032809868,
-      "grad_norm": 483.2486267089844,
-      "learning_rate": 4.94485980981102e-05,
-      "loss": 63.3755,
-      "step": 40070
-    },
-    {
-      "epoch": 0.16192827159346632,
-      "grad_norm": 2090.39208984375,
-      "learning_rate": 4.9447868774268166e-05,
-      "loss": 83.095,
-      "step": 40080
-    },
-    {
-      "epoch": 0.16196867285883393,
-      "grad_norm": 837.720947265625,
-      "learning_rate": 4.944713897380235e-05,
-      "loss": 68.0856,
-      "step": 40090
-    },
-    {
-      "epoch": 0.16200907412420157,
-      "grad_norm": 3449.74267578125,
-      "learning_rate": 4.9446408696726974e-05,
-      "loss": 80.644,
-      "step": 40100
-    },
-    {
-      "epoch": 0.1620494753895692,
-      "grad_norm": 532.7826538085938,
-      "learning_rate": 4.944567794305627e-05,
-      "loss": 72.2116,
-      "step": 40110
-    },
-    {
-      "epoch": 0.16208987665493682,
-      "grad_norm": 518.95556640625,
-      "learning_rate": 4.9444946712804494e-05,
-      "loss": 95.1311,
-      "step": 40120
-    },
-    {
-      "epoch": 0.16213027792030446,
-      "grad_norm": 2341.940673828125,
-      "learning_rate": 4.944421500598589e-05,
-      "loss": 101.21,
-      "step": 40130
-    },
-    {
-      "epoch": 0.1621706791856721,
-      "grad_norm": 694.1212158203125,
-      "learning_rate": 4.944348282261474e-05,
-      "loss": 124.4218,
-      "step": 40140
-    },
-    {
-      "epoch": 0.1622110804510397,
-      "grad_norm": 900.7708129882812,
-      "learning_rate": 4.9442750162705295e-05,
-      "loss": 89.3225,
-      "step": 40150
-    },
-    {
-      "epoch": 0.16225148171640735,
-      "grad_norm": 1264.3707275390625,
-      "learning_rate": 4.9442017026271864e-05,
-      "loss": 116.5199,
-      "step": 40160
-    },
-    {
-      "epoch": 0.162291882981775,
-      "grad_norm": 853.1660766601562,
-      "learning_rate": 4.944128341332872e-05,
-      "loss": 101.1801,
-      "step": 40170
-    },
-    {
-      "epoch": 0.16233228424714263,
-      "grad_norm": 696.7931518554688,
-      "learning_rate": 4.9440549323890176e-05,
-      "loss": 81.5634,
-      "step": 40180
-    },
-    {
-      "epoch": 0.16237268551251025,
-      "grad_norm": 1033.1129150390625,
-      "learning_rate": 4.9439814757970535e-05,
-      "loss": 79.6033,
-      "step": 40190
-    },
-    {
-      "epoch": 0.16241308677787789,
-      "grad_norm": 605.3699951171875,
-      "learning_rate": 4.9439079715584135e-05,
-      "loss": 107.7314,
-      "step": 40200
-    },
-    {
-      "epoch": 0.16245348804324553,
-      "grad_norm": 3663.65380859375,
-      "learning_rate": 4.943834419674529e-05,
-      "loss": 97.9108,
-      "step": 40210
-    },
-    {
-      "epoch": 0.16249388930861314,
-      "grad_norm": 504.4712829589844,
-      "learning_rate": 4.9437608201468336e-05,
-      "loss": 79.6529,
-      "step": 40220
-    },
-    {
-      "epoch": 0.16253429057398078,
-      "grad_norm": 759.597412109375,
-      "learning_rate": 4.9436871729767634e-05,
-      "loss": 65.1314,
-      "step": 40230
-    },
-    {
-      "epoch": 0.16257469183934842,
-      "grad_norm": 810.0010986328125,
-      "learning_rate": 4.943613478165753e-05,
-      "loss": 76.6986,
-      "step": 40240
-    },
-    {
-      "epoch": 0.16261509310471603,
-      "grad_norm": 701.2088623046875,
-      "learning_rate": 4.94353973571524e-05,
-      "loss": 59.842,
-      "step": 40250
-    },
-    {
-      "epoch": 0.16265549437008367,
-      "grad_norm": 460.05841064453125,
-      "learning_rate": 4.943465945626662e-05,
-      "loss": 74.4412,
-      "step": 40260
-    },
-    {
-      "epoch": 0.1626958956354513,
-      "grad_norm": 1123.1805419921875,
-      "learning_rate": 4.943392107901458e-05,
-      "loss": 141.1575,
-      "step": 40270
-    },
-    {
-      "epoch": 0.16273629690081892,
-      "grad_norm": 524.522705078125,
-      "learning_rate": 4.943318222541066e-05,
-      "loss": 117.3683,
-      "step": 40280
-    },
-    {
-      "epoch": 0.16277669816618656,
-      "grad_norm": 406.4377136230469,
-      "learning_rate": 4.943244289546928e-05,
-      "loss": 68.4648,
-      "step": 40290
-    },
-    {
-      "epoch": 0.1628170994315542,
-      "grad_norm": 752.1702880859375,
-      "learning_rate": 4.943170308920484e-05,
-      "loss": 74.2303,
-      "step": 40300
-    },
-    {
-      "epoch": 0.16285750069692181,
-      "grad_norm": 1674.6246337890625,
-      "learning_rate": 4.943096280663178e-05,
-      "loss": 123.305,
-      "step": 40310
-    },
-    {
-      "epoch": 0.16289790196228945,
-      "grad_norm": 1532.0059814453125,
-      "learning_rate": 4.9430222047764506e-05,
-      "loss": 82.7036,
-      "step": 40320
-    },
-    {
-      "epoch": 0.1629383032276571,
-      "grad_norm": 723.3096923828125,
-      "learning_rate": 4.942948081261749e-05,
-      "loss": 131.268,
-      "step": 40330
-    },
-    {
-      "epoch": 0.16297870449302473,
-      "grad_norm": 1614.462646484375,
-      "learning_rate": 4.942873910120516e-05,
-      "loss": 109.0338,
-      "step": 40340
-    },
-    {
-      "epoch": 0.16301910575839235,
-      "grad_norm": 720.302734375,
-      "learning_rate": 4.9427996913542e-05,
-      "loss": 63.5732,
-      "step": 40350
-    },
-    {
-      "epoch": 0.16305950702376,
-      "grad_norm": 1045.8563232421875,
-      "learning_rate": 4.9427254249642444e-05,
-      "loss": 96.6388,
-      "step": 40360
-    },
-    {
-      "epoch": 0.16309990828912763,
-      "grad_norm": 983.675048828125,
-      "learning_rate": 4.9426511109521e-05,
-      "loss": 66.8883,
-      "step": 40370
-    },
-    {
-      "epoch": 0.16314030955449524,
-      "grad_norm": 668.4130249023438,
-      "learning_rate": 4.9425767493192144e-05,
-      "loss": 107.8524,
-      "step": 40380
-    },
-    {
-      "epoch": 0.16318071081986288,
-      "grad_norm": 542.26318359375,
-      "learning_rate": 4.942502340067038e-05,
-      "loss": 64.179,
-      "step": 40390
-    },
-    {
-      "epoch": 0.16322111208523052,
-      "grad_norm": 931.4650268554688,
-      "learning_rate": 4.942427883197021e-05,
-      "loss": 120.4847,
-      "step": 40400
-    },
-    {
-      "epoch": 0.16326151335059813,
-      "grad_norm": 606.0546264648438,
-      "learning_rate": 4.942353378710614e-05,
-      "loss": 54.7097,
-      "step": 40410
-    },
-    {
-      "epoch": 0.16330191461596577,
-      "grad_norm": 1016.0202026367188,
-      "learning_rate": 4.9422788266092715e-05,
-      "loss": 108.11,
-      "step": 40420
-    },
-    {
-      "epoch": 0.1633423158813334,
-      "grad_norm": 1137.9029541015625,
-      "learning_rate": 4.942204226894445e-05,
-      "loss": 105.8649,
-      "step": 40430
-    },
-    {
-      "epoch": 0.16338271714670102,
-      "grad_norm": 1075.5977783203125,
-      "learning_rate": 4.94212957956759e-05,
-      "loss": 101.3763,
-      "step": 40440
-    },
-    {
-      "epoch": 0.16342311841206866,
-      "grad_norm": 674.2771606445312,
-      "learning_rate": 4.942054884630162e-05,
-      "loss": 81.1654,
-      "step": 40450
-    },
-    {
-      "epoch": 0.1634635196774363,
-      "grad_norm": 582.1860961914062,
-      "learning_rate": 4.941980142083617e-05,
-      "loss": 78.6935,
-      "step": 40460
-    },
-    {
-      "epoch": 0.16350392094280392,
-      "grad_norm": 688.1861572265625,
-      "learning_rate": 4.9419053519294115e-05,
-      "loss": 79.6108,
-      "step": 40470
-    },
-    {
-      "epoch": 0.16354432220817156,
-      "grad_norm": 1063.1949462890625,
-      "learning_rate": 4.941830514169004e-05,
-      "loss": 85.3794,
-      "step": 40480
-    },
-    {
-      "epoch": 0.1635847234735392,
-      "grad_norm": 802.294189453125,
-      "learning_rate": 4.941755628803853e-05,
-      "loss": 86.5771,
-      "step": 40490
-    },
-    {
-      "epoch": 0.16362512473890684,
-      "grad_norm": 2401.399169921875,
-      "learning_rate": 4.94168069583542e-05,
-      "loss": 123.1953,
-      "step": 40500
-    },
-    {
-      "epoch": 0.16366552600427445,
-      "grad_norm": 472.39874267578125,
-      "learning_rate": 4.941605715265164e-05,
-      "loss": 40.4517,
-      "step": 40510
-    },
-    {
-      "epoch": 0.1637059272696421,
-      "grad_norm": 502.11492919921875,
-      "learning_rate": 4.941530687094548e-05,
-      "loss": 93.8861,
-      "step": 40520
-    },
-    {
-      "epoch": 0.16374632853500973,
-      "grad_norm": 481.69085693359375,
-      "learning_rate": 4.9414556113250344e-05,
-      "loss": 100.2681,
-      "step": 40530
-    },
-    {
-      "epoch": 0.16378672980037734,
-      "grad_norm": 718.2989501953125,
-      "learning_rate": 4.941380487958086e-05,
-      "loss": 67.0155,
-      "step": 40540
-    },
-    {
-      "epoch": 0.16382713106574498,
-      "grad_norm": 707.5899658203125,
-      "learning_rate": 4.941305316995169e-05,
-      "loss": 80.2658,
-      "step": 40550
-    },
-    {
-      "epoch": 0.16386753233111262,
-      "grad_norm": 411.375244140625,
-      "learning_rate": 4.941230098437747e-05,
-      "loss": 86.6765,
-      "step": 40560
-    },
-    {
-      "epoch": 0.16390793359648023,
-      "grad_norm": 1048.686279296875,
-      "learning_rate": 4.941154832287288e-05,
-      "loss": 90.3216,
-      "step": 40570
-    },
-    {
-      "epoch": 0.16394833486184787,
-      "grad_norm": 951.8493041992188,
-      "learning_rate": 4.941079518545258e-05,
-      "loss": 64.0513,
-      "step": 40580
-    },
-    {
-      "epoch": 0.1639887361272155,
-      "grad_norm": 5584.05908203125,
-      "learning_rate": 4.9410041572131266e-05,
-      "loss": 109.0281,
-      "step": 40590
-    },
-    {
-      "epoch": 0.16402913739258312,
-      "grad_norm": 1825.2513427734375,
-      "learning_rate": 4.940928748292363e-05,
-      "loss": 83.6246,
-      "step": 40600
-    },
-    {
-      "epoch": 0.16406953865795076,
-      "grad_norm": 562.28076171875,
-      "learning_rate": 4.940853291784435e-05,
-      "loss": 72.4572,
-      "step": 40610
-    },
-    {
-      "epoch": 0.1641099399233184,
-      "grad_norm": 2325.668212890625,
-      "learning_rate": 4.9407777876908174e-05,
-      "loss": 104.7029,
-      "step": 40620
-    },
-    {
-      "epoch": 0.16415034118868602,
-      "grad_norm": 1462.803955078125,
-      "learning_rate": 4.9407022360129796e-05,
-      "loss": 113.1334,
-      "step": 40630
-    },
-    {
-      "epoch": 0.16419074245405366,
-      "grad_norm": 539.4075927734375,
-      "learning_rate": 4.9406266367523945e-05,
-      "loss": 87.7285,
-      "step": 40640
-    },
-    {
-      "epoch": 0.1642311437194213,
-      "grad_norm": 1116.6656494140625,
-      "learning_rate": 4.940550989910537e-05,
-      "loss": 127.9007,
-      "step": 40650
-    },
-    {
-      "epoch": 0.16427154498478894,
-      "grad_norm": 745.6444702148438,
-      "learning_rate": 4.9404752954888824e-05,
-      "loss": 95.8032,
-      "step": 40660
-    },
-    {
-      "epoch": 0.16431194625015655,
-      "grad_norm": 1984.2926025390625,
-      "learning_rate": 4.9403995534889044e-05,
-      "loss": 61.1674,
-      "step": 40670
-    },
-    {
-      "epoch": 0.1643523475155242,
-      "grad_norm": 553.41748046875,
-      "learning_rate": 4.9403237639120805e-05,
-      "loss": 87.0418,
-      "step": 40680
-    },
-    {
-      "epoch": 0.16439274878089183,
-      "grad_norm": 786.6685791015625,
-      "learning_rate": 4.9402479267598887e-05,
-      "loss": 93.9468,
-      "step": 40690
-    },
-    {
-      "epoch": 0.16443315004625944,
-      "grad_norm": 826.0059814453125,
-      "learning_rate": 4.940172042033808e-05,
-      "loss": 102.5501,
-      "step": 40700
-    },
-    {
-      "epoch": 0.16447355131162708,
-      "grad_norm": 1732.5660400390625,
-      "learning_rate": 4.9400961097353166e-05,
-      "loss": 110.0081,
-      "step": 40710
-    },
-    {
-      "epoch": 0.16451395257699472,
-      "grad_norm": 989.8864135742188,
-      "learning_rate": 4.940020129865895e-05,
-      "loss": 63.4497,
-      "step": 40720
-    },
-    {
-      "epoch": 0.16455435384236233,
-      "grad_norm": 484.3291931152344,
-      "learning_rate": 4.939944102427025e-05,
-      "loss": 111.996,
-      "step": 40730
-    },
-    {
-      "epoch": 0.16459475510772997,
-      "grad_norm": 218.7427520751953,
-      "learning_rate": 4.939868027420189e-05,
-      "loss": 83.2038,
-      "step": 40740
-    },
-    {
-      "epoch": 0.1646351563730976,
-      "grad_norm": 3197.0927734375,
-      "learning_rate": 4.939791904846869e-05,
-      "loss": 99.5802,
-      "step": 40750
-    },
-    {
-      "epoch": 0.16467555763846523,
-      "grad_norm": 805.2986450195312,
-      "learning_rate": 4.93971573470855e-05,
-      "loss": 85.1965,
-      "step": 40760
-    },
-    {
-      "epoch": 0.16471595890383287,
-      "grad_norm": 846.134765625,
-      "learning_rate": 4.939639517006717e-05,
-      "loss": 67.5336,
-      "step": 40770
-    },
-    {
-      "epoch": 0.1647563601692005,
-      "grad_norm": 780.6873779296875,
-      "learning_rate": 4.939563251742855e-05,
-      "loss": 85.52,
-      "step": 40780
-    },
-    {
-      "epoch": 0.16479676143456812,
-      "grad_norm": 1174.27685546875,
-      "learning_rate": 4.939486938918451e-05,
-      "loss": 97.5447,
-      "step": 40790
-    },
-    {
-      "epoch": 0.16483716269993576,
-      "grad_norm": 763.7103271484375,
-      "learning_rate": 4.9394105785349944e-05,
-      "loss": 50.4769,
-      "step": 40800
-    },
-    {
-      "epoch": 0.1648775639653034,
-      "grad_norm": 704.8574829101562,
-      "learning_rate": 4.939334170593972e-05,
-      "loss": 107.5855,
-      "step": 40810
-    },
-    {
-      "epoch": 0.16491796523067104,
-      "grad_norm": 695.6289672851562,
-      "learning_rate": 4.9392577150968745e-05,
-      "loss": 119.9477,
-      "step": 40820
-    },
-    {
-      "epoch": 0.16495836649603865,
-      "grad_norm": 960.3919677734375,
-      "learning_rate": 4.939181212045192e-05,
-      "loss": 75.6716,
-      "step": 40830
-    },
-    {
-      "epoch": 0.1649987677614063,
-      "grad_norm": 930.9579467773438,
-      "learning_rate": 4.939104661440415e-05,
-      "loss": 75.375,
-      "step": 40840
-    },
-    {
-      "epoch": 0.16503916902677393,
-      "grad_norm": 517.5274658203125,
-      "learning_rate": 4.939028063284038e-05,
-      "loss": 48.7727,
-      "step": 40850
-    },
-    {
-      "epoch": 0.16507957029214154,
-      "grad_norm": 559.23974609375,
-      "learning_rate": 4.938951417577552e-05,
-      "loss": 88.6447,
-      "step": 40860
-    },
-    {
-      "epoch": 0.16511997155750918,
-      "grad_norm": 1341.4295654296875,
-      "learning_rate": 4.938874724322454e-05,
-      "loss": 58.3951,
-      "step": 40870
-    },
-    {
-      "epoch": 0.16516037282287682,
-      "grad_norm": 818.9078369140625,
-      "learning_rate": 4.938797983520237e-05,
-      "loss": 90.6307,
-      "step": 40880
-    },
-    {
-      "epoch": 0.16520077408824443,
-      "grad_norm": 719.2857055664062,
-      "learning_rate": 4.938721195172398e-05,
-      "loss": 89.5419,
-      "step": 40890
-    },
-    {
-      "epoch": 0.16524117535361207,
-      "grad_norm": 1062.7843017578125,
-      "learning_rate": 4.938644359280433e-05,
-      "loss": 57.1088,
-      "step": 40900
-    },
-    {
-      "epoch": 0.16528157661897971,
-      "grad_norm": 713.6543579101562,
-      "learning_rate": 4.938567475845841e-05,
-      "loss": 98.1421,
-      "step": 40910
-    },
-    {
-      "epoch": 0.16532197788434733,
-      "grad_norm": 1030.1905517578125,
-      "learning_rate": 4.938490544870121e-05,
-      "loss": 115.5957,
-      "step": 40920
-    },
-    {
-      "epoch": 0.16536237914971497,
-      "grad_norm": 4530.09912109375,
-      "learning_rate": 4.938413566354772e-05,
-      "loss": 105.8919,
-      "step": 40930
-    },
-    {
-      "epoch": 0.1654027804150826,
-      "grad_norm": 1015.1611328125,
-      "learning_rate": 4.938336540301295e-05,
-      "loss": 130.0979,
-      "step": 40940
-    },
-    {
-      "epoch": 0.16544318168045022,
-      "grad_norm": 974.16064453125,
-      "learning_rate": 4.938259466711193e-05,
-      "loss": 61.6297,
-      "step": 40950
-    },
-    {
-      "epoch": 0.16548358294581786,
-      "grad_norm": 874.3550415039062,
-      "learning_rate": 4.938182345585966e-05,
-      "loss": 76.4001,
-      "step": 40960
-    },
-    {
-      "epoch": 0.1655239842111855,
-      "grad_norm": 2546.218017578125,
-      "learning_rate": 4.938105176927119e-05,
-      "loss": 78.0685,
-      "step": 40970
-    },
-    {
-      "epoch": 0.16556438547655314,
-      "grad_norm": 1073.3936767578125,
-      "learning_rate": 4.9380279607361575e-05,
-      "loss": 72.0243,
-      "step": 40980
-    },
-    {
-      "epoch": 0.16560478674192075,
-      "grad_norm": 1150.05029296875,
-      "learning_rate": 4.937950697014585e-05,
-      "loss": 76.5943,
-      "step": 40990
-    },
-    {
-      "epoch": 0.1656451880072884,
-      "grad_norm": 482.436279296875,
-      "learning_rate": 4.937873385763908e-05,
-      "loss": 75.9616,
-      "step": 41000
-    },
-    {
-      "epoch": 0.16568558927265603,
-      "grad_norm": 650.821044921875,
-      "learning_rate": 4.9377960269856346e-05,
-      "loss": 80.1524,
-      "step": 41010
-    },
-    {
-      "epoch": 0.16572599053802364,
-      "grad_norm": 486.8204040527344,
-      "learning_rate": 4.937718620681273e-05,
-      "loss": 73.0701,
-      "step": 41020
-    },
-    {
-      "epoch": 0.16576639180339128,
-      "grad_norm": 1333.5208740234375,
-      "learning_rate": 4.937641166852332e-05,
-      "loss": 73.1544,
-      "step": 41030
-    },
-    {
-      "epoch": 0.16580679306875892,
-      "grad_norm": 528.667236328125,
-      "learning_rate": 4.937563665500321e-05,
-      "loss": 89.639,
-      "step": 41040
-    },
-    {
-      "epoch": 0.16584719433412654,
-      "grad_norm": 924.5604858398438,
-      "learning_rate": 4.937486116626752e-05,
-      "loss": 122.8974,
-      "step": 41050
-    },
-    {
-      "epoch": 0.16588759559949418,
-      "grad_norm": 1802.9498291015625,
-      "learning_rate": 4.9374085202331354e-05,
-      "loss": 69.5255,
-      "step": 41060
-    },
-    {
-      "epoch": 0.16592799686486182,
-      "grad_norm": 519.0206909179688,
-      "learning_rate": 4.937330876320985e-05,
-      "loss": 89.9981,
-      "step": 41070
-    },
-    {
-      "epoch": 0.16596839813022943,
-      "grad_norm": 561.7305908203125,
-      "learning_rate": 4.9372531848918145e-05,
-      "loss": 76.0487,
-      "step": 41080
-    },
-    {
-      "epoch": 0.16600879939559707,
-      "grad_norm": 740.5826416015625,
-      "learning_rate": 4.9371754459471384e-05,
-      "loss": 115.638,
-      "step": 41090
-    },
-    {
-      "epoch": 0.1660492006609647,
-      "grad_norm": 674.3680419921875,
-      "learning_rate": 4.9370976594884723e-05,
-      "loss": 73.8446,
-      "step": 41100
-    },
-    {
-      "epoch": 0.16608960192633232,
-      "grad_norm": 539.9810180664062,
-      "learning_rate": 4.937019825517333e-05,
-      "loss": 57.1198,
-      "step": 41110
-    },
-    {
-      "epoch": 0.16613000319169996,
-      "grad_norm": 1040.4820556640625,
-      "learning_rate": 4.936941944035237e-05,
-      "loss": 144.8092,
-      "step": 41120
-    },
-    {
-      "epoch": 0.1661704044570676,
-      "grad_norm": 849.093994140625,
-      "learning_rate": 4.936864015043703e-05,
-      "loss": 52.7319,
-      "step": 41130
-    },
-    {
-      "epoch": 0.16621080572243524,
-      "grad_norm": 641.831787109375,
-      "learning_rate": 4.936786038544251e-05,
-      "loss": 59.0339,
-      "step": 41140
-    },
-    {
-      "epoch": 0.16625120698780285,
-      "grad_norm": 554.9229736328125,
-      "learning_rate": 4.9367080145384006e-05,
-      "loss": 95.6639,
-      "step": 41150
-    },
-    {
-      "epoch": 0.1662916082531705,
-      "grad_norm": 392.3255920410156,
-      "learning_rate": 4.936629943027672e-05,
-      "loss": 96.7075,
-      "step": 41160
-    },
-    {
-      "epoch": 0.16633200951853813,
-      "grad_norm": 448.32647705078125,
-      "learning_rate": 4.936551824013589e-05,
-      "loss": 65.1692,
-      "step": 41170
-    },
-    {
-      "epoch": 0.16637241078390574,
-      "grad_norm": 992.157470703125,
-      "learning_rate": 4.9364736574976736e-05,
-      "loss": 116.2756,
-      "step": 41180
-    },
-    {
-      "epoch": 0.16641281204927338,
-      "grad_norm": 383.3032531738281,
-      "learning_rate": 4.93639544348145e-05,
-      "loss": 82.3815,
-      "step": 41190
-    },
-    {
-      "epoch": 0.16645321331464102,
-      "grad_norm": 3468.877685546875,
-      "learning_rate": 4.9363171819664434e-05,
-      "loss": 124.7114,
-      "step": 41200
-    },
-    {
-      "epoch": 0.16649361458000864,
-      "grad_norm": 959.5894165039062,
-      "learning_rate": 4.936238872954178e-05,
-      "loss": 115.3659,
-      "step": 41210
-    },
-    {
-      "epoch": 0.16653401584537628,
-      "grad_norm": 654.4769287109375,
-      "learning_rate": 4.936160516446182e-05,
-      "loss": 84.5516,
-      "step": 41220
-    },
-    {
-      "epoch": 0.16657441711074392,
-      "grad_norm": 513.7091064453125,
-      "learning_rate": 4.936082112443983e-05,
-      "loss": 72.7509,
-      "step": 41230
-    },
-    {
-      "epoch": 0.16661481837611153,
-      "grad_norm": 1236.0504150390625,
-      "learning_rate": 4.936003660949108e-05,
-      "loss": 82.1154,
-      "step": 41240
-    },
-    {
-      "epoch": 0.16665521964147917,
-      "grad_norm": 1588.6417236328125,
-      "learning_rate": 4.9359251619630886e-05,
-      "loss": 76.5787,
-      "step": 41250
-    },
-    {
-      "epoch": 0.1666956209068468,
-      "grad_norm": 851.9100341796875,
-      "learning_rate": 4.935846615487453e-05,
-      "loss": 73.6734,
-      "step": 41260
-    },
-    {
-      "epoch": 0.16673602217221442,
-      "grad_norm": 439.6374816894531,
-      "learning_rate": 4.935768021523734e-05,
-      "loss": 75.6803,
-      "step": 41270
-    },
-    {
-      "epoch": 0.16677642343758206,
-      "grad_norm": 2391.953125,
-      "learning_rate": 4.935689380073464e-05,
-      "loss": 167.1796,
-      "step": 41280
-    },
-    {
-      "epoch": 0.1668168247029497,
-      "grad_norm": 591.0051879882812,
-      "learning_rate": 4.935610691138175e-05,
-      "loss": 89.2333,
-      "step": 41290
-    },
-    {
-      "epoch": 0.16685722596831734,
-      "grad_norm": 895.1334228515625,
-      "learning_rate": 4.9355319547194014e-05,
-      "loss": 110.6889,
-      "step": 41300
-    },
-    {
-      "epoch": 0.16689762723368495,
-      "grad_norm": 873.7748413085938,
-      "learning_rate": 4.935453170818679e-05,
-      "loss": 82.6634,
-      "step": 41310
-    },
-    {
-      "epoch": 0.1669380284990526,
-      "grad_norm": 582.4258422851562,
-      "learning_rate": 4.935374339437543e-05,
-      "loss": 82.0953,
-      "step": 41320
-    },
-    {
-      "epoch": 0.16697842976442023,
-      "grad_norm": 1372.591064453125,
-      "learning_rate": 4.9352954605775305e-05,
-      "loss": 110.3943,
-      "step": 41330
-    },
-    {
-      "epoch": 0.16701883102978785,
-      "grad_norm": 1086.363525390625,
-      "learning_rate": 4.935216534240179e-05,
-      "loss": 103.8,
-      "step": 41340
-    },
-    {
-      "epoch": 0.16705923229515549,
-      "grad_norm": 1426.1312255859375,
-      "learning_rate": 4.935137560427027e-05,
-      "loss": 99.1618,
-      "step": 41350
-    },
-    {
-      "epoch": 0.16709963356052313,
-      "grad_norm": 1669.37744140625,
-      "learning_rate": 4.935058539139615e-05,
-      "loss": 122.2307,
-      "step": 41360
-    },
-    {
-      "epoch": 0.16714003482589074,
-      "grad_norm": 766.4739379882812,
-      "learning_rate": 4.934979470379484e-05,
-      "loss": 77.6077,
-      "step": 41370
-    },
-    {
-      "epoch": 0.16718043609125838,
-      "grad_norm": 846.854248046875,
-      "learning_rate": 4.934900354148173e-05,
-      "loss": 87.1017,
-      "step": 41380
-    },
-    {
-      "epoch": 0.16722083735662602,
-      "grad_norm": 1178.765869140625,
-      "learning_rate": 4.934821190447228e-05,
-      "loss": 118.6451,
-      "step": 41390
-    },
-    {
-      "epoch": 0.16726123862199363,
-      "grad_norm": 622.7171020507812,
-      "learning_rate": 4.9347419792781876e-05,
-      "loss": 92.5647,
-      "step": 41400
-    },
-    {
-      "epoch": 0.16730163988736127,
-      "grad_norm": 871.9891967773438,
-      "learning_rate": 4.934662720642601e-05,
-      "loss": 78.3886,
-      "step": 41410
-    },
-    {
-      "epoch": 0.1673420411527289,
-      "grad_norm": 601.3154296875,
-      "learning_rate": 4.934583414542011e-05,
-      "loss": 69.2431,
-      "step": 41420
-    },
-    {
-      "epoch": 0.16738244241809652,
-      "grad_norm": 620.5543823242188,
-      "learning_rate": 4.9345040609779634e-05,
-      "loss": 74.3392,
-      "step": 41430
-    },
-    {
-      "epoch": 0.16742284368346416,
-      "grad_norm": 803.2666625976562,
-      "learning_rate": 4.934424659952006e-05,
-      "loss": 90.4268,
-      "step": 41440
-    },
-    {
-      "epoch": 0.1674632449488318,
-      "grad_norm": 1084.6322021484375,
-      "learning_rate": 4.934345211465686e-05,
-      "loss": 91.4339,
-      "step": 41450
-    },
-    {
-      "epoch": 0.16750364621419944,
-      "grad_norm": 570.4534912109375,
-      "learning_rate": 4.934265715520553e-05,
-      "loss": 67.0539,
-      "step": 41460
-    },
-    {
-      "epoch": 0.16754404747956705,
-      "grad_norm": 5001.0390625,
-      "learning_rate": 4.934186172118157e-05,
-      "loss": 115.2946,
-      "step": 41470
-    },
-    {
-      "epoch": 0.1675844487449347,
-      "grad_norm": 0.0,
-      "learning_rate": 4.934106581260049e-05,
-      "loss": 71.3725,
-      "step": 41480
-    },
-    {
-      "epoch": 0.16762485001030233,
-      "grad_norm": 674.1163940429688,
-      "learning_rate": 4.934026942947779e-05,
-      "loss": 67.2382,
-      "step": 41490
-    },
-    {
-      "epoch": 0.16766525127566995,
-      "grad_norm": 1031.8499755859375,
-      "learning_rate": 4.933947257182901e-05,
-      "loss": 81.7795,
-      "step": 41500
-    },
-    {
-      "epoch": 0.1677056525410376,
-      "grad_norm": 1388.8055419921875,
-      "learning_rate": 4.933867523966968e-05,
-      "loss": 67.2926,
-      "step": 41510
-    },
-    {
-      "epoch": 0.16774605380640523,
-      "grad_norm": 1057.6483154296875,
-      "learning_rate": 4.933787743301534e-05,
-      "loss": 78.5172,
-      "step": 41520
-    },
-    {
-      "epoch": 0.16778645507177284,
-      "grad_norm": 1191.138427734375,
-      "learning_rate": 4.933707915188156e-05,
-      "loss": 115.7021,
-      "step": 41530
-    },
-    {
-      "epoch": 0.16782685633714048,
-      "grad_norm": 803.36376953125,
-      "learning_rate": 4.933628039628389e-05,
-      "loss": 80.2134,
-      "step": 41540
-    },
-    {
-      "epoch": 0.16786725760250812,
-      "grad_norm": 434.1454772949219,
-      "learning_rate": 4.9335481166237904e-05,
-      "loss": 93.0307,
-      "step": 41550
-    },
-    {
-      "epoch": 0.16790765886787573,
-      "grad_norm": 1040.9764404296875,
-      "learning_rate": 4.933468146175918e-05,
-      "loss": 108.4676,
-      "step": 41560
-    },
-    {
-      "epoch": 0.16794806013324337,
-      "grad_norm": 2650.668701171875,
-      "learning_rate": 4.933388128286331e-05,
-      "loss": 124.0231,
-      "step": 41570
-    },
-    {
-      "epoch": 0.167988461398611,
-      "grad_norm": 850.1209716796875,
-      "learning_rate": 4.933308062956591e-05,
-      "loss": 98.8213,
-      "step": 41580
-    },
-    {
-      "epoch": 0.16802886266397862,
-      "grad_norm": 1043.2742919921875,
-      "learning_rate": 4.9332279501882564e-05,
-      "loss": 86.2264,
-      "step": 41590
-    },
-    {
-      "epoch": 0.16806926392934626,
-      "grad_norm": 1196.578369140625,
-      "learning_rate": 4.93314778998289e-05,
-      "loss": 80.5056,
-      "step": 41600
-    },
-    {
-      "epoch": 0.1681096651947139,
-      "grad_norm": 738.3115234375,
-      "learning_rate": 4.933067582342056e-05,
-      "loss": 111.9835,
-      "step": 41610
-    },
-    {
-      "epoch": 0.16815006646008154,
-      "grad_norm": 828.2302856445312,
-      "learning_rate": 4.932987327267316e-05,
-      "loss": 87.0005,
-      "step": 41620
-    },
-    {
-      "epoch": 0.16819046772544916,
-      "grad_norm": 474.343017578125,
-      "learning_rate": 4.932907024760236e-05,
-      "loss": 93.0332,
-      "step": 41630
-    },
-    {
-      "epoch": 0.1682308689908168,
-      "grad_norm": 718.5823364257812,
-      "learning_rate": 4.93282667482238e-05,
-      "loss": 66.0602,
-      "step": 41640
-    },
-    {
-      "epoch": 0.16827127025618444,
-      "grad_norm": 491.5641784667969,
-      "learning_rate": 4.9327462774553166e-05,
-      "loss": 93.1942,
-      "step": 41650
-    },
-    {
-      "epoch": 0.16831167152155205,
-      "grad_norm": 244.37391662597656,
-      "learning_rate": 4.9326658326606114e-05,
-      "loss": 80.5441,
-      "step": 41660
-    },
-    {
-      "epoch": 0.1683520727869197,
-      "grad_norm": 621.1357421875,
-      "learning_rate": 4.9325853404398337e-05,
-      "loss": 107.9278,
-      "step": 41670
-    },
-    {
-      "epoch": 0.16839247405228733,
-      "grad_norm": 1189.0419921875,
-      "learning_rate": 4.9325048007945526e-05,
-      "loss": 98.2764,
-      "step": 41680
-    },
-    {
-      "epoch": 0.16843287531765494,
-      "grad_norm": 1160.5760498046875,
-      "learning_rate": 4.9324242137263376e-05,
-      "loss": 85.1483,
-      "step": 41690
-    },
-    {
-      "epoch": 0.16847327658302258,
-      "grad_norm": 717.5256958007812,
-      "learning_rate": 4.93234357923676e-05,
-      "loss": 68.967,
-      "step": 41700
-    },
-    {
-      "epoch": 0.16851367784839022,
-      "grad_norm": 738.8279418945312,
-      "learning_rate": 4.932262897327393e-05,
-      "loss": 94.0461,
-      "step": 41710
-    },
-    {
-      "epoch": 0.16855407911375783,
-      "grad_norm": 372.3353576660156,
-      "learning_rate": 4.9321821679998074e-05,
-      "loss": 107.4272,
-      "step": 41720
-    },
-    {
-      "epoch": 0.16859448037912547,
-      "grad_norm": 778.2406616210938,
-      "learning_rate": 4.932101391255579e-05,
-      "loss": 127.5508,
-      "step": 41730
-    },
-    {
-      "epoch": 0.1686348816444931,
-      "grad_norm": 296.44610595703125,
-      "learning_rate": 4.9320205670962814e-05,
-      "loss": 68.8699,
-      "step": 41740
-    },
-    {
-      "epoch": 0.16867528290986072,
-      "grad_norm": 1934.7763671875,
-      "learning_rate": 4.931939695523492e-05,
-      "loss": 96.5082,
-      "step": 41750
-    },
-    {
-      "epoch": 0.16871568417522836,
-      "grad_norm": 920.8310546875,
-      "learning_rate": 4.9318587765387845e-05,
-      "loss": 84.8141,
-      "step": 41760
-    },
-    {
-      "epoch": 0.168756085440596,
-      "grad_norm": 785.7876586914062,
-      "learning_rate": 4.93177781014374e-05,
-      "loss": 69.8683,
-      "step": 41770
-    },
-    {
-      "epoch": 0.16879648670596364,
-      "grad_norm": 747.9173583984375,
-      "learning_rate": 4.9316967963399335e-05,
-      "loss": 87.638,
-      "step": 41780
-    },
-    {
-      "epoch": 0.16883688797133126,
-      "grad_norm": 1194.38037109375,
-      "learning_rate": 4.931615735128947e-05,
-      "loss": 66.9209,
-      "step": 41790
-    },
-    {
-      "epoch": 0.1688772892366989,
-      "grad_norm": 944.1591796875,
-      "learning_rate": 4.9315346265123594e-05,
-      "loss": 80.2107,
-      "step": 41800
-    },
-    {
-      "epoch": 0.16891769050206654,
-      "grad_norm": 1037.114013671875,
-      "learning_rate": 4.9314534704917525e-05,
-      "loss": 82.2505,
-      "step": 41810
-    },
-    {
-      "epoch": 0.16895809176743415,
-      "grad_norm": 723.7459716796875,
-      "learning_rate": 4.931372267068708e-05,
-      "loss": 65.5946,
-      "step": 41820
-    },
-    {
-      "epoch": 0.1689984930328018,
-      "grad_norm": 800.6956176757812,
-      "learning_rate": 4.93129101624481e-05,
-      "loss": 88.3285,
-      "step": 41830
-    },
-    {
-      "epoch": 0.16903889429816943,
-      "grad_norm": 634.937744140625,
-      "learning_rate": 4.9312097180216414e-05,
-      "loss": 85.8466,
-      "step": 41840
-    },
-    {
-      "epoch": 0.16907929556353704,
-      "grad_norm": 438.7951965332031,
-      "learning_rate": 4.9311283724007887e-05,
-      "loss": 59.9148,
-      "step": 41850
-    },
-    {
-      "epoch": 0.16911969682890468,
-      "grad_norm": 6580.62451171875,
-      "learning_rate": 4.931046979383835e-05,
-      "loss": 127.95,
-      "step": 41860
-    },
-    {
-      "epoch": 0.16916009809427232,
-      "grad_norm": 1237.6083984375,
-      "learning_rate": 4.9309655389723705e-05,
-      "loss": 81.8312,
-      "step": 41870
-    },
-    {
-      "epoch": 0.16920049935963993,
-      "grad_norm": 645.1431274414062,
-      "learning_rate": 4.9308840511679804e-05,
-      "loss": 83.625,
-      "step": 41880
-    },
-    {
-      "epoch": 0.16924090062500757,
-      "grad_norm": 808.107177734375,
-      "learning_rate": 4.930802515972255e-05,
-      "loss": 82.5783,
-      "step": 41890
-    },
-    {
-      "epoch": 0.1692813018903752,
-      "grad_norm": 542.672119140625,
-      "learning_rate": 4.930720933386782e-05,
-      "loss": 77.4106,
-      "step": 41900
-    },
-    {
-      "epoch": 0.16932170315574283,
-      "grad_norm": 1336.5858154296875,
-      "learning_rate": 4.930639303413154e-05,
-      "loss": 124.7732,
-      "step": 41910
-    },
-    {
-      "epoch": 0.16936210442111047,
-      "grad_norm": 591.5770263671875,
-      "learning_rate": 4.9305576260529607e-05,
-      "loss": 65.4329,
-      "step": 41920
-    },
-    {
-      "epoch": 0.1694025056864781,
-      "grad_norm": 804.681396484375,
-      "learning_rate": 4.930475901307795e-05,
-      "loss": 76.3269,
-      "step": 41930
-    },
-    {
-      "epoch": 0.16944290695184575,
-      "grad_norm": 987.9590454101562,
-      "learning_rate": 4.930394129179251e-05,
-      "loss": 71.606,
-      "step": 41940
-    },
-    {
-      "epoch": 0.16948330821721336,
-      "grad_norm": 766.5450439453125,
-      "learning_rate": 4.930312309668922e-05,
-      "loss": 109.5266,
-      "step": 41950
-    },
-    {
-      "epoch": 0.169523709482581,
-      "grad_norm": 337.4928894042969,
-      "learning_rate": 4.930230442778403e-05,
-      "loss": 74.3608,
-      "step": 41960
-    },
-    {
-      "epoch": 0.16956411074794864,
-      "grad_norm": 3450.510009765625,
-      "learning_rate": 4.930148528509291e-05,
-      "loss": 109.2802,
-      "step": 41970
-    },
-    {
-      "epoch": 0.16960451201331625,
-      "grad_norm": 864.0614624023438,
-      "learning_rate": 4.930066566863182e-05,
-      "loss": 78.3614,
-      "step": 41980
-    },
-    {
-      "epoch": 0.1696449132786839,
-      "grad_norm": 930.443115234375,
-      "learning_rate": 4.929984557841674e-05,
-      "loss": 126.5678,
-      "step": 41990
-    },
-    {
-      "epoch": 0.16968531454405153,
-      "grad_norm": 416.3900146484375,
-      "learning_rate": 4.929902501446366e-05,
-      "loss": 62.9766,
-      "step": 42000
-    },
-    {
-      "epoch": 0.16972571580941914,
-      "grad_norm": 533.2973022460938,
-      "learning_rate": 4.929820397678858e-05,
-      "loss": 64.3053,
-      "step": 42010
-    },
-    {
-      "epoch": 0.16976611707478678,
-      "grad_norm": 357.4167175292969,
-      "learning_rate": 4.92973824654075e-05,
-      "loss": 43.4001,
-      "step": 42020
-    },
-    {
-      "epoch": 0.16980651834015442,
-      "grad_norm": 1577.94384765625,
-      "learning_rate": 4.929656048033644e-05,
-      "loss": 72.9057,
-      "step": 42030
-    },
-    {
-      "epoch": 0.16984691960552203,
-      "grad_norm": 4723.44775390625,
-      "learning_rate": 4.929573802159143e-05,
-      "loss": 78.4315,
-      "step": 42040
-    },
-    {
-      "epoch": 0.16988732087088967,
-      "grad_norm": 607.1101684570312,
-      "learning_rate": 4.929491508918849e-05,
-      "loss": 83.3789,
-      "step": 42050
-    },
-    {
-      "epoch": 0.16992772213625731,
-      "grad_norm": 527.4968872070312,
-      "learning_rate": 4.929409168314368e-05,
-      "loss": 48.889,
-      "step": 42060
-    },
-    {
-      "epoch": 0.16996812340162493,
-      "grad_norm": 1297.30029296875,
-      "learning_rate": 4.9293267803473046e-05,
-      "loss": 85.3707,
-      "step": 42070
-    },
-    {
-      "epoch": 0.17000852466699257,
-      "grad_norm": 993.5496215820312,
-      "learning_rate": 4.9292443450192645e-05,
-      "loss": 103.5595,
-      "step": 42080
-    },
-    {
-      "epoch": 0.1700489259323602,
-      "grad_norm": 948.1973876953125,
-      "learning_rate": 4.929161862331855e-05,
-      "loss": 91.8254,
-      "step": 42090
-    },
-    {
-      "epoch": 0.17008932719772785,
-      "grad_norm": 740.2516479492188,
-      "learning_rate": 4.929079332286685e-05,
-      "loss": 65.2068,
-      "step": 42100
-    },
-    {
-      "epoch": 0.17012972846309546,
-      "grad_norm": 745.3865356445312,
-      "learning_rate": 4.9289967548853627e-05,
-      "loss": 49.6714,
-      "step": 42110
-    },
-    {
-      "epoch": 0.1701701297284631,
-      "grad_norm": 1320.7667236328125,
-      "learning_rate": 4.928914130129498e-05,
-      "loss": 86.5712,
-      "step": 42120
-    },
-    {
-      "epoch": 0.17021053099383074,
-      "grad_norm": 1176.5162353515625,
-      "learning_rate": 4.928831458020702e-05,
-      "loss": 113.8014,
-      "step": 42130
-    },
-    {
-      "epoch": 0.17025093225919835,
-      "grad_norm": 566.5439453125,
-      "learning_rate": 4.928748738560586e-05,
-      "loss": 72.2683,
-      "step": 42140
-    },
-    {
-      "epoch": 0.170291333524566,
-      "grad_norm": 1156.8316650390625,
-      "learning_rate": 4.9286659717507635e-05,
-      "loss": 76.2927,
-      "step": 42150
-    },
-    {
-      "epoch": 0.17033173478993363,
-      "grad_norm": 719.6638793945312,
-      "learning_rate": 4.9285831575928465e-05,
-      "loss": 69.5816,
-      "step": 42160
-    },
-    {
-      "epoch": 0.17037213605530124,
-      "grad_norm": 1295.44384765625,
-      "learning_rate": 4.9285002960884515e-05,
-      "loss": 84.5863,
-      "step": 42170
-    },
-    {
-      "epoch": 0.17041253732066888,
-      "grad_norm": 1051.867431640625,
-      "learning_rate": 4.9284173872391925e-05,
-      "loss": 82.4616,
-      "step": 42180
-    },
-    {
-      "epoch": 0.17045293858603652,
-      "grad_norm": 859.74609375,
-      "learning_rate": 4.928334431046686e-05,
-      "loss": 111.8195,
-      "step": 42190
-    },
-    {
-      "epoch": 0.17049333985140414,
-      "grad_norm": 1075.9024658203125,
-      "learning_rate": 4.92825142751255e-05,
-      "loss": 85.3877,
-      "step": 42200
-    },
-    {
-      "epoch": 0.17053374111677178,
-      "grad_norm": 1071.4013671875,
-      "learning_rate": 4.9281683766384026e-05,
-      "loss": 77.8994,
-      "step": 42210
-    },
-    {
-      "epoch": 0.17057414238213942,
-      "grad_norm": 841.3255615234375,
-      "learning_rate": 4.9280852784258624e-05,
-      "loss": 104.7014,
-      "step": 42220
-    },
-    {
-      "epoch": 0.17061454364750703,
-      "grad_norm": 1585.5380859375,
-      "learning_rate": 4.928002132876549e-05,
-      "loss": 73.6184,
-      "step": 42230
-    },
-    {
-      "epoch": 0.17065494491287467,
-      "grad_norm": 1064.724609375,
-      "learning_rate": 4.9279189399920844e-05,
-      "loss": 80.9862,
-      "step": 42240
-    },
-    {
-      "epoch": 0.1706953461782423,
-      "grad_norm": 1124.08642578125,
-      "learning_rate": 4.9278356997740904e-05,
-      "loss": 81.2137,
-      "step": 42250
-    },
-    {
-      "epoch": 0.17073574744360995,
-      "grad_norm": 1296.80078125,
-      "learning_rate": 4.9277524122241894e-05,
-      "loss": 76.6494,
-      "step": 42260
-    },
-    {
-      "epoch": 0.17077614870897756,
-      "grad_norm": 705.2342529296875,
-      "learning_rate": 4.927669077344005e-05,
-      "loss": 124.0301,
-      "step": 42270
-    },
-    {
-      "epoch": 0.1708165499743452,
-      "grad_norm": 699.51123046875,
-      "learning_rate": 4.927585695135162e-05,
-      "loss": 90.7621,
-      "step": 42280
-    },
-    {
-      "epoch": 0.17085695123971284,
-      "grad_norm": 1193.2637939453125,
-      "learning_rate": 4.9275022655992864e-05,
-      "loss": 77.3909,
-      "step": 42290
-    },
-    {
-      "epoch": 0.17089735250508045,
-      "grad_norm": 1224.87744140625,
-      "learning_rate": 4.927418788738004e-05,
-      "loss": 90.4889,
-      "step": 42300
-    },
-    {
-      "epoch": 0.1709377537704481,
-      "grad_norm": 566.4915771484375,
-      "learning_rate": 4.927335264552943e-05,
-      "loss": 76.954,
-      "step": 42310
-    },
-    {
-      "epoch": 0.17097815503581573,
-      "grad_norm": 485.48248291015625,
-      "learning_rate": 4.9272516930457314e-05,
-      "loss": 85.5353,
-      "step": 42320
-    },
-    {
-      "epoch": 0.17101855630118334,
-      "grad_norm": 510.9328918457031,
-      "learning_rate": 4.927168074217998e-05,
-      "loss": 78.9659,
-      "step": 42330
-    },
-    {
-      "epoch": 0.17105895756655098,
-      "grad_norm": 542.828125,
-      "learning_rate": 4.927084408071373e-05,
-      "loss": 77.3036,
-      "step": 42340
-    },
-    {
-      "epoch": 0.17109935883191862,
-      "grad_norm": 1816.97265625,
-      "learning_rate": 4.927000694607489e-05,
-      "loss": 89.3204,
-      "step": 42350
-    },
-    {
-      "epoch": 0.17113976009728624,
-      "grad_norm": 558.980712890625,
-      "learning_rate": 4.9269169338279766e-05,
-      "loss": 97.5531,
-      "step": 42360
-    },
-    {
-      "epoch": 0.17118016136265388,
-      "grad_norm": 2137.310791015625,
-      "learning_rate": 4.9268331257344685e-05,
-      "loss": 111.0955,
-      "step": 42370
-    },
-    {
-      "epoch": 0.17122056262802152,
-      "grad_norm": 349.211181640625,
-      "learning_rate": 4.9267492703286e-05,
-      "loss": 116.3871,
-      "step": 42380
-    },
-    {
-      "epoch": 0.17126096389338913,
-      "grad_norm": 1252.0584716796875,
-      "learning_rate": 4.926665367612005e-05,
-      "loss": 78.8891,
-      "step": 42390
-    },
-    {
-      "epoch": 0.17130136515875677,
-      "grad_norm": 587.7830200195312,
-      "learning_rate": 4.9265814175863186e-05,
-      "loss": 90.0841,
-      "step": 42400
-    },
-    {
-      "epoch": 0.1713417664241244,
-      "grad_norm": 1336.141845703125,
-      "learning_rate": 4.926497420253179e-05,
-      "loss": 99.6293,
-      "step": 42410
-    },
-    {
-      "epoch": 0.17138216768949205,
-      "grad_norm": 1955.26171875,
-      "learning_rate": 4.9264133756142224e-05,
-      "loss": 74.8807,
-      "step": 42420
-    },
-    {
-      "epoch": 0.17142256895485966,
-      "grad_norm": 1671.5946044921875,
-      "learning_rate": 4.926329283671088e-05,
-      "loss": 100.2369,
-      "step": 42430
-    },
-    {
-      "epoch": 0.1714629702202273,
-      "grad_norm": 489.6711120605469,
-      "learning_rate": 4.926245144425415e-05,
-      "loss": 92.0346,
-      "step": 42440
-    },
-    {
-      "epoch": 0.17150337148559494,
-      "grad_norm": 1527.400634765625,
-      "learning_rate": 4.9261609578788435e-05,
-      "loss": 104.6484,
-      "step": 42450
-    },
-    {
-      "epoch": 0.17154377275096255,
-      "grad_norm": 1441.6905517578125,
-      "learning_rate": 4.926076724033016e-05,
-      "loss": 118.5627,
-      "step": 42460
-    },
-    {
-      "epoch": 0.1715841740163302,
-      "grad_norm": 1650.8604736328125,
-      "learning_rate": 4.9259924428895734e-05,
-      "loss": 95.2558,
-      "step": 42470
-    },
-    {
-      "epoch": 0.17162457528169783,
-      "grad_norm": 1053.914794921875,
-      "learning_rate": 4.925908114450158e-05,
-      "loss": 92.1593,
-      "step": 42480
-    },
-    {
-      "epoch": 0.17166497654706545,
-      "grad_norm": 485.0592956542969,
-      "learning_rate": 4.925823738716416e-05,
-      "loss": 48.5927,
-      "step": 42490
-    },
-    {
-      "epoch": 0.17170537781243309,
-      "grad_norm": 551.2293701171875,
-      "learning_rate": 4.925739315689991e-05,
-      "loss": 111.4056,
-      "step": 42500
-    },
-    {
-      "epoch": 0.17174577907780073,
-      "grad_norm": 462.1259460449219,
-      "learning_rate": 4.92565484537253e-05,
-      "loss": 81.3713,
-      "step": 42510
-    },
-    {
-      "epoch": 0.17178618034316834,
-      "grad_norm": 976.673095703125,
-      "learning_rate": 4.925570327765678e-05,
-      "loss": 97.2899,
-      "step": 42520
-    },
-    {
-      "epoch": 0.17182658160853598,
-      "grad_norm": 1159.7283935546875,
-      "learning_rate": 4.9254857628710846e-05,
-      "loss": 72.4565,
-      "step": 42530
-    },
-    {
-      "epoch": 0.17186698287390362,
-      "grad_norm": 886.2018432617188,
-      "learning_rate": 4.9254011506903963e-05,
-      "loss": 75.7661,
-      "step": 42540
-    },
-    {
-      "epoch": 0.17190738413927123,
-      "grad_norm": 716.5491333007812,
-      "learning_rate": 4.925316491225265e-05,
-      "loss": 88.0013,
-      "step": 42550
-    },
-    {
-      "epoch": 0.17194778540463887,
-      "grad_norm": 741.2266235351562,
-      "learning_rate": 4.925231784477339e-05,
-      "loss": 106.4336,
-      "step": 42560
-    },
-    {
-      "epoch": 0.1719881866700065,
-      "grad_norm": 1328.2393798828125,
-      "learning_rate": 4.9251470304482716e-05,
-      "loss": 76.8043,
-      "step": 42570
-    },
-    {
-      "epoch": 0.17202858793537412,
-      "grad_norm": 3505.359375,
-      "learning_rate": 4.925062229139714e-05,
-      "loss": 113.7022,
-      "step": 42580
-    },
-    {
-      "epoch": 0.17206898920074176,
-      "grad_norm": 560.7538452148438,
-      "learning_rate": 4.924977380553321e-05,
-      "loss": 78.8128,
-      "step": 42590
-    },
-    {
-      "epoch": 0.1721093904661094,
-      "grad_norm": 936.1580810546875,
-      "learning_rate": 4.924892484690743e-05,
-      "loss": 86.6639,
-      "step": 42600
-    },
-    {
-      "epoch": 0.17214979173147704,
-      "grad_norm": 484.0994873046875,
-      "learning_rate": 4.924807541553639e-05,
-      "loss": 91.6706,
-      "step": 42610
-    },
-    {
-      "epoch": 0.17219019299684465,
-      "grad_norm": 490.91448974609375,
-      "learning_rate": 4.924722551143664e-05,
-      "loss": 52.3336,
-      "step": 42620
-    },
-    {
-      "epoch": 0.1722305942622123,
-      "grad_norm": 1924.5557861328125,
-      "learning_rate": 4.924637513462474e-05,
-      "loss": 98.4392,
-      "step": 42630
-    },
-    {
-      "epoch": 0.17227099552757993,
-      "grad_norm": 738.7843017578125,
-      "learning_rate": 4.9245524285117274e-05,
-      "loss": 90.864,
-      "step": 42640
-    },
-    {
-      "epoch": 0.17231139679294755,
-      "grad_norm": 805.4695434570312,
-      "learning_rate": 4.924467296293083e-05,
-      "loss": 102.7175,
-      "step": 42650
-    },
-    {
-      "epoch": 0.1723517980583152,
-      "grad_norm": 710.0062866210938,
-      "learning_rate": 4.924382116808201e-05,
-      "loss": 107.4096,
-      "step": 42660
-    },
-    {
-      "epoch": 0.17239219932368283,
-      "grad_norm": 703.4191284179688,
-      "learning_rate": 4.924296890058741e-05,
-      "loss": 82.6557,
-      "step": 42670
-    },
-    {
-      "epoch": 0.17243260058905044,
-      "grad_norm": 1078.1011962890625,
-      "learning_rate": 4.924211616046365e-05,
-      "loss": 86.0516,
-      "step": 42680
-    },
-    {
-      "epoch": 0.17247300185441808,
-      "grad_norm": 970.2047729492188,
-      "learning_rate": 4.924126294772735e-05,
-      "loss": 69.9036,
-      "step": 42690
-    },
-    {
-      "epoch": 0.17251340311978572,
-      "grad_norm": 1400.3450927734375,
-      "learning_rate": 4.924040926239515e-05,
-      "loss": 103.4677,
-      "step": 42700
-    },
-    {
-      "epoch": 0.17255380438515333,
-      "grad_norm": 1445.791259765625,
-      "learning_rate": 4.9239555104483695e-05,
-      "loss": 115.1777,
-      "step": 42710
-    },
-    {
-      "epoch": 0.17259420565052097,
-      "grad_norm": 570.3566284179688,
-      "learning_rate": 4.923870047400964e-05,
-      "loss": 83.1902,
-      "step": 42720
-    },
-    {
-      "epoch": 0.1726346069158886,
-      "grad_norm": 880.5896606445312,
-      "learning_rate": 4.923784537098963e-05,
-      "loss": 106.7852,
-      "step": 42730
-    },
-    {
-      "epoch": 0.17267500818125622,
-      "grad_norm": 13018.2724609375,
-      "learning_rate": 4.9236989795440346e-05,
-      "loss": 108.5354,
-      "step": 42740
-    },
-    {
-      "epoch": 0.17271540944662386,
-      "grad_norm": 1054.298583984375,
-      "learning_rate": 4.9236133747378475e-05,
-      "loss": 81.4467,
-      "step": 42750
-    },
-    {
-      "epoch": 0.1727558107119915,
-      "grad_norm": 1258.8382568359375,
-      "learning_rate": 4.9235277226820695e-05,
-      "loss": 115.5568,
-      "step": 42760
-    },
-    {
-      "epoch": 0.17279621197735914,
-      "grad_norm": 955.5006103515625,
-      "learning_rate": 4.923442023378371e-05,
-      "loss": 104.4606,
-      "step": 42770
-    },
-    {
-      "epoch": 0.17283661324272676,
-      "grad_norm": 797.49951171875,
-      "learning_rate": 4.9233562768284225e-05,
-      "loss": 75.7812,
-      "step": 42780
-    },
-    {
-      "epoch": 0.1728770145080944,
-      "grad_norm": 550.285400390625,
-      "learning_rate": 4.923270483033896e-05,
-      "loss": 90.7212,
-      "step": 42790
-    },
-    {
-      "epoch": 0.17291741577346204,
-      "grad_norm": 828.7791137695312,
-      "learning_rate": 4.923184641996463e-05,
-      "loss": 66.7497,
-      "step": 42800
-    },
-    {
-      "epoch": 0.17295781703882965,
-      "grad_norm": 862.7196044921875,
-      "learning_rate": 4.923098753717798e-05,
-      "loss": 74.5591,
-      "step": 42810
-    },
-    {
-      "epoch": 0.1729982183041973,
-      "grad_norm": 1932.47802734375,
-      "learning_rate": 4.923012818199576e-05,
-      "loss": 113.689,
-      "step": 42820
-    },
-    {
-      "epoch": 0.17303861956956493,
-      "grad_norm": 467.0488586425781,
-      "learning_rate": 4.922926835443472e-05,
-      "loss": 80.5391,
-      "step": 42830
-    },
-    {
-      "epoch": 0.17307902083493254,
-      "grad_norm": 1409.27734375,
-      "learning_rate": 4.922840805451161e-05,
-      "loss": 141.5631,
-      "step": 42840
-    },
-    {
-      "epoch": 0.17311942210030018,
-      "grad_norm": 470.03057861328125,
-      "learning_rate": 4.9227547282243214e-05,
-      "loss": 55.4408,
-      "step": 42850
-    },
-    {
-      "epoch": 0.17315982336566782,
-      "grad_norm": 399.5194091796875,
-      "learning_rate": 4.9226686037646314e-05,
-      "loss": 95.3332,
-      "step": 42860
-    },
-    {
-      "epoch": 0.17320022463103543,
-      "grad_norm": 1028.5482177734375,
-      "learning_rate": 4.92258243207377e-05,
-      "loss": 84.2015,
-      "step": 42870
-    },
-    {
-      "epoch": 0.17324062589640307,
-      "grad_norm": 936.9313354492188,
-      "learning_rate": 4.922496213153416e-05,
-      "loss": 85.3661,
-      "step": 42880
-    },
-    {
-      "epoch": 0.1732810271617707,
-      "grad_norm": 614.690673828125,
-      "learning_rate": 4.922409947005251e-05,
-      "loss": 49.2994,
-      "step": 42890
-    },
-    {
-      "epoch": 0.17332142842713832,
-      "grad_norm": 845.6972045898438,
-      "learning_rate": 4.922323633630958e-05,
-      "loss": 44.1013,
-      "step": 42900
-    },
-    {
-      "epoch": 0.17336182969250596,
-      "grad_norm": 1084.47900390625,
-      "learning_rate": 4.9222372730322176e-05,
-      "loss": 131.938,
-      "step": 42910
-    },
-    {
-      "epoch": 0.1734022309578736,
-      "grad_norm": 1015.8482666015625,
-      "learning_rate": 4.922150865210715e-05,
-      "loss": 90.8332,
-      "step": 42920
-    },
-    {
-      "epoch": 0.17344263222324124,
-      "grad_norm": 548.1792602539062,
-      "learning_rate": 4.922064410168134e-05,
-      "loss": 94.3187,
-      "step": 42930
-    },
-    {
-      "epoch": 0.17348303348860886,
-      "grad_norm": 625.8770751953125,
-      "learning_rate": 4.92197790790616e-05,
-      "loss": 86.6768,
-      "step": 42940
-    },
-    {
-      "epoch": 0.1735234347539765,
-      "grad_norm": 0.0,
-      "learning_rate": 4.9218913584264814e-05,
-      "loss": 86.3633,
-      "step": 42950
-    },
-    {
-      "epoch": 0.17356383601934414,
-      "grad_norm": 646.9606323242188,
-      "learning_rate": 4.9218047617307824e-05,
-      "loss": 88.7674,
-      "step": 42960
-    },
-    {
-      "epoch": 0.17360423728471175,
-      "grad_norm": 249.1079864501953,
-      "learning_rate": 4.9217181178207535e-05,
-      "loss": 61.6128,
-      "step": 42970
-    },
-    {
-      "epoch": 0.1736446385500794,
-      "grad_norm": 543.1034545898438,
-      "learning_rate": 4.9216314266980824e-05,
-      "loss": 79.8014,
-      "step": 42980
-    },
-    {
-      "epoch": 0.17368503981544703,
-      "grad_norm": 662.4139404296875,
-      "learning_rate": 4.921544688364461e-05,
-      "loss": 95.9196,
-      "step": 42990
-    },
-    {
-      "epoch": 0.17372544108081464,
-      "grad_norm": 422.02825927734375,
-      "learning_rate": 4.9214579028215776e-05,
-      "loss": 91.8231,
-      "step": 43000
-    },
-    {
-      "epoch": 0.17376584234618228,
-      "grad_norm": 891.07958984375,
-      "learning_rate": 4.921371070071127e-05,
-      "loss": 74.39,
-      "step": 43010
-    },
-    {
-      "epoch": 0.17380624361154992,
-      "grad_norm": 1176.2237548828125,
-      "learning_rate": 4.9212841901148e-05,
-      "loss": 90.9235,
-      "step": 43020
-    },
-    {
-      "epoch": 0.17384664487691753,
-      "grad_norm": 4489.8310546875,
-      "learning_rate": 4.9211972629542926e-05,
-      "loss": 124.854,
-      "step": 43030
-    },
-    {
-      "epoch": 0.17388704614228517,
-      "grad_norm": 639.3281860351562,
-      "learning_rate": 4.9211102885912965e-05,
-      "loss": 76.7055,
-      "step": 43040
-    },
-    {
-      "epoch": 0.1739274474076528,
-      "grad_norm": 1555.4083251953125,
-      "learning_rate": 4.9210232670275094e-05,
-      "loss": 110.0548,
-      "step": 43050
-    },
-    {
-      "epoch": 0.17396784867302043,
-      "grad_norm": 727.7996826171875,
-      "learning_rate": 4.920936198264627e-05,
-      "loss": 95.9471,
-      "step": 43060
-    },
-    {
-      "epoch": 0.17400824993838807,
-      "grad_norm": 1188.0120849609375,
-      "learning_rate": 4.920849082304347e-05,
-      "loss": 113.6858,
-      "step": 43070
-    },
-    {
-      "epoch": 0.1740486512037557,
-      "grad_norm": 1449.6580810546875,
-      "learning_rate": 4.920761919148369e-05,
-      "loss": 94.3256,
-      "step": 43080
-    },
-    {
-      "epoch": 0.17408905246912335,
-      "grad_norm": 454.728759765625,
-      "learning_rate": 4.9206747087983894e-05,
-      "loss": 78.3794,
-      "step": 43090
-    },
-    {
-      "epoch": 0.17412945373449096,
-      "grad_norm": 815.9066162109375,
-      "learning_rate": 4.9205874512561115e-05,
-      "loss": 89.8247,
-      "step": 43100
-    },
-    {
-      "epoch": 0.1741698549998586,
-      "grad_norm": 759.9569702148438,
-      "learning_rate": 4.920500146523234e-05,
-      "loss": 63.8019,
-      "step": 43110
-    },
-    {
-      "epoch": 0.17421025626522624,
-      "grad_norm": 2164.500732421875,
-      "learning_rate": 4.920412794601461e-05,
-      "loss": 100.4632,
-      "step": 43120
-    },
-    {
-      "epoch": 0.17425065753059385,
-      "grad_norm": 662.5512084960938,
-      "learning_rate": 4.920325395492493e-05,
-      "loss": 107.611,
-      "step": 43130
-    },
-    {
-      "epoch": 0.1742910587959615,
-      "grad_norm": 422.92645263671875,
-      "learning_rate": 4.920237949198037e-05,
-      "loss": 87.6762,
-      "step": 43140
-    },
-    {
-      "epoch": 0.17433146006132913,
-      "grad_norm": 915.8494873046875,
-      "learning_rate": 4.9201504557197955e-05,
-      "loss": 66.672,
-      "step": 43150
-    },
-    {
-      "epoch": 0.17437186132669674,
-      "grad_norm": 988.4778442382812,
-      "learning_rate": 4.9200629150594744e-05,
-      "loss": 91.4257,
-      "step": 43160
-    },
-    {
-      "epoch": 0.17441226259206438,
-      "grad_norm": 794.6514282226562,
-      "learning_rate": 4.919975327218781e-05,
-      "loss": 87.8521,
-      "step": 43170
-    },
-    {
-      "epoch": 0.17445266385743202,
-      "grad_norm": 884.67578125,
-      "learning_rate": 4.919887692199423e-05,
-      "loss": 74.1221,
-      "step": 43180
-    },
-    {
-      "epoch": 0.17449306512279963,
-      "grad_norm": 764.0117797851562,
-      "learning_rate": 4.919800010003108e-05,
-      "loss": 84.2991,
-      "step": 43190
-    },
-    {
-      "epoch": 0.17453346638816727,
-      "grad_norm": 1031.7420654296875,
-      "learning_rate": 4.919712280631547e-05,
-      "loss": 116.3033,
-      "step": 43200
-    },
-    {
-      "epoch": 0.17457386765353491,
-      "grad_norm": 593.7903442382812,
-      "learning_rate": 4.9196245040864486e-05,
-      "loss": 81.53,
-      "step": 43210
-    },
-    {
-      "epoch": 0.17461426891890253,
-      "grad_norm": 929.322021484375,
-      "learning_rate": 4.919536680369525e-05,
-      "loss": 87.4219,
-      "step": 43220
-    },
-    {
-      "epoch": 0.17465467018427017,
-      "grad_norm": 3907.416015625,
-      "learning_rate": 4.919448809482489e-05,
-      "loss": 81.7991,
-      "step": 43230
-    },
-    {
-      "epoch": 0.1746950714496378,
-      "grad_norm": 893.3704223632812,
-      "learning_rate": 4.9193608914270515e-05,
-      "loss": 96.2677,
-      "step": 43240
-    },
-    {
-      "epoch": 0.17473547271500545,
-      "grad_norm": 659.1729125976562,
-      "learning_rate": 4.919272926204929e-05,
-      "loss": 98.551,
-      "step": 43250
-    },
-    {
-      "epoch": 0.17477587398037306,
-      "grad_norm": 995.1451416015625,
-      "learning_rate": 4.9191849138178334e-05,
-      "loss": 71.9214,
-      "step": 43260
-    },
-    {
-      "epoch": 0.1748162752457407,
-      "grad_norm": 1777.868896484375,
-      "learning_rate": 4.919096854267484e-05,
-      "loss": 95.7281,
-      "step": 43270
-    },
-    {
-      "epoch": 0.17485667651110834,
-      "grad_norm": 779.3260498046875,
-      "learning_rate": 4.9190087475555955e-05,
-      "loss": 104.3814,
-      "step": 43280
-    },
-    {
-      "epoch": 0.17489707777647595,
-      "grad_norm": 939.8629760742188,
-      "learning_rate": 4.9189205936838864e-05,
-      "loss": 71.219,
-      "step": 43290
-    },
-    {
-      "epoch": 0.1749374790418436,
-      "grad_norm": 808.994140625,
-      "learning_rate": 4.9188323926540746e-05,
-      "loss": 63.1995,
-      "step": 43300
-    },
-    {
-      "epoch": 0.17497788030721123,
-      "grad_norm": 1255.6544189453125,
-      "learning_rate": 4.918744144467881e-05,
-      "loss": 73.3233,
-      "step": 43310
-    },
-    {
-      "epoch": 0.17501828157257884,
-      "grad_norm": 1129.1463623046875,
-      "learning_rate": 4.918655849127024e-05,
-      "loss": 80.2645,
-      "step": 43320
-    },
-    {
-      "epoch": 0.17505868283794648,
-      "grad_norm": 0.0,
-      "learning_rate": 4.918567506633226e-05,
-      "loss": 107.1662,
-      "step": 43330
-    },
-    {
-      "epoch": 0.17509908410331412,
-      "grad_norm": 219.4386444091797,
-      "learning_rate": 4.91847911698821e-05,
-      "loss": 66.5381,
-      "step": 43340
-    },
-    {
-      "epoch": 0.17513948536868174,
-      "grad_norm": 432.4940185546875,
-      "learning_rate": 4.918390680193698e-05,
-      "loss": 73.7235,
-      "step": 43350
-    },
-    {
-      "epoch": 0.17517988663404938,
-      "grad_norm": 645.31494140625,
-      "learning_rate": 4.918302196251415e-05,
-      "loss": 110.384,
-      "step": 43360
-    },
-    {
-      "epoch": 0.17522028789941702,
-      "grad_norm": 3361.419677734375,
-      "learning_rate": 4.918213665163085e-05,
-      "loss": 122.9304,
-      "step": 43370
-    },
-    {
-      "epoch": 0.17526068916478463,
-      "grad_norm": 898.6963500976562,
-      "learning_rate": 4.918125086930435e-05,
-      "loss": 79.2204,
-      "step": 43380
-    },
-    {
-      "epoch": 0.17530109043015227,
-      "grad_norm": 982.2764892578125,
-      "learning_rate": 4.918036461555192e-05,
-      "loss": 87.4724,
-      "step": 43390
-    },
-    {
-      "epoch": 0.1753414916955199,
-      "grad_norm": 423.7979431152344,
-      "learning_rate": 4.9179477890390825e-05,
-      "loss": 96.8271,
-      "step": 43400
-    },
-    {
-      "epoch": 0.17538189296088755,
-      "grad_norm": 994.8875732421875,
-      "learning_rate": 4.917859069383836e-05,
-      "loss": 65.6866,
-      "step": 43410
-    },
-    {
-      "epoch": 0.17542229422625516,
-      "grad_norm": 488.28973388671875,
-      "learning_rate": 4.9177703025911825e-05,
-      "loss": 67.0737,
-      "step": 43420
-    },
-    {
-      "epoch": 0.1754626954916228,
-      "grad_norm": 927.5771484375,
-      "learning_rate": 4.917681488662852e-05,
-      "loss": 96.7948,
-      "step": 43430
-    },
-    {
-      "epoch": 0.17550309675699044,
-      "grad_norm": 1078.08837890625,
-      "learning_rate": 4.917592627600577e-05,
-      "loss": 77.3857,
-      "step": 43440
-    },
-    {
-      "epoch": 0.17554349802235805,
-      "grad_norm": 1068.2432861328125,
-      "learning_rate": 4.917503719406088e-05,
-      "loss": 82.1481,
-      "step": 43450
-    },
-    {
-      "epoch": 0.1755838992877257,
-      "grad_norm": 622.7982177734375,
-      "learning_rate": 4.91741476408112e-05,
-      "loss": 88.3467,
-      "step": 43460
-    },
-    {
-      "epoch": 0.17562430055309333,
-      "grad_norm": 529.6117553710938,
-      "learning_rate": 4.917325761627406e-05,
-      "loss": 94.7772,
-      "step": 43470
-    },
-    {
-      "epoch": 0.17566470181846094,
-      "grad_norm": 527.8485107421875,
-      "learning_rate": 4.917236712046682e-05,
-      "loss": 49.5305,
-      "step": 43480
-    },
-    {
-      "epoch": 0.17570510308382858,
-      "grad_norm": 714.2109375,
-      "learning_rate": 4.917147615340684e-05,
-      "loss": 84.2201,
-      "step": 43490
-    },
-    {
-      "epoch": 0.17574550434919622,
-      "grad_norm": 836.039794921875,
-      "learning_rate": 4.917058471511149e-05,
-      "loss": 73.1889,
-      "step": 43500
-    },
-    {
-      "epoch": 0.17578590561456384,
-      "grad_norm": 1239.9703369140625,
-      "learning_rate": 4.9169692805598145e-05,
-      "loss": 94.0238,
-      "step": 43510
-    },
-    {
-      "epoch": 0.17582630687993148,
-      "grad_norm": 447.8238830566406,
-      "learning_rate": 4.916880042488419e-05,
-      "loss": 56.3747,
-      "step": 43520
-    },
-    {
-      "epoch": 0.17586670814529912,
-      "grad_norm": 646.1732788085938,
-      "learning_rate": 4.916790757298704e-05,
-      "loss": 73.1238,
-      "step": 43530
-    },
-    {
-      "epoch": 0.17590710941066673,
-      "grad_norm": 2080.33984375,
-      "learning_rate": 4.9167014249924075e-05,
-      "loss": 111.906,
-      "step": 43540
-    },
-    {
-      "epoch": 0.17594751067603437,
-      "grad_norm": 500.1817932128906,
-      "learning_rate": 4.9166120455712736e-05,
-      "loss": 55.3391,
-      "step": 43550
-    },
-    {
-      "epoch": 0.175987911941402,
-      "grad_norm": 495.2890319824219,
-      "learning_rate": 4.916522619037043e-05,
-      "loss": 111.2715,
-      "step": 43560
-    },
-    {
-      "epoch": 0.17602831320676965,
-      "grad_norm": 539.0352783203125,
-      "learning_rate": 4.91643314539146e-05,
-      "loss": 72.2024,
-      "step": 43570
-    },
-    {
-      "epoch": 0.17606871447213726,
-      "grad_norm": 1020.1885375976562,
-      "learning_rate": 4.916343624636269e-05,
-      "loss": 133.4901,
-      "step": 43580
-    },
-    {
-      "epoch": 0.1761091157375049,
-      "grad_norm": 463.427978515625,
-      "learning_rate": 4.916254056773215e-05,
-      "loss": 79.8422,
-      "step": 43590
-    },
-    {
-      "epoch": 0.17614951700287254,
-      "grad_norm": 1095.344970703125,
-      "learning_rate": 4.916164441804044e-05,
-      "loss": 79.1968,
-      "step": 43600
-    },
-    {
-      "epoch": 0.17618991826824015,
-      "grad_norm": 1455.2132568359375,
-      "learning_rate": 4.916074779730504e-05,
-      "loss": 79.4097,
-      "step": 43610
-    },
-    {
-      "epoch": 0.1762303195336078,
-      "grad_norm": 854.4801025390625,
-      "learning_rate": 4.915985070554341e-05,
-      "loss": 108.5478,
-      "step": 43620
-    },
-    {
-      "epoch": 0.17627072079897543,
-      "grad_norm": 1308.75341796875,
-      "learning_rate": 4.915895314277306e-05,
-      "loss": 64.448,
-      "step": 43630
-    },
-    {
-      "epoch": 0.17631112206434305,
-      "grad_norm": 755.5027465820312,
-      "learning_rate": 4.915805510901148e-05,
-      "loss": 99.9466,
-      "step": 43640
-    },
-    {
-      "epoch": 0.17635152332971069,
-      "grad_norm": 3085.270751953125,
-      "learning_rate": 4.9157156604276175e-05,
-      "loss": 99.849,
-      "step": 43650
-    },
-    {
-      "epoch": 0.17639192459507833,
-      "grad_norm": 1174.489013671875,
-      "learning_rate": 4.915625762858467e-05,
-      "loss": 71.8641,
-      "step": 43660
-    },
-    {
-      "epoch": 0.17643232586044594,
-      "grad_norm": 1123.778076171875,
-      "learning_rate": 4.9155358181954494e-05,
-      "loss": 92.842,
-      "step": 43670
-    },
-    {
-      "epoch": 0.17647272712581358,
-      "grad_norm": 424.877197265625,
-      "learning_rate": 4.915445826440316e-05,
-      "loss": 82.3663,
-      "step": 43680
-    },
-    {
-      "epoch": 0.17651312839118122,
-      "grad_norm": 2889.92041015625,
-      "learning_rate": 4.915355787594823e-05,
-      "loss": 81.4271,
-      "step": 43690
-    },
-    {
-      "epoch": 0.17655352965654883,
-      "grad_norm": 5192.76611328125,
-      "learning_rate": 4.915265701660726e-05,
-      "loss": 80.7691,
-      "step": 43700
-    },
-    {
-      "epoch": 0.17659393092191647,
-      "grad_norm": 995.813720703125,
-      "learning_rate": 4.9151755686397793e-05,
-      "loss": 75.4785,
-      "step": 43710
-    },
-    {
-      "epoch": 0.1766343321872841,
-      "grad_norm": 648.2084350585938,
-      "learning_rate": 4.9150853885337426e-05,
-      "loss": 93.2939,
-      "step": 43720
-    },
-    {
-      "epoch": 0.17667473345265175,
-      "grad_norm": 601.667724609375,
-      "learning_rate": 4.914995161344373e-05,
-      "loss": 68.4272,
-      "step": 43730
-    },
-    {
-      "epoch": 0.17671513471801936,
-      "grad_norm": 553.5189819335938,
-      "learning_rate": 4.9149048870734296e-05,
-      "loss": 75.1263,
-      "step": 43740
-    },
-    {
-      "epoch": 0.176755535983387,
-      "grad_norm": 816.7628784179688,
-      "learning_rate": 4.914814565722671e-05,
-      "loss": 100.5337,
-      "step": 43750
-    },
-    {
-      "epoch": 0.17679593724875464,
-      "grad_norm": 1510.35302734375,
-      "learning_rate": 4.9147241972938596e-05,
-      "loss": 94.5766,
-      "step": 43760
-    },
-    {
-      "epoch": 0.17683633851412225,
-      "grad_norm": 834.1815795898438,
-      "learning_rate": 4.9146337817887575e-05,
-      "loss": 70.5322,
-      "step": 43770
-    },
-    {
-      "epoch": 0.1768767397794899,
-      "grad_norm": 1132.84619140625,
-      "learning_rate": 4.914543319209126e-05,
-      "loss": 74.2719,
-      "step": 43780
-    },
-    {
-      "epoch": 0.17691714104485753,
-      "grad_norm": 0.0,
-      "learning_rate": 4.91445280955673e-05,
-      "loss": 91.3708,
-      "step": 43790
-    },
-    {
-      "epoch": 0.17695754231022515,
-      "grad_norm": 347.47705078125,
-      "learning_rate": 4.914362252833332e-05,
-      "loss": 70.5821,
-      "step": 43800
-    },
-    {
-      "epoch": 0.1769979435755928,
-      "grad_norm": 820.30419921875,
-      "learning_rate": 4.9142716490407e-05,
-      "loss": 122.0043,
-      "step": 43810
-    },
-    {
-      "epoch": 0.17703834484096043,
-      "grad_norm": 910.3120727539062,
-      "learning_rate": 4.9141809981805995e-05,
-      "loss": 86.2672,
-      "step": 43820
-    },
-    {
-      "epoch": 0.17707874610632804,
-      "grad_norm": 536.81982421875,
-      "learning_rate": 4.914090300254798e-05,
-      "loss": 67.7232,
-      "step": 43830
-    },
-    {
-      "epoch": 0.17711914737169568,
-      "grad_norm": 799.4282836914062,
-      "learning_rate": 4.913999555265062e-05,
-      "loss": 66.7534,
-      "step": 43840
-    },
-    {
-      "epoch": 0.17715954863706332,
-      "grad_norm": 978.2288208007812,
-      "learning_rate": 4.913908763213162e-05,
-      "loss": 97.7384,
-      "step": 43850
-    },
-    {
-      "epoch": 0.17719994990243093,
-      "grad_norm": 1155.240478515625,
-      "learning_rate": 4.913817924100869e-05,
-      "loss": 83.567,
-      "step": 43860
-    },
-    {
-      "epoch": 0.17724035116779857,
-      "grad_norm": 4742.7177734375,
-      "learning_rate": 4.913727037929952e-05,
-      "loss": 106.2576,
-      "step": 43870
-    },
-    {
-      "epoch": 0.1772807524331662,
-      "grad_norm": 725.6898193359375,
-      "learning_rate": 4.913636104702183e-05,
-      "loss": 95.1339,
-      "step": 43880
-    },
-    {
-      "epoch": 0.17732115369853385,
-      "grad_norm": 933.677978515625,
-      "learning_rate": 4.913545124419336e-05,
-      "loss": 77.6824,
-      "step": 43890
-    },
-    {
-      "epoch": 0.17736155496390146,
-      "grad_norm": 484.01092529296875,
-      "learning_rate": 4.913454097083185e-05,
-      "loss": 95.2952,
-      "step": 43900
-    },
-    {
-      "epoch": 0.1774019562292691,
-      "grad_norm": 4791.2919921875,
-      "learning_rate": 4.9133630226955026e-05,
-      "loss": 66.6884,
-      "step": 43910
-    },
-    {
-      "epoch": 0.17744235749463674,
-      "grad_norm": 750.3236694335938,
-      "learning_rate": 4.913271901258067e-05,
-      "loss": 71.9357,
-      "step": 43920
-    },
-    {
-      "epoch": 0.17748275876000436,
-      "grad_norm": 738.4391479492188,
-      "learning_rate": 4.913180732772652e-05,
-      "loss": 73.5072,
-      "step": 43930
-    },
-    {
-      "epoch": 0.177523160025372,
-      "grad_norm": 3437.857666015625,
-      "learning_rate": 4.913089517241037e-05,
-      "loss": 102.427,
-      "step": 43940
-    },
-    {
-      "epoch": 0.17756356129073964,
-      "grad_norm": 790.71044921875,
-      "learning_rate": 4.912998254665e-05,
-      "loss": 75.0368,
-      "step": 43950
-    },
-    {
-      "epoch": 0.17760396255610725,
-      "grad_norm": 417.4804382324219,
-      "learning_rate": 4.9129069450463186e-05,
-      "loss": 97.0161,
-      "step": 43960
-    },
-    {
-      "epoch": 0.1776443638214749,
-      "grad_norm": 988.0493774414062,
-      "learning_rate": 4.912815588386775e-05,
-      "loss": 67.218,
-      "step": 43970
-    },
-    {
-      "epoch": 0.17768476508684253,
-      "grad_norm": 867.89404296875,
-      "learning_rate": 4.912724184688149e-05,
-      "loss": 70.8273,
-      "step": 43980
-    },
-    {
-      "epoch": 0.17772516635221014,
-      "grad_norm": 678.370849609375,
-      "learning_rate": 4.9126327339522225e-05,
-      "loss": 79.2882,
-      "step": 43990
-    },
-    {
-      "epoch": 0.17776556761757778,
-      "grad_norm": 1945.7412109375,
-      "learning_rate": 4.912541236180779e-05,
-      "loss": 102.0449,
-      "step": 44000
-    },
-    {
-      "epoch": 0.17780596888294542,
-      "grad_norm": 874.1896362304688,
-      "learning_rate": 4.912449691375602e-05,
-      "loss": 73.3128,
-      "step": 44010
-    },
-    {
-      "epoch": 0.17784637014831303,
-      "grad_norm": 922.020751953125,
-      "learning_rate": 4.912358099538476e-05,
-      "loss": 105.76,
-      "step": 44020
-    },
-    {
-      "epoch": 0.17788677141368067,
-      "grad_norm": 295.5516662597656,
-      "learning_rate": 4.912266460671187e-05,
-      "loss": 71.2611,
-      "step": 44030
-    },
-    {
-      "epoch": 0.1779271726790483,
-      "grad_norm": 524.7007446289062,
-      "learning_rate": 4.912174774775522e-05,
-      "loss": 132.5543,
-      "step": 44040
-    },
-    {
-      "epoch": 0.17796757394441595,
-      "grad_norm": 1069.502685546875,
-      "learning_rate": 4.912083041853267e-05,
-      "loss": 49.895,
-      "step": 44050
-    },
-    {
-      "epoch": 0.17800797520978356,
-      "grad_norm": 1046.591796875,
-      "learning_rate": 4.911991261906212e-05,
-      "loss": 121.5496,
-      "step": 44060
-    },
-    {
-      "epoch": 0.1780483764751512,
-      "grad_norm": 1215.2974853515625,
-      "learning_rate": 4.9118994349361455e-05,
-      "loss": 67.8808,
-      "step": 44070
-    },
-    {
-      "epoch": 0.17808877774051884,
-      "grad_norm": 419.9411315917969,
-      "learning_rate": 4.911807560944858e-05,
-      "loss": 94.9882,
-      "step": 44080
-    },
-    {
-      "epoch": 0.17812917900588646,
-      "grad_norm": 555.5469360351562,
-      "learning_rate": 4.911715639934139e-05,
-      "loss": 75.9314,
-      "step": 44090
-    },
-    {
-      "epoch": 0.1781695802712541,
-      "grad_norm": 961.8922729492188,
-      "learning_rate": 4.911623671905784e-05,
-      "loss": 281.5219,
-      "step": 44100
-    },
-    {
-      "epoch": 0.17820998153662174,
-      "grad_norm": 443.88385009765625,
-      "learning_rate": 4.9115316568615824e-05,
-      "loss": 72.1933,
-      "step": 44110
-    },
-    {
-      "epoch": 0.17825038280198935,
-      "grad_norm": 670.9072265625,
-      "learning_rate": 4.9114395948033296e-05,
-      "loss": 59.2318,
-      "step": 44120
-    },
-    {
-      "epoch": 0.178290784067357,
-      "grad_norm": 951.0693359375,
-      "learning_rate": 4.911347485732821e-05,
-      "loss": 92.5108,
-      "step": 44130
-    },
-    {
-      "epoch": 0.17833118533272463,
-      "grad_norm": 742.0170288085938,
-      "learning_rate": 4.911255329651851e-05,
-      "loss": 81.8882,
-      "step": 44140
-    },
-    {
-      "epoch": 0.17837158659809224,
-      "grad_norm": 1449.71826171875,
-      "learning_rate": 4.9111631265622184e-05,
-      "loss": 99.7724,
-      "step": 44150
-    },
-    {
-      "epoch": 0.17841198786345988,
-      "grad_norm": 884.3943481445312,
-      "learning_rate": 4.911070876465719e-05,
-      "loss": 77.8581,
-      "step": 44160
-    },
-    {
-      "epoch": 0.17845238912882752,
-      "grad_norm": 930.3131103515625,
-      "learning_rate": 4.910978579364151e-05,
-      "loss": 75.9937,
-      "step": 44170
-    },
-    {
-      "epoch": 0.17849279039419513,
-      "grad_norm": 741.734130859375,
-      "learning_rate": 4.910886235259314e-05,
-      "loss": 49.2752,
-      "step": 44180
-    },
-    {
-      "epoch": 0.17853319165956277,
-      "grad_norm": 1187.31982421875,
-      "learning_rate": 4.910793844153009e-05,
-      "loss": 72.5356,
-      "step": 44190
-    },
-    {
-      "epoch": 0.1785735929249304,
-      "grad_norm": 1084.1396484375,
-      "learning_rate": 4.910701406047037e-05,
-      "loss": 54.3127,
-      "step": 44200
-    },
-    {
-      "epoch": 0.17861399419029805,
-      "grad_norm": 1090.561767578125,
-      "learning_rate": 4.910608920943199e-05,
-      "loss": 89.8081,
-      "step": 44210
-    },
-    {
-      "epoch": 0.17865439545566567,
-      "grad_norm": 926.6570434570312,
-      "learning_rate": 4.9105163888433e-05,
-      "loss": 102.4193,
-      "step": 44220
-    },
-    {
-      "epoch": 0.1786947967210333,
-      "grad_norm": 823.6111450195312,
-      "learning_rate": 4.910423809749143e-05,
-      "loss": 71.3021,
-      "step": 44230
-    },
-    {
-      "epoch": 0.17873519798640095,
-      "grad_norm": 1753.1083984375,
-      "learning_rate": 4.910331183662533e-05,
-      "loss": 73.3729,
-      "step": 44240
-    },
-    {
-      "epoch": 0.17877559925176856,
-      "grad_norm": 1302.8763427734375,
-      "learning_rate": 4.910238510585276e-05,
-      "loss": 73.3519,
-      "step": 44250
-    },
-    {
-      "epoch": 0.1788160005171362,
-      "grad_norm": 1096.9420166015625,
-      "learning_rate": 4.9101457905191774e-05,
-      "loss": 97.6709,
-      "step": 44260
-    },
-    {
-      "epoch": 0.17885640178250384,
-      "grad_norm": 1942.8685302734375,
-      "learning_rate": 4.910053023466046e-05,
-      "loss": 144.2541,
-      "step": 44270
-    },
-    {
-      "epoch": 0.17889680304787145,
-      "grad_norm": 784.9008178710938,
-      "learning_rate": 4.90996020942769e-05,
-      "loss": 80.2861,
-      "step": 44280
-    },
-    {
-      "epoch": 0.1789372043132391,
-      "grad_norm": 1027.322509765625,
-      "learning_rate": 4.9098673484059195e-05,
-      "loss": 69.3643,
-      "step": 44290
-    },
-    {
-      "epoch": 0.17897760557860673,
-      "grad_norm": 1583.6845703125,
-      "learning_rate": 4.9097744404025435e-05,
-      "loss": 92.1036,
-      "step": 44300
-    },
-    {
-      "epoch": 0.17901800684397434,
-      "grad_norm": 735.26806640625,
-      "learning_rate": 4.909681485419375e-05,
-      "loss": 60.5722,
-      "step": 44310
-    },
-    {
-      "epoch": 0.17905840810934198,
-      "grad_norm": 0.0,
-      "learning_rate": 4.909588483458225e-05,
-      "loss": 65.9718,
-      "step": 44320
-    },
-    {
-      "epoch": 0.17909880937470962,
-      "grad_norm": 676.1497802734375,
-      "learning_rate": 4.9094954345209075e-05,
-      "loss": 85.0213,
-      "step": 44330
-    },
-    {
-      "epoch": 0.17913921064007723,
-      "grad_norm": 873.891357421875,
-      "learning_rate": 4.909402338609236e-05,
-      "loss": 77.8588,
-      "step": 44340
-    },
-    {
-      "epoch": 0.17917961190544487,
-      "grad_norm": 693.5972900390625,
-      "learning_rate": 4.909309195725025e-05,
-      "loss": 124.6474,
-      "step": 44350
-    },
-    {
-      "epoch": 0.17922001317081251,
-      "grad_norm": 460.6761779785156,
-      "learning_rate": 4.90921600587009e-05,
-      "loss": 87.2825,
-      "step": 44360
-    },
-    {
-      "epoch": 0.17926041443618015,
-      "grad_norm": 1261.747314453125,
-      "learning_rate": 4.90912276904625e-05,
-      "loss": 70.8723,
-      "step": 44370
-    },
-    {
-      "epoch": 0.17930081570154777,
-      "grad_norm": 538.2672729492188,
-      "learning_rate": 4.909029485255321e-05,
-      "loss": 67.4372,
-      "step": 44380
-    },
-    {
-      "epoch": 0.1793412169669154,
-      "grad_norm": 1068.8714599609375,
-      "learning_rate": 4.9089361544991215e-05,
-      "loss": 68.0589,
-      "step": 44390
-    },
-    {
-      "epoch": 0.17938161823228305,
-      "grad_norm": 1119.72314453125,
-      "learning_rate": 4.908842776779472e-05,
-      "loss": 68.3467,
-      "step": 44400
-    },
-    {
-      "epoch": 0.17942201949765066,
-      "grad_norm": 568.3760986328125,
-      "learning_rate": 4.908749352098192e-05,
-      "loss": 73.3346,
-      "step": 44410
-    },
-    {
-      "epoch": 0.1794624207630183,
-      "grad_norm": 1218.2125244140625,
-      "learning_rate": 4.9086558804571034e-05,
-      "loss": 109.4271,
-      "step": 44420
-    },
-    {
-      "epoch": 0.17950282202838594,
-      "grad_norm": 582.5836181640625,
-      "learning_rate": 4.908562361858028e-05,
-      "loss": 72.2208,
-      "step": 44430
-    },
-    {
-      "epoch": 0.17954322329375355,
-      "grad_norm": 1486.9588623046875,
-      "learning_rate": 4.9084687963027894e-05,
-      "loss": 87.6006,
-      "step": 44440
-    },
-    {
-      "epoch": 0.1795836245591212,
-      "grad_norm": 646.9586791992188,
-      "learning_rate": 4.9083751837932126e-05,
-      "loss": 84.1968,
-      "step": 44450
-    },
-    {
-      "epoch": 0.17962402582448883,
-      "grad_norm": 665.9592895507812,
-      "learning_rate": 4.908281524331121e-05,
-      "loss": 61.7497,
-      "step": 44460
-    },
-    {
-      "epoch": 0.17966442708985644,
-      "grad_norm": 514.8817749023438,
-      "learning_rate": 4.908187817918341e-05,
-      "loss": 71.9887,
-      "step": 44470
-    },
-    {
-      "epoch": 0.17970482835522408,
-      "grad_norm": 883.0592651367188,
-      "learning_rate": 4.9080940645567e-05,
-      "loss": 72.1492,
-      "step": 44480
-    },
-    {
-      "epoch": 0.17974522962059172,
-      "grad_norm": 1375.27001953125,
-      "learning_rate": 4.908000264248025e-05,
-      "loss": 96.8598,
-      "step": 44490
-    },
-    {
-      "epoch": 0.17978563088595934,
-      "grad_norm": 645.9873046875,
-      "learning_rate": 4.907906416994146e-05,
-      "loss": 64.9413,
-      "step": 44500
-    },
-    {
-      "epoch": 0.17982603215132698,
-      "grad_norm": 1111.731201171875,
-      "learning_rate": 4.9078125227968904e-05,
-      "loss": 86.5026,
-      "step": 44510
-    },
-    {
-      "epoch": 0.17986643341669462,
-      "grad_norm": 934.5158081054688,
-      "learning_rate": 4.907718581658091e-05,
-      "loss": 110.799,
-      "step": 44520
-    },
-    {
-      "epoch": 0.17990683468206226,
-      "grad_norm": 920.4935913085938,
-      "learning_rate": 4.9076245935795786e-05,
-      "loss": 62.2967,
-      "step": 44530
-    },
-    {
-      "epoch": 0.17994723594742987,
-      "grad_norm": 892.095947265625,
-      "learning_rate": 4.9075305585631845e-05,
-      "loss": 82.56,
-      "step": 44540
-    },
-    {
-      "epoch": 0.1799876372127975,
-      "grad_norm": 1195.0810546875,
-      "learning_rate": 4.907436476610743e-05,
-      "loss": 48.5051,
-      "step": 44550
-    },
-    {
-      "epoch": 0.18002803847816515,
-      "grad_norm": 768.6405029296875,
-      "learning_rate": 4.907342347724087e-05,
-      "loss": 67.0793,
-      "step": 44560
-    },
-    {
-      "epoch": 0.18006843974353276,
-      "grad_norm": 1605.87060546875,
-      "learning_rate": 4.907248171905055e-05,
-      "loss": 114.835,
-      "step": 44570
-    },
-    {
-      "epoch": 0.1801088410089004,
-      "grad_norm": 718.5313720703125,
-      "learning_rate": 4.907153949155479e-05,
-      "loss": 100.6462,
-      "step": 44580
-    },
-    {
-      "epoch": 0.18014924227426804,
-      "grad_norm": 430.45037841796875,
-      "learning_rate": 4.907059679477197e-05,
-      "loss": 67.9887,
-      "step": 44590
-    },
-    {
-      "epoch": 0.18018964353963565,
-      "grad_norm": 234.9417266845703,
-      "learning_rate": 4.906965362872047e-05,
-      "loss": 77.6404,
-      "step": 44600
-    },
-    {
-      "epoch": 0.1802300448050033,
-      "grad_norm": 911.5045776367188,
-      "learning_rate": 4.906870999341869e-05,
-      "loss": 88.723,
-      "step": 44610
-    },
-    {
-      "epoch": 0.18027044607037093,
-      "grad_norm": 772.7266845703125,
-      "learning_rate": 4.906776588888502e-05,
-      "loss": 93.7803,
-      "step": 44620
-    },
-    {
-      "epoch": 0.18031084733573854,
-      "grad_norm": 1137.0054931640625,
-      "learning_rate": 4.9066821315137856e-05,
-      "loss": 110.4208,
-      "step": 44630
-    },
-    {
-      "epoch": 0.18035124860110618,
-      "grad_norm": 898.4769287109375,
-      "learning_rate": 4.906587627219562e-05,
-      "loss": 73.2961,
-      "step": 44640
-    },
-    {
-      "epoch": 0.18039164986647382,
-      "grad_norm": 958.9808959960938,
-      "learning_rate": 4.906493076007674e-05,
-      "loss": 57.1335,
-      "step": 44650
-    },
-    {
-      "epoch": 0.18043205113184144,
-      "grad_norm": 938.7362670898438,
-      "learning_rate": 4.9063984778799645e-05,
-      "loss": 101.8361,
-      "step": 44660
-    },
-    {
-      "epoch": 0.18047245239720908,
-      "grad_norm": 515.5078735351562,
-      "learning_rate": 4.906303832838278e-05,
-      "loss": 54.5177,
-      "step": 44670
-    },
-    {
-      "epoch": 0.18051285366257672,
-      "grad_norm": 531.8738403320312,
-      "learning_rate": 4.906209140884459e-05,
-      "loss": 82.9638,
-      "step": 44680
-    },
-    {
-      "epoch": 0.18055325492794436,
-      "grad_norm": 1305.5316162109375,
-      "learning_rate": 4.906114402020354e-05,
-      "loss": 95.9825,
-      "step": 44690
-    },
-    {
-      "epoch": 0.18059365619331197,
-      "grad_norm": 777.7756958007812,
-      "learning_rate": 4.90601961624781e-05,
-      "loss": 81.2786,
-      "step": 44700
-    },
-    {
-      "epoch": 0.1806340574586796,
-      "grad_norm": 1133.5772705078125,
-      "learning_rate": 4.905924783568675e-05,
-      "loss": 85.8932,
-      "step": 44710
-    },
-    {
-      "epoch": 0.18067445872404725,
-      "grad_norm": 599.99951171875,
-      "learning_rate": 4.9058299039847975e-05,
-      "loss": 101.0232,
-      "step": 44720
-    },
-    {
-      "epoch": 0.18071485998941486,
-      "grad_norm": 675.3793334960938,
-      "learning_rate": 4.9057349774980275e-05,
-      "loss": 56.9,
-      "step": 44730
-    },
-    {
-      "epoch": 0.1807552612547825,
-      "grad_norm": 961.1941528320312,
-      "learning_rate": 4.905640004110216e-05,
-      "loss": 84.8394,
-      "step": 44740
-    },
-    {
-      "epoch": 0.18079566252015014,
-      "grad_norm": 1229.0406494140625,
-      "learning_rate": 4.905544983823214e-05,
-      "loss": 65.7035,
-      "step": 44750
-    },
-    {
-      "epoch": 0.18083606378551775,
-      "grad_norm": 1367.959228515625,
-      "learning_rate": 4.905449916638873e-05,
-      "loss": 79.8659,
-      "step": 44760
-    },
-    {
-      "epoch": 0.1808764650508854,
-      "grad_norm": 705.6975708007812,
-      "learning_rate": 4.905354802559049e-05,
-      "loss": 109.2609,
-      "step": 44770
-    },
-    {
-      "epoch": 0.18091686631625303,
-      "grad_norm": 738.3800048828125,
-      "learning_rate": 4.905259641585594e-05,
-      "loss": 45.9301,
-      "step": 44780
-    },
-    {
-      "epoch": 0.18095726758162065,
-      "grad_norm": 746.9801025390625,
-      "learning_rate": 4.905164433720364e-05,
-      "loss": 65.5784,
-      "step": 44790
-    },
-    {
-      "epoch": 0.18099766884698829,
-      "grad_norm": 546.0762329101562,
-      "learning_rate": 4.905069178965215e-05,
-      "loss": 97.4426,
-      "step": 44800
-    },
-    {
-      "epoch": 0.18103807011235593,
-      "grad_norm": 714.2156372070312,
-      "learning_rate": 4.9049738773220046e-05,
-      "loss": 83.1105,
-      "step": 44810
-    },
-    {
-      "epoch": 0.18107847137772354,
-      "grad_norm": 862.12939453125,
-      "learning_rate": 4.9048785287925895e-05,
-      "loss": 56.6679,
-      "step": 44820
-    },
-    {
-      "epoch": 0.18111887264309118,
-      "grad_norm": 233.91058349609375,
-      "learning_rate": 4.9047831333788295e-05,
-      "loss": 82.0974,
-      "step": 44830
-    },
-    {
-      "epoch": 0.18115927390845882,
-      "grad_norm": 820.8073120117188,
-      "learning_rate": 4.904687691082585e-05,
-      "loss": 72.0476,
-      "step": 44840
-    },
-    {
-      "epoch": 0.18119967517382646,
-      "grad_norm": 1259.7579345703125,
-      "learning_rate": 4.9045922019057155e-05,
-      "loss": 83.561,
-      "step": 44850
-    },
-    {
-      "epoch": 0.18124007643919407,
-      "grad_norm": 879.6514282226562,
-      "learning_rate": 4.904496665850084e-05,
-      "loss": 82.4675,
-      "step": 44860
-    },
-    {
-      "epoch": 0.1812804777045617,
-      "grad_norm": 684.139892578125,
-      "learning_rate": 4.90440108291755e-05,
-      "loss": 90.4461,
-      "step": 44870
-    },
-    {
-      "epoch": 0.18132087896992935,
-      "grad_norm": 2262.56591796875,
-      "learning_rate": 4.904305453109981e-05,
-      "loss": 53.0787,
-      "step": 44880
-    },
-    {
-      "epoch": 0.18136128023529696,
-      "grad_norm": 1157.9019775390625,
-      "learning_rate": 4.9042097764292385e-05,
-      "loss": 83.7821,
-      "step": 44890
-    },
-    {
-      "epoch": 0.1814016815006646,
-      "grad_norm": 945.2867431640625,
-      "learning_rate": 4.904114052877188e-05,
-      "loss": 81.2998,
-      "step": 44900
-    },
-    {
-      "epoch": 0.18144208276603224,
-      "grad_norm": 1119.50048828125,
-      "learning_rate": 4.904018282455697e-05,
-      "loss": 49.5603,
-      "step": 44910
-    },
-    {
-      "epoch": 0.18148248403139985,
-      "grad_norm": 1104.3919677734375,
-      "learning_rate": 4.9039224651666325e-05,
-      "loss": 89.8467,
-      "step": 44920
-    },
-    {
-      "epoch": 0.1815228852967675,
-      "grad_norm": 852.224365234375,
-      "learning_rate": 4.903826601011861e-05,
-      "loss": 69.1575,
-      "step": 44930
-    },
-    {
-      "epoch": 0.18156328656213513,
-      "grad_norm": 473.6451110839844,
-      "learning_rate": 4.903730689993253e-05,
-      "loss": 92.4134,
-      "step": 44940
-    },
-    {
-      "epoch": 0.18160368782750275,
-      "grad_norm": 406.23199462890625,
-      "learning_rate": 4.903634732112678e-05,
-      "loss": 42.3387,
-      "step": 44950
-    },
-    {
-      "epoch": 0.1816440890928704,
-      "grad_norm": 1381.426025390625,
-      "learning_rate": 4.903538727372005e-05,
-      "loss": 81.0296,
-      "step": 44960
-    },
-    {
-      "epoch": 0.18168449035823803,
-      "grad_norm": 1150.0416259765625,
-      "learning_rate": 4.903442675773108e-05,
-      "loss": 67.7141,
-      "step": 44970
-    },
-    {
-      "epoch": 0.18172489162360564,
-      "grad_norm": 816.09619140625,
-      "learning_rate": 4.903346577317859e-05,
-      "loss": 85.8159,
-      "step": 44980
-    },
-    {
-      "epoch": 0.18176529288897328,
-      "grad_norm": 602.0270385742188,
-      "learning_rate": 4.90325043200813e-05,
-      "loss": 63.4132,
-      "step": 44990
-    },
-    {
-      "epoch": 0.18180569415434092,
-      "grad_norm": 815.766357421875,
-      "learning_rate": 4.9031542398457974e-05,
-      "loss": 88.5122,
-      "step": 45000
-    },
-    {
-      "epoch": 0.18184609541970856,
-      "grad_norm": 863.7614135742188,
-      "learning_rate": 4.9030580008327353e-05,
-      "loss": 91.4247,
-      "step": 45010
-    },
-    {
-      "epoch": 0.18188649668507617,
-      "grad_norm": 2180.32373046875,
-      "learning_rate": 4.902961714970821e-05,
-      "loss": 104.5198,
-      "step": 45020
-    },
-    {
-      "epoch": 0.1819268979504438,
-      "grad_norm": 527.2975463867188,
-      "learning_rate": 4.90286538226193e-05,
-      "loss": 89.9132,
-      "step": 45030
-    },
-    {
-      "epoch": 0.18196729921581145,
-      "grad_norm": 797.1927490234375,
-      "learning_rate": 4.902769002707942e-05,
-      "loss": 60.7632,
-      "step": 45040
-    },
-    {
-      "epoch": 0.18200770048117906,
-      "grad_norm": 1439.7625732421875,
-      "learning_rate": 4.902672576310735e-05,
-      "loss": 94.8452,
-      "step": 45050
-    },
-    {
-      "epoch": 0.1820481017465467,
-      "grad_norm": 962.4320678710938,
-      "learning_rate": 4.902576103072189e-05,
-      "loss": 93.4051,
-      "step": 45060
-    },
-    {
-      "epoch": 0.18208850301191434,
-      "grad_norm": 509.36468505859375,
-      "learning_rate": 4.902479582994185e-05,
-      "loss": 101.6452,
-      "step": 45070
-    },
-    {
-      "epoch": 0.18212890427728196,
-      "grad_norm": 674.908447265625,
-      "learning_rate": 4.902383016078605e-05,
-      "loss": 82.7371,
-      "step": 45080
-    },
-    {
-      "epoch": 0.1821693055426496,
-      "grad_norm": 581.795166015625,
-      "learning_rate": 4.902286402327331e-05,
-      "loss": 78.2122,
-      "step": 45090
-    },
-    {
-      "epoch": 0.18220970680801724,
-      "grad_norm": 1014.2080078125,
-      "learning_rate": 4.902189741742247e-05,
-      "loss": 93.4292,
-      "step": 45100
-    },
-    {
-      "epoch": 0.18225010807338485,
-      "grad_norm": 1033.72412109375,
-      "learning_rate": 4.902093034325237e-05,
-      "loss": 86.2924,
-      "step": 45110
-    },
-    {
-      "epoch": 0.1822905093387525,
-      "grad_norm": 553.49755859375,
-      "learning_rate": 4.901996280078186e-05,
-      "loss": 115.3292,
-      "step": 45120
-    },
-    {
-      "epoch": 0.18233091060412013,
-      "grad_norm": 867.1639404296875,
-      "learning_rate": 4.901899479002982e-05,
-      "loss": 64.9478,
-      "step": 45130
-    },
-    {
-      "epoch": 0.18237131186948774,
-      "grad_norm": 314.60528564453125,
-      "learning_rate": 4.901802631101511e-05,
-      "loss": 65.43,
-      "step": 45140
-    },
-    {
-      "epoch": 0.18241171313485538,
-      "grad_norm": 676.8156127929688,
-      "learning_rate": 4.90170573637566e-05,
-      "loss": 71.8611,
-      "step": 45150
-    },
-    {
-      "epoch": 0.18245211440022302,
-      "grad_norm": 947.8834228515625,
-      "learning_rate": 4.90160879482732e-05,
-      "loss": 94.2625,
-      "step": 45160
-    },
-    {
-      "epoch": 0.18249251566559066,
-      "grad_norm": 628.1900634765625,
-      "learning_rate": 4.901511806458381e-05,
-      "loss": 59.9568,
-      "step": 45170
-    },
-    {
-      "epoch": 0.18253291693095827,
-      "grad_norm": 1697.5426025390625,
-      "learning_rate": 4.9014147712707316e-05,
-      "loss": 78.6836,
-      "step": 45180
-    },
-    {
-      "epoch": 0.1825733181963259,
-      "grad_norm": 1006.1732177734375,
-      "learning_rate": 4.9013176892662654e-05,
-      "loss": 82.129,
-      "step": 45190
-    },
-    {
-      "epoch": 0.18261371946169355,
-      "grad_norm": 1150.0892333984375,
-      "learning_rate": 4.9012205604468744e-05,
-      "loss": 117.7069,
-      "step": 45200
-    },
-    {
-      "epoch": 0.18265412072706116,
-      "grad_norm": 707.0164184570312,
-      "learning_rate": 4.9011233848144525e-05,
-      "loss": 82.716,
-      "step": 45210
-    },
-    {
-      "epoch": 0.1826945219924288,
-      "grad_norm": 355.14959716796875,
-      "learning_rate": 4.9010261623708944e-05,
-      "loss": 47.5373,
-      "step": 45220
-    },
-    {
-      "epoch": 0.18273492325779644,
-      "grad_norm": 784.4992065429688,
-      "learning_rate": 4.9009288931180947e-05,
-      "loss": 82.2301,
-      "step": 45230
-    },
-    {
-      "epoch": 0.18277532452316406,
-      "grad_norm": 1023.0245361328125,
-      "learning_rate": 4.90083157705795e-05,
-      "loss": 67.702,
-      "step": 45240
-    },
-    {
-      "epoch": 0.1828157257885317,
-      "grad_norm": 801.0816040039062,
-      "learning_rate": 4.900734214192358e-05,
-      "loss": 76.0119,
-      "step": 45250
-    },
-    {
-      "epoch": 0.18285612705389934,
-      "grad_norm": 1161.0206298828125,
-      "learning_rate": 4.900636804523217e-05,
-      "loss": 75.7235,
-      "step": 45260
-    },
-    {
-      "epoch": 0.18289652831926695,
-      "grad_norm": 1342.0335693359375,
-      "learning_rate": 4.900539348052424e-05,
-      "loss": 82.4319,
-      "step": 45270
-    },
-    {
-      "epoch": 0.1829369295846346,
-      "grad_norm": 1176.3917236328125,
-      "learning_rate": 4.9004418447818815e-05,
-      "loss": 97.9844,
-      "step": 45280
-    },
-    {
-      "epoch": 0.18297733085000223,
-      "grad_norm": 1022.2837524414062,
-      "learning_rate": 4.9003442947134895e-05,
-      "loss": 78.1378,
-      "step": 45290
-    },
-    {
-      "epoch": 0.18301773211536984,
-      "grad_norm": 816.5732421875,
-      "learning_rate": 4.90024669784915e-05,
-      "loss": 70.9764,
-      "step": 45300
-    },
-    {
-      "epoch": 0.18305813338073748,
-      "grad_norm": 830.2484741210938,
-      "learning_rate": 4.9001490541907645e-05,
-      "loss": 62.789,
-      "step": 45310
-    },
-    {
-      "epoch": 0.18309853464610512,
-      "grad_norm": 1164.3995361328125,
-      "learning_rate": 4.900051363740238e-05,
-      "loss": 63.8086,
-      "step": 45320
-    },
-    {
-      "epoch": 0.18313893591147276,
-      "grad_norm": 1487.6907958984375,
-      "learning_rate": 4.899953626499475e-05,
-      "loss": 78.6914,
-      "step": 45330
-    },
-    {
-      "epoch": 0.18317933717684037,
-      "grad_norm": 570.93212890625,
-      "learning_rate": 4.89985584247038e-05,
-      "loss": 80.8218,
-      "step": 45340
-    },
-    {
-      "epoch": 0.183219738442208,
-      "grad_norm": 822.6448974609375,
-      "learning_rate": 4.8997580116548595e-05,
-      "loss": 105.5186,
-      "step": 45350
-    },
-    {
-      "epoch": 0.18326013970757565,
-      "grad_norm": 1853.758544921875,
-      "learning_rate": 4.8996601340548215e-05,
-      "loss": 116.5778,
-      "step": 45360
-    },
-    {
-      "epoch": 0.18330054097294327,
-      "grad_norm": 712.744140625,
-      "learning_rate": 4.899562209672174e-05,
-      "loss": 67.3311,
-      "step": 45370
-    },
-    {
-      "epoch": 0.1833409422383109,
-      "grad_norm": 752.5093994140625,
-      "learning_rate": 4.899464238508825e-05,
-      "loss": 75.8195,
-      "step": 45380
-    },
-    {
-      "epoch": 0.18338134350367855,
-      "grad_norm": 1256.2913818359375,
-      "learning_rate": 4.899366220566686e-05,
-      "loss": 112.7811,
-      "step": 45390
-    },
-    {
-      "epoch": 0.18342174476904616,
-      "grad_norm": 1340.5006103515625,
-      "learning_rate": 4.899268155847667e-05,
-      "loss": 103.8136,
-      "step": 45400
-    },
-    {
-      "epoch": 0.1834621460344138,
-      "grad_norm": 698.527587890625,
-      "learning_rate": 4.89917004435368e-05,
-      "loss": 85.698,
-      "step": 45410
-    },
-    {
-      "epoch": 0.18350254729978144,
-      "grad_norm": 697.1134033203125,
-      "learning_rate": 4.899071886086638e-05,
-      "loss": 95.746,
-      "step": 45420
-    },
-    {
-      "epoch": 0.18354294856514905,
-      "grad_norm": 860.381591796875,
-      "learning_rate": 4.898973681048454e-05,
-      "loss": 71.6687,
-      "step": 45430
-    },
-    {
-      "epoch": 0.1835833498305167,
-      "grad_norm": 300.5081481933594,
-      "learning_rate": 4.898875429241044e-05,
-      "loss": 57.1445,
-      "step": 45440
-    },
-    {
-      "epoch": 0.18362375109588433,
-      "grad_norm": 325.8232727050781,
-      "learning_rate": 4.898777130666322e-05,
-      "loss": 102.5872,
-      "step": 45450
-    },
-    {
-      "epoch": 0.18366415236125194,
-      "grad_norm": 714.4207153320312,
-      "learning_rate": 4.898678785326205e-05,
-      "loss": 96.0007,
-      "step": 45460
-    },
-    {
-      "epoch": 0.18370455362661958,
-      "grad_norm": 879.5983276367188,
-      "learning_rate": 4.8985803932226094e-05,
-      "loss": 91.5776,
-      "step": 45470
-    },
-    {
-      "epoch": 0.18374495489198722,
-      "grad_norm": 1637.9261474609375,
-      "learning_rate": 4.898481954357455e-05,
-      "loss": 108.3861,
-      "step": 45480
-    },
-    {
-      "epoch": 0.18378535615735486,
-      "grad_norm": 839.2368774414062,
-      "learning_rate": 4.8983834687326596e-05,
-      "loss": 96.1661,
-      "step": 45490
-    },
-    {
-      "epoch": 0.18382575742272247,
-      "grad_norm": 1169.771240234375,
-      "learning_rate": 4.898284936350144e-05,
-      "loss": 102.8537,
-      "step": 45500
-    },
-    {
-      "epoch": 0.18386615868809011,
-      "grad_norm": 760.4490966796875,
-      "learning_rate": 4.898186357211829e-05,
-      "loss": 74.1794,
-      "step": 45510
-    },
-    {
-      "epoch": 0.18390655995345775,
-      "grad_norm": 784.6276245117188,
-      "learning_rate": 4.898087731319636e-05,
-      "loss": 67.2023,
-      "step": 45520
-    },
-    {
-      "epoch": 0.18394696121882537,
-      "grad_norm": 1203.34228515625,
-      "learning_rate": 4.8979890586754875e-05,
-      "loss": 104.9335,
-      "step": 45530
-    },
-    {
-      "epoch": 0.183987362484193,
-      "grad_norm": 1361.6192626953125,
-      "learning_rate": 4.897890339281309e-05,
-      "loss": 83.1708,
-      "step": 45540
-    },
-    {
-      "epoch": 0.18402776374956065,
-      "grad_norm": 1109.7901611328125,
-      "learning_rate": 4.897791573139023e-05,
-      "loss": 89.0058,
-      "step": 45550
-    },
-    {
-      "epoch": 0.18406816501492826,
-      "grad_norm": 945.8731689453125,
-      "learning_rate": 4.897692760250556e-05,
-      "loss": 124.9025,
-      "step": 45560
-    },
-    {
-      "epoch": 0.1841085662802959,
-      "grad_norm": 931.4639282226562,
-      "learning_rate": 4.897593900617834e-05,
-      "loss": 87.5932,
-      "step": 45570
-    },
-    {
-      "epoch": 0.18414896754566354,
-      "grad_norm": 928.37646484375,
-      "learning_rate": 4.897494994242785e-05,
-      "loss": 121.7126,
-      "step": 45580
-    },
-    {
-      "epoch": 0.18418936881103115,
-      "grad_norm": 1160.150634765625,
-      "learning_rate": 4.8973960411273364e-05,
-      "loss": 88.2009,
-      "step": 45590
-    },
-    {
-      "epoch": 0.1842297700763988,
-      "grad_norm": 1830.1046142578125,
-      "learning_rate": 4.8972970412734176e-05,
-      "loss": 94.9112,
-      "step": 45600
-    },
-    {
-      "epoch": 0.18427017134176643,
-      "grad_norm": 760.2313232421875,
-      "learning_rate": 4.897197994682959e-05,
-      "loss": 66.2683,
-      "step": 45610
-    },
-    {
-      "epoch": 0.18431057260713404,
-      "grad_norm": 1265.816650390625,
-      "learning_rate": 4.897098901357891e-05,
-      "loss": 145.3422,
-      "step": 45620
-    },
-    {
-      "epoch": 0.18435097387250168,
-      "grad_norm": 1088.4708251953125,
-      "learning_rate": 4.896999761300146e-05,
-      "loss": 124.1203,
-      "step": 45630
-    },
-    {
-      "epoch": 0.18439137513786932,
-      "grad_norm": 2023.6080322265625,
-      "learning_rate": 4.896900574511657e-05,
-      "loss": 89.7067,
-      "step": 45640
-    },
-    {
-      "epoch": 0.18443177640323694,
-      "grad_norm": 467.109130859375,
-      "learning_rate": 4.896801340994357e-05,
-      "loss": 67.6761,
-      "step": 45650
-    },
-    {
-      "epoch": 0.18447217766860458,
-      "grad_norm": 1232.957763671875,
-      "learning_rate": 4.896702060750181e-05,
-      "loss": 87.1448,
-      "step": 45660
-    },
-    {
-      "epoch": 0.18451257893397222,
-      "grad_norm": 1274.385009765625,
-      "learning_rate": 4.896602733781065e-05,
-      "loss": 90.0393,
-      "step": 45670
-    },
-    {
-      "epoch": 0.18455298019933986,
-      "grad_norm": 913.9437255859375,
-      "learning_rate": 4.8965033600889435e-05,
-      "loss": 58.4704,
-      "step": 45680
-    },
-    {
-      "epoch": 0.18459338146470747,
-      "grad_norm": 1077.12744140625,
-      "learning_rate": 4.896403939675756e-05,
-      "loss": 101.7226,
-      "step": 45690
-    },
-    {
-      "epoch": 0.1846337827300751,
-      "grad_norm": 648.5958862304688,
-      "learning_rate": 4.89630447254344e-05,
-      "loss": 59.2482,
-      "step": 45700
-    },
-    {
-      "epoch": 0.18467418399544275,
-      "grad_norm": 1022.220703125,
-      "learning_rate": 4.896204958693934e-05,
-      "loss": 62.0977,
-      "step": 45710
-    },
-    {
-      "epoch": 0.18471458526081036,
-      "grad_norm": 902.6193237304688,
-      "learning_rate": 4.8961053981291795e-05,
-      "loss": 57.4871,
-      "step": 45720
-    },
-    {
-      "epoch": 0.184754986526178,
-      "grad_norm": 971.0794067382812,
-      "learning_rate": 4.896005790851116e-05,
-      "loss": 67.6352,
-      "step": 45730
-    },
-    {
-      "epoch": 0.18479538779154564,
-      "grad_norm": 696.0235595703125,
-      "learning_rate": 4.8959061368616863e-05,
-      "loss": 60.1176,
-      "step": 45740
-    },
-    {
-      "epoch": 0.18483578905691325,
-      "grad_norm": 403.1430969238281,
-      "learning_rate": 4.895806436162833e-05,
-      "loss": 92.4626,
-      "step": 45750
-    },
-    {
-      "epoch": 0.1848761903222809,
-      "grad_norm": 524.8573608398438,
-      "learning_rate": 4.8957066887565e-05,
-      "loss": 82.3987,
-      "step": 45760
-    },
-    {
-      "epoch": 0.18491659158764853,
-      "grad_norm": 671.1500244140625,
-      "learning_rate": 4.8956068946446314e-05,
-      "loss": 63.0742,
-      "step": 45770
-    },
-    {
-      "epoch": 0.18495699285301614,
-      "grad_norm": 1303.492431640625,
-      "learning_rate": 4.8955070538291735e-05,
-      "loss": 80.9564,
-      "step": 45780
-    },
-    {
-      "epoch": 0.18499739411838378,
-      "grad_norm": 433.896240234375,
-      "learning_rate": 4.8954071663120715e-05,
-      "loss": 58.4002,
-      "step": 45790
-    },
-    {
-      "epoch": 0.18503779538375142,
-      "grad_norm": 504.2688903808594,
-      "learning_rate": 4.8953072320952745e-05,
-      "loss": 73.8806,
-      "step": 45800
-    },
-    {
-      "epoch": 0.18507819664911904,
-      "grad_norm": 974.8590087890625,
-      "learning_rate": 4.895207251180729e-05,
-      "loss": 112.7232,
-      "step": 45810
-    },
-    {
-      "epoch": 0.18511859791448668,
-      "grad_norm": 1069.8795166015625,
-      "learning_rate": 4.8951072235703855e-05,
-      "loss": 108.1218,
-      "step": 45820
-    },
-    {
-      "epoch": 0.18515899917985432,
-      "grad_norm": 385.00994873046875,
-      "learning_rate": 4.895007149266193e-05,
-      "loss": 55.9683,
-      "step": 45830
-    },
-    {
-      "epoch": 0.18519940044522196,
-      "grad_norm": 680.8944091796875,
-      "learning_rate": 4.8949070282701034e-05,
-      "loss": 81.2045,
-      "step": 45840
-    },
-    {
-      "epoch": 0.18523980171058957,
-      "grad_norm": 885.607177734375,
-      "learning_rate": 4.8948068605840694e-05,
-      "loss": 83.6066,
-      "step": 45850
-    },
-    {
-      "epoch": 0.1852802029759572,
-      "grad_norm": 794.8469848632812,
-      "learning_rate": 4.894706646210041e-05,
-      "loss": 97.3249,
-      "step": 45860
-    },
-    {
-      "epoch": 0.18532060424132485,
-      "grad_norm": 865.634033203125,
-      "learning_rate": 4.8946063851499746e-05,
-      "loss": 51.3998,
-      "step": 45870
-    },
-    {
-      "epoch": 0.18536100550669246,
-      "grad_norm": 882.5504760742188,
-      "learning_rate": 4.894506077405824e-05,
-      "loss": 65.3006,
-      "step": 45880
-    },
-    {
-      "epoch": 0.1854014067720601,
-      "grad_norm": 712.5584106445312,
-      "learning_rate": 4.894405722979544e-05,
-      "loss": 60.6922,
-      "step": 45890
-    },
-    {
-      "epoch": 0.18544180803742774,
-      "grad_norm": 1048.72216796875,
-      "learning_rate": 4.894305321873092e-05,
-      "loss": 79.0462,
-      "step": 45900
-    },
-    {
-      "epoch": 0.18548220930279535,
-      "grad_norm": 757.8132934570312,
-      "learning_rate": 4.894204874088425e-05,
-      "loss": 82.1683,
-      "step": 45910
-    },
-    {
-      "epoch": 0.185522610568163,
-      "grad_norm": 946.23828125,
-      "learning_rate": 4.8941043796275015e-05,
-      "loss": 80.1746,
-      "step": 45920
-    },
-    {
-      "epoch": 0.18556301183353063,
-      "grad_norm": 1357.65380859375,
-      "learning_rate": 4.8940038384922806e-05,
-      "loss": 77.1498,
-      "step": 45930
-    },
-    {
-      "epoch": 0.18560341309889825,
-      "grad_norm": 1069.114990234375,
-      "learning_rate": 4.8939032506847224e-05,
-      "loss": 97.8804,
-      "step": 45940
-    },
-    {
-      "epoch": 0.18564381436426589,
-      "grad_norm": 1382.1502685546875,
-      "learning_rate": 4.893802616206787e-05,
-      "loss": 108.694,
-      "step": 45950
-    },
-    {
-      "epoch": 0.18568421562963353,
-      "grad_norm": 1493.73486328125,
-      "learning_rate": 4.893701935060439e-05,
-      "loss": 81.302,
-      "step": 45960
-    },
-    {
-      "epoch": 0.18572461689500114,
-      "grad_norm": 873.5044555664062,
-      "learning_rate": 4.893601207247638e-05,
-      "loss": 76.8872,
-      "step": 45970
-    },
-    {
-      "epoch": 0.18576501816036878,
-      "grad_norm": 2981.951416015625,
-      "learning_rate": 4.893500432770349e-05,
-      "loss": 68.5081,
-      "step": 45980
-    },
-    {
-      "epoch": 0.18580541942573642,
-      "grad_norm": 396.176513671875,
-      "learning_rate": 4.893399611630538e-05,
-      "loss": 65.5169,
-      "step": 45990
-    },
-    {
-      "epoch": 0.18584582069110406,
-      "grad_norm": 1273.2371826171875,
-      "learning_rate": 4.893298743830168e-05,
-      "loss": 69.0687,
-      "step": 46000
-    },
-    {
-      "epoch": 0.18588622195647167,
-      "grad_norm": 2934.45361328125,
-      "learning_rate": 4.8931978293712074e-05,
-      "loss": 67.4254,
-      "step": 46010
-    },
-    {
-      "epoch": 0.1859266232218393,
-      "grad_norm": 665.5234985351562,
-      "learning_rate": 4.8930968682556234e-05,
-      "loss": 49.5943,
-      "step": 46020
-    },
-    {
-      "epoch": 0.18596702448720695,
-      "grad_norm": 688.6306762695312,
-      "learning_rate": 4.892995860485384e-05,
-      "loss": 50.1611,
-      "step": 46030
-    },
-    {
-      "epoch": 0.18600742575257456,
-      "grad_norm": 544.0740356445312,
-      "learning_rate": 4.892894806062458e-05,
-      "loss": 90.3554,
-      "step": 46040
-    },
-    {
-      "epoch": 0.1860478270179422,
-      "grad_norm": 817.1072998046875,
-      "learning_rate": 4.892793704988816e-05,
-      "loss": 100.2068,
-      "step": 46050
-    },
-    {
-      "epoch": 0.18608822828330984,
-      "grad_norm": 901.2867431640625,
-      "learning_rate": 4.892692557266429e-05,
-      "loss": 82.2532,
-      "step": 46060
-    },
-    {
-      "epoch": 0.18612862954867745,
-      "grad_norm": 670.2236328125,
-      "learning_rate": 4.892591362897268e-05,
-      "loss": 113.8737,
-      "step": 46070
-    },
-    {
-      "epoch": 0.1861690308140451,
-      "grad_norm": 500.85870361328125,
-      "learning_rate": 4.892490121883306e-05,
-      "loss": 81.973,
-      "step": 46080
-    },
-    {
-      "epoch": 0.18620943207941273,
-      "grad_norm": 0.0,
-      "learning_rate": 4.892388834226519e-05,
-      "loss": 63.0598,
-      "step": 46090
-    },
-    {
-      "epoch": 0.18624983334478035,
-      "grad_norm": 1111.806396484375,
-      "learning_rate": 4.892287499928879e-05,
-      "loss": 70.0805,
-      "step": 46100
-    },
-    {
-      "epoch": 0.186290234610148,
-      "grad_norm": 1321.56884765625,
-      "learning_rate": 4.892186118992362e-05,
-      "loss": 118.3066,
-      "step": 46110
-    },
-    {
-      "epoch": 0.18633063587551563,
-      "grad_norm": 916.1724243164062,
-      "learning_rate": 4.892084691418947e-05,
-      "loss": 99.126,
-      "step": 46120
-    },
-    {
-      "epoch": 0.18637103714088324,
-      "grad_norm": 1164.2841796875,
-      "learning_rate": 4.891983217210607e-05,
-      "loss": 66.5533,
-      "step": 46130
-    },
-    {
-      "epoch": 0.18641143840625088,
-      "grad_norm": 589.622314453125,
-      "learning_rate": 4.891881696369325e-05,
-      "loss": 76.5307,
-      "step": 46140
-    },
-    {
-      "epoch": 0.18645183967161852,
-      "grad_norm": 1253.39306640625,
-      "learning_rate": 4.891780128897077e-05,
-      "loss": 72.5351,
-      "step": 46150
-    },
-    {
-      "epoch": 0.18649224093698616,
-      "grad_norm": 1563.49658203125,
-      "learning_rate": 4.891678514795843e-05,
-      "loss": 93.3726,
-      "step": 46160
-    },
-    {
-      "epoch": 0.18653264220235377,
-      "grad_norm": 3200.355712890625,
-      "learning_rate": 4.891576854067607e-05,
-      "loss": 124.7522,
-      "step": 46170
-    },
-    {
-      "epoch": 0.1865730434677214,
-      "grad_norm": 897.306640625,
-      "learning_rate": 4.891475146714347e-05,
-      "loss": 68.6371,
-      "step": 46180
-    },
-    {
-      "epoch": 0.18661344473308905,
-      "grad_norm": 992.8071899414062,
-      "learning_rate": 4.891373392738049e-05,
-      "loss": 80.4052,
-      "step": 46190
-    },
-    {
-      "epoch": 0.18665384599845666,
-      "grad_norm": 947.232421875,
-      "learning_rate": 4.891271592140695e-05,
-      "loss": 92.7935,
-      "step": 46200
-    },
-    {
-      "epoch": 0.1866942472638243,
-      "grad_norm": 1065.53759765625,
-      "learning_rate": 4.891169744924271e-05,
-      "loss": 180.0239,
-      "step": 46210
-    },
-    {
-      "epoch": 0.18673464852919194,
-      "grad_norm": 960.1724243164062,
-      "learning_rate": 4.8910678510907606e-05,
-      "loss": 61.9903,
-      "step": 46220
-    },
-    {
-      "epoch": 0.18677504979455956,
-      "grad_norm": 1468.2706298828125,
-      "learning_rate": 4.890965910642152e-05,
-      "loss": 79.8836,
-      "step": 46230
-    },
-    {
-      "epoch": 0.1868154510599272,
-      "grad_norm": 0.0,
-      "learning_rate": 4.8908639235804324e-05,
-      "loss": 87.0793,
-      "step": 46240
-    },
-    {
-      "epoch": 0.18685585232529484,
-      "grad_norm": 0.0,
-      "learning_rate": 4.890761889907589e-05,
-      "loss": 78.9375,
-      "step": 46250
-    },
-    {
-      "epoch": 0.18689625359066245,
-      "grad_norm": 1007.9310302734375,
-      "learning_rate": 4.890659809625612e-05,
-      "loss": 102.4231,
-      "step": 46260
-    },
-    {
-      "epoch": 0.1869366548560301,
-      "grad_norm": 794.476318359375,
-      "learning_rate": 4.890557682736491e-05,
-      "loss": 84.0354,
-      "step": 46270
-    },
-    {
-      "epoch": 0.18697705612139773,
-      "grad_norm": 530.2271118164062,
-      "learning_rate": 4.890455509242218e-05,
-      "loss": 68.9714,
-      "step": 46280
-    },
-    {
-      "epoch": 0.18701745738676534,
-      "grad_norm": 1666.1671142578125,
-      "learning_rate": 4.8903532891447836e-05,
-      "loss": 84.9628,
-      "step": 46290
-    },
-    {
-      "epoch": 0.18705785865213298,
-      "grad_norm": 787.04833984375,
-      "learning_rate": 4.890251022446181e-05,
-      "loss": 75.6712,
-      "step": 46300
-    },
-    {
-      "epoch": 0.18709825991750062,
-      "grad_norm": 2045.2965087890625,
-      "learning_rate": 4.890148709148404e-05,
-      "loss": 98.4277,
-      "step": 46310
-    },
-    {
-      "epoch": 0.18713866118286826,
-      "grad_norm": 949.1669311523438,
-      "learning_rate": 4.890046349253448e-05,
-      "loss": 63.0205,
-      "step": 46320
-    },
-    {
-      "epoch": 0.18717906244823587,
-      "grad_norm": 578.8233642578125,
-      "learning_rate": 4.8899439427633076e-05,
-      "loss": 47.0746,
-      "step": 46330
-    },
-    {
-      "epoch": 0.1872194637136035,
-      "grad_norm": 406.12615966796875,
-      "learning_rate": 4.88984148967998e-05,
-      "loss": 53.2635,
-      "step": 46340
-    },
-    {
-      "epoch": 0.18725986497897115,
-      "grad_norm": 1593.3697509765625,
-      "learning_rate": 4.889738990005462e-05,
-      "loss": 93.6755,
-      "step": 46350
-    },
-    {
-      "epoch": 0.18730026624433876,
-      "grad_norm": 722.7239990234375,
-      "learning_rate": 4.889636443741752e-05,
-      "loss": 77.3317,
-      "step": 46360
-    },
-    {
-      "epoch": 0.1873406675097064,
-      "grad_norm": 2375.264892578125,
-      "learning_rate": 4.88953385089085e-05,
-      "loss": 88.7548,
-      "step": 46370
-    },
-    {
-      "epoch": 0.18738106877507404,
-      "grad_norm": 641.643798828125,
-      "learning_rate": 4.8894312114547535e-05,
-      "loss": 95.3137,
-      "step": 46380
-    },
-    {
-      "epoch": 0.18742147004044166,
-      "grad_norm": 1006.8427734375,
-      "learning_rate": 4.889328525435467e-05,
-      "loss": 74.9918,
-      "step": 46390
-    },
-    {
-      "epoch": 0.1874618713058093,
-      "grad_norm": 638.9917602539062,
-      "learning_rate": 4.889225792834991e-05,
-      "loss": 73.9908,
-      "step": 46400
-    },
-    {
-      "epoch": 0.18750227257117694,
-      "grad_norm": 2413.532470703125,
-      "learning_rate": 4.889123013655327e-05,
-      "loss": 129.9344,
-      "step": 46410
-    },
-    {
-      "epoch": 0.18754267383654455,
-      "grad_norm": 447.9295349121094,
-      "learning_rate": 4.8890201878984796e-05,
-      "loss": 79.6906,
-      "step": 46420
-    },
-    {
-      "epoch": 0.1875830751019122,
-      "grad_norm": 1658.515380859375,
-      "learning_rate": 4.888917315566455e-05,
-      "loss": 74.6684,
-      "step": 46430
-    },
-    {
-      "epoch": 0.18762347636727983,
-      "grad_norm": 814.1004028320312,
-      "learning_rate": 4.888814396661256e-05,
-      "loss": 64.3334,
-      "step": 46440
-    },
-    {
-      "epoch": 0.18766387763264744,
-      "grad_norm": 723.90673828125,
-      "learning_rate": 4.8887114311848915e-05,
-      "loss": 98.6432,
-      "step": 46450
-    },
-    {
-      "epoch": 0.18770427889801508,
-      "grad_norm": 535.9703369140625,
-      "learning_rate": 4.8886084191393677e-05,
-      "loss": 50.3761,
-      "step": 46460
-    },
-    {
-      "epoch": 0.18774468016338272,
-      "grad_norm": 452.6556701660156,
-      "learning_rate": 4.888505360526693e-05,
-      "loss": 109.0327,
-      "step": 46470
-    },
-    {
-      "epoch": 0.18778508142875036,
-      "grad_norm": 691.0087280273438,
-      "learning_rate": 4.888402255348876e-05,
-      "loss": 101.6102,
-      "step": 46480
-    },
-    {
-      "epoch": 0.18782548269411797,
-      "grad_norm": 391.9655456542969,
-      "learning_rate": 4.888299103607928e-05,
-      "loss": 82.483,
-      "step": 46490
-    },
-    {
-      "epoch": 0.1878658839594856,
-      "grad_norm": 898.39892578125,
-      "learning_rate": 4.888195905305859e-05,
-      "loss": 58.2394,
-      "step": 46500
-    },
-    {
-      "epoch": 0.18790628522485325,
-      "grad_norm": 902.1316528320312,
-      "learning_rate": 4.888092660444682e-05,
-      "loss": 48.3732,
-      "step": 46510
-    },
-    {
-      "epoch": 0.18794668649022087,
-      "grad_norm": 873.6351928710938,
-      "learning_rate": 4.887989369026409e-05,
-      "loss": 55.4155,
-      "step": 46520
-    },
-    {
-      "epoch": 0.1879870877555885,
-      "grad_norm": 684.0804443359375,
-      "learning_rate": 4.887886031053053e-05,
-      "loss": 80.0627,
-      "step": 46530
-    },
-    {
-      "epoch": 0.18802748902095615,
-      "grad_norm": 932.9844360351562,
-      "learning_rate": 4.887782646526631e-05,
-      "loss": 62.5363,
-      "step": 46540
-    },
-    {
-      "epoch": 0.18806789028632376,
-      "grad_norm": 676.0230102539062,
-      "learning_rate": 4.8876792154491556e-05,
-      "loss": 67.6069,
-      "step": 46550
-    },
-    {
-      "epoch": 0.1881082915516914,
-      "grad_norm": 673.8368530273438,
-      "learning_rate": 4.887575737822645e-05,
-      "loss": 62.4173,
-      "step": 46560
-    },
-    {
-      "epoch": 0.18814869281705904,
-      "grad_norm": 1231.91943359375,
-      "learning_rate": 4.8874722136491155e-05,
-      "loss": 58.3462,
-      "step": 46570
-    },
-    {
-      "epoch": 0.18818909408242665,
-      "grad_norm": 669.4678344726562,
-      "learning_rate": 4.887368642930588e-05,
-      "loss": 140.696,
-      "step": 46580
-    },
-    {
-      "epoch": 0.1882294953477943,
-      "grad_norm": 797.5696411132812,
-      "learning_rate": 4.887265025669078e-05,
-      "loss": 68.3853,
-      "step": 46590
-    },
-    {
-      "epoch": 0.18826989661316193,
-      "grad_norm": 722.3099975585938,
-      "learning_rate": 4.887161361866608e-05,
-      "loss": 101.0911,
-      "step": 46600
-    },
-    {
-      "epoch": 0.18831029787852954,
-      "grad_norm": 1002.6024169921875,
-      "learning_rate": 4.887057651525198e-05,
-      "loss": 109.1899,
-      "step": 46610
-    },
-    {
-      "epoch": 0.18835069914389718,
-      "grad_norm": 950.6636352539062,
-      "learning_rate": 4.8869538946468694e-05,
-      "loss": 59.6831,
-      "step": 46620
-    },
-    {
-      "epoch": 0.18839110040926482,
-      "grad_norm": 1075.5361328125,
-      "learning_rate": 4.8868500912336465e-05,
-      "loss": 73.4534,
-      "step": 46630
-    },
-    {
-      "epoch": 0.18843150167463246,
-      "grad_norm": 371.59368896484375,
-      "learning_rate": 4.8867462412875526e-05,
-      "loss": 94.6471,
-      "step": 46640
-    },
-    {
-      "epoch": 0.18847190294000007,
-      "grad_norm": 1247.0185546875,
-      "learning_rate": 4.886642344810611e-05,
-      "loss": 86.5103,
-      "step": 46650
-    },
-    {
-      "epoch": 0.18851230420536771,
-      "grad_norm": 86.08003234863281,
-      "learning_rate": 4.8865384018048494e-05,
-      "loss": 77.3341,
-      "step": 46660
-    },
-    {
-      "epoch": 0.18855270547073535,
-      "grad_norm": 517.982421875,
-      "learning_rate": 4.886434412272293e-05,
-      "loss": 58.2183,
-      "step": 46670
-    },
-    {
-      "epoch": 0.18859310673610297,
-      "grad_norm": 772.2702026367188,
-      "learning_rate": 4.886330376214968e-05,
-      "loss": 90.4057,
-      "step": 46680
-    },
-    {
-      "epoch": 0.1886335080014706,
-      "grad_norm": 827.40234375,
-      "learning_rate": 4.886226293634904e-05,
-      "loss": 62.3181,
-      "step": 46690
-    },
-    {
-      "epoch": 0.18867390926683825,
-      "grad_norm": 662.444580078125,
-      "learning_rate": 4.886122164534131e-05,
-      "loss": 72.7424,
-      "step": 46700
-    },
-    {
-      "epoch": 0.18871431053220586,
-      "grad_norm": 423.5435791015625,
-      "learning_rate": 4.886017988914676e-05,
-      "loss": 132.9958,
-      "step": 46710
-    },
-    {
-      "epoch": 0.1887547117975735,
-      "grad_norm": 773.7642822265625,
-      "learning_rate": 4.8859137667785735e-05,
-      "loss": 112.7814,
-      "step": 46720
-    },
-    {
-      "epoch": 0.18879511306294114,
-      "grad_norm": 942.0858154296875,
-      "learning_rate": 4.8858094981278524e-05,
-      "loss": 74.1822,
-      "step": 46730
-    },
-    {
-      "epoch": 0.18883551432830875,
-      "grad_norm": 623.146484375,
-      "learning_rate": 4.8857051829645485e-05,
-      "loss": 67.047,
-      "step": 46740
-    },
-    {
-      "epoch": 0.1888759155936764,
-      "grad_norm": 2191.332275390625,
-      "learning_rate": 4.8856008212906925e-05,
-      "loss": 74.4607,
-      "step": 46750
-    },
-    {
-      "epoch": 0.18891631685904403,
-      "grad_norm": 796.4672241210938,
-      "learning_rate": 4.88549641310832e-05,
-      "loss": 65.4805,
-      "step": 46760
-    },
-    {
-      "epoch": 0.18895671812441164,
-      "grad_norm": 897.36181640625,
-      "learning_rate": 4.885391958419468e-05,
-      "loss": 104.9449,
-      "step": 46770
-    },
-    {
-      "epoch": 0.18899711938977928,
-      "grad_norm": 4131.201171875,
-      "learning_rate": 4.885287457226172e-05,
-      "loss": 109.5929,
-      "step": 46780
-    },
-    {
-      "epoch": 0.18903752065514692,
-      "grad_norm": 1081.7864990234375,
-      "learning_rate": 4.885182909530468e-05,
-      "loss": 101.8282,
-      "step": 46790
-    },
-    {
-      "epoch": 0.18907792192051456,
-      "grad_norm": 761.68408203125,
-      "learning_rate": 4.885078315334395e-05,
-      "loss": 59.1006,
-      "step": 46800
-    },
-    {
-      "epoch": 0.18911832318588218,
-      "grad_norm": 1586.106689453125,
-      "learning_rate": 4.884973674639993e-05,
-      "loss": 71.1058,
-      "step": 46810
-    },
-    {
-      "epoch": 0.18915872445124982,
-      "grad_norm": 1164.787841796875,
-      "learning_rate": 4.884868987449301e-05,
-      "loss": 64.9034,
-      "step": 46820
-    },
-    {
-      "epoch": 0.18919912571661746,
-      "grad_norm": 629.8527221679688,
-      "learning_rate": 4.8847642537643604e-05,
-      "loss": 71.8605,
-      "step": 46830
-    },
-    {
-      "epoch": 0.18923952698198507,
-      "grad_norm": 824.6557006835938,
-      "learning_rate": 4.884659473587213e-05,
-      "loss": 91.3435,
-      "step": 46840
-    },
-    {
-      "epoch": 0.1892799282473527,
-      "grad_norm": 213.20458984375,
-      "learning_rate": 4.884554646919901e-05,
-      "loss": 64.7744,
-      "step": 46850
-    },
-    {
-      "epoch": 0.18932032951272035,
-      "grad_norm": 957.3148803710938,
-      "learning_rate": 4.884449773764469e-05,
-      "loss": 68.4271,
-      "step": 46860
-    },
-    {
-      "epoch": 0.18936073077808796,
-      "grad_norm": 1541.1634521484375,
-      "learning_rate": 4.884344854122961e-05,
-      "loss": 93.5629,
-      "step": 46870
-    },
-    {
-      "epoch": 0.1894011320434556,
-      "grad_norm": 1493.8668212890625,
-      "learning_rate": 4.884239887997423e-05,
-      "loss": 80.8613,
-      "step": 46880
-    },
-    {
-      "epoch": 0.18944153330882324,
-      "grad_norm": 678.560546875,
-      "learning_rate": 4.8841348753899e-05,
-      "loss": 87.8491,
-      "step": 46890
-    },
-    {
-      "epoch": 0.18948193457419085,
-      "grad_norm": 711.7306518554688,
-      "learning_rate": 4.88402981630244e-05,
-      "loss": 93.6465,
-      "step": 46900
-    },
-    {
-      "epoch": 0.1895223358395585,
-      "grad_norm": 554.8831176757812,
-      "learning_rate": 4.883924710737092e-05,
-      "loss": 73.9221,
-      "step": 46910
-    },
-    {
-      "epoch": 0.18956273710492613,
-      "grad_norm": 1330.556884765625,
-      "learning_rate": 4.8838195586959046e-05,
-      "loss": 79.4741,
-      "step": 46920
-    },
-    {
-      "epoch": 0.18960313837029374,
-      "grad_norm": 598.888427734375,
-      "learning_rate": 4.883714360180927e-05,
-      "loss": 66.7534,
-      "step": 46930
-    },
-    {
-      "epoch": 0.18964353963566138,
-      "grad_norm": 614.5370483398438,
-      "learning_rate": 4.883609115194211e-05,
-      "loss": 90.6562,
-      "step": 46940
-    },
-    {
-      "epoch": 0.18968394090102902,
-      "grad_norm": 1026.0899658203125,
-      "learning_rate": 4.883503823737808e-05,
-      "loss": 86.146,
-      "step": 46950
-    },
-    {
-      "epoch": 0.18972434216639666,
-      "grad_norm": 1076.3065185546875,
-      "learning_rate": 4.8833984858137715e-05,
-      "loss": 74.9594,
-      "step": 46960
-    },
-    {
-      "epoch": 0.18976474343176428,
-      "grad_norm": 555.6848754882812,
-      "learning_rate": 4.8832931014241534e-05,
-      "loss": 69.4619,
-      "step": 46970
-    },
-    {
-      "epoch": 0.18980514469713192,
-      "grad_norm": 1051.6075439453125,
-      "learning_rate": 4.88318767057101e-05,
-      "loss": 67.6875,
-      "step": 46980
-    },
-    {
-      "epoch": 0.18984554596249956,
-      "grad_norm": 1332.5283203125,
-      "learning_rate": 4.883082193256397e-05,
-      "loss": 73.1094,
-      "step": 46990
-    },
-    {
-      "epoch": 0.18988594722786717,
-      "grad_norm": 1304.3714599609375,
-      "learning_rate": 4.882976669482367e-05,
-      "loss": 91.2835,
-      "step": 47000
-    },
-    {
-      "epoch": 0.1899263484932348,
-      "grad_norm": 1050.1197509765625,
-      "learning_rate": 4.882871099250982e-05,
-      "loss": 96.9666,
-      "step": 47010
-    },
-    {
-      "epoch": 0.18996674975860245,
-      "grad_norm": 2204.67626953125,
-      "learning_rate": 4.882765482564298e-05,
-      "loss": 79.6932,
-      "step": 47020
-    },
-    {
-      "epoch": 0.19000715102397006,
-      "grad_norm": 308.5193786621094,
-      "learning_rate": 4.882659819424374e-05,
-      "loss": 60.4664,
-      "step": 47030
-    },
-    {
-      "epoch": 0.1900475522893377,
-      "grad_norm": 760.51953125,
-      "learning_rate": 4.8825541098332706e-05,
-      "loss": 93.623,
-      "step": 47040
-    },
-    {
-      "epoch": 0.19008795355470534,
-      "grad_norm": 1053.7843017578125,
-      "learning_rate": 4.882448353793048e-05,
-      "loss": 72.0362,
-      "step": 47050
-    },
-    {
-      "epoch": 0.19012835482007295,
-      "grad_norm": 510.9100341796875,
-      "learning_rate": 4.8823425513057674e-05,
-      "loss": 76.0767,
-      "step": 47060
-    },
-    {
-      "epoch": 0.1901687560854406,
-      "grad_norm": 1290.3902587890625,
-      "learning_rate": 4.8822367023734925e-05,
-      "loss": 76.7432,
-      "step": 47070
-    },
-    {
-      "epoch": 0.19020915735080823,
-      "grad_norm": 669.0762939453125,
-      "learning_rate": 4.8821308069982867e-05,
-      "loss": 72.9376,
-      "step": 47080
-    },
-    {
-      "epoch": 0.19024955861617585,
-      "grad_norm": 708.4724731445312,
-      "learning_rate": 4.8820248651822145e-05,
-      "loss": 74.2905,
-      "step": 47090
-    },
-    {
-      "epoch": 0.19028995988154349,
-      "grad_norm": 810.4485473632812,
-      "learning_rate": 4.8819188769273414e-05,
-      "loss": 85.8345,
-      "step": 47100
-    },
-    {
-      "epoch": 0.19033036114691113,
-      "grad_norm": 1206.7943115234375,
-      "learning_rate": 4.8818128422357335e-05,
-      "loss": 62.0077,
-      "step": 47110
-    },
-    {
-      "epoch": 0.19037076241227877,
-      "grad_norm": 763.1806030273438,
-      "learning_rate": 4.881706761109458e-05,
-      "loss": 80.4839,
-      "step": 47120
-    },
-    {
-      "epoch": 0.19041116367764638,
-      "grad_norm": 1197.156005859375,
-      "learning_rate": 4.8816006335505825e-05,
-      "loss": 130.0592,
-      "step": 47130
-    },
-    {
-      "epoch": 0.19045156494301402,
-      "grad_norm": 522.46484375,
-      "learning_rate": 4.8814944595611776e-05,
-      "loss": 71.9689,
-      "step": 47140
-    },
-    {
-      "epoch": 0.19049196620838166,
-      "grad_norm": 653.2218627929688,
-      "learning_rate": 4.881388239143311e-05,
-      "loss": 88.4613,
-      "step": 47150
-    },
-    {
-      "epoch": 0.19053236747374927,
-      "grad_norm": 696.4855346679688,
-      "learning_rate": 4.881281972299055e-05,
-      "loss": 99.2132,
-      "step": 47160
-    },
-    {
-      "epoch": 0.1905727687391169,
-      "grad_norm": 779.6743774414062,
-      "learning_rate": 4.8811756590304815e-05,
-      "loss": 54.9467,
-      "step": 47170
-    },
-    {
-      "epoch": 0.19061317000448455,
-      "grad_norm": 489.7877197265625,
-      "learning_rate": 4.881069299339662e-05,
-      "loss": 98.4642,
-      "step": 47180
-    },
-    {
-      "epoch": 0.19065357126985216,
-      "grad_norm": 1307.5743408203125,
-      "learning_rate": 4.880962893228671e-05,
-      "loss": 77.5308,
-      "step": 47190
-    },
-    {
-      "epoch": 0.1906939725352198,
-      "grad_norm": 1113.280517578125,
-      "learning_rate": 4.880856440699582e-05,
-      "loss": 65.3198,
-      "step": 47200
-    },
-    {
-      "epoch": 0.19073437380058744,
-      "grad_norm": 859.1810913085938,
-      "learning_rate": 4.880749941754471e-05,
-      "loss": 119.8937,
-      "step": 47210
-    },
-    {
-      "epoch": 0.19077477506595505,
-      "grad_norm": 676.0623779296875,
-      "learning_rate": 4.8806433963954154e-05,
-      "loss": 88.6286,
-      "step": 47220
-    },
-    {
-      "epoch": 0.1908151763313227,
-      "grad_norm": 922.7902221679688,
-      "learning_rate": 4.880536804624491e-05,
-      "loss": 102.1759,
-      "step": 47230
-    },
-    {
-      "epoch": 0.19085557759669033,
-      "grad_norm": 1272.20068359375,
-      "learning_rate": 4.880430166443775e-05,
-      "loss": 107.2287,
-      "step": 47240
-    },
-    {
-      "epoch": 0.19089597886205795,
-      "grad_norm": 1099.0933837890625,
-      "learning_rate": 4.880323481855347e-05,
-      "loss": 71.5676,
-      "step": 47250
-    },
-    {
-      "epoch": 0.1909363801274256,
-      "grad_norm": 958.9481201171875,
-      "learning_rate": 4.880216750861288e-05,
-      "loss": 51.0081,
-      "step": 47260
-    },
-    {
-      "epoch": 0.19097678139279323,
-      "grad_norm": 850.7756958007812,
-      "learning_rate": 4.880109973463678e-05,
-      "loss": 102.9255,
-      "step": 47270
-    },
-    {
-      "epoch": 0.19101718265816087,
-      "grad_norm": 522.798095703125,
-      "learning_rate": 4.880003149664599e-05,
-      "loss": 73.2628,
-      "step": 47280
-    },
-    {
-      "epoch": 0.19105758392352848,
-      "grad_norm": 1138.0860595703125,
-      "learning_rate": 4.879896279466133e-05,
-      "loss": 93.8309,
-      "step": 47290
-    },
-    {
-      "epoch": 0.19109798518889612,
-      "grad_norm": 877.7420654296875,
-      "learning_rate": 4.8797893628703635e-05,
-      "loss": 68.0124,
-      "step": 47300
-    },
-    {
-      "epoch": 0.19113838645426376,
-      "grad_norm": 372.2701416015625,
-      "learning_rate": 4.879682399879375e-05,
-      "loss": 83.2249,
-      "step": 47310
-    },
-    {
-      "epoch": 0.19117878771963137,
-      "grad_norm": 600.6828002929688,
-      "learning_rate": 4.8795753904952534e-05,
-      "loss": 69.8246,
-      "step": 47320
-    },
-    {
-      "epoch": 0.191219188984999,
-      "grad_norm": 1143.0029296875,
-      "learning_rate": 4.879468334720085e-05,
-      "loss": 61.1958,
-      "step": 47330
-    },
-    {
-      "epoch": 0.19125959025036665,
-      "grad_norm": 1139.1126708984375,
-      "learning_rate": 4.879361232555956e-05,
-      "loss": 79.5726,
-      "step": 47340
-    },
-    {
-      "epoch": 0.19129999151573426,
-      "grad_norm": 1177.59033203125,
-      "learning_rate": 4.879254084004955e-05,
-      "loss": 84.1079,
-      "step": 47350
-    },
-    {
-      "epoch": 0.1913403927811019,
-      "grad_norm": 643.5814819335938,
-      "learning_rate": 4.8791468890691696e-05,
-      "loss": 89.2118,
-      "step": 47360
-    },
-    {
-      "epoch": 0.19138079404646954,
-      "grad_norm": 273.9512939453125,
-      "learning_rate": 4.879039647750692e-05,
-      "loss": 72.0723,
-      "step": 47370
-    },
-    {
-      "epoch": 0.19142119531183716,
-      "grad_norm": 670.9833984375,
-      "learning_rate": 4.8789323600516104e-05,
-      "loss": 79.4604,
-      "step": 47380
-    },
-    {
-      "epoch": 0.1914615965772048,
-      "grad_norm": 573.1339721679688,
-      "learning_rate": 4.8788250259740185e-05,
-      "loss": 57.4144,
-      "step": 47390
-    },
-    {
-      "epoch": 0.19150199784257244,
-      "grad_norm": 746.1387939453125,
-      "learning_rate": 4.878717645520008e-05,
-      "loss": 71.1703,
-      "step": 47400
-    },
-    {
-      "epoch": 0.19154239910794005,
-      "grad_norm": 770.6351318359375,
-      "learning_rate": 4.878610218691673e-05,
-      "loss": 71.4196,
-      "step": 47410
-    },
-    {
-      "epoch": 0.1915828003733077,
-      "grad_norm": 419.6976318359375,
-      "learning_rate": 4.878502745491106e-05,
-      "loss": 55.445,
-      "step": 47420
-    },
-    {
-      "epoch": 0.19162320163867533,
-      "grad_norm": 4431.18115234375,
-      "learning_rate": 4.8783952259204036e-05,
-      "loss": 75.606,
-      "step": 47430
-    },
-    {
-      "epoch": 0.19166360290404297,
-      "grad_norm": 534.2245483398438,
-      "learning_rate": 4.878287659981662e-05,
-      "loss": 74.2982,
-      "step": 47440
-    },
-    {
-      "epoch": 0.19170400416941058,
-      "grad_norm": 1988.7900390625,
-      "learning_rate": 4.878180047676978e-05,
-      "loss": 100.465,
-      "step": 47450
-    },
-    {
-      "epoch": 0.19174440543477822,
-      "grad_norm": 605.2179565429688,
-      "learning_rate": 4.87807238900845e-05,
-      "loss": 54.1781,
-      "step": 47460
-    },
-    {
-      "epoch": 0.19178480670014586,
-      "grad_norm": 1887.808837890625,
-      "learning_rate": 4.8779646839781765e-05,
-      "loss": 103.9908,
-      "step": 47470
-    },
-    {
-      "epoch": 0.19182520796551347,
-      "grad_norm": 1397.8228759765625,
-      "learning_rate": 4.877856932588257e-05,
-      "loss": 112.4378,
-      "step": 47480
-    },
-    {
-      "epoch": 0.1918656092308811,
-      "grad_norm": 1550.006103515625,
-      "learning_rate": 4.877749134840792e-05,
-      "loss": 104.1645,
-      "step": 47490
-    },
-    {
-      "epoch": 0.19190601049624875,
-      "grad_norm": 555.1295166015625,
-      "learning_rate": 4.877641290737884e-05,
-      "loss": 61.3664,
-      "step": 47500
-    },
-    {
-      "epoch": 0.19194641176161636,
-      "grad_norm": 1102.827392578125,
-      "learning_rate": 4.877533400281635e-05,
-      "loss": 119.7094,
-      "step": 47510
-    },
-    {
-      "epoch": 0.191986813026984,
-      "grad_norm": 890.612548828125,
-      "learning_rate": 4.877425463474148e-05,
-      "loss": 75.4402,
-      "step": 47520
-    },
-    {
-      "epoch": 0.19202721429235164,
-      "grad_norm": 586.08056640625,
-      "learning_rate": 4.877317480317528e-05,
-      "loss": 87.8783,
-      "step": 47530
-    },
-    {
-      "epoch": 0.19206761555771926,
-      "grad_norm": 574.8908081054688,
-      "learning_rate": 4.8772094508138796e-05,
-      "loss": 88.9533,
-      "step": 47540
-    },
-    {
-      "epoch": 0.1921080168230869,
-      "grad_norm": 761.5361938476562,
-      "learning_rate": 4.877101374965308e-05,
-      "loss": 62.7146,
-      "step": 47550
-    },
-    {
-      "epoch": 0.19214841808845454,
-      "grad_norm": 709.9255981445312,
-      "learning_rate": 4.8769932527739225e-05,
-      "loss": 78.0876,
-      "step": 47560
-    },
-    {
-      "epoch": 0.19218881935382215,
-      "grad_norm": 1276.592529296875,
-      "learning_rate": 4.87688508424183e-05,
-      "loss": 74.2328,
-      "step": 47570
-    },
-    {
-      "epoch": 0.1922292206191898,
-      "grad_norm": 1980.7711181640625,
-      "learning_rate": 4.876776869371139e-05,
-      "loss": 84.7106,
-      "step": 47580
-    },
-    {
-      "epoch": 0.19226962188455743,
-      "grad_norm": 1390.198486328125,
-      "learning_rate": 4.876668608163959e-05,
-      "loss": 89.8972,
-      "step": 47590
-    },
-    {
-      "epoch": 0.19231002314992507,
-      "grad_norm": 413.9287109375,
-      "learning_rate": 4.8765603006224006e-05,
-      "loss": 95.8701,
-      "step": 47600
-    },
-    {
-      "epoch": 0.19235042441529268,
-      "grad_norm": 1712.273193359375,
-      "learning_rate": 4.876451946748576e-05,
-      "loss": 123.683,
-      "step": 47610
-    },
-    {
-      "epoch": 0.19239082568066032,
-      "grad_norm": 1209.84423828125,
-      "learning_rate": 4.8763435465445964e-05,
-      "loss": 102.8881,
-      "step": 47620
-    },
-    {
-      "epoch": 0.19243122694602796,
-      "grad_norm": 919.6061401367188,
-      "learning_rate": 4.8762351000125766e-05,
-      "loss": 83.7829,
-      "step": 47630
-    },
-    {
-      "epoch": 0.19247162821139557,
-      "grad_norm": 938.0872192382812,
-      "learning_rate": 4.87612660715463e-05,
-      "loss": 43.0572,
-      "step": 47640
-    },
-    {
-      "epoch": 0.1925120294767632,
-      "grad_norm": 1763.4127197265625,
-      "learning_rate": 4.876018067972872e-05,
-      "loss": 91.6115,
-      "step": 47650
-    },
-    {
-      "epoch": 0.19255243074213085,
-      "grad_norm": 2353.800537109375,
-      "learning_rate": 4.8759094824694184e-05,
-      "loss": 79.1209,
-      "step": 47660
-    },
-    {
-      "epoch": 0.19259283200749847,
-      "grad_norm": 1034.469482421875,
-      "learning_rate": 4.875800850646387e-05,
-      "loss": 63.04,
-      "step": 47670
-    },
-    {
-      "epoch": 0.1926332332728661,
-      "grad_norm": 497.1041564941406,
-      "learning_rate": 4.8756921725058934e-05,
-      "loss": 78.9805,
-      "step": 47680
-    },
-    {
-      "epoch": 0.19267363453823375,
-      "grad_norm": 410.94970703125,
-      "learning_rate": 4.875583448050059e-05,
-      "loss": 77.002,
-      "step": 47690
-    },
-    {
-      "epoch": 0.19271403580360136,
-      "grad_norm": 1801.6767578125,
-      "learning_rate": 4.875474677281002e-05,
-      "loss": 80.726,
-      "step": 47700
-    },
-    {
-      "epoch": 0.192754437068969,
-      "grad_norm": 946.7315673828125,
-      "learning_rate": 4.8753658602008425e-05,
-      "loss": 64.131,
-      "step": 47710
-    },
-    {
-      "epoch": 0.19279483833433664,
-      "grad_norm": 1104.990478515625,
-      "learning_rate": 4.875256996811703e-05,
-      "loss": 63.1742,
-      "step": 47720
-    },
-    {
-      "epoch": 0.19283523959970425,
-      "grad_norm": 889.265869140625,
-      "learning_rate": 4.875148087115706e-05,
-      "loss": 121.2112,
-      "step": 47730
-    },
-    {
-      "epoch": 0.1928756408650719,
-      "grad_norm": 502.1475830078125,
-      "learning_rate": 4.875039131114975e-05,
-      "loss": 81.0771,
-      "step": 47740
-    },
-    {
-      "epoch": 0.19291604213043953,
-      "grad_norm": 2143.898193359375,
-      "learning_rate": 4.874930128811631e-05,
-      "loss": 87.0424,
-      "step": 47750
-    },
-    {
-      "epoch": 0.19295644339580717,
-      "grad_norm": 792.10693359375,
-      "learning_rate": 4.874821080207803e-05,
-      "loss": 73.0605,
-      "step": 47760
-    },
-    {
-      "epoch": 0.19299684466117478,
-      "grad_norm": 483.91546630859375,
-      "learning_rate": 4.8747119853056156e-05,
-      "loss": 88.888,
-      "step": 47770
-    },
-    {
-      "epoch": 0.19303724592654242,
-      "grad_norm": 1531.7535400390625,
-      "learning_rate": 4.8746028441071943e-05,
-      "loss": 59.7669,
-      "step": 47780
-    },
-    {
-      "epoch": 0.19307764719191006,
-      "grad_norm": 677.5689086914062,
-      "learning_rate": 4.874493656614669e-05,
-      "loss": 74.3335,
-      "step": 47790
-    },
-    {
-      "epoch": 0.19311804845727767,
-      "grad_norm": 515.4765625,
-      "learning_rate": 4.874384422830167e-05,
-      "loss": 56.0753,
-      "step": 47800
-    },
-    {
-      "epoch": 0.19315844972264531,
-      "grad_norm": 763.7691040039062,
-      "learning_rate": 4.8742751427558186e-05,
-      "loss": 77.0294,
-      "step": 47810
-    },
-    {
-      "epoch": 0.19319885098801295,
-      "grad_norm": 1666.2574462890625,
-      "learning_rate": 4.874165816393754e-05,
-      "loss": 46.0998,
-      "step": 47820
-    },
-    {
-      "epoch": 0.19323925225338057,
-      "grad_norm": 753.572021484375,
-      "learning_rate": 4.874056443746104e-05,
-      "loss": 55.663,
-      "step": 47830
-    },
-    {
-      "epoch": 0.1932796535187482,
-      "grad_norm": 664.1289672851562,
-      "learning_rate": 4.873947024815002e-05,
-      "loss": 145.2848,
-      "step": 47840
-    },
-    {
-      "epoch": 0.19332005478411585,
-      "grad_norm": 910.7630615234375,
-      "learning_rate": 4.87383755960258e-05,
-      "loss": 100.3147,
-      "step": 47850
-    },
-    {
-      "epoch": 0.19336045604948346,
-      "grad_norm": 1756.9093017578125,
-      "learning_rate": 4.8737280481109724e-05,
-      "loss": 103.6948,
-      "step": 47860
-    },
-    {
-      "epoch": 0.1934008573148511,
-      "grad_norm": 587.9384155273438,
-      "learning_rate": 4.8736184903423155e-05,
-      "loss": 90.1571,
-      "step": 47870
-    },
-    {
-      "epoch": 0.19344125858021874,
-      "grad_norm": 852.3531494140625,
-      "learning_rate": 4.873508886298743e-05,
-      "loss": 72.187,
-      "step": 47880
-    },
-    {
-      "epoch": 0.19348165984558635,
-      "grad_norm": 844.1437377929688,
-      "learning_rate": 4.8733992359823936e-05,
-      "loss": 76.1646,
-      "step": 47890
-    },
-    {
-      "epoch": 0.193522061110954,
-      "grad_norm": 1602.298095703125,
-      "learning_rate": 4.8732895393954036e-05,
-      "loss": 84.8224,
-      "step": 47900
-    },
-    {
-      "epoch": 0.19356246237632163,
-      "grad_norm": 870.3892211914062,
-      "learning_rate": 4.8731797965399125e-05,
-      "loss": 115.4123,
-      "step": 47910
-    },
-    {
-      "epoch": 0.19360286364168927,
-      "grad_norm": 1025.730224609375,
-      "learning_rate": 4.873070007418059e-05,
-      "loss": 74.2097,
-      "step": 47920
-    },
-    {
-      "epoch": 0.19364326490705688,
-      "grad_norm": 1289.6861572265625,
-      "learning_rate": 4.8729601720319845e-05,
-      "loss": 112.8718,
-      "step": 47930
-    },
-    {
-      "epoch": 0.19368366617242452,
-      "grad_norm": 592.2373657226562,
-      "learning_rate": 4.8728502903838295e-05,
-      "loss": 63.9621,
-      "step": 47940
-    },
-    {
-      "epoch": 0.19372406743779216,
-      "grad_norm": 5277.1982421875,
-      "learning_rate": 4.8727403624757365e-05,
-      "loss": 100.8859,
-      "step": 47950
-    },
-    {
-      "epoch": 0.19376446870315978,
-      "grad_norm": 816.9944458007812,
-      "learning_rate": 4.872630388309849e-05,
-      "loss": 82.8986,
-      "step": 47960
-    },
-    {
-      "epoch": 0.19380486996852742,
-      "grad_norm": 181.56996154785156,
-      "learning_rate": 4.8725203678883104e-05,
-      "loss": 65.7631,
-      "step": 47970
-    },
-    {
-      "epoch": 0.19384527123389506,
-      "grad_norm": 1124.836181640625,
-      "learning_rate": 4.872410301213265e-05,
-      "loss": 81.7168,
-      "step": 47980
-    },
-    {
-      "epoch": 0.19388567249926267,
-      "grad_norm": 1639.8275146484375,
-      "learning_rate": 4.8723001882868604e-05,
-      "loss": 103.2546,
-      "step": 47990
-    },
-    {
-      "epoch": 0.1939260737646303,
-      "grad_norm": 633.078857421875,
-      "learning_rate": 4.8721900291112415e-05,
-      "loss": 98.4941,
-      "step": 48000
-    },
-    {
-      "epoch": 0.19396647502999795,
-      "grad_norm": 766.0423583984375,
-      "learning_rate": 4.872079823688557e-05,
-      "loss": 51.5216,
-      "step": 48010
-    },
-    {
-      "epoch": 0.19400687629536556,
-      "grad_norm": 755.6162719726562,
-      "learning_rate": 4.871969572020955e-05,
-      "loss": 76.6551,
-      "step": 48020
-    },
-    {
-      "epoch": 0.1940472775607332,
-      "grad_norm": 8129.7333984375,
-      "learning_rate": 4.871859274110585e-05,
-      "loss": 132.5075,
-      "step": 48030
-    },
-    {
-      "epoch": 0.19408767882610084,
-      "grad_norm": 1120.7578125,
-      "learning_rate": 4.871748929959598e-05,
-      "loss": 87.0844,
-      "step": 48040
-    },
-    {
-      "epoch": 0.19412808009146845,
-      "grad_norm": 651.3961181640625,
-      "learning_rate": 4.8716385395701435e-05,
-      "loss": 61.7545,
-      "step": 48050
-    },
-    {
-      "epoch": 0.1941684813568361,
-      "grad_norm": 646.9969482421875,
-      "learning_rate": 4.871528102944376e-05,
-      "loss": 100.9905,
-      "step": 48060
-    },
-    {
-      "epoch": 0.19420888262220373,
-      "grad_norm": 478.8792419433594,
-      "learning_rate": 4.8714176200844464e-05,
-      "loss": 70.1561,
-      "step": 48070
-    },
-    {
-      "epoch": 0.19424928388757137,
-      "grad_norm": 617.067626953125,
-      "learning_rate": 4.8713070909925094e-05,
-      "loss": 87.2428,
-      "step": 48080
-    },
-    {
-      "epoch": 0.19428968515293898,
-      "grad_norm": 773.7765502929688,
-      "learning_rate": 4.8711965156707195e-05,
-      "loss": 78.5379,
-      "step": 48090
-    },
-    {
-      "epoch": 0.19433008641830662,
-      "grad_norm": 1457.695068359375,
-      "learning_rate": 4.871085894121233e-05,
-      "loss": 106.1761,
-      "step": 48100
-    },
-    {
-      "epoch": 0.19437048768367426,
-      "grad_norm": 1464.89013671875,
-      "learning_rate": 4.8709752263462064e-05,
-      "loss": 59.2071,
-      "step": 48110
-    },
-    {
-      "epoch": 0.19441088894904188,
-      "grad_norm": 1820.7398681640625,
-      "learning_rate": 4.870864512347797e-05,
-      "loss": 71.2924,
-      "step": 48120
-    },
-    {
-      "epoch": 0.19445129021440952,
-      "grad_norm": 596.8287353515625,
-      "learning_rate": 4.8707537521281635e-05,
-      "loss": 60.9779,
-      "step": 48130
-    },
-    {
-      "epoch": 0.19449169147977716,
-      "grad_norm": 1118.587890625,
-      "learning_rate": 4.870642945689465e-05,
-      "loss": 97.0064,
-      "step": 48140
-    },
-    {
-      "epoch": 0.19453209274514477,
-      "grad_norm": 1756.2308349609375,
-      "learning_rate": 4.8705320930338615e-05,
-      "loss": 63.2261,
-      "step": 48150
-    },
-    {
-      "epoch": 0.1945724940105124,
-      "grad_norm": 846.0660400390625,
-      "learning_rate": 4.870421194163515e-05,
-      "loss": 118.8603,
-      "step": 48160
-    },
-    {
-      "epoch": 0.19461289527588005,
-      "grad_norm": 1581.7791748046875,
-      "learning_rate": 4.8703102490805865e-05,
-      "loss": 122.0613,
-      "step": 48170
-    },
-    {
-      "epoch": 0.19465329654124766,
-      "grad_norm": 1016.5093383789062,
-      "learning_rate": 4.87019925778724e-05,
-      "loss": 60.2842,
-      "step": 48180
-    },
-    {
-      "epoch": 0.1946936978066153,
-      "grad_norm": 717.7647094726562,
-      "learning_rate": 4.870088220285638e-05,
-      "loss": 79.8649,
-      "step": 48190
-    },
-    {
-      "epoch": 0.19473409907198294,
-      "grad_norm": 1193.3416748046875,
-      "learning_rate": 4.8699771365779453e-05,
-      "loss": 60.5539,
-      "step": 48200
-    },
-    {
-      "epoch": 0.19477450033735055,
-      "grad_norm": 2215.272216796875,
-      "learning_rate": 4.8698660066663294e-05,
-      "loss": 87.0759,
-      "step": 48210
-    },
-    {
-      "epoch": 0.1948149016027182,
-      "grad_norm": 1735.73828125,
-      "learning_rate": 4.869754830552956e-05,
-      "loss": 96.5737,
-      "step": 48220
-    },
-    {
-      "epoch": 0.19485530286808583,
-      "grad_norm": 375.5717468261719,
-      "learning_rate": 4.869643608239991e-05,
-      "loss": 88.8928,
-      "step": 48230
-    },
-    {
-      "epoch": 0.19489570413345347,
-      "grad_norm": 612.527587890625,
-      "learning_rate": 4.8695323397296044e-05,
-      "loss": 66.2606,
-      "step": 48240
-    },
-    {
-      "epoch": 0.19493610539882109,
-      "grad_norm": 1690.7991943359375,
-      "learning_rate": 4.869421025023965e-05,
-      "loss": 92.1369,
-      "step": 48250
-    },
-    {
-      "epoch": 0.19497650666418873,
-      "grad_norm": 1483.668212890625,
-      "learning_rate": 4.8693096641252424e-05,
-      "loss": 65.0718,
-      "step": 48260
-    },
-    {
-      "epoch": 0.19501690792955637,
-      "grad_norm": 601.1848754882812,
-      "learning_rate": 4.8691982570356084e-05,
-      "loss": 49.7445,
-      "step": 48270
-    },
-    {
-      "epoch": 0.19505730919492398,
-      "grad_norm": 1077.0008544921875,
-      "learning_rate": 4.8690868037572346e-05,
-      "loss": 71.7525,
-      "step": 48280
-    },
-    {
-      "epoch": 0.19509771046029162,
-      "grad_norm": 1292.0035400390625,
-      "learning_rate": 4.8689753042922935e-05,
-      "loss": 75.1074,
-      "step": 48290
-    },
-    {
-      "epoch": 0.19513811172565926,
-      "grad_norm": 1112.646484375,
-      "learning_rate": 4.8688637586429595e-05,
-      "loss": 82.0207,
-      "step": 48300
-    },
-    {
-      "epoch": 0.19517851299102687,
-      "grad_norm": 1545.366943359375,
-      "learning_rate": 4.8687521668114064e-05,
-      "loss": 102.4248,
-      "step": 48310
-    },
-    {
-      "epoch": 0.1952189142563945,
-      "grad_norm": 814.1812744140625,
-      "learning_rate": 4.8686405287998116e-05,
-      "loss": 111.8968,
-      "step": 48320
-    },
-    {
-      "epoch": 0.19525931552176215,
-      "grad_norm": 632.5703735351562,
-      "learning_rate": 4.8685288446103495e-05,
-      "loss": 78.0891,
-      "step": 48330
-    },
-    {
-      "epoch": 0.19529971678712976,
-      "grad_norm": 2291.05859375,
-      "learning_rate": 4.8684171142451986e-05,
-      "loss": 63.0176,
-      "step": 48340
-    },
-    {
-      "epoch": 0.1953401180524974,
-      "grad_norm": 954.6275024414062,
-      "learning_rate": 4.8683053377065356e-05,
-      "loss": 113.7088,
-      "step": 48350
-    },
-    {
-      "epoch": 0.19538051931786504,
-      "grad_norm": 619.459228515625,
-      "learning_rate": 4.8681935149965416e-05,
-      "loss": 66.4437,
-      "step": 48360
-    },
-    {
-      "epoch": 0.19542092058323265,
-      "grad_norm": 981.1046142578125,
-      "learning_rate": 4.868081646117395e-05,
-      "loss": 81.0435,
-      "step": 48370
-    },
-    {
-      "epoch": 0.1954613218486003,
-      "grad_norm": 714.146728515625,
-      "learning_rate": 4.867969731071279e-05,
-      "loss": 101.9001,
-      "step": 48380
-    },
-    {
-      "epoch": 0.19550172311396793,
-      "grad_norm": 1931.5303955078125,
-      "learning_rate": 4.8678577698603734e-05,
-      "loss": 95.807,
-      "step": 48390
-    },
-    {
-      "epoch": 0.19554212437933557,
-      "grad_norm": 3275.3095703125,
-      "learning_rate": 4.867745762486861e-05,
-      "loss": 73.3455,
-      "step": 48400
-    },
-    {
-      "epoch": 0.1955825256447032,
-      "grad_norm": 1230.974609375,
-      "learning_rate": 4.867633708952926e-05,
-      "loss": 77.9758,
-      "step": 48410
-    },
-    {
-      "epoch": 0.19562292691007083,
-      "grad_norm": 849.453125,
-      "learning_rate": 4.867521609260754e-05,
-      "loss": 73.4305,
-      "step": 48420
-    },
-    {
-      "epoch": 0.19566332817543847,
-      "grad_norm": 903.135009765625,
-      "learning_rate": 4.867409463412528e-05,
-      "loss": 56.7117,
-      "step": 48430
-    },
-    {
-      "epoch": 0.19570372944080608,
-      "grad_norm": 1139.8895263671875,
-      "learning_rate": 4.8672972714104357e-05,
-      "loss": 67.9619,
-      "step": 48440
-    },
-    {
-      "epoch": 0.19574413070617372,
-      "grad_norm": 908.2789916992188,
-      "learning_rate": 4.867185033256665e-05,
-      "loss": 62.4522,
-      "step": 48450
-    },
-    {
-      "epoch": 0.19578453197154136,
-      "grad_norm": 1399.5394287109375,
-      "learning_rate": 4.8670727489534034e-05,
-      "loss": 113.5989,
-      "step": 48460
-    },
-    {
-      "epoch": 0.19582493323690897,
-      "grad_norm": 2880.655029296875,
-      "learning_rate": 4.8669604185028394e-05,
-      "loss": 107.0346,
-      "step": 48470
-    },
-    {
-      "epoch": 0.1958653345022766,
-      "grad_norm": 689.0923461914062,
-      "learning_rate": 4.866848041907164e-05,
-      "loss": 89.6595,
-      "step": 48480
-    },
-    {
-      "epoch": 0.19590573576764425,
-      "grad_norm": 2163.55029296875,
-      "learning_rate": 4.866735619168568e-05,
-      "loss": 87.9743,
-      "step": 48490
-    },
-    {
-      "epoch": 0.19594613703301186,
-      "grad_norm": 816.2272338867188,
-      "learning_rate": 4.8666231502892415e-05,
-      "loss": 102.1195,
-      "step": 48500
-    },
-    {
-      "epoch": 0.1959865382983795,
-      "grad_norm": 3147.00830078125,
-      "learning_rate": 4.866510635271379e-05,
-      "loss": 98.8874,
-      "step": 48510
-    },
-    {
-      "epoch": 0.19602693956374714,
-      "grad_norm": 693.9053955078125,
-      "learning_rate": 4.8663980741171724e-05,
-      "loss": 76.9899,
-      "step": 48520
-    },
-    {
-      "epoch": 0.19606734082911476,
-      "grad_norm": 833.5950317382812,
-      "learning_rate": 4.866285466828817e-05,
-      "loss": 56.6829,
-      "step": 48530
-    },
-    {
-      "epoch": 0.1961077420944824,
-      "grad_norm": 945.19873046875,
-      "learning_rate": 4.86617281340851e-05,
-      "loss": 82.8859,
-      "step": 48540
-    },
-    {
-      "epoch": 0.19614814335985004,
-      "grad_norm": 607.2847290039062,
-      "learning_rate": 4.866060113858444e-05,
-      "loss": 71.8742,
-      "step": 48550
-    },
-    {
-      "epoch": 0.19618854462521768,
-      "grad_norm": 787.0588989257812,
-      "learning_rate": 4.865947368180818e-05,
-      "loss": 81.9591,
-      "step": 48560
-    },
-    {
-      "epoch": 0.1962289458905853,
-      "grad_norm": 406.9506530761719,
-      "learning_rate": 4.865834576377831e-05,
-      "loss": 42.3592,
-      "step": 48570
-    },
-    {
-      "epoch": 0.19626934715595293,
-      "grad_norm": 525.7552490234375,
-      "learning_rate": 4.86572173845168e-05,
-      "loss": 109.6562,
-      "step": 48580
-    },
-    {
-      "epoch": 0.19630974842132057,
-      "grad_norm": 1523.6407470703125,
-      "learning_rate": 4.865608854404566e-05,
-      "loss": 133.584,
-      "step": 48590
-    },
-    {
-      "epoch": 0.19635014968668818,
-      "grad_norm": 620.2904052734375,
-      "learning_rate": 4.8654959242386896e-05,
-      "loss": 72.112,
-      "step": 48600
-    },
-    {
-      "epoch": 0.19639055095205582,
-      "grad_norm": 926.3845825195312,
-      "learning_rate": 4.865382947956253e-05,
-      "loss": 106.2173,
-      "step": 48610
-    },
-    {
-      "epoch": 0.19643095221742346,
-      "grad_norm": 1164.56982421875,
-      "learning_rate": 4.865269925559457e-05,
-      "loss": 75.1725,
-      "step": 48620
-    },
-    {
-      "epoch": 0.19647135348279107,
-      "grad_norm": 946.3316650390625,
-      "learning_rate": 4.865156857050507e-05,
-      "loss": 91.8206,
-      "step": 48630
-    },
-    {
-      "epoch": 0.1965117547481587,
-      "grad_norm": 1000.8588256835938,
-      "learning_rate": 4.865043742431605e-05,
-      "loss": 68.0882,
-      "step": 48640
-    },
-    {
-      "epoch": 0.19655215601352635,
-      "grad_norm": 1161.449951171875,
-      "learning_rate": 4.8649305817049596e-05,
-      "loss": 70.0598,
-      "step": 48650
-    },
-    {
-      "epoch": 0.19659255727889396,
-      "grad_norm": 1116.1414794921875,
-      "learning_rate": 4.864817374872773e-05,
-      "loss": 100.4134,
-      "step": 48660
-    },
-    {
-      "epoch": 0.1966329585442616,
-      "grad_norm": 1149.848388671875,
-      "learning_rate": 4.864704121937256e-05,
-      "loss": 59.9185,
-      "step": 48670
-    },
-    {
-      "epoch": 0.19667335980962924,
-      "grad_norm": 932.8038940429688,
-      "learning_rate": 4.8645908229006135e-05,
-      "loss": 49.7766,
-      "step": 48680
-    },
-    {
-      "epoch": 0.19671376107499686,
-      "grad_norm": 592.9822387695312,
-      "learning_rate": 4.864477477765056e-05,
-      "loss": 46.122,
-      "step": 48690
-    },
-    {
-      "epoch": 0.1967541623403645,
-      "grad_norm": 484.1387023925781,
-      "learning_rate": 4.864364086532792e-05,
-      "loss": 60.0866,
-      "step": 48700
-    },
-    {
-      "epoch": 0.19679456360573214,
-      "grad_norm": 1089.63818359375,
-      "learning_rate": 4.8642506492060335e-05,
-      "loss": 102.3337,
-      "step": 48710
-    },
-    {
-      "epoch": 0.19683496487109975,
-      "grad_norm": 1087.8228759765625,
-      "learning_rate": 4.8641371657869916e-05,
-      "loss": 83.8455,
-      "step": 48720
-    },
-    {
-      "epoch": 0.1968753661364674,
-      "grad_norm": 465.7780456542969,
-      "learning_rate": 4.864023636277878e-05,
-      "loss": 58.4958,
-      "step": 48730
-    },
-    {
-      "epoch": 0.19691576740183503,
-      "grad_norm": 1844.8414306640625,
-      "learning_rate": 4.863910060680907e-05,
-      "loss": 84.4384,
-      "step": 48740
-    },
-    {
-      "epoch": 0.19695616866720267,
-      "grad_norm": 1077.40478515625,
-      "learning_rate": 4.8637964389982926e-05,
-      "loss": 134.2446,
-      "step": 48750
-    },
-    {
-      "epoch": 0.19699656993257028,
-      "grad_norm": 755.9237060546875,
-      "learning_rate": 4.863682771232248e-05,
-      "loss": 101.4888,
-      "step": 48760
-    },
-    {
-      "epoch": 0.19703697119793792,
-      "grad_norm": 947.1353149414062,
-      "learning_rate": 4.8635690573849926e-05,
-      "loss": 99.0996,
-      "step": 48770
-    },
-    {
-      "epoch": 0.19707737246330556,
-      "grad_norm": 758.8486938476562,
-      "learning_rate": 4.8634552974587414e-05,
-      "loss": 65.2204,
-      "step": 48780
-    },
-    {
-      "epoch": 0.19711777372867317,
-      "grad_norm": 2005.01025390625,
-      "learning_rate": 4.863341491455712e-05,
-      "loss": 119.6658,
-      "step": 48790
-    },
-    {
-      "epoch": 0.1971581749940408,
-      "grad_norm": 980.075439453125,
-      "learning_rate": 4.863227639378124e-05,
-      "loss": 92.1868,
-      "step": 48800
-    },
-    {
-      "epoch": 0.19719857625940845,
-      "grad_norm": 511.9769592285156,
-      "learning_rate": 4.8631137412281954e-05,
-      "loss": 54.6814,
-      "step": 48810
-    },
-    {
-      "epoch": 0.19723897752477607,
-      "grad_norm": 993.7817993164062,
-      "learning_rate": 4.862999797008149e-05,
-      "loss": 90.118,
-      "step": 48820
-    },
-    {
-      "epoch": 0.1972793787901437,
-      "grad_norm": 687.139404296875,
-      "learning_rate": 4.8628858067202045e-05,
-      "loss": 98.846,
-      "step": 48830
-    },
-    {
-      "epoch": 0.19731978005551135,
-      "grad_norm": 1727.8497314453125,
-      "learning_rate": 4.862771770366584e-05,
-      "loss": 91.3292,
-      "step": 48840
-    },
-    {
-      "epoch": 0.19736018132087896,
-      "grad_norm": 527.6170043945312,
-      "learning_rate": 4.862657687949512e-05,
-      "loss": 84.25,
-      "step": 48850
-    },
-    {
-      "epoch": 0.1974005825862466,
-      "grad_norm": 657.8152465820312,
-      "learning_rate": 4.862543559471212e-05,
-      "loss": 49.7685,
-      "step": 48860
-    },
-    {
-      "epoch": 0.19744098385161424,
-      "grad_norm": 537.2709350585938,
-      "learning_rate": 4.8624293849339095e-05,
-      "loss": 61.4911,
-      "step": 48870
-    },
-    {
-      "epoch": 0.19748138511698185,
-      "grad_norm": 598.13671875,
-      "learning_rate": 4.862315164339829e-05,
-      "loss": 82.7801,
-      "step": 48880
-    },
-    {
-      "epoch": 0.1975217863823495,
-      "grad_norm": 480.8949279785156,
-      "learning_rate": 4.862200897691199e-05,
-      "loss": 58.1646,
-      "step": 48890
-    },
-    {
-      "epoch": 0.19756218764771713,
-      "grad_norm": 868.3237915039062,
-      "learning_rate": 4.8620865849902456e-05,
-      "loss": 73.8393,
-      "step": 48900
-    },
-    {
-      "epoch": 0.19760258891308477,
-      "grad_norm": 451.89483642578125,
-      "learning_rate": 4.861972226239199e-05,
-      "loss": 63.7517,
-      "step": 48910
-    },
-    {
-      "epoch": 0.19764299017845238,
-      "grad_norm": 598.8991088867188,
-      "learning_rate": 4.861857821440287e-05,
-      "loss": 76.1932,
-      "step": 48920
-    },
-    {
-      "epoch": 0.19768339144382002,
-      "grad_norm": 717.087646484375,
-      "learning_rate": 4.861743370595741e-05,
-      "loss": 68.1542,
-      "step": 48930
-    },
-    {
-      "epoch": 0.19772379270918766,
-      "grad_norm": 1464.67333984375,
-      "learning_rate": 4.861628873707792e-05,
-      "loss": 86.0041,
-      "step": 48940
-    },
-    {
-      "epoch": 0.19776419397455527,
-      "grad_norm": 442.174560546875,
-      "learning_rate": 4.861514330778673e-05,
-      "loss": 81.9746,
-      "step": 48950
-    },
-    {
-      "epoch": 0.19780459523992291,
-      "grad_norm": 1277.0028076171875,
-      "learning_rate": 4.861399741810615e-05,
-      "loss": 87.9091,
-      "step": 48960
-    },
-    {
-      "epoch": 0.19784499650529055,
-      "grad_norm": 485.0671081542969,
-      "learning_rate": 4.8612851068058544e-05,
-      "loss": 57.811,
-      "step": 48970
-    },
-    {
-      "epoch": 0.19788539777065817,
-      "grad_norm": 1858.374267578125,
-      "learning_rate": 4.861170425766625e-05,
-      "loss": 95.383,
-      "step": 48980
-    },
-    {
-      "epoch": 0.1979257990360258,
-      "grad_norm": 944.0939331054688,
-      "learning_rate": 4.861055698695162e-05,
-      "loss": 104.2803,
-      "step": 48990
-    },
-    {
-      "epoch": 0.19796620030139345,
-      "grad_norm": 970.7718505859375,
-      "learning_rate": 4.860940925593703e-05,
-      "loss": 100.1417,
-      "step": 49000
-    },
-    {
-      "epoch": 0.19800660156676106,
-      "grad_norm": 1792.8084716796875,
-      "learning_rate": 4.860826106464484e-05,
-      "loss": 86.6734,
-      "step": 49010
-    },
-    {
-      "epoch": 0.1980470028321287,
-      "grad_norm": 1215.9122314453125,
-      "learning_rate": 4.8607112413097464e-05,
-      "loss": 88.0753,
-      "step": 49020
-    },
-    {
-      "epoch": 0.19808740409749634,
-      "grad_norm": 484.51873779296875,
-      "learning_rate": 4.860596330131727e-05,
-      "loss": 78.6937,
-      "step": 49030
-    },
-    {
-      "epoch": 0.19812780536286395,
-      "grad_norm": 945.8279418945312,
-      "learning_rate": 4.860481372932667e-05,
-      "loss": 69.6853,
-      "step": 49040
-    },
-    {
-      "epoch": 0.1981682066282316,
-      "grad_norm": 507.302978515625,
-      "learning_rate": 4.860366369714807e-05,
-      "loss": 91.7122,
-      "step": 49050
-    },
-    {
-      "epoch": 0.19820860789359923,
-      "grad_norm": 921.6329956054688,
-      "learning_rate": 4.8602513204803896e-05,
-      "loss": 103.2615,
-      "step": 49060
-    },
-    {
-      "epoch": 0.19824900915896687,
-      "grad_norm": 596.2329711914062,
-      "learning_rate": 4.8601362252316574e-05,
-      "loss": 103.5869,
-      "step": 49070
-    },
-    {
-      "epoch": 0.19828941042433448,
-      "grad_norm": 1672.80615234375,
-      "learning_rate": 4.860021083970855e-05,
-      "loss": 83.6912,
-      "step": 49080
-    },
-    {
-      "epoch": 0.19832981168970212,
-      "grad_norm": 271.271728515625,
-      "learning_rate": 4.8599058967002254e-05,
-      "loss": 74.6612,
-      "step": 49090
-    },
-    {
-      "epoch": 0.19837021295506976,
-      "grad_norm": 497.50048828125,
-      "learning_rate": 4.859790663422016e-05,
-      "loss": 69.9331,
-      "step": 49100
-    },
-    {
-      "epoch": 0.19841061422043738,
-      "grad_norm": 573.849365234375,
-      "learning_rate": 4.8596753841384735e-05,
-      "loss": 54.5267,
-      "step": 49110
-    },
-    {
-      "epoch": 0.19845101548580502,
-      "grad_norm": 1481.6771240234375,
-      "learning_rate": 4.859560058851844e-05,
-      "loss": 63.2541,
-      "step": 49120
-    },
-    {
-      "epoch": 0.19849141675117266,
-      "grad_norm": 3614.8828125,
-      "learning_rate": 4.859444687564376e-05,
-      "loss": 98.0111,
-      "step": 49130
-    },
-    {
-      "epoch": 0.19853181801654027,
-      "grad_norm": 1466.5594482421875,
-      "learning_rate": 4.859329270278319e-05,
-      "loss": 67.2758,
-      "step": 49140
-    },
-    {
-      "epoch": 0.1985722192819079,
-      "grad_norm": 611.2523803710938,
-      "learning_rate": 4.859213806995924e-05,
-      "loss": 99.23,
-      "step": 49150
-    },
-    {
-      "epoch": 0.19861262054727555,
-      "grad_norm": 791.8298950195312,
-      "learning_rate": 4.85909829771944e-05,
-      "loss": 92.561,
-      "step": 49160
-    },
-    {
-      "epoch": 0.19865302181264316,
-      "grad_norm": 1123.1046142578125,
-      "learning_rate": 4.8589827424511216e-05,
-      "loss": 90.6359,
-      "step": 49170
-    },
-    {
-      "epoch": 0.1986934230780108,
-      "grad_norm": 0.0,
-      "learning_rate": 4.858867141193219e-05,
-      "loss": 75.7769,
-      "step": 49180
-    },
-    {
-      "epoch": 0.19873382434337844,
-      "grad_norm": 654.7210693359375,
-      "learning_rate": 4.858751493947987e-05,
-      "loss": 39.3021,
-      "step": 49190
-    },
-    {
-      "epoch": 0.19877422560874605,
-      "grad_norm": 1822.922607421875,
-      "learning_rate": 4.858635800717681e-05,
-      "loss": 112.5853,
-      "step": 49200
-    },
-    {
-      "epoch": 0.1988146268741137,
-      "grad_norm": 422.10791015625,
-      "learning_rate": 4.8585200615045555e-05,
-      "loss": 66.1211,
-      "step": 49210
-    },
-    {
-      "epoch": 0.19885502813948133,
-      "grad_norm": 1020.6856689453125,
-      "learning_rate": 4.8584042763108675e-05,
-      "loss": 70.5579,
-      "step": 49220
-    },
-    {
-      "epoch": 0.19889542940484897,
-      "grad_norm": 628.3194580078125,
-      "learning_rate": 4.858288445138873e-05,
-      "loss": 110.4609,
-      "step": 49230
-    },
-    {
-      "epoch": 0.19893583067021658,
-      "grad_norm": 1215.436767578125,
-      "learning_rate": 4.8581725679908317e-05,
-      "loss": 145.442,
-      "step": 49240
-    },
-    {
-      "epoch": 0.19897623193558422,
-      "grad_norm": 255.2740478515625,
-      "learning_rate": 4.858056644869002e-05,
-      "loss": 74.1351,
-      "step": 49250
-    },
-    {
-      "epoch": 0.19901663320095186,
-      "grad_norm": 686.521484375,
-      "learning_rate": 4.8579406757756455e-05,
-      "loss": 57.5146,
-      "step": 49260
-    },
-    {
-      "epoch": 0.19905703446631948,
-      "grad_norm": 452.3636169433594,
-      "learning_rate": 4.85782466071302e-05,
-      "loss": 68.5748,
-      "step": 49270
-    },
-    {
-      "epoch": 0.19909743573168712,
-      "grad_norm": 2242.72412109375,
-      "learning_rate": 4.857708599683389e-05,
-      "loss": 86.5858,
-      "step": 49280
-    },
-    {
-      "epoch": 0.19913783699705476,
-      "grad_norm": 625.6463623046875,
-      "learning_rate": 4.8575924926890145e-05,
-      "loss": 50.7092,
-      "step": 49290
-    },
-    {
-      "epoch": 0.19917823826242237,
-      "grad_norm": 558.0465698242188,
-      "learning_rate": 4.8574763397321614e-05,
-      "loss": 53.2504,
-      "step": 49300
-    },
-    {
-      "epoch": 0.19921863952779,
-      "grad_norm": 932.68212890625,
-      "learning_rate": 4.857360140815093e-05,
-      "loss": 65.5452,
-      "step": 49310
-    },
-    {
-      "epoch": 0.19925904079315765,
-      "grad_norm": 965.7637939453125,
-      "learning_rate": 4.857243895940076e-05,
-      "loss": 68.6537,
-      "step": 49320
-    },
-    {
-      "epoch": 0.19929944205852526,
-      "grad_norm": 676.4684448242188,
-      "learning_rate": 4.857127605109374e-05,
-      "loss": 80.4764,
-      "step": 49330
-    },
-    {
-      "epoch": 0.1993398433238929,
-      "grad_norm": 499.67083740234375,
-      "learning_rate": 4.8570112683252565e-05,
-      "loss": 96.8114,
-      "step": 49340
-    },
-    {
-      "epoch": 0.19938024458926054,
-      "grad_norm": 1086.6888427734375,
-      "learning_rate": 4.856894885589991e-05,
-      "loss": 98.3765,
-      "step": 49350
-    },
-    {
-      "epoch": 0.19942064585462815,
-      "grad_norm": 626.739990234375,
-      "learning_rate": 4.856778456905846e-05,
-      "loss": 64.8752,
-      "step": 49360
-    },
-    {
-      "epoch": 0.1994610471199958,
-      "grad_norm": 464.48565673828125,
-      "learning_rate": 4.856661982275093e-05,
-      "loss": 52.9684,
-      "step": 49370
-    },
-    {
-      "epoch": 0.19950144838536343,
-      "grad_norm": 536.5781860351562,
-      "learning_rate": 4.8565454617e-05,
-      "loss": 71.4869,
-      "step": 49380
-    },
-    {
-      "epoch": 0.19954184965073107,
-      "grad_norm": 3326.29833984375,
-      "learning_rate": 4.85642889518284e-05,
-      "loss": 165.3745,
-      "step": 49390
-    },
-    {
-      "epoch": 0.19958225091609869,
-      "grad_norm": 2225.459716796875,
-      "learning_rate": 4.856312282725886e-05,
-      "loss": 92.2177,
-      "step": 49400
-    },
-    {
-      "epoch": 0.19962265218146633,
-      "grad_norm": 611.5123901367188,
-      "learning_rate": 4.85619562433141e-05,
-      "loss": 53.4561,
-      "step": 49410
-    },
-    {
-      "epoch": 0.19966305344683397,
-      "grad_norm": 763.9044189453125,
-      "learning_rate": 4.8560789200016884e-05,
-      "loss": 65.2781,
-      "step": 49420
-    },
-    {
-      "epoch": 0.19970345471220158,
-      "grad_norm": 1733.5440673828125,
-      "learning_rate": 4.8559621697389946e-05,
-      "loss": 149.0616,
-      "step": 49430
-    },
-    {
-      "epoch": 0.19974385597756922,
-      "grad_norm": 411.6968688964844,
-      "learning_rate": 4.855845373545605e-05,
-      "loss": 71.7242,
-      "step": 49440
-    },
-    {
-      "epoch": 0.19978425724293686,
-      "grad_norm": 258.44677734375,
-      "learning_rate": 4.855728531423798e-05,
-      "loss": 71.4265,
-      "step": 49450
-    },
-    {
-      "epoch": 0.19982465850830447,
-      "grad_norm": 1239.368896484375,
-      "learning_rate": 4.85561164337585e-05,
-      "loss": 75.7033,
-      "step": 49460
-    },
-    {
-      "epoch": 0.1998650597736721,
-      "grad_norm": 1232.6944580078125,
-      "learning_rate": 4.85549470940404e-05,
-      "loss": 54.9064,
-      "step": 49470
-    },
-    {
-      "epoch": 0.19990546103903975,
-      "grad_norm": 698.7216796875,
-      "learning_rate": 4.855377729510648e-05,
-      "loss": 64.2161,
-      "step": 49480
-    },
-    {
-      "epoch": 0.19994586230440736,
-      "grad_norm": 590.6932983398438,
-      "learning_rate": 4.8552607036979553e-05,
-      "loss": 59.4658,
-      "step": 49490
-    },
-    {
-      "epoch": 0.199986263569775,
-      "grad_norm": 1102.7227783203125,
-      "learning_rate": 4.855143631968242e-05,
-      "loss": 64.1912,
-      "step": 49500
-    },
-    {
-      "epoch": 0.20002666483514264,
-      "grad_norm": 2396.97119140625,
-      "learning_rate": 4.855026514323792e-05,
-      "loss": 101.3902,
-      "step": 49510
-    },
-    {
-      "epoch": 0.20006706610051025,
-      "grad_norm": 622.559326171875,
-      "learning_rate": 4.8549093507668865e-05,
-      "loss": 91.4989,
-      "step": 49520
-    },
-    {
-      "epoch": 0.2001074673658779,
-      "grad_norm": 1117.0147705078125,
-      "learning_rate": 4.854792141299811e-05,
-      "loss": 52.4889,
-      "step": 49530
-    },
-    {
-      "epoch": 0.20014786863124553,
-      "grad_norm": 726.4915771484375,
-      "learning_rate": 4.85467488592485e-05,
-      "loss": 69.4123,
-      "step": 49540
-    },
-    {
-      "epoch": 0.20018826989661317,
-      "grad_norm": 563.3633422851562,
-      "learning_rate": 4.85455758464429e-05,
-      "loss": 79.5041,
-      "step": 49550
-    },
-    {
-      "epoch": 0.2002286711619808,
-      "grad_norm": 1249.004150390625,
-      "learning_rate": 4.854440237460418e-05,
-      "loss": 113.2019,
-      "step": 49560
-    },
-    {
-      "epoch": 0.20026907242734843,
-      "grad_norm": 1637.2099609375,
-      "learning_rate": 4.854322844375522e-05,
-      "loss": 94.2531,
-      "step": 49570
-    },
-    {
-      "epoch": 0.20030947369271607,
-      "grad_norm": 916.0922241210938,
-      "learning_rate": 4.85420540539189e-05,
-      "loss": 74.7405,
-      "step": 49580
-    },
-    {
-      "epoch": 0.20034987495808368,
-      "grad_norm": 660.0616455078125,
-      "learning_rate": 4.8540879205118106e-05,
-      "loss": 63.3157,
-      "step": 49590
-    },
-    {
-      "epoch": 0.20039027622345132,
-      "grad_norm": 1970.4896240234375,
-      "learning_rate": 4.8539703897375755e-05,
-      "loss": 93.7624,
-      "step": 49600
-    },
-    {
-      "epoch": 0.20043067748881896,
-      "grad_norm": 765.5314331054688,
-      "learning_rate": 4.853852813071476e-05,
-      "loss": 83.952,
-      "step": 49610
-    },
-    {
-      "epoch": 0.20047107875418657,
-      "grad_norm": 883.2976684570312,
-      "learning_rate": 4.853735190515804e-05,
-      "loss": 100.3142,
-      "step": 49620
-    },
-    {
-      "epoch": 0.2005114800195542,
-      "grad_norm": 1285.146240234375,
-      "learning_rate": 4.853617522072853e-05,
-      "loss": 84.2881,
-      "step": 49630
-    },
-    {
-      "epoch": 0.20055188128492185,
-      "grad_norm": 561.2415771484375,
-      "learning_rate": 4.853499807744916e-05,
-      "loss": 67.512,
-      "step": 49640
-    },
-    {
-      "epoch": 0.20059228255028946,
-      "grad_norm": 759.5241088867188,
-      "learning_rate": 4.85338204753429e-05,
-      "loss": 70.0843,
-      "step": 49650
-    },
-    {
-      "epoch": 0.2006326838156571,
-      "grad_norm": 1011.6978759765625,
-      "learning_rate": 4.8532642414432674e-05,
-      "loss": 91.1174,
-      "step": 49660
-    },
-    {
-      "epoch": 0.20067308508102474,
-      "grad_norm": 710.1160888671875,
-      "learning_rate": 4.853146389474148e-05,
-      "loss": 78.4036,
-      "step": 49670
-    },
-    {
-      "epoch": 0.20071348634639236,
-      "grad_norm": 1832.636474609375,
-      "learning_rate": 4.853028491629228e-05,
-      "loss": 69.5313,
-      "step": 49680
-    },
-    {
-      "epoch": 0.20075388761176,
-      "grad_norm": 630.8060913085938,
-      "learning_rate": 4.852910547910806e-05,
-      "loss": 72.0492,
-      "step": 49690
-    },
-    {
-      "epoch": 0.20079428887712764,
-      "grad_norm": 891.0122680664062,
-      "learning_rate": 4.852792558321182e-05,
-      "loss": 88.1902,
-      "step": 49700
-    },
-    {
-      "epoch": 0.20083469014249528,
-      "grad_norm": 493.8849182128906,
-      "learning_rate": 4.852674522862656e-05,
-      "loss": 82.2126,
-      "step": 49710
-    },
-    {
-      "epoch": 0.2008750914078629,
-      "grad_norm": 916.6065063476562,
-      "learning_rate": 4.852556441537528e-05,
-      "loss": 73.7906,
-      "step": 49720
-    },
-    {
-      "epoch": 0.20091549267323053,
-      "grad_norm": 1729.7864990234375,
-      "learning_rate": 4.852438314348101e-05,
-      "loss": 98.6484,
-      "step": 49730
-    },
-    {
-      "epoch": 0.20095589393859817,
-      "grad_norm": 971.2913818359375,
-      "learning_rate": 4.852320141296679e-05,
-      "loss": 74.1662,
-      "step": 49740
-    },
-    {
-      "epoch": 0.20099629520396578,
-      "grad_norm": 673.4738159179688,
-      "learning_rate": 4.852201922385564e-05,
-      "loss": 87.5311,
-      "step": 49750
-    },
-    {
-      "epoch": 0.20103669646933342,
-      "grad_norm": 700.3085327148438,
-      "learning_rate": 4.852083657617061e-05,
-      "loss": 119.7827,
-      "step": 49760
-    },
-    {
-      "epoch": 0.20107709773470106,
-      "grad_norm": 574.7796630859375,
-      "learning_rate": 4.851965346993478e-05,
-      "loss": 118.024,
-      "step": 49770
-    },
-    {
-      "epoch": 0.20111749900006867,
-      "grad_norm": 1227.26904296875,
-      "learning_rate": 4.851846990517118e-05,
-      "loss": 96.7121,
-      "step": 49780
-    },
-    {
-      "epoch": 0.2011579002654363,
-      "grad_norm": 1115.0523681640625,
-      "learning_rate": 4.8517285881902904e-05,
-      "loss": 69.7559,
-      "step": 49790
-    },
-    {
-      "epoch": 0.20119830153080395,
-      "grad_norm": 820.30126953125,
-      "learning_rate": 4.851610140015304e-05,
-      "loss": 94.3278,
-      "step": 49800
-    },
-    {
-      "epoch": 0.20123870279617156,
-      "grad_norm": 3716.187744140625,
-      "learning_rate": 4.8514916459944666e-05,
-      "loss": 136.7065,
-      "step": 49810
-    },
-    {
-      "epoch": 0.2012791040615392,
-      "grad_norm": 1158.3758544921875,
-      "learning_rate": 4.8513731061300887e-05,
-      "loss": 88.3688,
-      "step": 49820
-    },
-    {
-      "epoch": 0.20131950532690684,
-      "grad_norm": 2444.747314453125,
-      "learning_rate": 4.851254520424482e-05,
-      "loss": 58.1214,
-      "step": 49830
-    },
-    {
-      "epoch": 0.20135990659227446,
-      "grad_norm": 594.2299194335938,
-      "learning_rate": 4.851135888879958e-05,
-      "loss": 50.3362,
-      "step": 49840
-    },
-    {
-      "epoch": 0.2014003078576421,
-      "grad_norm": 422.1976318359375,
-      "learning_rate": 4.851017211498829e-05,
-      "loss": 114.6325,
-      "step": 49850
-    },
-    {
-      "epoch": 0.20144070912300974,
-      "grad_norm": 348.53741455078125,
-      "learning_rate": 4.85089848828341e-05,
-      "loss": 81.1028,
-      "step": 49860
-    },
-    {
-      "epoch": 0.20148111038837738,
-      "grad_norm": 629.5589599609375,
-      "learning_rate": 4.8507797192360134e-05,
-      "loss": 47.3838,
-      "step": 49870
-    },
-    {
-      "epoch": 0.201521511653745,
-      "grad_norm": 719.5811157226562,
-      "learning_rate": 4.850660904358956e-05,
-      "loss": 57.1744,
-      "step": 49880
-    },
-    {
-      "epoch": 0.20156191291911263,
-      "grad_norm": 497.17913818359375,
-      "learning_rate": 4.850542043654555e-05,
-      "loss": 75.8821,
-      "step": 49890
-    },
-    {
-      "epoch": 0.20160231418448027,
-      "grad_norm": 569.685546875,
-      "learning_rate": 4.8504231371251255e-05,
-      "loss": 59.8405,
-      "step": 49900
-    },
-    {
-      "epoch": 0.20164271544984788,
-      "grad_norm": 503.3626708984375,
-      "learning_rate": 4.850304184772988e-05,
-      "loss": 64.7331,
-      "step": 49910
-    },
-    {
-      "epoch": 0.20168311671521552,
-      "grad_norm": 444.8017883300781,
-      "learning_rate": 4.85018518660046e-05,
-      "loss": 132.0521,
-      "step": 49920
-    },
-    {
-      "epoch": 0.20172351798058316,
-      "grad_norm": 557.6558227539062,
-      "learning_rate": 4.850066142609862e-05,
-      "loss": 66.8273,
-      "step": 49930
-    },
-    {
-      "epoch": 0.20176391924595077,
-      "grad_norm": 625.9427490234375,
-      "learning_rate": 4.849947052803514e-05,
-      "loss": 97.3501,
-      "step": 49940
-    },
-    {
-      "epoch": 0.2018043205113184,
-      "grad_norm": 413.3929748535156,
-      "learning_rate": 4.849827917183739e-05,
-      "loss": 104.1396,
-      "step": 49950
-    },
-    {
-      "epoch": 0.20184472177668605,
-      "grad_norm": 1487.6632080078125,
-      "learning_rate": 4.849708735752859e-05,
-      "loss": 84.6355,
-      "step": 49960
-    },
-    {
-      "epoch": 0.20188512304205367,
-      "grad_norm": 399.767578125,
-      "learning_rate": 4.849589508513197e-05,
-      "loss": 71.0594,
-      "step": 49970
-    },
-    {
-      "epoch": 0.2019255243074213,
-      "grad_norm": 739.0697631835938,
-      "learning_rate": 4.849470235467078e-05,
-      "loss": 91.7866,
-      "step": 49980
-    },
-    {
-      "epoch": 0.20196592557278895,
-      "grad_norm": 962.0905151367188,
-      "learning_rate": 4.849350916616827e-05,
-      "loss": 142.8859,
-      "step": 49990
-    },
-    {
-      "epoch": 0.20200632683815656,
-      "grad_norm": 556.9208374023438,
-      "learning_rate": 4.849231551964771e-05,
-      "loss": 69.2205,
-      "step": 50000
-    },
-    {
-      "epoch": 0.2020467281035242,
-      "grad_norm": 994.56298828125,
-      "learning_rate": 4.849112141513236e-05,
-      "loss": 98.4823,
-      "step": 50010
-    },
-    {
-      "epoch": 0.20208712936889184,
-      "grad_norm": 446.2982482910156,
-      "learning_rate": 4.8489926852645505e-05,
-      "loss": 91.0375,
-      "step": 50020
-    },
-    {
-      "epoch": 0.20212753063425948,
-      "grad_norm": 2271.33349609375,
-      "learning_rate": 4.848873183221043e-05,
-      "loss": 89.0778,
-      "step": 50030
-    },
-    {
-      "epoch": 0.2021679318996271,
-      "grad_norm": 1082.55810546875,
-      "learning_rate": 4.8487536353850444e-05,
-      "loss": 109.4579,
-      "step": 50040
-    },
-    {
-      "epoch": 0.20220833316499473,
-      "grad_norm": 2170.5400390625,
-      "learning_rate": 4.8486340417588835e-05,
-      "loss": 84.2964,
-      "step": 50050
-    },
-    {
-      "epoch": 0.20224873443036237,
-      "grad_norm": 454.11248779296875,
-      "learning_rate": 4.8485144023448936e-05,
-      "loss": 62.0007,
-      "step": 50060
-    },
-    {
-      "epoch": 0.20228913569572998,
-      "grad_norm": 494.5736083984375,
-      "learning_rate": 4.848394717145406e-05,
-      "loss": 87.3821,
-      "step": 50070
-    },
-    {
-      "epoch": 0.20232953696109762,
-      "grad_norm": 1059.5152587890625,
-      "learning_rate": 4.848274986162754e-05,
-      "loss": 68.3891,
-      "step": 50080
-    },
-    {
-      "epoch": 0.20236993822646526,
-      "grad_norm": 778.6331176757812,
-      "learning_rate": 4.848155209399272e-05,
-      "loss": 72.0611,
-      "step": 50090
-    },
-    {
-      "epoch": 0.20241033949183287,
-      "grad_norm": 819.3021240234375,
-      "learning_rate": 4.848035386857296e-05,
-      "loss": 67.9465,
-      "step": 50100
-    },
-    {
-      "epoch": 0.20245074075720051,
-      "grad_norm": 1114.4420166015625,
-      "learning_rate": 4.847915518539161e-05,
-      "loss": 75.596,
-      "step": 50110
-    },
-    {
-      "epoch": 0.20249114202256815,
-      "grad_norm": 1062.3887939453125,
-      "learning_rate": 4.847795604447204e-05,
-      "loss": 102.3436,
-      "step": 50120
-    },
-    {
-      "epoch": 0.20253154328793577,
-      "grad_norm": 1471.7073974609375,
-      "learning_rate": 4.847675644583764e-05,
-      "loss": 129.1764,
-      "step": 50130
-    },
-    {
-      "epoch": 0.2025719445533034,
-      "grad_norm": 1061.678955078125,
-      "learning_rate": 4.847555638951177e-05,
-      "loss": 84.9815,
-      "step": 50140
-    },
-    {
-      "epoch": 0.20261234581867105,
-      "grad_norm": 1374.27001953125,
-      "learning_rate": 4.8474355875517854e-05,
-      "loss": 88.13,
-      "step": 50150
-    },
-    {
-      "epoch": 0.20265274708403866,
-      "grad_norm": 1075.933837890625,
-      "learning_rate": 4.8473154903879276e-05,
-      "loss": 94.9892,
-      "step": 50160
-    },
-    {
-      "epoch": 0.2026931483494063,
-      "grad_norm": 314.6831359863281,
-      "learning_rate": 4.8471953474619466e-05,
-      "loss": 83.2022,
-      "step": 50170
-    },
-    {
-      "epoch": 0.20273354961477394,
-      "grad_norm": 1138.607421875,
-      "learning_rate": 4.847075158776183e-05,
-      "loss": 78.0873,
-      "step": 50180
-    },
-    {
-      "epoch": 0.20277395088014158,
-      "grad_norm": 561.6530151367188,
-      "learning_rate": 4.846954924332981e-05,
-      "loss": 57.0055,
-      "step": 50190
-    },
-    {
-      "epoch": 0.2028143521455092,
-      "grad_norm": 1306.6680908203125,
-      "learning_rate": 4.846834644134686e-05,
-      "loss": 69.6428,
-      "step": 50200
-    },
-    {
-      "epoch": 0.20285475341087683,
-      "grad_norm": 418.66046142578125,
-      "learning_rate": 4.846714318183639e-05,
-      "loss": 87.8257,
-      "step": 50210
-    },
-    {
-      "epoch": 0.20289515467624447,
-      "grad_norm": 1192.7034912109375,
-      "learning_rate": 4.84659394648219e-05,
-      "loss": 93.4679,
-      "step": 50220
-    },
-    {
-      "epoch": 0.20293555594161208,
-      "grad_norm": 1370.776611328125,
-      "learning_rate": 4.846473529032684e-05,
-      "loss": 112.6145,
-      "step": 50230
-    },
-    {
-      "epoch": 0.20297595720697972,
-      "grad_norm": 555.2098388671875,
-      "learning_rate": 4.846353065837467e-05,
-      "loss": 104.6596,
-      "step": 50240
-    },
-    {
-      "epoch": 0.20301635847234736,
-      "grad_norm": 714.0413208007812,
-      "learning_rate": 4.84623255689889e-05,
-      "loss": 63.1508,
-      "step": 50250
-    },
-    {
-      "epoch": 0.20305675973771498,
-      "grad_norm": 779.4690551757812,
-      "learning_rate": 4.846112002219301e-05,
-      "loss": 77.2726,
-      "step": 50260
-    },
-    {
-      "epoch": 0.20309716100308262,
-      "grad_norm": 857.3240966796875,
-      "learning_rate": 4.845991401801051e-05,
-      "loss": 86.2318,
-      "step": 50270
-    },
-    {
-      "epoch": 0.20313756226845026,
-      "grad_norm": 700.0541381835938,
-      "learning_rate": 4.845870755646491e-05,
-      "loss": 72.9242,
-      "step": 50280
-    },
-    {
-      "epoch": 0.20317796353381787,
-      "grad_norm": 357.010009765625,
-      "learning_rate": 4.8457500637579726e-05,
-      "loss": 53.2458,
-      "step": 50290
-    },
-    {
-      "epoch": 0.2032183647991855,
-      "grad_norm": 1076.287841796875,
-      "learning_rate": 4.845629326137849e-05,
-      "loss": 60.6979,
-      "step": 50300
-    },
-    {
-      "epoch": 0.20325876606455315,
-      "grad_norm": 1700.1820068359375,
-      "learning_rate": 4.845508542788474e-05,
-      "loss": 92.155,
-      "step": 50310
-    },
-    {
-      "epoch": 0.20329916732992076,
-      "grad_norm": 800.1776123046875,
-      "learning_rate": 4.845387713712203e-05,
-      "loss": 89.9958,
-      "step": 50320
-    },
-    {
-      "epoch": 0.2033395685952884,
-      "grad_norm": 953.5119018554688,
-      "learning_rate": 4.8452668389113895e-05,
-      "loss": 103.765,
-      "step": 50330
-    },
-    {
-      "epoch": 0.20337996986065604,
-      "grad_norm": 558.611083984375,
-      "learning_rate": 4.845145918388393e-05,
-      "loss": 48.235,
-      "step": 50340
-    },
-    {
-      "epoch": 0.20342037112602368,
-      "grad_norm": 106.23624420166016,
-      "learning_rate": 4.8450249521455695e-05,
-      "loss": 87.897,
-      "step": 50350
-    },
-    {
-      "epoch": 0.2034607723913913,
-      "grad_norm": 1008.41748046875,
-      "learning_rate": 4.844903940185276e-05,
-      "loss": 55.0348,
-      "step": 50360
-    },
-    {
-      "epoch": 0.20350117365675893,
-      "grad_norm": 1108.453369140625,
-      "learning_rate": 4.844782882509874e-05,
-      "loss": 79.4438,
-      "step": 50370
-    },
-    {
-      "epoch": 0.20354157492212657,
-      "grad_norm": 985.6517944335938,
-      "learning_rate": 4.844661779121723e-05,
-      "loss": 61.7789,
-      "step": 50380
-    },
-    {
-      "epoch": 0.20358197618749418,
-      "grad_norm": 863.3707885742188,
-      "learning_rate": 4.844540630023182e-05,
-      "loss": 75.1732,
-      "step": 50390
-    },
-    {
-      "epoch": 0.20362237745286182,
-      "grad_norm": 843.4656372070312,
-      "learning_rate": 4.844419435216615e-05,
-      "loss": 61.1795,
-      "step": 50400
-    },
-    {
-      "epoch": 0.20366277871822946,
-      "grad_norm": 739.8035888671875,
-      "learning_rate": 4.844298194704384e-05,
-      "loss": 89.2932,
-      "step": 50410
-    },
-    {
-      "epoch": 0.20370317998359708,
-      "grad_norm": 455.288818359375,
-      "learning_rate": 4.8441769084888534e-05,
-      "loss": 58.3124,
-      "step": 50420
-    },
-    {
-      "epoch": 0.20374358124896472,
-      "grad_norm": 1115.42333984375,
-      "learning_rate": 4.844055576572387e-05,
-      "loss": 70.2969,
-      "step": 50430
-    },
-    {
-      "epoch": 0.20378398251433236,
-      "grad_norm": 1393.1253662109375,
-      "learning_rate": 4.84393419895735e-05,
-      "loss": 89.79,
-      "step": 50440
-    },
-    {
-      "epoch": 0.20382438377969997,
-      "grad_norm": 1216.634033203125,
-      "learning_rate": 4.84381277564611e-05,
-      "loss": 72.9308,
-      "step": 50450
-    },
-    {
-      "epoch": 0.2038647850450676,
-      "grad_norm": 548.3377075195312,
-      "learning_rate": 4.8436913066410316e-05,
-      "loss": 73.1693,
-      "step": 50460
-    },
-    {
-      "epoch": 0.20390518631043525,
-      "grad_norm": 1117.44091796875,
-      "learning_rate": 4.843569791944486e-05,
-      "loss": 121.1625,
-      "step": 50470
-    },
-    {
-      "epoch": 0.20394558757580286,
-      "grad_norm": 1097.303955078125,
-      "learning_rate": 4.843448231558839e-05,
-      "loss": 103.4902,
-      "step": 50480
-    },
-    {
-      "epoch": 0.2039859888411705,
-      "grad_norm": 797.4445190429688,
-      "learning_rate": 4.843326625486464e-05,
-      "loss": 78.5437,
-      "step": 50490
-    },
-    {
-      "epoch": 0.20402639010653814,
-      "grad_norm": 876.6442260742188,
-      "learning_rate": 4.843204973729729e-05,
-      "loss": 72.4397,
-      "step": 50500
-    },
-    {
-      "epoch": 0.20406679137190578,
-      "grad_norm": 696.5914306640625,
-      "learning_rate": 4.843083276291007e-05,
-      "loss": 75.2865,
-      "step": 50510
-    },
-    {
-      "epoch": 0.2041071926372734,
-      "grad_norm": 202.0533447265625,
-      "learning_rate": 4.84296153317267e-05,
-      "loss": 42.6595,
-      "step": 50520
-    },
-    {
-      "epoch": 0.20414759390264103,
-      "grad_norm": 711.9306030273438,
-      "learning_rate": 4.8428397443770926e-05,
-      "loss": 76.4253,
-      "step": 50530
-    },
-    {
-      "epoch": 0.20418799516800867,
-      "grad_norm": 1101.085205078125,
-      "learning_rate": 4.842717909906647e-05,
-      "loss": 91.274,
-      "step": 50540
-    },
-    {
-      "epoch": 0.20422839643337629,
-      "grad_norm": 934.9185180664062,
-      "learning_rate": 4.84259602976371e-05,
-      "loss": 99.1536,
-      "step": 50550
-    },
-    {
-      "epoch": 0.20426879769874393,
-      "grad_norm": 1069.0614013671875,
-      "learning_rate": 4.8424741039506575e-05,
-      "loss": 76.847,
-      "step": 50560
-    },
-    {
-      "epoch": 0.20430919896411157,
-      "grad_norm": 5626.541015625,
-      "learning_rate": 4.842352132469867e-05,
-      "loss": 85.2173,
-      "step": 50570
-    },
-    {
-      "epoch": 0.20434960022947918,
-      "grad_norm": 467.4293212890625,
-      "learning_rate": 4.8422301153237145e-05,
-      "loss": 57.4749,
-      "step": 50580
-    },
-    {
-      "epoch": 0.20439000149484682,
-      "grad_norm": 721.87353515625,
-      "learning_rate": 4.842108052514581e-05,
-      "loss": 67.7757,
-      "step": 50590
-    },
-    {
-      "epoch": 0.20443040276021446,
-      "grad_norm": 1129.6400146484375,
-      "learning_rate": 4.841985944044845e-05,
-      "loss": 69.0111,
-      "step": 50600
-    },
-    {
-      "epoch": 0.20447080402558207,
-      "grad_norm": 566.9580688476562,
-      "learning_rate": 4.8418637899168874e-05,
-      "loss": 96.5319,
-      "step": 50610
-    },
-    {
-      "epoch": 0.2045112052909497,
-      "grad_norm": 1207.4417724609375,
-      "learning_rate": 4.8417415901330886e-05,
-      "loss": 86.0825,
-      "step": 50620
-    },
-    {
-      "epoch": 0.20455160655631735,
-      "grad_norm": 912.6641235351562,
-      "learning_rate": 4.841619344695833e-05,
-      "loss": 75.0874,
-      "step": 50630
-    },
-    {
-      "epoch": 0.20459200782168496,
-      "grad_norm": 619.9535522460938,
-      "learning_rate": 4.8414970536075024e-05,
-      "loss": 82.5721,
-      "step": 50640
-    },
-    {
-      "epoch": 0.2046324090870526,
-      "grad_norm": 522.5993041992188,
-      "learning_rate": 4.841374716870481e-05,
-      "loss": 95.127,
-      "step": 50650
-    },
-    {
-      "epoch": 0.20467281035242024,
-      "grad_norm": 368.2784729003906,
-      "learning_rate": 4.841252334487154e-05,
-      "loss": 61.5005,
-      "step": 50660
-    },
-    {
-      "epoch": 0.20471321161778788,
-      "grad_norm": 1060.747802734375,
-      "learning_rate": 4.841129906459908e-05,
-      "loss": 69.1981,
-      "step": 50670
-    },
-    {
-      "epoch": 0.2047536128831555,
-      "grad_norm": 1138.5040283203125,
-      "learning_rate": 4.841007432791129e-05,
-      "loss": 56.1066,
-      "step": 50680
-    },
-    {
-      "epoch": 0.20479401414852313,
-      "grad_norm": 1501.6522216796875,
-      "learning_rate": 4.840884913483204e-05,
-      "loss": 108.5628,
-      "step": 50690
-    },
-    {
-      "epoch": 0.20483441541389077,
-      "grad_norm": 1450.2125244140625,
-      "learning_rate": 4.8407623485385234e-05,
-      "loss": 89.9987,
-      "step": 50700
-    },
-    {
-      "epoch": 0.2048748166792584,
-      "grad_norm": 503.9439697265625,
-      "learning_rate": 4.840639737959476e-05,
-      "loss": 69.8462,
-      "step": 50710
-    },
-    {
-      "epoch": 0.20491521794462603,
-      "grad_norm": 702.1473388671875,
-      "learning_rate": 4.8405170817484515e-05,
-      "loss": 102.1393,
-      "step": 50720
-    },
-    {
-      "epoch": 0.20495561920999367,
-      "grad_norm": 557.0831909179688,
-      "learning_rate": 4.840394379907841e-05,
-      "loss": 70.9738,
-      "step": 50730
-    },
-    {
-      "epoch": 0.20499602047536128,
-      "grad_norm": 905.9644775390625,
-      "learning_rate": 4.840271632440038e-05,
-      "loss": 63.7002,
-      "step": 50740
-    },
-    {
-      "epoch": 0.20503642174072892,
-      "grad_norm": 1401.463623046875,
-      "learning_rate": 4.840148839347434e-05,
-      "loss": 82.3691,
-      "step": 50750
-    },
-    {
-      "epoch": 0.20507682300609656,
-      "grad_norm": 457.6974182128906,
-      "learning_rate": 4.8400260006324235e-05,
-      "loss": 64.2692,
-      "step": 50760
-    },
-    {
-      "epoch": 0.20511722427146417,
-      "grad_norm": 1235.328857421875,
-      "learning_rate": 4.839903116297401e-05,
-      "loss": 76.285,
-      "step": 50770
-    },
-    {
-      "epoch": 0.2051576255368318,
-      "grad_norm": 1098.241943359375,
-      "learning_rate": 4.8397801863447635e-05,
-      "loss": 98.041,
-      "step": 50780
-    },
-    {
-      "epoch": 0.20519802680219945,
-      "grad_norm": 1030.691650390625,
-      "learning_rate": 4.8396572107769066e-05,
-      "loss": 81.4219,
-      "step": 50790
-    },
-    {
-      "epoch": 0.20523842806756706,
-      "grad_norm": 546.6036987304688,
-      "learning_rate": 4.839534189596228e-05,
-      "loss": 62.711,
-      "step": 50800
-    },
-    {
-      "epoch": 0.2052788293329347,
-      "grad_norm": 595.7460327148438,
-      "learning_rate": 4.839411122805125e-05,
-      "loss": 78.6366,
-      "step": 50810
-    },
-    {
-      "epoch": 0.20531923059830234,
-      "grad_norm": 435.6410217285156,
-      "learning_rate": 4.839288010405998e-05,
-      "loss": 54.1102,
-      "step": 50820
-    },
-    {
-      "epoch": 0.20535963186366998,
-      "grad_norm": 756.9534301757812,
-      "learning_rate": 4.839164852401247e-05,
-      "loss": 55.358,
-      "step": 50830
-    },
-    {
-      "epoch": 0.2054000331290376,
-      "grad_norm": 1443.138671875,
-      "learning_rate": 4.8390416487932733e-05,
-      "loss": 76.8449,
-      "step": 50840
-    },
-    {
-      "epoch": 0.20544043439440524,
-      "grad_norm": 666.9951171875,
-      "learning_rate": 4.8389183995844785e-05,
-      "loss": 98.6886,
-      "step": 50850
-    },
-    {
-      "epoch": 0.20548083565977288,
-      "grad_norm": 720.92919921875,
-      "learning_rate": 4.838795104777265e-05,
-      "loss": 101.151,
-      "step": 50860
-    },
-    {
-      "epoch": 0.2055212369251405,
-      "grad_norm": 929.8797607421875,
-      "learning_rate": 4.8386717643740366e-05,
-      "loss": 87.1545,
-      "step": 50870
-    },
-    {
-      "epoch": 0.20556163819050813,
-      "grad_norm": 1356.6619873046875,
-      "learning_rate": 4.8385483783771986e-05,
-      "loss": 92.7927,
-      "step": 50880
-    },
-    {
-      "epoch": 0.20560203945587577,
-      "grad_norm": 2006.3673095703125,
-      "learning_rate": 4.838424946789156e-05,
-      "loss": 75.9321,
-      "step": 50890
-    },
-    {
-      "epoch": 0.20564244072124338,
-      "grad_norm": 741.2835693359375,
-      "learning_rate": 4.8383014696123144e-05,
-      "loss": 74.4236,
-      "step": 50900
-    },
-    {
-      "epoch": 0.20568284198661102,
-      "grad_norm": 719.8021850585938,
-      "learning_rate": 4.838177946849083e-05,
-      "loss": 100.4028,
-      "step": 50910
-    },
-    {
-      "epoch": 0.20572324325197866,
-      "grad_norm": 410.4776306152344,
-      "learning_rate": 4.8380543785018677e-05,
-      "loss": 78.5173,
-      "step": 50920
-    },
-    {
-      "epoch": 0.20576364451734627,
-      "grad_norm": 563.4188842773438,
-      "learning_rate": 4.8379307645730795e-05,
-      "loss": 116.8776,
-      "step": 50930
-    },
-    {
-      "epoch": 0.2058040457827139,
-      "grad_norm": 882.4771728515625,
-      "learning_rate": 4.837807105065127e-05,
-      "loss": 63.1256,
-      "step": 50940
-    },
-    {
-      "epoch": 0.20584444704808155,
-      "grad_norm": 576.0557861328125,
-      "learning_rate": 4.837683399980421e-05,
-      "loss": 100.8891,
-      "step": 50950
-    },
-    {
-      "epoch": 0.20588484831344916,
-      "grad_norm": 2097.596923828125,
-      "learning_rate": 4.837559649321374e-05,
-      "loss": 107.397,
-      "step": 50960
-    },
-    {
-      "epoch": 0.2059252495788168,
-      "grad_norm": 1427.4742431640625,
-      "learning_rate": 4.837435853090398e-05,
-      "loss": 68.9581,
-      "step": 50970
-    },
-    {
-      "epoch": 0.20596565084418444,
-      "grad_norm": 1769.10107421875,
-      "learning_rate": 4.837312011289907e-05,
-      "loss": 77.3765,
-      "step": 50980
-    },
-    {
-      "epoch": 0.20600605210955208,
-      "grad_norm": 745.1818237304688,
-      "learning_rate": 4.837188123922314e-05,
-      "loss": 58.5138,
-      "step": 50990
-    },
-    {
-      "epoch": 0.2060464533749197,
-      "grad_norm": 1252.1639404296875,
-      "learning_rate": 4.837064190990036e-05,
-      "loss": 87.0167,
-      "step": 51000
-    },
-    {
-      "epoch": 0.20608685464028734,
-      "grad_norm": 702.9400024414062,
-      "learning_rate": 4.836940212495489e-05,
-      "loss": 100.8639,
-      "step": 51010
-    },
-    {
-      "epoch": 0.20612725590565498,
-      "grad_norm": 1918.0904541015625,
-      "learning_rate": 4.836816188441089e-05,
-      "loss": 67.2167,
-      "step": 51020
-    },
-    {
-      "epoch": 0.2061676571710226,
-      "grad_norm": 1665.779296875,
-      "learning_rate": 4.8366921188292534e-05,
-      "loss": 70.5055,
-      "step": 51030
-    },
-    {
-      "epoch": 0.20620805843639023,
-      "grad_norm": 1276.0400390625,
-      "learning_rate": 4.8365680036624026e-05,
-      "loss": 69.3662,
-      "step": 51040
-    },
-    {
-      "epoch": 0.20624845970175787,
-      "grad_norm": 1376.8558349609375,
-      "learning_rate": 4.836443842942956e-05,
-      "loss": 88.9127,
-      "step": 51050
-    },
-    {
-      "epoch": 0.20628886096712548,
-      "grad_norm": 1290.6573486328125,
-      "learning_rate": 4.836319636673334e-05,
-      "loss": 114.1035,
-      "step": 51060
-    },
-    {
-      "epoch": 0.20632926223249312,
-      "grad_norm": 789.5421142578125,
-      "learning_rate": 4.836195384855957e-05,
-      "loss": 67.4084,
-      "step": 51070
-    },
-    {
-      "epoch": 0.20636966349786076,
-      "grad_norm": 843.2109985351562,
-      "learning_rate": 4.8360710874932485e-05,
-      "loss": 77.7251,
-      "step": 51080
-    },
-    {
-      "epoch": 0.20641006476322837,
-      "grad_norm": 873.6146850585938,
-      "learning_rate": 4.8359467445876314e-05,
-      "loss": 111.2144,
-      "step": 51090
-    },
-    {
-      "epoch": 0.206450466028596,
-      "grad_norm": 1589.0201416015625,
-      "learning_rate": 4.8358223561415304e-05,
-      "loss": 95.8929,
-      "step": 51100
-    },
-    {
-      "epoch": 0.20649086729396365,
-      "grad_norm": 914.6998291015625,
-      "learning_rate": 4.8356979221573696e-05,
-      "loss": 58.4896,
-      "step": 51110
-    },
-    {
-      "epoch": 0.20653126855933127,
-      "grad_norm": 566.5862426757812,
-      "learning_rate": 4.8355734426375753e-05,
-      "loss": 82.7033,
-      "step": 51120
-    },
-    {
-      "epoch": 0.2065716698246989,
-      "grad_norm": 1035.8038330078125,
-      "learning_rate": 4.835448917584574e-05,
-      "loss": 93.2203,
-      "step": 51130
-    },
-    {
-      "epoch": 0.20661207109006655,
-      "grad_norm": 645.23388671875,
-      "learning_rate": 4.8353243470007944e-05,
-      "loss": 84.6093,
-      "step": 51140
-    },
-    {
-      "epoch": 0.20665247235543419,
-      "grad_norm": 919.4197998046875,
-      "learning_rate": 4.835199730888664e-05,
-      "loss": 70.3317,
-      "step": 51150
-    },
-    {
-      "epoch": 0.2066928736208018,
-      "grad_norm": 647.1675415039062,
-      "learning_rate": 4.835075069250613e-05,
-      "loss": 70.5222,
-      "step": 51160
-    },
-    {
-      "epoch": 0.20673327488616944,
-      "grad_norm": 689.3214111328125,
-      "learning_rate": 4.8349503620890705e-05,
-      "loss": 73.7844,
-      "step": 51170
-    },
-    {
-      "epoch": 0.20677367615153708,
-      "grad_norm": 974.9696044921875,
-      "learning_rate": 4.8348256094064695e-05,
-      "loss": 113.3237,
-      "step": 51180
-    },
-    {
-      "epoch": 0.2068140774169047,
-      "grad_norm": 595.2847290039062,
-      "learning_rate": 4.834700811205241e-05,
-      "loss": 65.6414,
-      "step": 51190
-    },
-    {
-      "epoch": 0.20685447868227233,
-      "grad_norm": 1128.4539794921875,
-      "learning_rate": 4.834575967487817e-05,
-      "loss": 105.4969,
-      "step": 51200
-    },
-    {
-      "epoch": 0.20689487994763997,
-      "grad_norm": 445.4708557128906,
-      "learning_rate": 4.834451078256634e-05,
-      "loss": 62.121,
-      "step": 51210
-    },
-    {
-      "epoch": 0.20693528121300758,
-      "grad_norm": 308.1874694824219,
-      "learning_rate": 4.8343261435141244e-05,
-      "loss": 74.7087,
-      "step": 51220
-    },
-    {
-      "epoch": 0.20697568247837522,
-      "grad_norm": 579.1897583007812,
-      "learning_rate": 4.8342011632627254e-05,
-      "loss": 115.5755,
-      "step": 51230
-    },
-    {
-      "epoch": 0.20701608374374286,
-      "grad_norm": 1018.190185546875,
-      "learning_rate": 4.834076137504873e-05,
-      "loss": 77.1754,
-      "step": 51240
-    },
-    {
-      "epoch": 0.20705648500911047,
-      "grad_norm": 560.5921020507812,
-      "learning_rate": 4.8339510662430046e-05,
-      "loss": 89.5999,
-      "step": 51250
-    },
-    {
-      "epoch": 0.20709688627447811,
-      "grad_norm": 554.189453125,
-      "learning_rate": 4.833825949479558e-05,
-      "loss": 60.0793,
-      "step": 51260
-    },
-    {
-      "epoch": 0.20713728753984575,
-      "grad_norm": 666.8161010742188,
-      "learning_rate": 4.8337007872169735e-05,
-      "loss": 79.2109,
-      "step": 51270
-    },
-    {
-      "epoch": 0.20717768880521337,
-      "grad_norm": 758.923828125,
-      "learning_rate": 4.833575579457691e-05,
-      "loss": 71.3804,
-      "step": 51280
-    },
-    {
-      "epoch": 0.207218090070581,
-      "grad_norm": 768.92236328125,
-      "learning_rate": 4.8334503262041505e-05,
-      "loss": 54.9832,
-      "step": 51290
-    },
-    {
-      "epoch": 0.20725849133594865,
-      "grad_norm": 453.7713317871094,
-      "learning_rate": 4.833325027458795e-05,
-      "loss": 86.5132,
-      "step": 51300
-    },
-    {
-      "epoch": 0.2072988926013163,
-      "grad_norm": 688.4905395507812,
-      "learning_rate": 4.8331996832240675e-05,
-      "loss": 62.2691,
-      "step": 51310
-    },
-    {
-      "epoch": 0.2073392938666839,
-      "grad_norm": 941.0972900390625,
-      "learning_rate": 4.83307429350241e-05,
-      "loss": 68.9129,
-      "step": 51320
-    },
-    {
-      "epoch": 0.20737969513205154,
-      "grad_norm": 553.0337524414062,
-      "learning_rate": 4.832948858296268e-05,
-      "loss": 70.7225,
-      "step": 51330
-    },
-    {
-      "epoch": 0.20742009639741918,
-      "grad_norm": 736.6832885742188,
-      "learning_rate": 4.832823377608087e-05,
-      "loss": 90.5628,
-      "step": 51340
-    },
-    {
-      "epoch": 0.2074604976627868,
-      "grad_norm": 1483.525390625,
-      "learning_rate": 4.832697851440313e-05,
-      "loss": 60.9628,
-      "step": 51350
-    },
-    {
-      "epoch": 0.20750089892815443,
-      "grad_norm": 1290.9232177734375,
-      "learning_rate": 4.8325722797953945e-05,
-      "loss": 56.6402,
-      "step": 51360
-    },
-    {
-      "epoch": 0.20754130019352207,
-      "grad_norm": 1492.0635986328125,
-      "learning_rate": 4.8324466626757775e-05,
-      "loss": 88.9837,
-      "step": 51370
-    },
-    {
-      "epoch": 0.20758170145888968,
-      "grad_norm": 1464.2392578125,
-      "learning_rate": 4.8323210000839124e-05,
-      "loss": 82.0623,
-      "step": 51380
-    },
-    {
-      "epoch": 0.20762210272425732,
-      "grad_norm": 771.36767578125,
-      "learning_rate": 4.832195292022249e-05,
-      "loss": 99.1351,
-      "step": 51390
-    },
-    {
-      "epoch": 0.20766250398962496,
-      "grad_norm": 513.316650390625,
-      "learning_rate": 4.832069538493237e-05,
-      "loss": 54.0263,
-      "step": 51400
-    },
-    {
-      "epoch": 0.20770290525499258,
-      "grad_norm": 813.9430541992188,
-      "learning_rate": 4.831943739499328e-05,
-      "loss": 79.7442,
-      "step": 51410
-    },
-    {
-      "epoch": 0.20774330652036022,
-      "grad_norm": 1018.91650390625,
-      "learning_rate": 4.831817895042977e-05,
-      "loss": 103.3693,
-      "step": 51420
-    },
-    {
-      "epoch": 0.20778370778572786,
-      "grad_norm": 1169.8072509765625,
-      "learning_rate": 4.8316920051266343e-05,
-      "loss": 68.9814,
-      "step": 51430
-    },
-    {
-      "epoch": 0.20782410905109547,
-      "grad_norm": 739.5846557617188,
-      "learning_rate": 4.8315660697527566e-05,
-      "loss": 47.6177,
-      "step": 51440
-    },
-    {
-      "epoch": 0.2078645103164631,
-      "grad_norm": 996.7125854492188,
-      "learning_rate": 4.831440088923797e-05,
-      "loss": 72.8957,
-      "step": 51450
-    },
-    {
-      "epoch": 0.20790491158183075,
-      "grad_norm": 765.3619384765625,
-      "learning_rate": 4.8313140626422125e-05,
-      "loss": 56.8616,
-      "step": 51460
-    },
-    {
-      "epoch": 0.2079453128471984,
-      "grad_norm": 2611.9541015625,
-      "learning_rate": 4.831187990910461e-05,
-      "loss": 117.1232,
-      "step": 51470
-    },
-    {
-      "epoch": 0.207985714112566,
-      "grad_norm": 1099.97314453125,
-      "learning_rate": 4.831061873730999e-05,
-      "loss": 65.8406,
-      "step": 51480
-    },
-    {
-      "epoch": 0.20802611537793364,
-      "grad_norm": 672.7211303710938,
-      "learning_rate": 4.8309357111062856e-05,
-      "loss": 101.9847,
-      "step": 51490
-    },
-    {
-      "epoch": 0.20806651664330128,
-      "grad_norm": 559.6053466796875,
-      "learning_rate": 4.830809503038781e-05,
-      "loss": 60.1119,
-      "step": 51500
-    },
-    {
-      "epoch": 0.2081069179086689,
-      "grad_norm": 611.0339965820312,
-      "learning_rate": 4.8306832495309445e-05,
-      "loss": 77.4869,
-      "step": 51510
-    },
-    {
-      "epoch": 0.20814731917403653,
-      "grad_norm": 955.992919921875,
-      "learning_rate": 4.830556950585238e-05,
-      "loss": 94.2499,
-      "step": 51520
-    },
-    {
-      "epoch": 0.20818772043940417,
-      "grad_norm": 1458.7508544921875,
-      "learning_rate": 4.830430606204125e-05,
-      "loss": 107.4706,
-      "step": 51530
-    },
-    {
-      "epoch": 0.20822812170477178,
-      "grad_norm": 550.4443359375,
-      "learning_rate": 4.830304216390066e-05,
-      "loss": 58.3049,
-      "step": 51540
-    },
-    {
-      "epoch": 0.20826852297013942,
-      "grad_norm": 758.5188598632812,
-      "learning_rate": 4.8301777811455276e-05,
-      "loss": 82.5834,
-      "step": 51550
-    },
-    {
-      "epoch": 0.20830892423550706,
-      "grad_norm": 715.7566528320312,
-      "learning_rate": 4.8300513004729735e-05,
-      "loss": 47.3867,
-      "step": 51560
-    },
-    {
-      "epoch": 0.20834932550087468,
-      "grad_norm": 517.6349487304688,
-      "learning_rate": 4.82992477437487e-05,
-      "loss": 60.1919,
-      "step": 51570
-    },
-    {
-      "epoch": 0.20838972676624232,
-      "grad_norm": 1060.4306640625,
-      "learning_rate": 4.8297982028536826e-05,
-      "loss": 106.7914,
-      "step": 51580
-    },
-    {
-      "epoch": 0.20843012803160996,
-      "grad_norm": 603.7183227539062,
-      "learning_rate": 4.82967158591188e-05,
-      "loss": 59.8083,
-      "step": 51590
-    },
-    {
-      "epoch": 0.20847052929697757,
-      "grad_norm": 1019.84326171875,
-      "learning_rate": 4.829544923551931e-05,
-      "loss": 63.1892,
-      "step": 51600
-    },
-    {
-      "epoch": 0.2085109305623452,
-      "grad_norm": 648.5969848632812,
-      "learning_rate": 4.8294182157763044e-05,
-      "loss": 95.7722,
-      "step": 51610
-    },
-    {
-      "epoch": 0.20855133182771285,
-      "grad_norm": 2398.117431640625,
-      "learning_rate": 4.82929146258747e-05,
-      "loss": 65.186,
-      "step": 51620
-    },
-    {
-      "epoch": 0.20859173309308046,
-      "grad_norm": 342.774658203125,
-      "learning_rate": 4.8291646639878995e-05,
-      "loss": 81.3063,
-      "step": 51630
-    },
-    {
-      "epoch": 0.2086321343584481,
-      "grad_norm": 1235.5203857421875,
-      "learning_rate": 4.829037819980065e-05,
-      "loss": 44.4151,
-      "step": 51640
-    },
-    {
-      "epoch": 0.20867253562381574,
-      "grad_norm": 470.635498046875,
-      "learning_rate": 4.828910930566439e-05,
-      "loss": 62.5535,
-      "step": 51650
-    },
-    {
-      "epoch": 0.20871293688918338,
-      "grad_norm": 830.7872924804688,
-      "learning_rate": 4.828783995749495e-05,
-      "loss": 80.7936,
-      "step": 51660
-    },
-    {
-      "epoch": 0.208753338154551,
-      "grad_norm": 1270.7835693359375,
-      "learning_rate": 4.828657015531709e-05,
-      "loss": 89.139,
-      "step": 51670
-    },
-    {
-      "epoch": 0.20879373941991863,
-      "grad_norm": 726.8765258789062,
-      "learning_rate": 4.828529989915555e-05,
-      "loss": 53.5606,
-      "step": 51680
-    },
-    {
-      "epoch": 0.20883414068528627,
-      "grad_norm": 2210.846435546875,
-      "learning_rate": 4.8284029189035094e-05,
-      "loss": 97.6787,
-      "step": 51690
-    },
-    {
-      "epoch": 0.20887454195065389,
-      "grad_norm": 0.0,
-      "learning_rate": 4.828275802498051e-05,
-      "loss": 71.8659,
-      "step": 51700
-    },
-    {
-      "epoch": 0.20891494321602153,
-      "grad_norm": 935.220458984375,
-      "learning_rate": 4.828148640701657e-05,
-      "loss": 65.6444,
-      "step": 51710
-    },
-    {
-      "epoch": 0.20895534448138917,
-      "grad_norm": 287.4408264160156,
-      "learning_rate": 4.828021433516806e-05,
-      "loss": 61.6178,
-      "step": 51720
-    },
-    {
-      "epoch": 0.20899574574675678,
-      "grad_norm": 477.3240966796875,
-      "learning_rate": 4.82789418094598e-05,
-      "loss": 93.1937,
-      "step": 51730
-    },
-    {
-      "epoch": 0.20903614701212442,
-      "grad_norm": 516.9016723632812,
-      "learning_rate": 4.827766882991657e-05,
-      "loss": 77.4756,
-      "step": 51740
-    },
-    {
-      "epoch": 0.20907654827749206,
-      "grad_norm": 1626.0362548828125,
-      "learning_rate": 4.827639539656321e-05,
-      "loss": 100.3807,
-      "step": 51750
-    },
-    {
-      "epoch": 0.20911694954285967,
-      "grad_norm": 530.7556762695312,
-      "learning_rate": 4.827512150942454e-05,
-      "loss": 96.0756,
-      "step": 51760
-    },
-    {
-      "epoch": 0.2091573508082273,
-      "grad_norm": 865.0823974609375,
-      "learning_rate": 4.827384716852539e-05,
-      "loss": 50.7269,
-      "step": 51770
-    },
-    {
-      "epoch": 0.20919775207359495,
-      "grad_norm": 1538.5042724609375,
-      "learning_rate": 4.82725723738906e-05,
-      "loss": 147.1135,
-      "step": 51780
-    },
-    {
-      "epoch": 0.20923815333896256,
-      "grad_norm": 574.5728149414062,
-      "learning_rate": 4.827129712554504e-05,
-      "loss": 60.3232,
-      "step": 51790
-    },
-    {
-      "epoch": 0.2092785546043302,
-      "grad_norm": 674.2813110351562,
-      "learning_rate": 4.8270021423513554e-05,
-      "loss": 58.9629,
-      "step": 51800
-    },
-    {
-      "epoch": 0.20931895586969784,
-      "grad_norm": 937.812255859375,
-      "learning_rate": 4.826874526782103e-05,
-      "loss": 83.8374,
-      "step": 51810
-    },
-    {
-      "epoch": 0.20935935713506548,
-      "grad_norm": 713.9767456054688,
-      "learning_rate": 4.8267468658492335e-05,
-      "loss": 63.4093,
-      "step": 51820
-    },
-    {
-      "epoch": 0.2093997584004331,
-      "grad_norm": 1410.48583984375,
-      "learning_rate": 4.826619159555236e-05,
-      "loss": 77.9979,
-      "step": 51830
-    },
-    {
-      "epoch": 0.20944015966580073,
-      "grad_norm": 845.0786743164062,
-      "learning_rate": 4.826491407902599e-05,
-      "loss": 86.6751,
-      "step": 51840
-    },
-    {
-      "epoch": 0.20948056093116837,
-      "grad_norm": 2135.817626953125,
-      "learning_rate": 4.8263636108938156e-05,
-      "loss": 96.0103,
-      "step": 51850
-    },
-    {
-      "epoch": 0.209520962196536,
-      "grad_norm": 1159.19775390625,
-      "learning_rate": 4.8262357685313754e-05,
-      "loss": 135.6379,
-      "step": 51860
-    },
-    {
-      "epoch": 0.20956136346190363,
-      "grad_norm": 962.7410278320312,
-      "learning_rate": 4.826107880817771e-05,
-      "loss": 65.2435,
-      "step": 51870
-    },
-    {
-      "epoch": 0.20960176472727127,
-      "grad_norm": 958.7470703125,
-      "learning_rate": 4.8259799477554965e-05,
-      "loss": 99.5246,
-      "step": 51880
-    },
-    {
-      "epoch": 0.20964216599263888,
-      "grad_norm": 675.7080078125,
-      "learning_rate": 4.825851969347045e-05,
-      "loss": 61.0314,
-      "step": 51890
-    },
-    {
-      "epoch": 0.20968256725800652,
-      "grad_norm": 577.6461791992188,
-      "learning_rate": 4.8257239455949124e-05,
-      "loss": 58.5346,
-      "step": 51900
-    },
-    {
-      "epoch": 0.20972296852337416,
-      "grad_norm": 1043.0440673828125,
-      "learning_rate": 4.825595876501593e-05,
-      "loss": 90.7892,
-      "step": 51910
-    },
-    {
-      "epoch": 0.20976336978874177,
-      "grad_norm": 597.583984375,
-      "learning_rate": 4.825467762069585e-05,
-      "loss": 89.8225,
-      "step": 51920
-    },
-    {
-      "epoch": 0.2098037710541094,
-      "grad_norm": 588.63427734375,
-      "learning_rate": 4.825339602301387e-05,
-      "loss": 76.6868,
-      "step": 51930
-    },
-    {
-      "epoch": 0.20984417231947705,
-      "grad_norm": 1137.5469970703125,
-      "learning_rate": 4.825211397199495e-05,
-      "loss": 77.4429,
-      "step": 51940
-    },
-    {
-      "epoch": 0.20988457358484466,
-      "grad_norm": 1342.8388671875,
-      "learning_rate": 4.82508314676641e-05,
-      "loss": 87.7724,
-      "step": 51950
-    },
-    {
-      "epoch": 0.2099249748502123,
-      "grad_norm": 549.8995971679688,
-      "learning_rate": 4.824954851004633e-05,
-      "loss": 62.8191,
-      "step": 51960
-    },
-    {
-      "epoch": 0.20996537611557994,
-      "grad_norm": 347.0243835449219,
-      "learning_rate": 4.8248265099166634e-05,
-      "loss": 68.7288,
-      "step": 51970
-    },
-    {
-      "epoch": 0.21000577738094758,
-      "grad_norm": 938.3527221679688,
-      "learning_rate": 4.824698123505004e-05,
-      "loss": 57.8288,
-      "step": 51980
-    },
-    {
-      "epoch": 0.2100461786463152,
-      "grad_norm": 869.6561279296875,
-      "learning_rate": 4.824569691772158e-05,
-      "loss": 82.7138,
-      "step": 51990
-    },
-    {
-      "epoch": 0.21008657991168284,
-      "grad_norm": 940.8391723632812,
-      "learning_rate": 4.8244412147206284e-05,
-      "loss": 87.2308,
-      "step": 52000
-    },
-    {
-      "epoch": 0.21012698117705048,
-      "grad_norm": 1123.21826171875,
-      "learning_rate": 4.8243126923529214e-05,
-      "loss": 62.064,
-      "step": 52010
-    },
-    {
-      "epoch": 0.2101673824424181,
-      "grad_norm": 919.0745849609375,
-      "learning_rate": 4.824184124671542e-05,
-      "loss": 80.0747,
-      "step": 52020
-    },
-    {
-      "epoch": 0.21020778370778573,
-      "grad_norm": 1110.951904296875,
-      "learning_rate": 4.8240555116789964e-05,
-      "loss": 58.8613,
-      "step": 52030
-    },
-    {
-      "epoch": 0.21024818497315337,
-      "grad_norm": 1179.7130126953125,
-      "learning_rate": 4.823926853377791e-05,
-      "loss": 85.1923,
-      "step": 52040
-    },
-    {
-      "epoch": 0.21028858623852098,
-      "grad_norm": 451.0613708496094,
-      "learning_rate": 4.8237981497704365e-05,
-      "loss": 63.9672,
-      "step": 52050
-    },
-    {
-      "epoch": 0.21032898750388862,
-      "grad_norm": 1503.4912109375,
-      "learning_rate": 4.8236694008594405e-05,
-      "loss": 85.0293,
-      "step": 52060
-    },
-    {
-      "epoch": 0.21036938876925626,
-      "grad_norm": 691.0103149414062,
-      "learning_rate": 4.823540606647313e-05,
-      "loss": 83.7888,
-      "step": 52070
-    },
-    {
-      "epoch": 0.21040979003462387,
-      "grad_norm": 393.8356628417969,
-      "learning_rate": 4.823411767136565e-05,
-      "loss": 71.635,
-      "step": 52080
-    },
-    {
-      "epoch": 0.2104501912999915,
-      "grad_norm": 828.97119140625,
-      "learning_rate": 4.8232828823297085e-05,
-      "loss": 88.108,
-      "step": 52090
-    },
-    {
-      "epoch": 0.21049059256535915,
-      "grad_norm": 1106.6741943359375,
-      "learning_rate": 4.8231539522292564e-05,
-      "loss": 95.024,
-      "step": 52100
-    },
-    {
-      "epoch": 0.21053099383072676,
-      "grad_norm": 996.2150268554688,
-      "learning_rate": 4.823024976837721e-05,
-      "loss": 63.8951,
-      "step": 52110
-    },
-    {
-      "epoch": 0.2105713950960944,
-      "grad_norm": 1018.7054443359375,
-      "learning_rate": 4.822895956157619e-05,
-      "loss": 83.2238,
-      "step": 52120
-    },
-    {
-      "epoch": 0.21061179636146204,
-      "grad_norm": 648.1104736328125,
-      "learning_rate": 4.8227668901914636e-05,
-      "loss": 123.0206,
-      "step": 52130
-    },
-    {
-      "epoch": 0.21065219762682968,
-      "grad_norm": 781.1571655273438,
-      "learning_rate": 4.822637778941772e-05,
-      "loss": 43.9732,
-      "step": 52140
-    },
-    {
-      "epoch": 0.2106925988921973,
-      "grad_norm": 539.1663208007812,
-      "learning_rate": 4.8225086224110615e-05,
-      "loss": 77.1991,
-      "step": 52150
-    },
-    {
-      "epoch": 0.21073300015756494,
-      "grad_norm": 654.8760986328125,
-      "learning_rate": 4.822379420601849e-05,
-      "loss": 95.2022,
-      "step": 52160
-    },
-    {
-      "epoch": 0.21077340142293258,
-      "grad_norm": 666.3115234375,
-      "learning_rate": 4.822250173516655e-05,
-      "loss": 101.8235,
-      "step": 52170
-    },
-    {
-      "epoch": 0.2108138026883002,
-      "grad_norm": 8625.4052734375,
-      "learning_rate": 4.822120881157998e-05,
-      "loss": 73.1473,
-      "step": 52180
-    },
-    {
-      "epoch": 0.21085420395366783,
-      "grad_norm": 1041.82470703125,
-      "learning_rate": 4.821991543528398e-05,
-      "loss": 101.8729,
-      "step": 52190
-    },
-    {
-      "epoch": 0.21089460521903547,
-      "grad_norm": 0.0,
-      "learning_rate": 4.821862160630378e-05,
-      "loss": 49.5271,
-      "step": 52200
-    },
-    {
-      "epoch": 0.21093500648440308,
-      "grad_norm": 347.1128234863281,
-      "learning_rate": 4.8217327324664595e-05,
-      "loss": 109.6646,
-      "step": 52210
-    },
-    {
-      "epoch": 0.21097540774977072,
-      "grad_norm": 786.9452514648438,
-      "learning_rate": 4.821603259039167e-05,
-      "loss": 69.5838,
-      "step": 52220
-    },
-    {
-      "epoch": 0.21101580901513836,
-      "grad_norm": 308.2260437011719,
-      "learning_rate": 4.821473740351023e-05,
-      "loss": 41.9356,
-      "step": 52230
-    },
-    {
-      "epoch": 0.21105621028050597,
-      "grad_norm": 730.7586669921875,
-      "learning_rate": 4.821344176404554e-05,
-      "loss": 64.5068,
-      "step": 52240
-    },
-    {
-      "epoch": 0.2110966115458736,
-      "grad_norm": 724.8195190429688,
-      "learning_rate": 4.8212145672022844e-05,
-      "loss": 75.1824,
-      "step": 52250
-    },
-    {
-      "epoch": 0.21113701281124125,
-      "grad_norm": 1213.788818359375,
-      "learning_rate": 4.821084912746742e-05,
-      "loss": 99.653,
-      "step": 52260
-    },
-    {
-      "epoch": 0.21117741407660887,
-      "grad_norm": 886.8688354492188,
-      "learning_rate": 4.820955213040454e-05,
-      "loss": 85.2024,
-      "step": 52270
-    },
-    {
-      "epoch": 0.2112178153419765,
-      "grad_norm": 746.9656372070312,
-      "learning_rate": 4.8208254680859494e-05,
-      "loss": 62.4096,
-      "step": 52280
-    },
-    {
-      "epoch": 0.21125821660734415,
-      "grad_norm": 695.8228759765625,
-      "learning_rate": 4.820695677885757e-05,
-      "loss": 123.5433,
-      "step": 52290
-    },
-    {
-      "epoch": 0.21129861787271179,
-      "grad_norm": 642.0867309570312,
-      "learning_rate": 4.820565842442408e-05,
-      "loss": 82.3376,
-      "step": 52300
-    },
-    {
-      "epoch": 0.2113390191380794,
-      "grad_norm": 448.39202880859375,
-      "learning_rate": 4.8204359617584336e-05,
-      "loss": 58.77,
-      "step": 52310
-    },
-    {
-      "epoch": 0.21137942040344704,
-      "grad_norm": 906.605712890625,
-      "learning_rate": 4.820306035836365e-05,
-      "loss": 79.2194,
-      "step": 52320
-    },
-    {
-      "epoch": 0.21141982166881468,
-      "grad_norm": 478.28643798828125,
-      "learning_rate": 4.8201760646787366e-05,
-      "loss": 56.0789,
-      "step": 52330
-    },
-    {
-      "epoch": 0.2114602229341823,
-      "grad_norm": 1141.816650390625,
-      "learning_rate": 4.82004604828808e-05,
-      "loss": 110.6768,
-      "step": 52340
-    },
-    {
-      "epoch": 0.21150062419954993,
-      "grad_norm": 280.15692138671875,
-      "learning_rate": 4.819915986666932e-05,
-      "loss": 93.5777,
-      "step": 52350
-    },
-    {
-      "epoch": 0.21154102546491757,
-      "grad_norm": 1224.0416259765625,
-      "learning_rate": 4.819785879817827e-05,
-      "loss": 104.5719,
-      "step": 52360
-    },
-    {
-      "epoch": 0.21158142673028518,
-      "grad_norm": 960.8833618164062,
-      "learning_rate": 4.819655727743302e-05,
-      "loss": 88.1145,
-      "step": 52370
-    },
-    {
-      "epoch": 0.21162182799565282,
-      "grad_norm": 717.3423461914062,
-      "learning_rate": 4.8195255304458945e-05,
-      "loss": 138.9197,
-      "step": 52380
-    },
-    {
-      "epoch": 0.21166222926102046,
-      "grad_norm": 902.6900634765625,
-      "learning_rate": 4.819395287928143e-05,
-      "loss": 72.1466,
-      "step": 52390
-    },
-    {
-      "epoch": 0.21170263052638807,
-      "grad_norm": 988.1195068359375,
-      "learning_rate": 4.8192650001925855e-05,
-      "loss": 63.8846,
-      "step": 52400
-    },
-    {
-      "epoch": 0.21174303179175571,
-      "grad_norm": 1469.6943359375,
-      "learning_rate": 4.8191346672417633e-05,
-      "loss": 68.6236,
-      "step": 52410
-    },
-    {
-      "epoch": 0.21178343305712335,
-      "grad_norm": 529.160888671875,
-      "learning_rate": 4.819004289078217e-05,
-      "loss": 71.5383,
-      "step": 52420
-    },
-    {
-      "epoch": 0.21182383432249097,
-      "grad_norm": 1089.6658935546875,
-      "learning_rate": 4.818873865704487e-05,
-      "loss": 101.7536,
-      "step": 52430
-    },
-    {
-      "epoch": 0.2118642355878586,
-      "grad_norm": 1817.579833984375,
-      "learning_rate": 4.818743397123119e-05,
-      "loss": 111.1048,
-      "step": 52440
-    },
-    {
-      "epoch": 0.21190463685322625,
-      "grad_norm": 1221.6204833984375,
-      "learning_rate": 4.818612883336654e-05,
-      "loss": 59.6956,
-      "step": 52450
-    },
-    {
-      "epoch": 0.2119450381185939,
-      "grad_norm": 572.4197998046875,
-      "learning_rate": 4.8184823243476364e-05,
-      "loss": 69.0064,
-      "step": 52460
-    },
-    {
-      "epoch": 0.2119854393839615,
-      "grad_norm": 617.7356567382812,
-      "learning_rate": 4.818351720158613e-05,
-      "loss": 50.4712,
-      "step": 52470
-    },
-    {
-      "epoch": 0.21202584064932914,
-      "grad_norm": 530.05224609375,
-      "learning_rate": 4.8182210707721284e-05,
-      "loss": 63.9792,
-      "step": 52480
-    },
-    {
-      "epoch": 0.21206624191469678,
-      "grad_norm": 2283.126953125,
-      "learning_rate": 4.8180903761907315e-05,
-      "loss": 76.8574,
-      "step": 52490
-    },
-    {
-      "epoch": 0.2121066431800644,
-      "grad_norm": 721.9110717773438,
-      "learning_rate": 4.817959636416969e-05,
-      "loss": 61.0024,
-      "step": 52500
-    },
-    {
-      "epoch": 0.21214704444543203,
-      "grad_norm": 993.3989868164062,
-      "learning_rate": 4.81782885145339e-05,
-      "loss": 97.0186,
-      "step": 52510
-    },
-    {
-      "epoch": 0.21218744571079967,
-      "grad_norm": 897.0438232421875,
-      "learning_rate": 4.8176980213025434e-05,
-      "loss": 83.1266,
-      "step": 52520
-    },
-    {
-      "epoch": 0.21222784697616728,
-      "grad_norm": 677.8610229492188,
-      "learning_rate": 4.817567145966982e-05,
-      "loss": 46.0433,
-      "step": 52530
-    },
-    {
-      "epoch": 0.21226824824153492,
-      "grad_norm": 893.5686645507812,
-      "learning_rate": 4.817436225449255e-05,
-      "loss": 66.1131,
-      "step": 52540
-    },
-    {
-      "epoch": 0.21230864950690256,
-      "grad_norm": 879.1597290039062,
-      "learning_rate": 4.817305259751916e-05,
-      "loss": 56.4979,
-      "step": 52550
-    },
-    {
-      "epoch": 0.21234905077227018,
-      "grad_norm": 1273.606689453125,
-      "learning_rate": 4.817174248877518e-05,
-      "loss": 68.6022,
-      "step": 52560
-    },
-    {
-      "epoch": 0.21238945203763782,
-      "grad_norm": 958.6312255859375,
-      "learning_rate": 4.8170431928286155e-05,
-      "loss": 73.1606,
-      "step": 52570
-    },
-    {
-      "epoch": 0.21242985330300546,
-      "grad_norm": 2562.6357421875,
-      "learning_rate": 4.816912091607762e-05,
-      "loss": 97.7864,
-      "step": 52580
-    },
-    {
-      "epoch": 0.21247025456837307,
-      "grad_norm": 305.3066711425781,
-      "learning_rate": 4.816780945217515e-05,
-      "loss": 77.4546,
-      "step": 52590
-    },
-    {
-      "epoch": 0.2125106558337407,
-      "grad_norm": 721.4779663085938,
-      "learning_rate": 4.81664975366043e-05,
-      "loss": 87.7981,
-      "step": 52600
-    },
-    {
-      "epoch": 0.21255105709910835,
-      "grad_norm": 1225.9923095703125,
-      "learning_rate": 4.816518516939067e-05,
-      "loss": 82.7573,
-      "step": 52610
-    },
-    {
-      "epoch": 0.212591458364476,
-      "grad_norm": 729.4354248046875,
-      "learning_rate": 4.8163872350559816e-05,
-      "loss": 66.5677,
-      "step": 52620
-    },
-    {
-      "epoch": 0.2126318596298436,
-      "grad_norm": 1146.287109375,
-      "learning_rate": 4.8162559080137346e-05,
-      "loss": 108.5152,
-      "step": 52630
-    },
-    {
-      "epoch": 0.21267226089521124,
-      "grad_norm": 1008.0455322265625,
-      "learning_rate": 4.8161245358148866e-05,
-      "loss": 91.5322,
-      "step": 52640
-    },
-    {
-      "epoch": 0.21271266216057888,
-      "grad_norm": 193.93905639648438,
-      "learning_rate": 4.815993118461998e-05,
-      "loss": 40.234,
-      "step": 52650
-    },
-    {
-      "epoch": 0.2127530634259465,
-      "grad_norm": 1779.3536376953125,
-      "learning_rate": 4.815861655957632e-05,
-      "loss": 82.0843,
-      "step": 52660
-    },
-    {
-      "epoch": 0.21279346469131413,
-      "grad_norm": 761.0687255859375,
-      "learning_rate": 4.81573014830435e-05,
-      "loss": 98.4412,
-      "step": 52670
-    },
-    {
-      "epoch": 0.21283386595668177,
-      "grad_norm": 639.0739135742188,
-      "learning_rate": 4.815598595504717e-05,
-      "loss": 103.2698,
-      "step": 52680
-    },
-    {
-      "epoch": 0.21287426722204938,
-      "grad_norm": 1339.623291015625,
-      "learning_rate": 4.8154669975612966e-05,
-      "loss": 97.1118,
-      "step": 52690
-    },
-    {
-      "epoch": 0.21291466848741702,
-      "grad_norm": 1258.3170166015625,
-      "learning_rate": 4.8153353544766553e-05,
-      "loss": 73.1002,
-      "step": 52700
-    },
-    {
-      "epoch": 0.21295506975278466,
-      "grad_norm": 783.255859375,
-      "learning_rate": 4.815203666253359e-05,
-      "loss": 60.4868,
-      "step": 52710
-    },
-    {
-      "epoch": 0.21299547101815228,
-      "grad_norm": 621.6766357421875,
-      "learning_rate": 4.8150719328939755e-05,
-      "loss": 53.9103,
-      "step": 52720
-    },
-    {
-      "epoch": 0.21303587228351992,
-      "grad_norm": 899.034912109375,
-      "learning_rate": 4.814940154401073e-05,
-      "loss": 90.3724,
-      "step": 52730
-    },
-    {
-      "epoch": 0.21307627354888756,
-      "grad_norm": 1244.728515625,
-      "learning_rate": 4.81480833077722e-05,
-      "loss": 57.8157,
-      "step": 52740
-    },
-    {
-      "epoch": 0.21311667481425517,
-      "grad_norm": 809.1812744140625,
-      "learning_rate": 4.814676462024988e-05,
-      "loss": 110.7006,
-      "step": 52750
-    },
-    {
-      "epoch": 0.2131570760796228,
-      "grad_norm": 3029.4775390625,
-      "learning_rate": 4.814544548146945e-05,
-      "loss": 63.5398,
-      "step": 52760
-    },
-    {
-      "epoch": 0.21319747734499045,
-      "grad_norm": 392.19464111328125,
-      "learning_rate": 4.814412589145665e-05,
-      "loss": 84.5926,
-      "step": 52770
-    },
-    {
-      "epoch": 0.2132378786103581,
-      "grad_norm": 883.9444580078125,
-      "learning_rate": 4.814280585023721e-05,
-      "loss": 73.6628,
-      "step": 52780
-    },
-    {
-      "epoch": 0.2132782798757257,
-      "grad_norm": 1597.8800048828125,
-      "learning_rate": 4.814148535783684e-05,
-      "loss": 137.1192,
-      "step": 52790
-    },
-    {
-      "epoch": 0.21331868114109334,
-      "grad_norm": 390.61407470703125,
-      "learning_rate": 4.8140164414281306e-05,
-      "loss": 58.5544,
-      "step": 52800
-    },
-    {
-      "epoch": 0.21335908240646098,
-      "grad_norm": 1327.6656494140625,
-      "learning_rate": 4.813884301959635e-05,
-      "loss": 62.8842,
-      "step": 52810
-    },
-    {
-      "epoch": 0.2133994836718286,
-      "grad_norm": 1165.1925048828125,
-      "learning_rate": 4.813752117380774e-05,
-      "loss": 73.2491,
-      "step": 52820
-    },
-    {
-      "epoch": 0.21343988493719623,
-      "grad_norm": 974.8809204101562,
-      "learning_rate": 4.813619887694124e-05,
-      "loss": 86.8438,
-      "step": 52830
-    },
-    {
-      "epoch": 0.21348028620256387,
-      "grad_norm": 658.647216796875,
-      "learning_rate": 4.813487612902264e-05,
-      "loss": 75.7994,
-      "step": 52840
-    },
-    {
-      "epoch": 0.21352068746793149,
-      "grad_norm": 619.7402954101562,
-      "learning_rate": 4.8133552930077716e-05,
-      "loss": 47.0964,
-      "step": 52850
-    },
-    {
-      "epoch": 0.21356108873329913,
-      "grad_norm": 865.9132690429688,
-      "learning_rate": 4.813222928013226e-05,
-      "loss": 47.7462,
-      "step": 52860
-    },
-    {
-      "epoch": 0.21360148999866677,
-      "grad_norm": 983.5340576171875,
-      "learning_rate": 4.813090517921209e-05,
-      "loss": 48.6633,
-      "step": 52870
-    },
-    {
-      "epoch": 0.21364189126403438,
-      "grad_norm": 607.2757568359375,
-      "learning_rate": 4.812958062734302e-05,
-      "loss": 93.2731,
-      "step": 52880
-    },
-    {
-      "epoch": 0.21368229252940202,
-      "grad_norm": 732.9096069335938,
-      "learning_rate": 4.812825562455086e-05,
-      "loss": 83.7247,
-      "step": 52890
-    },
-    {
-      "epoch": 0.21372269379476966,
-      "grad_norm": 493.807373046875,
-      "learning_rate": 4.812693017086145e-05,
-      "loss": 71.3714,
-      "step": 52900
-    },
-    {
-      "epoch": 0.21376309506013727,
-      "grad_norm": 880.7445678710938,
-      "learning_rate": 4.8125604266300636e-05,
-      "loss": 69.891,
-      "step": 52910
-    },
-    {
-      "epoch": 0.2138034963255049,
-      "grad_norm": 1000.0642700195312,
-      "learning_rate": 4.812427791089426e-05,
-      "loss": 106.0305,
-      "step": 52920
-    },
-    {
-      "epoch": 0.21384389759087255,
-      "grad_norm": 649.3963012695312,
-      "learning_rate": 4.812295110466817e-05,
-      "loss": 71.4997,
-      "step": 52930
-    },
-    {
-      "epoch": 0.2138842988562402,
-      "grad_norm": 539.8707885742188,
-      "learning_rate": 4.812162384764826e-05,
-      "loss": 81.3335,
-      "step": 52940
-    },
-    {
-      "epoch": 0.2139247001216078,
-      "grad_norm": 822.1153564453125,
-      "learning_rate": 4.8120296139860376e-05,
-      "loss": 67.337,
-      "step": 52950
-    },
-    {
-      "epoch": 0.21396510138697544,
-      "grad_norm": 2130.726806640625,
-      "learning_rate": 4.811896798133042e-05,
-      "loss": 111.1284,
-      "step": 52960
-    },
-    {
-      "epoch": 0.21400550265234308,
-      "grad_norm": 467.0498046875,
-      "learning_rate": 4.811763937208428e-05,
-      "loss": 84.292,
-      "step": 52970
-    },
-    {
-      "epoch": 0.2140459039177107,
-      "grad_norm": 1501.177978515625,
-      "learning_rate": 4.811631031214786e-05,
-      "loss": 63.4714,
-      "step": 52980
-    },
-    {
-      "epoch": 0.21408630518307833,
-      "grad_norm": 628.075439453125,
-      "learning_rate": 4.811498080154707e-05,
-      "loss": 113.7706,
-      "step": 52990
-    },
-    {
-      "epoch": 0.21412670644844597,
-      "grad_norm": 792.900146484375,
-      "learning_rate": 4.8113650840307834e-05,
-      "loss": 55.3442,
-      "step": 53000
-    },
-    {
-      "epoch": 0.2141671077138136,
-      "grad_norm": 989.729736328125,
-      "learning_rate": 4.811232042845607e-05,
-      "loss": 88.5752,
-      "step": 53010
-    },
-    {
-      "epoch": 0.21420750897918123,
-      "grad_norm": 1206.408447265625,
-      "learning_rate": 4.8110989566017716e-05,
-      "loss": 50.2796,
-      "step": 53020
-    },
-    {
-      "epoch": 0.21424791024454887,
-      "grad_norm": 854.8396606445312,
-      "learning_rate": 4.810965825301873e-05,
-      "loss": 85.1104,
-      "step": 53030
-    },
-    {
-      "epoch": 0.21428831150991648,
-      "grad_norm": 841.41357421875,
-      "learning_rate": 4.810832648948505e-05,
-      "loss": 64.3325,
-      "step": 53040
-    },
-    {
-      "epoch": 0.21432871277528412,
-      "grad_norm": 1686.3604736328125,
-      "learning_rate": 4.810699427544265e-05,
-      "loss": 73.6608,
-      "step": 53050
-    },
-    {
-      "epoch": 0.21436911404065176,
-      "grad_norm": 1173.083740234375,
-      "learning_rate": 4.810566161091751e-05,
-      "loss": 80.8808,
-      "step": 53060
-    },
-    {
-      "epoch": 0.21440951530601937,
-      "grad_norm": 1014.2672119140625,
-      "learning_rate": 4.810432849593559e-05,
-      "loss": 109.5157,
-      "step": 53070
-    },
-    {
-      "epoch": 0.214449916571387,
-      "grad_norm": 848.516357421875,
-      "learning_rate": 4.810299493052289e-05,
-      "loss": 86.9813,
-      "step": 53080
-    },
-    {
-      "epoch": 0.21449031783675465,
-      "grad_norm": 2046.2830810546875,
-      "learning_rate": 4.810166091470542e-05,
-      "loss": 71.7423,
-      "step": 53090
-    },
-    {
-      "epoch": 0.2145307191021223,
-      "grad_norm": 856.9232177734375,
-      "learning_rate": 4.810032644850917e-05,
-      "loss": 50.4441,
-      "step": 53100
-    },
-    {
-      "epoch": 0.2145711203674899,
-      "grad_norm": 868.9803466796875,
-      "learning_rate": 4.809899153196017e-05,
-      "loss": 115.6349,
-      "step": 53110
-    },
-    {
-      "epoch": 0.21461152163285754,
-      "grad_norm": 0.0,
-      "learning_rate": 4.809765616508443e-05,
-      "loss": 59.3985,
-      "step": 53120
-    },
-    {
-      "epoch": 0.21465192289822518,
-      "grad_norm": 2540.765625,
-      "learning_rate": 4.8096320347908e-05,
-      "loss": 75.7832,
-      "step": 53130
-    },
-    {
-      "epoch": 0.2146923241635928,
-      "grad_norm": 927.6019897460938,
-      "learning_rate": 4.8094984080456904e-05,
-      "loss": 65.1414,
-      "step": 53140
-    },
-    {
-      "epoch": 0.21473272542896044,
-      "grad_norm": 863.6876831054688,
-      "learning_rate": 4.8093647362757206e-05,
-      "loss": 163.4252,
-      "step": 53150
-    },
-    {
-      "epoch": 0.21477312669432808,
-      "grad_norm": 542.0109252929688,
-      "learning_rate": 4.809231019483497e-05,
-      "loss": 81.4207,
-      "step": 53160
-    },
-    {
-      "epoch": 0.2148135279596957,
-      "grad_norm": 738.8974609375,
-      "learning_rate": 4.809097257671625e-05,
-      "loss": 71.7718,
-      "step": 53170
-    },
-    {
-      "epoch": 0.21485392922506333,
-      "grad_norm": 1300.409423828125,
-      "learning_rate": 4.808963450842713e-05,
-      "loss": 63.6445,
-      "step": 53180
-    },
-    {
-      "epoch": 0.21489433049043097,
-      "grad_norm": 710.06103515625,
-      "learning_rate": 4.80882959899937e-05,
-      "loss": 75.5958,
-      "step": 53190
-    },
-    {
-      "epoch": 0.21493473175579858,
-      "grad_norm": 1211.570556640625,
-      "learning_rate": 4.808695702144206e-05,
-      "loss": 96.9039,
-      "step": 53200
-    },
-    {
-      "epoch": 0.21497513302116622,
-      "grad_norm": 4932.537109375,
-      "learning_rate": 4.808561760279831e-05,
-      "loss": 62.6395,
-      "step": 53210
-    },
-    {
-      "epoch": 0.21501553428653386,
-      "grad_norm": 1449.601806640625,
-      "learning_rate": 4.8084277734088544e-05,
-      "loss": 76.1311,
-      "step": 53220
-    },
-    {
-      "epoch": 0.21505593555190147,
-      "grad_norm": 1062.4901123046875,
-      "learning_rate": 4.808293741533891e-05,
-      "loss": 94.2731,
-      "step": 53230
-    },
-    {
-      "epoch": 0.2150963368172691,
-      "grad_norm": 766.841552734375,
-      "learning_rate": 4.808159664657552e-05,
-      "loss": 59.1123,
-      "step": 53240
-    },
-    {
-      "epoch": 0.21513673808263675,
-      "grad_norm": 503.95635986328125,
-      "learning_rate": 4.808025542782453e-05,
-      "loss": 71.4319,
-      "step": 53250
-    },
-    {
-      "epoch": 0.2151771393480044,
-      "grad_norm": 2950.0439453125,
-      "learning_rate": 4.8078913759112066e-05,
-      "loss": 90.5065,
-      "step": 53260
-    },
-    {
-      "epoch": 0.215217540613372,
-      "grad_norm": 1027.7640380859375,
-      "learning_rate": 4.80775716404643e-05,
-      "loss": 56.2461,
-      "step": 53270
-    },
-    {
-      "epoch": 0.21525794187873964,
-      "grad_norm": 507.6009521484375,
-      "learning_rate": 4.8076229071907397e-05,
-      "loss": 66.3295,
-      "step": 53280
-    },
-    {
-      "epoch": 0.21529834314410728,
-      "grad_norm": 665.6806640625,
-      "learning_rate": 4.807488605346753e-05,
-      "loss": 80.3044,
-      "step": 53290
-    },
-    {
-      "epoch": 0.2153387444094749,
-      "grad_norm": 989.35791015625,
-      "learning_rate": 4.8073542585170877e-05,
-      "loss": 68.0304,
-      "step": 53300
-    },
-    {
-      "epoch": 0.21537914567484254,
-      "grad_norm": 6701.48046875,
-      "learning_rate": 4.8072198667043635e-05,
-      "loss": 130.3542,
-      "step": 53310
-    },
-    {
-      "epoch": 0.21541954694021018,
-      "grad_norm": 626.8556518554688,
-      "learning_rate": 4.8070854299111994e-05,
-      "loss": 69.6508,
-      "step": 53320
-    },
-    {
-      "epoch": 0.2154599482055778,
-      "grad_norm": 889.5938720703125,
-      "learning_rate": 4.806950948140217e-05,
-      "loss": 94.4536,
-      "step": 53330
-    },
-    {
-      "epoch": 0.21550034947094543,
-      "grad_norm": 470.2935485839844,
-      "learning_rate": 4.8068164213940393e-05,
-      "loss": 65.35,
-      "step": 53340
-    },
-    {
-      "epoch": 0.21554075073631307,
-      "grad_norm": 1064.505859375,
-      "learning_rate": 4.8066818496752875e-05,
-      "loss": 88.5015,
-      "step": 53350
-    },
-    {
-      "epoch": 0.21558115200168068,
-      "grad_norm": 673.2662963867188,
-      "learning_rate": 4.8065472329865854e-05,
-      "loss": 63.5901,
-      "step": 53360
-    },
-    {
-      "epoch": 0.21562155326704832,
-      "grad_norm": 880.4266967773438,
-      "learning_rate": 4.806412571330557e-05,
-      "loss": 63.0849,
-      "step": 53370
-    },
-    {
-      "epoch": 0.21566195453241596,
-      "grad_norm": 761.7822875976562,
-      "learning_rate": 4.8062778647098284e-05,
-      "loss": 78.9925,
-      "step": 53380
-    },
-    {
-      "epoch": 0.21570235579778357,
-      "grad_norm": 974.259521484375,
-      "learning_rate": 4.806143113127025e-05,
-      "loss": 93.895,
-      "step": 53390
-    },
-    {
-      "epoch": 0.2157427570631512,
-      "grad_norm": 555.7957153320312,
-      "learning_rate": 4.8060083165847754e-05,
-      "loss": 74.5153,
-      "step": 53400
-    },
-    {
-      "epoch": 0.21578315832851885,
-      "grad_norm": 925.432373046875,
-      "learning_rate": 4.805873475085706e-05,
-      "loss": 119.6515,
-      "step": 53410
-    },
-    {
-      "epoch": 0.2158235595938865,
-      "grad_norm": 817.2948608398438,
-      "learning_rate": 4.805738588632446e-05,
-      "loss": 90.0423,
-      "step": 53420
-    },
-    {
-      "epoch": 0.2158639608592541,
-      "grad_norm": 921.9326782226562,
-      "learning_rate": 4.805603657227625e-05,
-      "loss": 106.7092,
-      "step": 53430
-    },
-    {
-      "epoch": 0.21590436212462175,
-      "grad_norm": 1191.692626953125,
-      "learning_rate": 4.805468680873874e-05,
-      "loss": 57.5349,
-      "step": 53440
-    },
-    {
-      "epoch": 0.21594476338998939,
-      "grad_norm": 327.62908935546875,
-      "learning_rate": 4.8053336595738236e-05,
-      "loss": 84.3568,
-      "step": 53450
-    },
-    {
-      "epoch": 0.215985164655357,
-      "grad_norm": 752.6683349609375,
-      "learning_rate": 4.805198593330107e-05,
-      "loss": 63.1536,
-      "step": 53460
-    },
-    {
-      "epoch": 0.21602556592072464,
-      "grad_norm": 711.6024780273438,
-      "learning_rate": 4.8050634821453565e-05,
-      "loss": 62.8601,
-      "step": 53470
-    },
-    {
-      "epoch": 0.21606596718609228,
-      "grad_norm": 466.1304931640625,
-      "learning_rate": 4.8049283260222075e-05,
-      "loss": 77.9144,
-      "step": 53480
-    },
-    {
-      "epoch": 0.2161063684514599,
-      "grad_norm": 1117.0677490234375,
-      "learning_rate": 4.804793124963294e-05,
-      "loss": 76.5366,
-      "step": 53490
-    },
-    {
-      "epoch": 0.21614676971682753,
-      "grad_norm": 637.1599731445312,
-      "learning_rate": 4.8046578789712515e-05,
-      "loss": 67.9231,
-      "step": 53500
-    },
-    {
-      "epoch": 0.21618717098219517,
-      "grad_norm": 568.0106811523438,
-      "learning_rate": 4.804522588048718e-05,
-      "loss": 75.5774,
-      "step": 53510
-    },
-    {
-      "epoch": 0.21622757224756278,
-      "grad_norm": 432.543212890625,
-      "learning_rate": 4.8043872521983294e-05,
-      "loss": 59.4265,
-      "step": 53520
-    },
-    {
-      "epoch": 0.21626797351293042,
-      "grad_norm": 454.713134765625,
-      "learning_rate": 4.804251871422725e-05,
-      "loss": 96.0723,
-      "step": 53530
-    },
-    {
-      "epoch": 0.21630837477829806,
-      "grad_norm": 1082.2203369140625,
-      "learning_rate": 4.804116445724543e-05,
-      "loss": 79.7681,
-      "step": 53540
-    },
-    {
-      "epoch": 0.21634877604366567,
-      "grad_norm": 621.678466796875,
-      "learning_rate": 4.803980975106427e-05,
-      "loss": 85.7473,
-      "step": 53550
-    },
-    {
-      "epoch": 0.21638917730903331,
-      "grad_norm": 890.781005859375,
-      "learning_rate": 4.803845459571014e-05,
-      "loss": 86.9005,
-      "step": 53560
-    },
-    {
-      "epoch": 0.21642957857440095,
-      "grad_norm": 649.1585083007812,
-      "learning_rate": 4.8037098991209484e-05,
-      "loss": 61.2427,
-      "step": 53570
-    },
-    {
-      "epoch": 0.2164699798397686,
-      "grad_norm": 850.4962768554688,
-      "learning_rate": 4.8035742937588724e-05,
-      "loss": 59.2555,
-      "step": 53580
-    },
-    {
-      "epoch": 0.2165103811051362,
-      "grad_norm": 787.1389770507812,
-      "learning_rate": 4.803438643487429e-05,
-      "loss": 82.7354,
-      "step": 53590
-    },
-    {
-      "epoch": 0.21655078237050385,
-      "grad_norm": 743.615234375,
-      "learning_rate": 4.803302948309264e-05,
-      "loss": 72.4777,
-      "step": 53600
-    },
-    {
-      "epoch": 0.2165911836358715,
-      "grad_norm": 1553.5562744140625,
-      "learning_rate": 4.8031672082270216e-05,
-      "loss": 85.1181,
-      "step": 53610
-    },
-    {
-      "epoch": 0.2166315849012391,
-      "grad_norm": 525.2081298828125,
-      "learning_rate": 4.803031423243349e-05,
-      "loss": 65.5073,
-      "step": 53620
-    },
-    {
-      "epoch": 0.21667198616660674,
-      "grad_norm": 718.9132080078125,
-      "learning_rate": 4.802895593360893e-05,
-      "loss": 169.8341,
-      "step": 53630
-    },
-    {
-      "epoch": 0.21671238743197438,
-      "grad_norm": 663.7149658203125,
-      "learning_rate": 4.8027597185823016e-05,
-      "loss": 54.949,
-      "step": 53640
-    },
-    {
-      "epoch": 0.216752788697342,
-      "grad_norm": 558.2113037109375,
-      "learning_rate": 4.802623798910224e-05,
-      "loss": 72.456,
-      "step": 53650
-    },
-    {
-      "epoch": 0.21679318996270963,
-      "grad_norm": 3290.45068359375,
-      "learning_rate": 4.802487834347311e-05,
-      "loss": 80.3636,
-      "step": 53660
-    },
-    {
-      "epoch": 0.21683359122807727,
-      "grad_norm": 601.9519653320312,
-      "learning_rate": 4.802351824896211e-05,
-      "loss": 82.0123,
-      "step": 53670
-    },
-    {
-      "epoch": 0.21687399249344488,
-      "grad_norm": 0.0,
-      "learning_rate": 4.802215770559577e-05,
-      "loss": 58.6228,
-      "step": 53680
-    },
-    {
-      "epoch": 0.21691439375881252,
-      "grad_norm": 610.6953125,
-      "learning_rate": 4.802079671340062e-05,
-      "loss": 60.981,
-      "step": 53690
-    },
-    {
-      "epoch": 0.21695479502418016,
-      "grad_norm": 774.2489624023438,
-      "learning_rate": 4.801943527240318e-05,
-      "loss": 94.8428,
-      "step": 53700
-    },
-    {
-      "epoch": 0.21699519628954778,
-      "grad_norm": 1123.4874267578125,
-      "learning_rate": 4.801807338263e-05,
-      "loss": 76.9702,
-      "step": 53710
-    },
-    {
-      "epoch": 0.21703559755491542,
-      "grad_norm": 1223.6903076171875,
-      "learning_rate": 4.801671104410763e-05,
-      "loss": 101.9046,
-      "step": 53720
-    },
-    {
-      "epoch": 0.21707599882028306,
-      "grad_norm": 1136.135009765625,
-      "learning_rate": 4.801534825686263e-05,
-      "loss": 98.1998,
-      "step": 53730
-    },
-    {
-      "epoch": 0.2171164000856507,
-      "grad_norm": 1007.6103515625,
-      "learning_rate": 4.801398502092156e-05,
-      "loss": 79.4994,
-      "step": 53740
-    },
-    {
-      "epoch": 0.2171568013510183,
-      "grad_norm": 1269.972900390625,
-      "learning_rate": 4.8012621336311016e-05,
-      "loss": 63.9924,
-      "step": 53750
-    },
-    {
-      "epoch": 0.21719720261638595,
-      "grad_norm": 643.7221069335938,
-      "learning_rate": 4.8011257203057556e-05,
-      "loss": 55.9485,
-      "step": 53760
-    },
-    {
-      "epoch": 0.2172376038817536,
-      "grad_norm": 1297.72412109375,
-      "learning_rate": 4.80098926211878e-05,
-      "loss": 92.5376,
-      "step": 53770
-    },
-    {
-      "epoch": 0.2172780051471212,
-      "grad_norm": 1075.8739013671875,
-      "learning_rate": 4.800852759072833e-05,
-      "loss": 59.2101,
-      "step": 53780
-    },
-    {
-      "epoch": 0.21731840641248884,
-      "grad_norm": 1902.2833251953125,
-      "learning_rate": 4.800716211170578e-05,
-      "loss": 95.2259,
-      "step": 53790
-    },
-    {
-      "epoch": 0.21735880767785648,
-      "grad_norm": 1468.7894287109375,
-      "learning_rate": 4.800579618414676e-05,
-      "loss": 94.4642,
-      "step": 53800
-    },
-    {
-      "epoch": 0.2173992089432241,
-      "grad_norm": 640.5420532226562,
-      "learning_rate": 4.80044298080779e-05,
-      "loss": 86.4085,
-      "step": 53810
-    },
-    {
-      "epoch": 0.21743961020859173,
-      "grad_norm": 872.8704833984375,
-      "learning_rate": 4.800306298352583e-05,
-      "loss": 64.3737,
-      "step": 53820
-    },
-    {
-      "epoch": 0.21748001147395937,
-      "grad_norm": 586.101806640625,
-      "learning_rate": 4.800169571051721e-05,
-      "loss": 89.3042,
-      "step": 53830
-    },
-    {
-      "epoch": 0.21752041273932698,
-      "grad_norm": 966.0181884765625,
-      "learning_rate": 4.800032798907869e-05,
-      "loss": 90.1948,
-      "step": 53840
-    },
-    {
-      "epoch": 0.21756081400469462,
-      "grad_norm": 795.524658203125,
-      "learning_rate": 4.799895981923693e-05,
-      "loss": 110.5656,
-      "step": 53850
-    },
-    {
-      "epoch": 0.21760121527006226,
-      "grad_norm": 496.3986511230469,
-      "learning_rate": 4.799759120101861e-05,
-      "loss": 51.413,
-      "step": 53860
-    },
-    {
-      "epoch": 0.21764161653542988,
-      "grad_norm": 596.7861328125,
-      "learning_rate": 4.799622213445041e-05,
-      "loss": 44.7856,
-      "step": 53870
-    },
-    {
-      "epoch": 0.21768201780079752,
-      "grad_norm": 1061.037353515625,
-      "learning_rate": 4.7994852619559016e-05,
-      "loss": 93.6382,
-      "step": 53880
-    },
-    {
-      "epoch": 0.21772241906616516,
-      "grad_norm": 386.0953063964844,
-      "learning_rate": 4.7993482656371135e-05,
-      "loss": 70.8924,
-      "step": 53890
-    },
-    {
-      "epoch": 0.2177628203315328,
-      "grad_norm": 1275.1336669921875,
-      "learning_rate": 4.799211224491348e-05,
-      "loss": 77.7109,
-      "step": 53900
-    },
-    {
-      "epoch": 0.2178032215969004,
-      "grad_norm": 534.1307373046875,
-      "learning_rate": 4.799074138521274e-05,
-      "loss": 47.0972,
-      "step": 53910
-    },
-    {
-      "epoch": 0.21784362286226805,
-      "grad_norm": 915.685791015625,
-      "learning_rate": 4.798937007729568e-05,
-      "loss": 67.5173,
-      "step": 53920
-    },
-    {
-      "epoch": 0.2178840241276357,
-      "grad_norm": 965.9025268554688,
-      "learning_rate": 4.7987998321189e-05,
-      "loss": 60.5081,
-      "step": 53930
-    },
-    {
-      "epoch": 0.2179244253930033,
-      "grad_norm": 2893.699462890625,
-      "learning_rate": 4.798662611691947e-05,
-      "loss": 109.9214,
-      "step": 53940
-    },
-    {
-      "epoch": 0.21796482665837094,
-      "grad_norm": 468.1808776855469,
-      "learning_rate": 4.7985253464513825e-05,
-      "loss": 56.8471,
-      "step": 53950
-    },
-    {
-      "epoch": 0.21800522792373858,
-      "grad_norm": 547.9169311523438,
-      "learning_rate": 4.798388036399883e-05,
-      "loss": 57.3491,
-      "step": 53960
-    },
-    {
-      "epoch": 0.2180456291891062,
-      "grad_norm": 482.74652099609375,
-      "learning_rate": 4.7982506815401254e-05,
-      "loss": 69.7557,
-      "step": 53970
-    },
-    {
-      "epoch": 0.21808603045447383,
-      "grad_norm": 1082.1453857421875,
-      "learning_rate": 4.7981132818747876e-05,
-      "loss": 96.0472,
-      "step": 53980
-    },
-    {
-      "epoch": 0.21812643171984147,
-      "grad_norm": 1212.4945068359375,
-      "learning_rate": 4.797975837406547e-05,
-      "loss": 63.2927,
-      "step": 53990
-    },
-    {
-      "epoch": 0.21816683298520909,
-      "grad_norm": 921.6978149414062,
-      "learning_rate": 4.797838348138086e-05,
-      "loss": 57.7244,
-      "step": 54000
-    },
-    {
-      "epoch": 0.21820723425057673,
-      "grad_norm": 873.8997802734375,
-      "learning_rate": 4.797700814072083e-05,
-      "loss": 66.1426,
-      "step": 54010
-    },
-    {
-      "epoch": 0.21824763551594437,
-      "grad_norm": 760.1358032226562,
-      "learning_rate": 4.7975632352112195e-05,
-      "loss": 77.4176,
-      "step": 54020
-    },
-    {
-      "epoch": 0.21828803678131198,
-      "grad_norm": 1324.942626953125,
-      "learning_rate": 4.7974256115581785e-05,
-      "loss": 69.078,
-      "step": 54030
-    },
-    {
-      "epoch": 0.21832843804667962,
-      "grad_norm": 1516.5565185546875,
-      "learning_rate": 4.797287943115641e-05,
-      "loss": 64.1911,
-      "step": 54040
-    },
-    {
-      "epoch": 0.21836883931204726,
-      "grad_norm": 1137.977294921875,
-      "learning_rate": 4.7971502298862936e-05,
-      "loss": 102.8754,
-      "step": 54050
-    },
-    {
-      "epoch": 0.2184092405774149,
-      "grad_norm": 813.7706909179688,
-      "learning_rate": 4.7970124718728193e-05,
-      "loss": 82.1068,
-      "step": 54060
-    },
-    {
-      "epoch": 0.2184496418427825,
-      "grad_norm": 635.2130737304688,
-      "learning_rate": 4.7968746690779044e-05,
-      "loss": 58.6373,
-      "step": 54070
-    },
-    {
-      "epoch": 0.21849004310815015,
-      "grad_norm": 642.5101928710938,
-      "learning_rate": 4.796736821504235e-05,
-      "loss": 74.7753,
-      "step": 54080
-    },
-    {
-      "epoch": 0.2185304443735178,
-      "grad_norm": 849.4306030273438,
-      "learning_rate": 4.7965989291545e-05,
-      "loss": 91.7941,
-      "step": 54090
-    },
-    {
-      "epoch": 0.2185708456388854,
-      "grad_norm": 740.3421630859375,
-      "learning_rate": 4.796460992031385e-05,
-      "loss": 72.5141,
-      "step": 54100
-    },
-    {
-      "epoch": 0.21861124690425304,
-      "grad_norm": 606.4213256835938,
-      "learning_rate": 4.7963230101375814e-05,
-      "loss": 90.6276,
-      "step": 54110
-    },
-    {
-      "epoch": 0.21865164816962068,
-      "grad_norm": 534.52197265625,
-      "learning_rate": 4.7961849834757786e-05,
-      "loss": 141.8116,
-      "step": 54120
-    },
-    {
-      "epoch": 0.2186920494349883,
-      "grad_norm": 2459.29443359375,
-      "learning_rate": 4.7960469120486674e-05,
-      "loss": 104.6196,
-      "step": 54130
-    },
-    {
-      "epoch": 0.21873245070035593,
-      "grad_norm": 804.312744140625,
-      "learning_rate": 4.7959087958589386e-05,
-      "loss": 71.0338,
-      "step": 54140
-    },
-    {
-      "epoch": 0.21877285196572357,
-      "grad_norm": 884.2997436523438,
-      "learning_rate": 4.7957706349092865e-05,
-      "loss": 80.0751,
-      "step": 54150
-    },
-    {
-      "epoch": 0.2188132532310912,
-      "grad_norm": 1428.2335205078125,
-      "learning_rate": 4.795632429202405e-05,
-      "loss": 73.554,
-      "step": 54160
-    },
-    {
-      "epoch": 0.21885365449645883,
-      "grad_norm": 555.4033203125,
-      "learning_rate": 4.795494178740986e-05,
-      "loss": 52.6659,
-      "step": 54170
-    },
-    {
-      "epoch": 0.21889405576182647,
-      "grad_norm": 479.893310546875,
-      "learning_rate": 4.795355883527727e-05,
-      "loss": 81.4304,
-      "step": 54180
-    },
-    {
-      "epoch": 0.21893445702719408,
-      "grad_norm": 738.9581909179688,
-      "learning_rate": 4.7952175435653226e-05,
-      "loss": 83.4163,
-      "step": 54190
-    },
-    {
-      "epoch": 0.21897485829256172,
-      "grad_norm": 599.0054321289062,
-      "learning_rate": 4.79507915885647e-05,
-      "loss": 62.3013,
-      "step": 54200
-    },
-    {
-      "epoch": 0.21901525955792936,
-      "grad_norm": 672.7980346679688,
-      "learning_rate": 4.794940729403869e-05,
-      "loss": 73.0649,
-      "step": 54210
-    },
-    {
-      "epoch": 0.219055660823297,
-      "grad_norm": 1021.8695068359375,
-      "learning_rate": 4.794802255210217e-05,
-      "loss": 74.2551,
-      "step": 54220
-    },
-    {
-      "epoch": 0.2190960620886646,
-      "grad_norm": 2235.082763671875,
-      "learning_rate": 4.794663736278212e-05,
-      "loss": 100.521,
-      "step": 54230
-    },
-    {
-      "epoch": 0.21913646335403225,
-      "grad_norm": 454.6265563964844,
-      "learning_rate": 4.794525172610558e-05,
-      "loss": 64.2982,
-      "step": 54240
-    },
-    {
-      "epoch": 0.2191768646193999,
-      "grad_norm": 2061.634765625,
-      "learning_rate": 4.794386564209953e-05,
-      "loss": 80.8011,
-      "step": 54250
-    },
-    {
-      "epoch": 0.2192172658847675,
-      "grad_norm": 1272.87451171875,
-      "learning_rate": 4.7942479110791015e-05,
-      "loss": 75.4505,
-      "step": 54260
-    },
-    {
-      "epoch": 0.21925766715013514,
-      "grad_norm": 732.8099975585938,
-      "learning_rate": 4.7941092132207056e-05,
-      "loss": 76.4685,
-      "step": 54270
-    },
-    {
-      "epoch": 0.21929806841550278,
-      "grad_norm": 1131.0069580078125,
-      "learning_rate": 4.793970470637469e-05,
-      "loss": 95.1893,
-      "step": 54280
-    },
-    {
-      "epoch": 0.2193384696808704,
-      "grad_norm": 707.0789794921875,
-      "learning_rate": 4.793831683332098e-05,
-      "loss": 79.5281,
-      "step": 54290
-    },
-    {
-      "epoch": 0.21937887094623804,
-      "grad_norm": 1276.7493896484375,
-      "learning_rate": 4.7936928513072964e-05,
-      "loss": 63.4697,
-      "step": 54300
-    },
-    {
-      "epoch": 0.21941927221160568,
-      "grad_norm": 169.8773193359375,
-      "learning_rate": 4.793553974565773e-05,
-      "loss": 65.7283,
-      "step": 54310
-    },
-    {
-      "epoch": 0.2194596734769733,
-      "grad_norm": 554.336181640625,
-      "learning_rate": 4.793415053110233e-05,
-      "loss": 56.5391,
-      "step": 54320
-    },
-    {
-      "epoch": 0.21950007474234093,
-      "grad_norm": 848.8544921875,
-      "learning_rate": 4.7932760869433865e-05,
-      "loss": 66.7205,
-      "step": 54330
-    },
-    {
-      "epoch": 0.21954047600770857,
-      "grad_norm": 1150.58349609375,
-      "learning_rate": 4.793137076067942e-05,
-      "loss": 70.9918,
-      "step": 54340
-    },
-    {
-      "epoch": 0.21958087727307618,
-      "grad_norm": 1347.6571044921875,
-      "learning_rate": 4.792998020486609e-05,
-      "loss": 75.1297,
-      "step": 54350
-    },
-    {
-      "epoch": 0.21962127853844382,
-      "grad_norm": 796.1422119140625,
-      "learning_rate": 4.792858920202099e-05,
-      "loss": 91.2722,
-      "step": 54360
-    },
-    {
-      "epoch": 0.21966167980381146,
-      "grad_norm": 978.01025390625,
-      "learning_rate": 4.792719775217124e-05,
-      "loss": 79.3633,
-      "step": 54370
-    },
-    {
-      "epoch": 0.2197020810691791,
-      "grad_norm": 676.4722900390625,
-      "learning_rate": 4.7925805855343975e-05,
-      "loss": 76.3162,
-      "step": 54380
-    },
-    {
-      "epoch": 0.2197424823345467,
-      "grad_norm": 820.4541625976562,
-      "learning_rate": 4.7924413511566315e-05,
-      "loss": 43.4046,
-      "step": 54390
-    },
-    {
-      "epoch": 0.21978288359991435,
-      "grad_norm": 598.8165893554688,
-      "learning_rate": 4.7923020720865414e-05,
-      "loss": 57.7346,
-      "step": 54400
-    },
-    {
-      "epoch": 0.219823284865282,
-      "grad_norm": 2260.6591796875,
-      "learning_rate": 4.792162748326841e-05,
-      "loss": 97.9996,
-      "step": 54410
-    },
-    {
-      "epoch": 0.2198636861306496,
-      "grad_norm": 1380.8861083984375,
-      "learning_rate": 4.792023379880249e-05,
-      "loss": 88.7846,
-      "step": 54420
-    },
-    {
-      "epoch": 0.21990408739601724,
-      "grad_norm": 1240.204833984375,
-      "learning_rate": 4.791883966749482e-05,
-      "loss": 78.3506,
-      "step": 54430
-    },
-    {
-      "epoch": 0.21994448866138488,
-      "grad_norm": 436.11798095703125,
-      "learning_rate": 4.791744508937256e-05,
-      "loss": 80.6678,
-      "step": 54440
-    },
-    {
-      "epoch": 0.2199848899267525,
-      "grad_norm": 1142.2906494140625,
-      "learning_rate": 4.791605006446291e-05,
-      "loss": 63.2783,
-      "step": 54450
-    },
-    {
-      "epoch": 0.22002529119212014,
-      "grad_norm": 815.8292846679688,
-      "learning_rate": 4.7914654592793065e-05,
-      "loss": 77.567,
-      "step": 54460
-    },
-    {
-      "epoch": 0.22006569245748778,
-      "grad_norm": 920.00439453125,
-      "learning_rate": 4.791325867439024e-05,
-      "loss": 57.8406,
-      "step": 54470
-    },
-    {
-      "epoch": 0.2201060937228554,
-      "grad_norm": 752.0264282226562,
-      "learning_rate": 4.791186230928163e-05,
-      "loss": 94.0675,
-      "step": 54480
-    },
-    {
-      "epoch": 0.22014649498822303,
-      "grad_norm": 1893.0584716796875,
-      "learning_rate": 4.7910465497494474e-05,
-      "loss": 136.9929,
-      "step": 54490
-    },
-    {
-      "epoch": 0.22018689625359067,
-      "grad_norm": 634.4692993164062,
-      "learning_rate": 4.790906823905599e-05,
-      "loss": 69.6565,
-      "step": 54500
-    },
-    {
-      "epoch": 0.22022729751895828,
-      "grad_norm": 395.0136413574219,
-      "learning_rate": 4.790767053399343e-05,
-      "loss": 33.5299,
-      "step": 54510
-    },
-    {
-      "epoch": 0.22026769878432592,
-      "grad_norm": 1267.1798095703125,
-      "learning_rate": 4.790627238233405e-05,
-      "loss": 52.7564,
-      "step": 54520
-    },
-    {
-      "epoch": 0.22030810004969356,
-      "grad_norm": 1663.51611328125,
-      "learning_rate": 4.790487378410509e-05,
-      "loss": 97.8893,
-      "step": 54530
-    },
-    {
-      "epoch": 0.2203485013150612,
-      "grad_norm": 279.7654113769531,
-      "learning_rate": 4.790347473933382e-05,
-      "loss": 49.2227,
-      "step": 54540
-    },
-    {
-      "epoch": 0.2203889025804288,
-      "grad_norm": 1237.12255859375,
-      "learning_rate": 4.7902075248047515e-05,
-      "loss": 93.2389,
-      "step": 54550
-    },
-    {
-      "epoch": 0.22042930384579645,
-      "grad_norm": 1792.2801513671875,
-      "learning_rate": 4.7900675310273466e-05,
-      "loss": 65.3986,
-      "step": 54560
-    },
-    {
-      "epoch": 0.2204697051111641,
-      "grad_norm": 923.925048828125,
-      "learning_rate": 4.7899274926038976e-05,
-      "loss": 60.5506,
-      "step": 54570
-    },
-    {
-      "epoch": 0.2205101063765317,
-      "grad_norm": 624.0264282226562,
-      "learning_rate": 4.789787409537131e-05,
-      "loss": 46.6642,
-      "step": 54580
-    },
-    {
-      "epoch": 0.22055050764189935,
-      "grad_norm": 732.738037109375,
-      "learning_rate": 4.789647281829781e-05,
-      "loss": 118.6461,
-      "step": 54590
-    },
-    {
-      "epoch": 0.22059090890726699,
-      "grad_norm": 654.3661499023438,
-      "learning_rate": 4.789507109484579e-05,
-      "loss": 108.6401,
-      "step": 54600
-    },
-    {
-      "epoch": 0.2206313101726346,
-      "grad_norm": 628.9659423828125,
-      "learning_rate": 4.7893668925042565e-05,
-      "loss": 56.0631,
-      "step": 54610
-    },
-    {
-      "epoch": 0.22067171143800224,
-      "grad_norm": 1165.976806640625,
-      "learning_rate": 4.789226630891548e-05,
-      "loss": 67.4176,
-      "step": 54620
-    },
-    {
-      "epoch": 0.22071211270336988,
-      "grad_norm": 2275.9443359375,
-      "learning_rate": 4.789086324649187e-05,
-      "loss": 81.9733,
-      "step": 54630
-    },
-    {
-      "epoch": 0.2207525139687375,
-      "grad_norm": 1239.550537109375,
-      "learning_rate": 4.78894597377991e-05,
-      "loss": 81.8693,
-      "step": 54640
-    },
-    {
-      "epoch": 0.22079291523410513,
-      "grad_norm": 474.0306701660156,
-      "learning_rate": 4.788805578286454e-05,
-      "loss": 70.7014,
-      "step": 54650
-    },
-    {
-      "epoch": 0.22083331649947277,
-      "grad_norm": 1004.7076416015625,
-      "learning_rate": 4.788665138171553e-05,
-      "loss": 79.816,
-      "step": 54660
-    },
-    {
-      "epoch": 0.22087371776484038,
-      "grad_norm": 536.8333129882812,
-      "learning_rate": 4.788524653437948e-05,
-      "loss": 107.741,
-      "step": 54670
-    },
-    {
-      "epoch": 0.22091411903020802,
-      "grad_norm": 1520.8243408203125,
-      "learning_rate": 4.7883841240883766e-05,
-      "loss": 67.9091,
-      "step": 54680
-    },
-    {
-      "epoch": 0.22095452029557566,
-      "grad_norm": 978.5899047851562,
-      "learning_rate": 4.7882435501255785e-05,
-      "loss": 61.7931,
-      "step": 54690
-    },
-    {
-      "epoch": 0.22099492156094327,
-      "grad_norm": 2291.6357421875,
-      "learning_rate": 4.788102931552294e-05,
-      "loss": 66.0947,
-      "step": 54700
-    },
-    {
-      "epoch": 0.22103532282631091,
-      "grad_norm": 1223.5504150390625,
-      "learning_rate": 4.787962268371266e-05,
-      "loss": 96.6182,
-      "step": 54710
-    },
-    {
-      "epoch": 0.22107572409167855,
-      "grad_norm": 753.4441528320312,
-      "learning_rate": 4.7878215605852336e-05,
-      "loss": 78.4969,
-      "step": 54720
-    },
-    {
-      "epoch": 0.2211161253570462,
-      "grad_norm": 2007.7423095703125,
-      "learning_rate": 4.7876808081969436e-05,
-      "loss": 88.9049,
-      "step": 54730
-    },
-    {
-      "epoch": 0.2211565266224138,
-      "grad_norm": 520.70361328125,
-      "learning_rate": 4.787540011209138e-05,
-      "loss": 55.8599,
-      "step": 54740
-    },
-    {
-      "epoch": 0.22119692788778145,
-      "grad_norm": 509.8031005859375,
-      "learning_rate": 4.7873991696245624e-05,
-      "loss": 68.4668,
-      "step": 54750
-    },
-    {
-      "epoch": 0.2212373291531491,
-      "grad_norm": 1531.3848876953125,
-      "learning_rate": 4.787258283445962e-05,
-      "loss": 109.6678,
-      "step": 54760
-    },
-    {
-      "epoch": 0.2212777304185167,
-      "grad_norm": 1034.49755859375,
-      "learning_rate": 4.7871173526760835e-05,
-      "loss": 82.1401,
-      "step": 54770
-    },
-    {
-      "epoch": 0.22131813168388434,
-      "grad_norm": 1044.2027587890625,
-      "learning_rate": 4.7869763773176756e-05,
-      "loss": 55.1931,
-      "step": 54780
-    },
-    {
-      "epoch": 0.22135853294925198,
-      "grad_norm": 937.3899536132812,
-      "learning_rate": 4.786835357373486e-05,
-      "loss": 86.6882,
-      "step": 54790
-    },
-    {
-      "epoch": 0.2213989342146196,
-      "grad_norm": 816.9603881835938,
-      "learning_rate": 4.7866942928462625e-05,
-      "loss": 68.9065,
-      "step": 54800
-    },
-    {
-      "epoch": 0.22143933547998723,
-      "grad_norm": 1069.583984375,
-      "learning_rate": 4.7865531837387576e-05,
-      "loss": 79.5692,
-      "step": 54810
-    },
-    {
-      "epoch": 0.22147973674535487,
-      "grad_norm": 859.8712768554688,
-      "learning_rate": 4.7864120300537206e-05,
-      "loss": 52.4455,
-      "step": 54820
-    },
-    {
-      "epoch": 0.22152013801072248,
-      "grad_norm": 861.6782836914062,
-      "learning_rate": 4.786270831793904e-05,
-      "loss": 67.6935,
-      "step": 54830
-    },
-    {
-      "epoch": 0.22156053927609012,
-      "grad_norm": 382.9268798828125,
-      "learning_rate": 4.786129588962061e-05,
-      "loss": 85.1282,
-      "step": 54840
-    },
-    {
-      "epoch": 0.22160094054145776,
-      "grad_norm": 1050.475830078125,
-      "learning_rate": 4.785988301560944e-05,
-      "loss": 103.8309,
-      "step": 54850
-    },
-    {
-      "epoch": 0.22164134180682538,
-      "grad_norm": 691.6054077148438,
-      "learning_rate": 4.785846969593308e-05,
-      "loss": 68.233,
-      "step": 54860
-    },
-    {
-      "epoch": 0.22168174307219302,
-      "grad_norm": 2130.216064453125,
-      "learning_rate": 4.785705593061909e-05,
-      "loss": 68.9335,
-      "step": 54870
-    },
-    {
-      "epoch": 0.22172214433756066,
-      "grad_norm": 804.7174072265625,
-      "learning_rate": 4.7855641719695023e-05,
-      "loss": 60.727,
-      "step": 54880
-    },
-    {
-      "epoch": 0.2217625456029283,
-      "grad_norm": 634.82080078125,
-      "learning_rate": 4.785422706318846e-05,
-      "loss": 88.999,
-      "step": 54890
-    },
-    {
-      "epoch": 0.2218029468682959,
-      "grad_norm": 597.0457763671875,
-      "learning_rate": 4.785281196112698e-05,
-      "loss": 64.0293,
-      "step": 54900
-    },
-    {
-      "epoch": 0.22184334813366355,
-      "grad_norm": 1046.802001953125,
-      "learning_rate": 4.785139641353815e-05,
-      "loss": 70.1819,
-      "step": 54910
-    },
-    {
-      "epoch": 0.2218837493990312,
-      "grad_norm": 753.67236328125,
-      "learning_rate": 4.7849980420449594e-05,
-      "loss": 66.1747,
-      "step": 54920
-    },
-    {
-      "epoch": 0.2219241506643988,
-      "grad_norm": 1037.9305419921875,
-      "learning_rate": 4.7848563981888893e-05,
-      "loss": 108.7891,
-      "step": 54930
-    },
-    {
-      "epoch": 0.22196455192976644,
-      "grad_norm": 1170.2239990234375,
-      "learning_rate": 4.784714709788368e-05,
-      "loss": 102.6976,
-      "step": 54940
-    },
-    {
-      "epoch": 0.22200495319513408,
-      "grad_norm": 894.5665283203125,
-      "learning_rate": 4.7845729768461576e-05,
-      "loss": 54.2852,
-      "step": 54950
-    },
-    {
-      "epoch": 0.2220453544605017,
-      "grad_norm": 584.92431640625,
-      "learning_rate": 4.7844311993650205e-05,
-      "loss": 75.9472,
-      "step": 54960
-    },
-    {
-      "epoch": 0.22208575572586933,
-      "grad_norm": 1176.026123046875,
-      "learning_rate": 4.784289377347721e-05,
-      "loss": 88.7713,
-      "step": 54970
-    },
-    {
-      "epoch": 0.22212615699123697,
-      "grad_norm": 834.7158203125,
-      "learning_rate": 4.7841475107970244e-05,
-      "loss": 86.6488,
-      "step": 54980
-    },
-    {
-      "epoch": 0.22216655825660458,
-      "grad_norm": 1956.4654541015625,
-      "learning_rate": 4.784005599715696e-05,
-      "loss": 53.2461,
-      "step": 54990
-    },
-    {
-      "epoch": 0.22220695952197222,
-      "grad_norm": 1810.177734375,
-      "learning_rate": 4.783863644106502e-05,
-      "loss": 141.2733,
-      "step": 55000
-    },
-    {
-      "epoch": 0.22224736078733986,
-      "grad_norm": 641.3799438476562,
-      "learning_rate": 4.783721643972211e-05,
-      "loss": 62.0612,
-      "step": 55010
-    },
-    {
-      "epoch": 0.22228776205270748,
-      "grad_norm": 1118.1512451171875,
-      "learning_rate": 4.783579599315591e-05,
-      "loss": 80.4928,
-      "step": 55020
-    },
-    {
-      "epoch": 0.22232816331807512,
-      "grad_norm": 1295.068115234375,
-      "learning_rate": 4.783437510139411e-05,
-      "loss": 75.4781,
-      "step": 55030
-    },
-    {
-      "epoch": 0.22236856458344276,
-      "grad_norm": 489.80078125,
-      "learning_rate": 4.7832953764464405e-05,
-      "loss": 67.9461,
-      "step": 55040
-    },
-    {
-      "epoch": 0.2224089658488104,
-      "grad_norm": 432.6492614746094,
-      "learning_rate": 4.783153198239452e-05,
-      "loss": 66.1955,
-      "step": 55050
-    },
-    {
-      "epoch": 0.222449367114178,
-      "grad_norm": 0.0,
-      "learning_rate": 4.783010975521216e-05,
-      "loss": 61.2556,
-      "step": 55060
-    },
-    {
-      "epoch": 0.22248976837954565,
-      "grad_norm": 656.0430908203125,
-      "learning_rate": 4.7828687082945054e-05,
-      "loss": 50.6344,
-      "step": 55070
-    },
-    {
-      "epoch": 0.2225301696449133,
-      "grad_norm": 799.9983520507812,
-      "learning_rate": 4.782726396562094e-05,
-      "loss": 63.7389,
-      "step": 55080
-    },
-    {
-      "epoch": 0.2225705709102809,
-      "grad_norm": 848.791748046875,
-      "learning_rate": 4.782584040326757e-05,
-      "loss": 83.2984,
-      "step": 55090
-    },
-    {
-      "epoch": 0.22261097217564854,
-      "grad_norm": 749.4911499023438,
-      "learning_rate": 4.7824416395912686e-05,
-      "loss": 84.4783,
-      "step": 55100
-    },
-    {
-      "epoch": 0.22265137344101618,
-      "grad_norm": 546.4697265625,
-      "learning_rate": 4.782299194358405e-05,
-      "loss": 80.4849,
-      "step": 55110
-    },
-    {
-      "epoch": 0.2226917747063838,
-      "grad_norm": 587.6947021484375,
-      "learning_rate": 4.782156704630944e-05,
-      "loss": 79.875,
-      "step": 55120
-    },
-    {
-      "epoch": 0.22273217597175143,
-      "grad_norm": 1178.3653564453125,
-      "learning_rate": 4.782014170411663e-05,
-      "loss": 77.4586,
-      "step": 55130
-    },
-    {
-      "epoch": 0.22277257723711907,
-      "grad_norm": 641.4794921875,
-      "learning_rate": 4.781871591703341e-05,
-      "loss": 70.9194,
-      "step": 55140
-    },
-    {
-      "epoch": 0.22281297850248669,
-      "grad_norm": 450.2964172363281,
-      "learning_rate": 4.7817289685087577e-05,
-      "loss": 92.5157,
-      "step": 55150
-    },
-    {
-      "epoch": 0.22285337976785433,
-      "grad_norm": 1155.9171142578125,
-      "learning_rate": 4.781586300830693e-05,
-      "loss": 125.689,
-      "step": 55160
-    },
-    {
-      "epoch": 0.22289378103322197,
-      "grad_norm": 684.2454833984375,
-      "learning_rate": 4.781443588671929e-05,
-      "loss": 61.9113,
-      "step": 55170
-    },
-    {
-      "epoch": 0.22293418229858958,
-      "grad_norm": 0.0,
-      "learning_rate": 4.781300832035247e-05,
-      "loss": 66.496,
-      "step": 55180
-    },
-    {
-      "epoch": 0.22297458356395722,
-      "grad_norm": 546.8045043945312,
-      "learning_rate": 4.7811580309234314e-05,
-      "loss": 74.1194,
-      "step": 55190
-    },
-    {
-      "epoch": 0.22301498482932486,
-      "grad_norm": 559.2355346679688,
-      "learning_rate": 4.781015185339266e-05,
-      "loss": 63.7682,
-      "step": 55200
-    },
-    {
-      "epoch": 0.2230553860946925,
-      "grad_norm": 848.5513916015625,
-      "learning_rate": 4.7808722952855344e-05,
-      "loss": 59.4205,
-      "step": 55210
-    },
-    {
-      "epoch": 0.2230957873600601,
-      "grad_norm": 513.669677734375,
-      "learning_rate": 4.780729360765024e-05,
-      "loss": 69.9661,
-      "step": 55220
-    },
-    {
-      "epoch": 0.22313618862542775,
-      "grad_norm": 1898.05419921875,
-      "learning_rate": 4.7805863817805196e-05,
-      "loss": 60.6413,
-      "step": 55230
-    },
-    {
-      "epoch": 0.2231765898907954,
-      "grad_norm": 304.1728820800781,
-      "learning_rate": 4.78044335833481e-05,
-      "loss": 65.169,
-      "step": 55240
-    },
-    {
-      "epoch": 0.223216991156163,
-      "grad_norm": 1311.47265625,
-      "learning_rate": 4.780300290430682e-05,
-      "loss": 75.9994,
-      "step": 55250
-    },
-    {
-      "epoch": 0.22325739242153064,
-      "grad_norm": 867.0023193359375,
-      "learning_rate": 4.780157178070928e-05,
-      "loss": 70.7438,
-      "step": 55260
-    },
-    {
-      "epoch": 0.22329779368689828,
-      "grad_norm": 749.3482055664062,
-      "learning_rate": 4.780014021258334e-05,
-      "loss": 87.707,
-      "step": 55270
-    },
-    {
-      "epoch": 0.2233381949522659,
-      "grad_norm": 717.660888671875,
-      "learning_rate": 4.779870819995694e-05,
-      "loss": 90.9197,
-      "step": 55280
-    },
-    {
-      "epoch": 0.22337859621763353,
-      "grad_norm": 677.76708984375,
-      "learning_rate": 4.779727574285798e-05,
-      "loss": 69.9488,
-      "step": 55290
-    },
-    {
-      "epoch": 0.22341899748300117,
-      "grad_norm": 781.3521728515625,
-      "learning_rate": 4.77958428413144e-05,
-      "loss": 83.5576,
-      "step": 55300
-    },
-    {
-      "epoch": 0.2234593987483688,
-      "grad_norm": 678.9473266601562,
-      "learning_rate": 4.779440949535412e-05,
-      "loss": 74.7966,
-      "step": 55310
-    },
-    {
-      "epoch": 0.22349980001373643,
-      "grad_norm": 1800.738037109375,
-      "learning_rate": 4.779297570500509e-05,
-      "loss": 97.8668,
-      "step": 55320
-    },
-    {
-      "epoch": 0.22354020127910407,
-      "grad_norm": 642.32568359375,
-      "learning_rate": 4.779154147029527e-05,
-      "loss": 38.1043,
-      "step": 55330
-    },
-    {
-      "epoch": 0.22358060254447168,
-      "grad_norm": 955.9810180664062,
-      "learning_rate": 4.7790106791252614e-05,
-      "loss": 99.5178,
-      "step": 55340
-    },
-    {
-      "epoch": 0.22362100380983932,
-      "grad_norm": 2772.03564453125,
-      "learning_rate": 4.7788671667905096e-05,
-      "loss": 103.198,
-      "step": 55350
-    },
-    {
-      "epoch": 0.22366140507520696,
-      "grad_norm": 896.0523071289062,
-      "learning_rate": 4.7787236100280685e-05,
-      "loss": 104.9025,
-      "step": 55360
-    },
-    {
-      "epoch": 0.2237018063405746,
-      "grad_norm": 657.1527709960938,
-      "learning_rate": 4.7785800088407376e-05,
-      "loss": 64.4315,
-      "step": 55370
-    },
-    {
-      "epoch": 0.2237422076059422,
-      "grad_norm": 1311.0645751953125,
-      "learning_rate": 4.7784363632313166e-05,
-      "loss": 82.5705,
-      "step": 55380
-    },
-    {
-      "epoch": 0.22378260887130985,
-      "grad_norm": 636.6400146484375,
-      "learning_rate": 4.778292673202606e-05,
-      "loss": 69.7723,
-      "step": 55390
-    },
-    {
-      "epoch": 0.2238230101366775,
-      "grad_norm": 768.77392578125,
-      "learning_rate": 4.778148938757406e-05,
-      "loss": 88.6552,
-      "step": 55400
-    },
-    {
-      "epoch": 0.2238634114020451,
-      "grad_norm": 494.320068359375,
-      "learning_rate": 4.7780051598985196e-05,
-      "loss": 78.5235,
-      "step": 55410
-    },
-    {
-      "epoch": 0.22390381266741274,
-      "grad_norm": 525.7529907226562,
-      "learning_rate": 4.7778613366287505e-05,
-      "loss": 106.9901,
-      "step": 55420
-    },
-    {
-      "epoch": 0.22394421393278038,
-      "grad_norm": 574.625732421875,
-      "learning_rate": 4.7777174689509006e-05,
-      "loss": 81.0395,
-      "step": 55430
-    },
-    {
-      "epoch": 0.223984615198148,
-      "grad_norm": 599.45947265625,
-      "learning_rate": 4.7775735568677775e-05,
-      "loss": 68.9037,
-      "step": 55440
-    },
-    {
-      "epoch": 0.22402501646351564,
-      "grad_norm": 653.6656494140625,
-      "learning_rate": 4.777429600382185e-05,
-      "loss": 79.6398,
-      "step": 55450
-    },
-    {
-      "epoch": 0.22406541772888328,
-      "grad_norm": 372.3857116699219,
-      "learning_rate": 4.777285599496929e-05,
-      "loss": 47.1804,
-      "step": 55460
-    },
-    {
-      "epoch": 0.2241058189942509,
-      "grad_norm": 444.17919921875,
-      "learning_rate": 4.777141554214819e-05,
-      "loss": 79.5894,
-      "step": 55470
-    },
-    {
-      "epoch": 0.22414622025961853,
-      "grad_norm": 952.4342041015625,
-      "learning_rate": 4.776997464538662e-05,
-      "loss": 79.2442,
-      "step": 55480
-    },
-    {
-      "epoch": 0.22418662152498617,
-      "grad_norm": 1239.525146484375,
-      "learning_rate": 4.776853330471266e-05,
-      "loss": 70.9746,
-      "step": 55490
-    },
-    {
-      "epoch": 0.22422702279035378,
-      "grad_norm": 5761.12841796875,
-      "learning_rate": 4.776709152015443e-05,
-      "loss": 110.6402,
-      "step": 55500
-    },
-    {
-      "epoch": 0.22426742405572142,
-      "grad_norm": 5974.0244140625,
-      "learning_rate": 4.776564929174003e-05,
-      "loss": 88.7858,
-      "step": 55510
-    },
-    {
-      "epoch": 0.22430782532108906,
-      "grad_norm": 702.3343505859375,
-      "learning_rate": 4.776420661949758e-05,
-      "loss": 80.6855,
-      "step": 55520
-    },
-    {
-      "epoch": 0.2243482265864567,
-      "grad_norm": 736.3425903320312,
-      "learning_rate": 4.776276350345519e-05,
-      "loss": 74.7216,
-      "step": 55530
-    },
-    {
-      "epoch": 0.2243886278518243,
-      "grad_norm": 412.60272216796875,
-      "learning_rate": 4.776131994364102e-05,
-      "loss": 80.698,
-      "step": 55540
-    },
-    {
-      "epoch": 0.22442902911719195,
-      "grad_norm": 723.0787353515625,
-      "learning_rate": 4.775987594008319e-05,
-      "loss": 106.2656,
-      "step": 55550
-    },
-    {
-      "epoch": 0.2244694303825596,
-      "grad_norm": 598.0833129882812,
-      "learning_rate": 4.775843149280986e-05,
-      "loss": 60.9806,
-      "step": 55560
-    },
-    {
-      "epoch": 0.2245098316479272,
-      "grad_norm": 1235.12060546875,
-      "learning_rate": 4.775698660184919e-05,
-      "loss": 80.7508,
-      "step": 55570
-    },
-    {
-      "epoch": 0.22455023291329484,
-      "grad_norm": 874.450927734375,
-      "learning_rate": 4.775554126722935e-05,
-      "loss": 42.9898,
-      "step": 55580
-    },
-    {
-      "epoch": 0.22459063417866248,
-      "grad_norm": 589.5454711914062,
-      "learning_rate": 4.775409548897853e-05,
-      "loss": 94.5234,
-      "step": 55590
-    },
-    {
-      "epoch": 0.2246310354440301,
-      "grad_norm": 918.5469360351562,
-      "learning_rate": 4.775264926712489e-05,
-      "loss": 48.5788,
-      "step": 55600
-    },
-    {
-      "epoch": 0.22467143670939774,
-      "grad_norm": 1047.6563720703125,
-      "learning_rate": 4.775120260169665e-05,
-      "loss": 88.8342,
-      "step": 55610
-    },
-    {
-      "epoch": 0.22471183797476538,
-      "grad_norm": 932.34912109375,
-      "learning_rate": 4.774975549272199e-05,
-      "loss": 48.4354,
-      "step": 55620
-    },
-    {
-      "epoch": 0.224752239240133,
-      "grad_norm": 614.5720825195312,
-      "learning_rate": 4.774830794022915e-05,
-      "loss": 93.6599,
-      "step": 55630
-    },
-    {
-      "epoch": 0.22479264050550063,
-      "grad_norm": 503.9685363769531,
-      "learning_rate": 4.7746859944246325e-05,
-      "loss": 85.0896,
-      "step": 55640
-    },
-    {
-      "epoch": 0.22483304177086827,
-      "grad_norm": 662.5595703125,
-      "learning_rate": 4.774541150480175e-05,
-      "loss": 103.3519,
-      "step": 55650
-    },
-    {
-      "epoch": 0.22487344303623588,
-      "grad_norm": 982.870849609375,
-      "learning_rate": 4.7743962621923674e-05,
-      "loss": 82.5962,
-      "step": 55660
-    },
-    {
-      "epoch": 0.22491384430160352,
-      "grad_norm": 1002.5994873046875,
-      "learning_rate": 4.774251329564034e-05,
-      "loss": 64.0428,
-      "step": 55670
-    },
-    {
-      "epoch": 0.22495424556697116,
-      "grad_norm": 360.1722412109375,
-      "learning_rate": 4.7741063525980004e-05,
-      "loss": 75.0625,
-      "step": 55680
-    },
-    {
-      "epoch": 0.2249946468323388,
-      "grad_norm": 1272.3038330078125,
-      "learning_rate": 4.773961331297092e-05,
-      "loss": 114.2109,
-      "step": 55690
-    },
-    {
-      "epoch": 0.2250350480977064,
-      "grad_norm": 6723.4404296875,
-      "learning_rate": 4.773816265664136e-05,
-      "loss": 103.1604,
-      "step": 55700
-    },
-    {
-      "epoch": 0.22507544936307405,
-      "grad_norm": 2476.1787109375,
-      "learning_rate": 4.7736711557019617e-05,
-      "loss": 101.4048,
-      "step": 55710
-    },
-    {
-      "epoch": 0.2251158506284417,
-      "grad_norm": 1094.4051513671875,
-      "learning_rate": 4.7735260014133986e-05,
-      "loss": 80.1582,
-      "step": 55720
-    },
-    {
-      "epoch": 0.2251562518938093,
-      "grad_norm": 990.44677734375,
-      "learning_rate": 4.773380802801275e-05,
-      "loss": 80.2596,
-      "step": 55730
-    },
-    {
-      "epoch": 0.22519665315917695,
-      "grad_norm": 700.4454956054688,
-      "learning_rate": 4.773235559868422e-05,
-      "loss": 78.9232,
-      "step": 55740
-    },
-    {
-      "epoch": 0.22523705442454459,
-      "grad_norm": 1065.6510009765625,
-      "learning_rate": 4.773090272617672e-05,
-      "loss": 85.6272,
-      "step": 55750
-    },
-    {
-      "epoch": 0.2252774556899122,
-      "grad_norm": 810.5206909179688,
-      "learning_rate": 4.772944941051856e-05,
-      "loss": 83.4951,
-      "step": 55760
-    },
-    {
-      "epoch": 0.22531785695527984,
-      "grad_norm": 508.2962341308594,
-      "learning_rate": 4.772799565173809e-05,
-      "loss": 63.031,
-      "step": 55770
-    },
-    {
-      "epoch": 0.22535825822064748,
-      "grad_norm": 371.3773498535156,
-      "learning_rate": 4.772654144986364e-05,
-      "loss": 96.568,
-      "step": 55780
-    },
-    {
-      "epoch": 0.2253986594860151,
-      "grad_norm": 492.9206237792969,
-      "learning_rate": 4.772508680492356e-05,
-      "loss": 76.7317,
-      "step": 55790
-    },
-    {
-      "epoch": 0.22543906075138273,
-      "grad_norm": 368.9469909667969,
-      "learning_rate": 4.772363171694622e-05,
-      "loss": 62.3136,
-      "step": 55800
-    },
-    {
-      "epoch": 0.22547946201675037,
-      "grad_norm": 445.32501220703125,
-      "learning_rate": 4.7722176185959974e-05,
-      "loss": 47.5343,
-      "step": 55810
-    },
-    {
-      "epoch": 0.22551986328211798,
-      "grad_norm": 385.32098388671875,
-      "learning_rate": 4.772072021199321e-05,
-      "loss": 86.4635,
-      "step": 55820
-    },
-    {
-      "epoch": 0.22556026454748562,
-      "grad_norm": 828.4990844726562,
-      "learning_rate": 4.771926379507431e-05,
-      "loss": 79.7598,
-      "step": 55830
-    },
-    {
-      "epoch": 0.22560066581285326,
-      "grad_norm": 881.7483520507812,
-      "learning_rate": 4.7717806935231665e-05,
-      "loss": 54.1686,
-      "step": 55840
-    },
-    {
-      "epoch": 0.2256410670782209,
-      "grad_norm": 882.0201416015625,
-      "learning_rate": 4.7716349632493674e-05,
-      "loss": 61.8202,
-      "step": 55850
-    },
-    {
-      "epoch": 0.22568146834358851,
-      "grad_norm": 1207.105712890625,
-      "learning_rate": 4.7714891886888756e-05,
-      "loss": 78.0542,
-      "step": 55860
-    },
-    {
-      "epoch": 0.22572186960895615,
-      "grad_norm": 1286.038330078125,
-      "learning_rate": 4.771343369844532e-05,
-      "loss": 84.5764,
-      "step": 55870
-    },
-    {
-      "epoch": 0.2257622708743238,
-      "grad_norm": 518.7008056640625,
-      "learning_rate": 4.771197506719181e-05,
-      "loss": 68.4155,
-      "step": 55880
-    },
-    {
-      "epoch": 0.2258026721396914,
-      "grad_norm": 1427.2728271484375,
-      "learning_rate": 4.7710515993156645e-05,
-      "loss": 89.9567,
-      "step": 55890
-    },
-    {
-      "epoch": 0.22584307340505905,
-      "grad_norm": 964.0196533203125,
-      "learning_rate": 4.770905647636828e-05,
-      "loss": 68.5106,
-      "step": 55900
-    },
-    {
-      "epoch": 0.2258834746704267,
-      "grad_norm": 452.55487060546875,
-      "learning_rate": 4.770759651685517e-05,
-      "loss": 56.08,
-      "step": 55910
-    },
-    {
-      "epoch": 0.2259238759357943,
-      "grad_norm": 1233.5595703125,
-      "learning_rate": 4.770613611464577e-05,
-      "loss": 74.0022,
-      "step": 55920
-    },
-    {
-      "epoch": 0.22596427720116194,
-      "grad_norm": 1455.181640625,
-      "learning_rate": 4.7704675269768565e-05,
-      "loss": 91.9318,
-      "step": 55930
-    },
-    {
-      "epoch": 0.22600467846652958,
-      "grad_norm": 735.1796264648438,
-      "learning_rate": 4.7703213982252016e-05,
-      "loss": 85.3958,
-      "step": 55940
-    },
-    {
-      "epoch": 0.2260450797318972,
-      "grad_norm": 670.2793579101562,
-      "learning_rate": 4.770175225212463e-05,
-      "loss": 60.0693,
-      "step": 55950
-    },
-    {
-      "epoch": 0.22608548099726483,
-      "grad_norm": 820.1773071289062,
-      "learning_rate": 4.7700290079414896e-05,
-      "loss": 75.0173,
-      "step": 55960
-    },
-    {
-      "epoch": 0.22612588226263247,
-      "grad_norm": 640.2984619140625,
-      "learning_rate": 4.769882746415132e-05,
-      "loss": 58.8079,
-      "step": 55970
-    },
-    {
-      "epoch": 0.22616628352800008,
-      "grad_norm": 1904.8134765625,
-      "learning_rate": 4.769736440636241e-05,
-      "loss": 59.5396,
-      "step": 55980
-    },
-    {
-      "epoch": 0.22620668479336772,
-      "grad_norm": 764.6589965820312,
-      "learning_rate": 4.76959009060767e-05,
-      "loss": 47.2034,
-      "step": 55990
-    },
-    {
-      "epoch": 0.22624708605873536,
-      "grad_norm": 312.55859375,
-      "learning_rate": 4.769443696332272e-05,
-      "loss": 62.7167,
-      "step": 56000
-    },
-    {
-      "epoch": 0.226287487324103,
-      "grad_norm": 560.5413818359375,
-      "learning_rate": 4.7692972578129005e-05,
-      "loss": 79.5518,
-      "step": 56010
-    },
-    {
-      "epoch": 0.22632788858947062,
-      "grad_norm": 664.6395874023438,
-      "learning_rate": 4.769150775052411e-05,
-      "loss": 63.6301,
-      "step": 56020
-    },
-    {
-      "epoch": 0.22636828985483826,
-      "grad_norm": 1466.82666015625,
-      "learning_rate": 4.769004248053658e-05,
-      "loss": 63.513,
-      "step": 56030
-    },
-    {
-      "epoch": 0.2264086911202059,
-      "grad_norm": 994.0587768554688,
-      "learning_rate": 4.7688576768194994e-05,
-      "loss": 59.1224,
-      "step": 56040
-    },
-    {
-      "epoch": 0.2264490923855735,
-      "grad_norm": 700.6870727539062,
-      "learning_rate": 4.7687110613527926e-05,
-      "loss": 85.2521,
-      "step": 56050
-    },
-    {
-      "epoch": 0.22648949365094115,
-      "grad_norm": 1606.5682373046875,
-      "learning_rate": 4.7685644016563956e-05,
-      "loss": 97.3078,
-      "step": 56060
-    },
-    {
-      "epoch": 0.2265298949163088,
-      "grad_norm": 559.5364379882812,
-      "learning_rate": 4.7684176977331674e-05,
-      "loss": 75.0282,
-      "step": 56070
-    },
-    {
-      "epoch": 0.2265702961816764,
-      "grad_norm": 750.7931518554688,
-      "learning_rate": 4.768270949585968e-05,
-      "loss": 89.0941,
-      "step": 56080
-    },
-    {
-      "epoch": 0.22661069744704404,
-      "grad_norm": 843.3683471679688,
-      "learning_rate": 4.7681241572176596e-05,
-      "loss": 76.073,
-      "step": 56090
-    },
-    {
-      "epoch": 0.22665109871241168,
-      "grad_norm": 759.1117553710938,
-      "learning_rate": 4.767977320631103e-05,
-      "loss": 72.5775,
-      "step": 56100
-    },
-    {
-      "epoch": 0.2266914999777793,
-      "grad_norm": 687.8123168945312,
-      "learning_rate": 4.76783043982916e-05,
-      "loss": 53.4036,
-      "step": 56110
-    },
-    {
-      "epoch": 0.22673190124314693,
-      "grad_norm": 996.044921875,
-      "learning_rate": 4.767683514814696e-05,
-      "loss": 56.8859,
-      "step": 56120
-    },
-    {
-      "epoch": 0.22677230250851457,
-      "grad_norm": 821.6911010742188,
-      "learning_rate": 4.767536545590574e-05,
-      "loss": 66.1622,
-      "step": 56130
-    },
-    {
-      "epoch": 0.22681270377388218,
-      "grad_norm": 1311.89892578125,
-      "learning_rate": 4.767389532159659e-05,
-      "loss": 69.1287,
-      "step": 56140
-    },
-    {
-      "epoch": 0.22685310503924982,
-      "grad_norm": 734.8349609375,
-      "learning_rate": 4.7672424745248176e-05,
-      "loss": 54.9413,
-      "step": 56150
-    },
-    {
-      "epoch": 0.22689350630461746,
-      "grad_norm": 854.7682495117188,
-      "learning_rate": 4.767095372688918e-05,
-      "loss": 98.0047,
-      "step": 56160
-    },
-    {
-      "epoch": 0.2269339075699851,
-      "grad_norm": 969.171630859375,
-      "learning_rate": 4.7669482266548264e-05,
-      "loss": 98.2503,
-      "step": 56170
-    },
-    {
-      "epoch": 0.22697430883535272,
-      "grad_norm": 1126.5001220703125,
-      "learning_rate": 4.7668010364254124e-05,
-      "loss": 63.528,
-      "step": 56180
-    },
-    {
-      "epoch": 0.22701471010072036,
-      "grad_norm": 1452.1922607421875,
-      "learning_rate": 4.7666538020035445e-05,
-      "loss": 68.9039,
-      "step": 56190
-    },
-    {
-      "epoch": 0.227055111366088,
-      "grad_norm": 1715.3773193359375,
-      "learning_rate": 4.7665065233920945e-05,
-      "loss": 109.966,
-      "step": 56200
-    },
-    {
-      "epoch": 0.2270955126314556,
-      "grad_norm": 631.9252319335938,
-      "learning_rate": 4.766359200593933e-05,
-      "loss": 49.2157,
-      "step": 56210
-    },
-    {
-      "epoch": 0.22713591389682325,
-      "grad_norm": 1373.4630126953125,
-      "learning_rate": 4.766211833611931e-05,
-      "loss": 74.802,
-      "step": 56220
-    },
-    {
-      "epoch": 0.2271763151621909,
-      "grad_norm": 2058.58740234375,
-      "learning_rate": 4.766064422448964e-05,
-      "loss": 75.4964,
-      "step": 56230
-    },
-    {
-      "epoch": 0.2272167164275585,
-      "grad_norm": 2785.7197265625,
-      "learning_rate": 4.765916967107903e-05,
-      "loss": 77.155,
-      "step": 56240
-    },
-    {
-      "epoch": 0.22725711769292614,
-      "grad_norm": 593.6171264648438,
-      "learning_rate": 4.765769467591625e-05,
-      "loss": 60.3693,
-      "step": 56250
-    },
-    {
-      "epoch": 0.22729751895829378,
-      "grad_norm": 1533.2698974609375,
-      "learning_rate": 4.7656219239030046e-05,
-      "loss": 95.2811,
-      "step": 56260
-    },
-    {
-      "epoch": 0.2273379202236614,
-      "grad_norm": 806.5111083984375,
-      "learning_rate": 4.7654743360449186e-05,
-      "loss": 65.6008,
-      "step": 56270
-    },
-    {
-      "epoch": 0.22737832148902903,
-      "grad_norm": 1836.31787109375,
-      "learning_rate": 4.7653267040202436e-05,
-      "loss": 80.5326,
-      "step": 56280
-    },
-    {
-      "epoch": 0.22741872275439667,
-      "grad_norm": 733.4277954101562,
-      "learning_rate": 4.765179027831858e-05,
-      "loss": 81.6788,
-      "step": 56290
-    },
-    {
-      "epoch": 0.22745912401976429,
-      "grad_norm": 1102.4857177734375,
-      "learning_rate": 4.7650313074826425e-05,
-      "loss": 63.4363,
-      "step": 56300
-    },
-    {
-      "epoch": 0.22749952528513193,
-      "grad_norm": 1184.0565185546875,
-      "learning_rate": 4.764883542975475e-05,
-      "loss": 113.9402,
-      "step": 56310
-    },
-    {
-      "epoch": 0.22753992655049957,
-      "grad_norm": 901.2548828125,
-      "learning_rate": 4.764735734313236e-05,
-      "loss": 61.9935,
-      "step": 56320
-    },
-    {
-      "epoch": 0.2275803278158672,
-      "grad_norm": 750.1239624023438,
-      "learning_rate": 4.7645878814988075e-05,
-      "loss": 82.1202,
-      "step": 56330
-    },
-    {
-      "epoch": 0.22762072908123482,
-      "grad_norm": 1717.6053466796875,
-      "learning_rate": 4.764439984535074e-05,
-      "loss": 54.0498,
-      "step": 56340
-    },
-    {
-      "epoch": 0.22766113034660246,
-      "grad_norm": 755.3112182617188,
-      "learning_rate": 4.764292043424916e-05,
-      "loss": 46.0645,
-      "step": 56350
-    },
-    {
-      "epoch": 0.2277015316119701,
-      "grad_norm": 819.234375,
-      "learning_rate": 4.764144058171219e-05,
-      "loss": 68.3072,
-      "step": 56360
-    },
-    {
-      "epoch": 0.2277419328773377,
-      "grad_norm": 1272.74462890625,
-      "learning_rate": 4.763996028776868e-05,
-      "loss": 81.0632,
-      "step": 56370
-    },
-    {
-      "epoch": 0.22778233414270535,
-      "grad_norm": 970.59765625,
-      "learning_rate": 4.763847955244749e-05,
-      "loss": 45.2182,
-      "step": 56380
-    },
-    {
-      "epoch": 0.227822735408073,
-      "grad_norm": 895.1033325195312,
-      "learning_rate": 4.7636998375777486e-05,
-      "loss": 78.1244,
-      "step": 56390
-    },
-    {
-      "epoch": 0.2278631366734406,
-      "grad_norm": 1142.0018310546875,
-      "learning_rate": 4.763551675778755e-05,
-      "loss": 93.2416,
-      "step": 56400
-    },
-    {
-      "epoch": 0.22790353793880824,
-      "grad_norm": 694.2943725585938,
-      "learning_rate": 4.7634034698506545e-05,
-      "loss": 66.8523,
-      "step": 56410
-    },
-    {
-      "epoch": 0.22794393920417588,
-      "grad_norm": 649.4854736328125,
-      "learning_rate": 4.76325521979634e-05,
-      "loss": 94.2294,
-      "step": 56420
-    },
-    {
-      "epoch": 0.2279843404695435,
-      "grad_norm": 1363.1820068359375,
-      "learning_rate": 4.7631069256186986e-05,
-      "loss": 68.8548,
-      "step": 56430
-    },
-    {
-      "epoch": 0.22802474173491113,
-      "grad_norm": 766.2236328125,
-      "learning_rate": 4.7629585873206226e-05,
-      "loss": 98.7785,
-      "step": 56440
-    },
-    {
-      "epoch": 0.22806514300027877,
-      "grad_norm": 1092.222900390625,
-      "learning_rate": 4.7628102049050036e-05,
-      "loss": 68.331,
-      "step": 56450
-    },
-    {
-      "epoch": 0.2281055442656464,
-      "grad_norm": 745.677978515625,
-      "learning_rate": 4.7626617783747364e-05,
-      "loss": 86.9224,
-      "step": 56460
-    },
-    {
-      "epoch": 0.22814594553101403,
-      "grad_norm": 848.4603881835938,
-      "learning_rate": 4.762513307732711e-05,
-      "loss": 97.7969,
-      "step": 56470
-    },
-    {
-      "epoch": 0.22818634679638167,
-      "grad_norm": 924.7987670898438,
-      "learning_rate": 4.762364792981825e-05,
-      "loss": 68.205,
-      "step": 56480
-    },
-    {
-      "epoch": 0.2282267480617493,
-      "grad_norm": 442.6836242675781,
-      "learning_rate": 4.762216234124972e-05,
-      "loss": 125.6712,
-      "step": 56490
-    },
-    {
-      "epoch": 0.22826714932711692,
-      "grad_norm": 716.2772827148438,
-      "learning_rate": 4.762067631165049e-05,
-      "loss": 126.1072,
-      "step": 56500
-    },
-    {
-      "epoch": 0.22830755059248456,
-      "grad_norm": 956.81005859375,
-      "learning_rate": 4.761918984104953e-05,
-      "loss": 51.0658,
-      "step": 56510
-    },
-    {
-      "epoch": 0.2283479518578522,
-      "grad_norm": 379.62493896484375,
-      "learning_rate": 4.761770292947582e-05,
-      "loss": 97.7945,
-      "step": 56520
-    },
-    {
-      "epoch": 0.2283883531232198,
-      "grad_norm": 1116.5791015625,
-      "learning_rate": 4.761621557695834e-05,
-      "loss": 77.2991,
-      "step": 56530
-    },
-    {
-      "epoch": 0.22842875438858745,
-      "grad_norm": 544.430419921875,
-      "learning_rate": 4.76147277835261e-05,
-      "loss": 94.4459,
-      "step": 56540
-    },
-    {
-      "epoch": 0.2284691556539551,
-      "grad_norm": 827.6028442382812,
-      "learning_rate": 4.7613239549208106e-05,
-      "loss": 53.5641,
-      "step": 56550
-    },
-    {
-      "epoch": 0.2285095569193227,
-      "grad_norm": 1755.43505859375,
-      "learning_rate": 4.7611750874033356e-05,
-      "loss": 97.6019,
-      "step": 56560
-    },
-    {
-      "epoch": 0.22854995818469034,
-      "grad_norm": 456.3388671875,
-      "learning_rate": 4.7610261758030886e-05,
-      "loss": 65.4111,
-      "step": 56570
-    },
-    {
-      "epoch": 0.22859035945005798,
-      "grad_norm": 710.4873046875,
-      "learning_rate": 4.760877220122971e-05,
-      "loss": 70.0858,
-      "step": 56580
-    },
-    {
-      "epoch": 0.2286307607154256,
-      "grad_norm": 717.0205688476562,
-      "learning_rate": 4.76072822036589e-05,
-      "loss": 80.4853,
-      "step": 56590
-    },
-    {
-      "epoch": 0.22867116198079324,
-      "grad_norm": 813.551513671875,
-      "learning_rate": 4.760579176534747e-05,
-      "loss": 72.0974,
-      "step": 56600
-    },
-    {
-      "epoch": 0.22871156324616088,
-      "grad_norm": 1653.39111328125,
-      "learning_rate": 4.7604300886324496e-05,
-      "loss": 37.6728,
-      "step": 56610
-    },
-    {
-      "epoch": 0.2287519645115285,
-      "grad_norm": 1388.64501953125,
-      "learning_rate": 4.760280956661903e-05,
-      "loss": 102.1589,
-      "step": 56620
-    },
-    {
-      "epoch": 0.22879236577689613,
-      "grad_norm": 1519.29931640625,
-      "learning_rate": 4.760131780626017e-05,
-      "loss": 107.9805,
-      "step": 56630
-    },
-    {
-      "epoch": 0.22883276704226377,
-      "grad_norm": 967.7742919921875,
-      "learning_rate": 4.759982560527698e-05,
-      "loss": 86.072,
-      "step": 56640
-    },
-    {
-      "epoch": 0.2288731683076314,
-      "grad_norm": 1597.58642578125,
-      "learning_rate": 4.7598332963698545e-05,
-      "loss": 95.1602,
-      "step": 56650
-    },
-    {
-      "epoch": 0.22891356957299902,
-      "grad_norm": 616.4041137695312,
-      "learning_rate": 4.7596839881553976e-05,
-      "loss": 83.347,
-      "step": 56660
-    },
-    {
-      "epoch": 0.22895397083836666,
-      "grad_norm": 728.1845703125,
-      "learning_rate": 4.75953463588724e-05,
-      "loss": 58.7279,
-      "step": 56670
-    },
-    {
-      "epoch": 0.2289943721037343,
-      "grad_norm": 1369.0509033203125,
-      "learning_rate": 4.759385239568289e-05,
-      "loss": 104.2516,
-      "step": 56680
-    },
-    {
-      "epoch": 0.2290347733691019,
-      "grad_norm": 445.80267333984375,
-      "learning_rate": 4.75923579920146e-05,
-      "loss": 69.9564,
-      "step": 56690
-    },
-    {
-      "epoch": 0.22907517463446955,
-      "grad_norm": 1323.3516845703125,
-      "learning_rate": 4.7590863147896666e-05,
-      "loss": 57.2469,
-      "step": 56700
-    },
-    {
-      "epoch": 0.2291155758998372,
-      "grad_norm": 1134.4810791015625,
-      "learning_rate": 4.7589367863358225e-05,
-      "loss": 74.4222,
-      "step": 56710
-    },
-    {
-      "epoch": 0.2291559771652048,
-      "grad_norm": 1246.8629150390625,
-      "learning_rate": 4.758787213842842e-05,
-      "loss": 97.5686,
-      "step": 56720
-    },
-    {
-      "epoch": 0.22919637843057244,
-      "grad_norm": 1165.90478515625,
-      "learning_rate": 4.758637597313642e-05,
-      "loss": 84.3428,
-      "step": 56730
-    },
-    {
-      "epoch": 0.22923677969594008,
-      "grad_norm": 683.56689453125,
-      "learning_rate": 4.7584879367511395e-05,
-      "loss": 61.9244,
-      "step": 56740
-    },
-    {
-      "epoch": 0.2292771809613077,
-      "grad_norm": 2286.635498046875,
-      "learning_rate": 4.758338232158252e-05,
-      "loss": 90.8525,
-      "step": 56750
-    },
-    {
-      "epoch": 0.22931758222667534,
-      "grad_norm": 1139.4610595703125,
-      "learning_rate": 4.758188483537898e-05,
-      "loss": 93.7091,
-      "step": 56760
-    },
-    {
-      "epoch": 0.22935798349204298,
-      "grad_norm": 517.6661376953125,
-      "learning_rate": 4.758038690892997e-05,
-      "loss": 62.2874,
-      "step": 56770
-    },
-    {
-      "epoch": 0.2293983847574106,
-      "grad_norm": 657.8548583984375,
-      "learning_rate": 4.7578888542264686e-05,
-      "loss": 90.596,
-      "step": 56780
-    },
-    {
-      "epoch": 0.22943878602277823,
-      "grad_norm": 1066.094482421875,
-      "learning_rate": 4.757738973541236e-05,
-      "loss": 54.5622,
-      "step": 56790
-    },
-    {
-      "epoch": 0.22947918728814587,
-      "grad_norm": 1410.8668212890625,
-      "learning_rate": 4.7575890488402185e-05,
-      "loss": 44.4098,
-      "step": 56800
-    },
-    {
-      "epoch": 0.2295195885535135,
-      "grad_norm": 1515.7193603515625,
-      "learning_rate": 4.75743908012634e-05,
-      "loss": 89.9203,
-      "step": 56810
-    },
-    {
-      "epoch": 0.22955998981888112,
-      "grad_norm": 1256.6802978515625,
-      "learning_rate": 4.757289067402525e-05,
-      "loss": 65.3287,
-      "step": 56820
-    },
-    {
-      "epoch": 0.22960039108424876,
-      "grad_norm": 520.7095947265625,
-      "learning_rate": 4.757139010671697e-05,
-      "loss": 67.0209,
-      "step": 56830
-    },
-    {
-      "epoch": 0.2296407923496164,
-      "grad_norm": 1852.7618408203125,
-      "learning_rate": 4.7569889099367824e-05,
-      "loss": 90.3647,
-      "step": 56840
-    },
-    {
-      "epoch": 0.229681193614984,
-      "grad_norm": 888.56787109375,
-      "learning_rate": 4.7568387652007075e-05,
-      "loss": 97.766,
-      "step": 56850
-    },
-    {
-      "epoch": 0.22972159488035165,
-      "grad_norm": 515.2661743164062,
-      "learning_rate": 4.756688576466398e-05,
-      "loss": 52.9156,
-      "step": 56860
-    },
-    {
-      "epoch": 0.2297619961457193,
-      "grad_norm": 779.8905639648438,
-      "learning_rate": 4.756538343736784e-05,
-      "loss": 85.2727,
-      "step": 56870
-    },
-    {
-      "epoch": 0.2298023974110869,
-      "grad_norm": 808.3165893554688,
-      "learning_rate": 4.756388067014792e-05,
-      "loss": 89.8556,
-      "step": 56880
-    },
-    {
-      "epoch": 0.22984279867645455,
-      "grad_norm": 1065.549072265625,
-      "learning_rate": 4.7562377463033536e-05,
-      "loss": 81.1974,
-      "step": 56890
-    },
-    {
-      "epoch": 0.22988319994182219,
-      "grad_norm": 1067.1673583984375,
-      "learning_rate": 4.7560873816053984e-05,
-      "loss": 61.5579,
-      "step": 56900
-    },
-    {
-      "epoch": 0.2299236012071898,
-      "grad_norm": 768.210205078125,
-      "learning_rate": 4.755936972923859e-05,
-      "loss": 75.4602,
-      "step": 56910
-    },
-    {
-      "epoch": 0.22996400247255744,
-      "grad_norm": 333.94232177734375,
-      "learning_rate": 4.7557865202616656e-05,
-      "loss": 65.8191,
-      "step": 56920
-    },
-    {
-      "epoch": 0.23000440373792508,
-      "grad_norm": 1091.112548828125,
-      "learning_rate": 4.7556360236217534e-05,
-      "loss": 112.1934,
-      "step": 56930
-    },
-    {
-      "epoch": 0.2300448050032927,
-      "grad_norm": 1078.4996337890625,
-      "learning_rate": 4.755485483007056e-05,
-      "loss": 80.1494,
-      "step": 56940
-    },
-    {
-      "epoch": 0.23008520626866033,
-      "grad_norm": 663.4862670898438,
-      "learning_rate": 4.755334898420507e-05,
-      "loss": 81.2759,
-      "step": 56950
-    },
-    {
-      "epoch": 0.23012560753402797,
-      "grad_norm": 1670.8721923828125,
-      "learning_rate": 4.7551842698650436e-05,
-      "loss": 98.7399,
-      "step": 56960
-    },
-    {
-      "epoch": 0.2301660087993956,
-      "grad_norm": 707.2112426757812,
-      "learning_rate": 4.755033597343602e-05,
-      "loss": 84.5847,
-      "step": 56970
-    },
-    {
-      "epoch": 0.23020641006476322,
-      "grad_norm": 913.306884765625,
-      "learning_rate": 4.7548828808591195e-05,
-      "loss": 78.6086,
-      "step": 56980
-    },
-    {
-      "epoch": 0.23024681133013086,
-      "grad_norm": 390.49578857421875,
-      "learning_rate": 4.754732120414534e-05,
-      "loss": 81.2955,
-      "step": 56990
-    },
-    {
-      "epoch": 0.2302872125954985,
-      "grad_norm": 593.7713623046875,
-      "learning_rate": 4.754581316012785e-05,
-      "loss": 59.7465,
-      "step": 57000
-    },
-    {
-      "epoch": 0.23032761386086611,
-      "grad_norm": 704.7691650390625,
-      "learning_rate": 4.754430467656812e-05,
-      "loss": 61.9324,
-      "step": 57010
-    },
-    {
-      "epoch": 0.23036801512623375,
-      "grad_norm": 770.904296875,
-      "learning_rate": 4.7542795753495574e-05,
-      "loss": 68.9715,
-      "step": 57020
-    },
-    {
-      "epoch": 0.2304084163916014,
-      "grad_norm": 630.1139526367188,
-      "learning_rate": 4.754128639093961e-05,
-      "loss": 82.8383,
-      "step": 57030
-    },
-    {
-      "epoch": 0.230448817656969,
-      "grad_norm": 1761.17626953125,
-      "learning_rate": 4.753977658892967e-05,
-      "loss": 59.8425,
-      "step": 57040
-    },
-    {
-      "epoch": 0.23048921892233665,
-      "grad_norm": 1344.968017578125,
-      "learning_rate": 4.753826634749518e-05,
-      "loss": 89.4451,
-      "step": 57050
-    },
-    {
-      "epoch": 0.2305296201877043,
-      "grad_norm": 448.2859802246094,
-      "learning_rate": 4.753675566666558e-05,
-      "loss": 79.9184,
-      "step": 57060
-    },
-    {
-      "epoch": 0.2305700214530719,
-      "grad_norm": 999.4277954101562,
-      "learning_rate": 4.7535244546470325e-05,
-      "loss": 112.2555,
-      "step": 57070
-    },
-    {
-      "epoch": 0.23061042271843954,
-      "grad_norm": 1350.092041015625,
-      "learning_rate": 4.753373298693888e-05,
-      "loss": 84.6386,
-      "step": 57080
-    },
-    {
-      "epoch": 0.23065082398380718,
-      "grad_norm": 318.74359130859375,
-      "learning_rate": 4.753222098810071e-05,
-      "loss": 48.7075,
-      "step": 57090
-    },
-    {
-      "epoch": 0.2306912252491748,
-      "grad_norm": 1049.048095703125,
-      "learning_rate": 4.7530708549985287e-05,
-      "loss": 55.4899,
-      "step": 57100
-    },
-    {
-      "epoch": 0.23073162651454243,
-      "grad_norm": 2994.60107421875,
-      "learning_rate": 4.75291956726221e-05,
-      "loss": 82.0188,
-      "step": 57110
-    },
-    {
-      "epoch": 0.23077202777991007,
-      "grad_norm": 1358.6580810546875,
-      "learning_rate": 4.752768235604065e-05,
-      "loss": 84.3785,
-      "step": 57120
-    },
-    {
-      "epoch": 0.2308124290452777,
-      "grad_norm": 309.6406555175781,
-      "learning_rate": 4.7526168600270435e-05,
-      "loss": 73.9946,
-      "step": 57130
-    },
-    {
-      "epoch": 0.23085283031064532,
-      "grad_norm": 1002.0283203125,
-      "learning_rate": 4.752465440534096e-05,
-      "loss": 104.0093,
-      "step": 57140
-    },
-    {
-      "epoch": 0.23089323157601296,
-      "grad_norm": 697.1533813476562,
-      "learning_rate": 4.752313977128175e-05,
-      "loss": 45.6771,
-      "step": 57150
-    },
-    {
-      "epoch": 0.2309336328413806,
-      "grad_norm": 1255.6212158203125,
-      "learning_rate": 4.752162469812234e-05,
-      "loss": 77.5493,
-      "step": 57160
-    },
-    {
-      "epoch": 0.23097403410674822,
-      "grad_norm": 5676.9970703125,
-      "learning_rate": 4.752010918589226e-05,
-      "loss": 64.4288,
-      "step": 57170
-    },
-    {
-      "epoch": 0.23101443537211586,
-      "grad_norm": 564.4259033203125,
-      "learning_rate": 4.7518593234621056e-05,
-      "loss": 98.8771,
-      "step": 57180
-    },
-    {
-      "epoch": 0.2310548366374835,
-      "grad_norm": 1186.0013427734375,
-      "learning_rate": 4.7517076844338285e-05,
-      "loss": 67.5268,
-      "step": 57190
-    },
-    {
-      "epoch": 0.2310952379028511,
-      "grad_norm": 1269.7178955078125,
-      "learning_rate": 4.7515560015073514e-05,
-      "loss": 89.8433,
-      "step": 57200
-    },
-    {
-      "epoch": 0.23113563916821875,
-      "grad_norm": 0.0,
-      "learning_rate": 4.75140427468563e-05,
-      "loss": 60.9059,
-      "step": 57210
-    },
-    {
-      "epoch": 0.2311760404335864,
-      "grad_norm": 1260.9530029296875,
-      "learning_rate": 4.751252503971624e-05,
-      "loss": 68.3741,
-      "step": 57220
-    },
-    {
-      "epoch": 0.231216441698954,
-      "grad_norm": 1468.294677734375,
-      "learning_rate": 4.75110068936829e-05,
-      "loss": 86.1376,
-      "step": 57230
-    },
-    {
-      "epoch": 0.23125684296432164,
-      "grad_norm": 631.3265991210938,
-      "learning_rate": 4.7509488308785905e-05,
-      "loss": 92.6364,
-      "step": 57240
-    },
-    {
-      "epoch": 0.23129724422968928,
-      "grad_norm": 391.2497863769531,
-      "learning_rate": 4.7507969285054845e-05,
-      "loss": 57.7668,
-      "step": 57250
-    },
-    {
-      "epoch": 0.2313376454950569,
-      "grad_norm": 1009.0247192382812,
-      "learning_rate": 4.750644982251933e-05,
-      "loss": 60.969,
-      "step": 57260
-    },
-    {
-      "epoch": 0.23137804676042453,
-      "grad_norm": 787.4730834960938,
-      "learning_rate": 4.7504929921208984e-05,
-      "loss": 105.2391,
-      "step": 57270
-    },
-    {
-      "epoch": 0.23141844802579217,
-      "grad_norm": 649.8621215820312,
-      "learning_rate": 4.750340958115346e-05,
-      "loss": 62.9108,
-      "step": 57280
-    },
-    {
-      "epoch": 0.2314588492911598,
-      "grad_norm": 1550.597900390625,
-      "learning_rate": 4.7501888802382365e-05,
-      "loss": 96.6346,
-      "step": 57290
-    },
-    {
-      "epoch": 0.23149925055652743,
-      "grad_norm": 1971.5574951171875,
-      "learning_rate": 4.750036758492537e-05,
-      "loss": 85.7584,
-      "step": 57300
-    },
-    {
-      "epoch": 0.23153965182189506,
-      "grad_norm": 386.6150817871094,
-      "learning_rate": 4.749884592881212e-05,
-      "loss": 109.9105,
-      "step": 57310
-    },
-    {
-      "epoch": 0.2315800530872627,
-      "grad_norm": 1582.1597900390625,
-      "learning_rate": 4.749732383407229e-05,
-      "loss": 112.0731,
-      "step": 57320
-    },
-    {
-      "epoch": 0.23162045435263032,
-      "grad_norm": 0.0,
-      "learning_rate": 4.7495801300735554e-05,
-      "loss": 52.8345,
-      "step": 57330
-    },
-    {
-      "epoch": 0.23166085561799796,
-      "grad_norm": 489.4527893066406,
-      "learning_rate": 4.7494278328831584e-05,
-      "loss": 69.7233,
-      "step": 57340
-    },
-    {
-      "epoch": 0.2317012568833656,
-      "grad_norm": 483.29364013671875,
-      "learning_rate": 4.7492754918390074e-05,
-      "loss": 89.4249,
-      "step": 57350
-    },
-    {
-      "epoch": 0.2317416581487332,
-      "grad_norm": 2289.152099609375,
-      "learning_rate": 4.749123106944073e-05,
-      "loss": 75.3741,
-      "step": 57360
-    },
-    {
-      "epoch": 0.23178205941410085,
-      "grad_norm": 730.8201904296875,
-      "learning_rate": 4.748970678201326e-05,
-      "loss": 76.5618,
-      "step": 57370
-    },
-    {
-      "epoch": 0.2318224606794685,
-      "grad_norm": 604.216064453125,
-      "learning_rate": 4.7488182056137374e-05,
-      "loss": 64.0703,
-      "step": 57380
-    },
-    {
-      "epoch": 0.2318628619448361,
-      "grad_norm": 850.9320068359375,
-      "learning_rate": 4.74866568918428e-05,
-      "loss": 82.7347,
-      "step": 57390
-    },
-    {
-      "epoch": 0.23190326321020374,
-      "grad_norm": 948.7066650390625,
-      "learning_rate": 4.7485131289159276e-05,
-      "loss": 76.1124,
-      "step": 57400
-    },
-    {
-      "epoch": 0.23194366447557138,
-      "grad_norm": 985.8638916015625,
-      "learning_rate": 4.7483605248116544e-05,
-      "loss": 90.0347,
-      "step": 57410
-    },
-    {
-      "epoch": 0.231984065740939,
-      "grad_norm": 322.2492370605469,
-      "learning_rate": 4.7482078768744345e-05,
-      "loss": 97.9077,
-      "step": 57420
-    },
-    {
-      "epoch": 0.23202446700630663,
-      "grad_norm": 1223.1875,
-      "learning_rate": 4.7480551851072454e-05,
-      "loss": 53.8952,
-      "step": 57430
-    },
-    {
-      "epoch": 0.23206486827167427,
-      "grad_norm": 455.7870788574219,
-      "learning_rate": 4.747902449513063e-05,
-      "loss": 78.5205,
-      "step": 57440
-    },
-    {
-      "epoch": 0.23210526953704191,
-      "grad_norm": 1016.0787963867188,
-      "learning_rate": 4.747749670094864e-05,
-      "loss": 86.3612,
-      "step": 57450
-    },
-    {
-      "epoch": 0.23214567080240953,
-      "grad_norm": 1260.7244873046875,
-      "learning_rate": 4.7475968468556295e-05,
-      "loss": 84.6346,
-      "step": 57460
-    },
-    {
-      "epoch": 0.23218607206777717,
-      "grad_norm": 437.6191711425781,
-      "learning_rate": 4.7474439797983364e-05,
-      "loss": 107.3411,
-      "step": 57470
-    },
-    {
-      "epoch": 0.2322264733331448,
-      "grad_norm": 516.1616821289062,
-      "learning_rate": 4.7472910689259655e-05,
-      "loss": 71.8174,
-      "step": 57480
-    },
-    {
-      "epoch": 0.23226687459851242,
-      "grad_norm": 662.67822265625,
-      "learning_rate": 4.747138114241499e-05,
-      "loss": 73.287,
-      "step": 57490
-    },
-    {
-      "epoch": 0.23230727586388006,
-      "grad_norm": 1739.5025634765625,
-      "learning_rate": 4.7469851157479177e-05,
-      "loss": 92.4851,
-      "step": 57500
-    },
-    {
-      "epoch": 0.2323476771292477,
-      "grad_norm": 682.4597778320312,
-      "learning_rate": 4.746832073448205e-05,
-      "loss": 88.4044,
-      "step": 57510
-    },
-    {
-      "epoch": 0.2323880783946153,
-      "grad_norm": 797.5489501953125,
-      "learning_rate": 4.7466789873453444e-05,
-      "loss": 86.4066,
-      "step": 57520
-    },
-    {
-      "epoch": 0.23242847965998295,
-      "grad_norm": 423.3626708984375,
-      "learning_rate": 4.74652585744232e-05,
-      "loss": 62.8509,
-      "step": 57530
-    },
-    {
-      "epoch": 0.2324688809253506,
-      "grad_norm": 712.4830322265625,
-      "learning_rate": 4.746372683742117e-05,
-      "loss": 42.4057,
-      "step": 57540
-    },
-    {
-      "epoch": 0.2325092821907182,
-      "grad_norm": 799.061767578125,
-      "learning_rate": 4.746219466247722e-05,
-      "loss": 52.0032,
-      "step": 57550
-    },
-    {
-      "epoch": 0.23254968345608584,
-      "grad_norm": 945.8649291992188,
-      "learning_rate": 4.746066204962123e-05,
-      "loss": 88.8774,
-      "step": 57560
-    },
-    {
-      "epoch": 0.23259008472145348,
-      "grad_norm": 803.7813110351562,
-      "learning_rate": 4.745912899888306e-05,
-      "loss": 93.842,
-      "step": 57570
-    },
-    {
-      "epoch": 0.2326304859868211,
-      "grad_norm": 696.2904663085938,
-      "learning_rate": 4.745759551029261e-05,
-      "loss": 101.2722,
-      "step": 57580
-    },
-    {
-      "epoch": 0.23267088725218874,
-      "grad_norm": 857.1948852539062,
-      "learning_rate": 4.745606158387978e-05,
-      "loss": 61.6568,
-      "step": 57590
-    },
-    {
-      "epoch": 0.23271128851755638,
-      "grad_norm": 548.427734375,
-      "learning_rate": 4.745452721967446e-05,
-      "loss": 96.5383,
-      "step": 57600
-    },
-    {
-      "epoch": 0.23275168978292402,
-      "grad_norm": 1423.155517578125,
-      "learning_rate": 4.745299241770658e-05,
-      "loss": 85.5254,
-      "step": 57610
-    },
-    {
-      "epoch": 0.23279209104829163,
-      "grad_norm": 496.381591796875,
-      "learning_rate": 4.745145717800605e-05,
-      "loss": 67.1594,
-      "step": 57620
-    },
-    {
-      "epoch": 0.23283249231365927,
-      "grad_norm": 629.1244506835938,
-      "learning_rate": 4.74499215006028e-05,
-      "loss": 64.9573,
-      "step": 57630
-    },
-    {
-      "epoch": 0.2328728935790269,
-      "grad_norm": 903.4159545898438,
-      "learning_rate": 4.744838538552677e-05,
-      "loss": 76.1402,
-      "step": 57640
-    },
-    {
-      "epoch": 0.23291329484439452,
-      "grad_norm": 504.8460693359375,
-      "learning_rate": 4.744684883280792e-05,
-      "loss": 82.0837,
-      "step": 57650
-    },
-    {
-      "epoch": 0.23295369610976216,
-      "grad_norm": 923.25732421875,
-      "learning_rate": 4.744531184247619e-05,
-      "loss": 61.5699,
-      "step": 57660
-    },
-    {
-      "epoch": 0.2329940973751298,
-      "grad_norm": 869.8147583007812,
-      "learning_rate": 4.744377441456155e-05,
-      "loss": 68.7058,
-      "step": 57670
-    },
-    {
-      "epoch": 0.2330344986404974,
-      "grad_norm": 924.66845703125,
-      "learning_rate": 4.744223654909397e-05,
-      "loss": 109.6333,
-      "step": 57680
-    },
-    {
-      "epoch": 0.23307489990586505,
-      "grad_norm": 1206.5703125,
-      "learning_rate": 4.744069824610344e-05,
-      "loss": 95.6685,
-      "step": 57690
-    },
-    {
-      "epoch": 0.2331153011712327,
-      "grad_norm": 777.4462280273438,
-      "learning_rate": 4.743915950561994e-05,
-      "loss": 65.7932,
-      "step": 57700
-    },
-    {
-      "epoch": 0.2331557024366003,
-      "grad_norm": 741.0814819335938,
-      "learning_rate": 4.743762032767348e-05,
-      "loss": 51.2344,
-      "step": 57710
-    },
-    {
-      "epoch": 0.23319610370196794,
-      "grad_norm": 898.6174926757812,
-      "learning_rate": 4.743608071229405e-05,
-      "loss": 97.9571,
-      "step": 57720
-    },
-    {
-      "epoch": 0.23323650496733558,
-      "grad_norm": 949.9400024414062,
-      "learning_rate": 4.743454065951168e-05,
-      "loss": 88.1819,
-      "step": 57730
-    },
-    {
-      "epoch": 0.2332769062327032,
-      "grad_norm": 6401.75244140625,
-      "learning_rate": 4.743300016935639e-05,
-      "loss": 86.2424,
-      "step": 57740
-    },
-    {
-      "epoch": 0.23331730749807084,
-      "grad_norm": 1315.59130859375,
-      "learning_rate": 4.743145924185821e-05,
-      "loss": 78.7651,
-      "step": 57750
-    },
-    {
-      "epoch": 0.23335770876343848,
-      "grad_norm": 968.3670043945312,
-      "learning_rate": 4.742991787704719e-05,
-      "loss": 53.8725,
-      "step": 57760
-    },
-    {
-      "epoch": 0.2333981100288061,
-      "grad_norm": 281.5350036621094,
-      "learning_rate": 4.7428376074953365e-05,
-      "loss": 88.1397,
-      "step": 57770
-    },
-    {
-      "epoch": 0.23343851129417373,
-      "grad_norm": 1783.01318359375,
-      "learning_rate": 4.7426833835606806e-05,
-      "loss": 111.0655,
-      "step": 57780
-    },
-    {
-      "epoch": 0.23347891255954137,
-      "grad_norm": 971.6173095703125,
-      "learning_rate": 4.7425291159037575e-05,
-      "loss": 71.0971,
-      "step": 57790
-    },
-    {
-      "epoch": 0.233519313824909,
-      "grad_norm": 966.53662109375,
-      "learning_rate": 4.742374804527575e-05,
-      "loss": 85.1464,
-      "step": 57800
-    },
-    {
-      "epoch": 0.23355971509027662,
-      "grad_norm": 1798.930419921875,
-      "learning_rate": 4.742220449435141e-05,
-      "loss": 84.1092,
-      "step": 57810
-    },
-    {
-      "epoch": 0.23360011635564426,
-      "grad_norm": 1772.7415771484375,
-      "learning_rate": 4.742066050629465e-05,
-      "loss": 50.9059,
-      "step": 57820
-    },
-    {
-      "epoch": 0.2336405176210119,
-      "grad_norm": 942.0059814453125,
-      "learning_rate": 4.741911608113557e-05,
-      "loss": 77.9877,
-      "step": 57830
-    },
-    {
-      "epoch": 0.2336809188863795,
-      "grad_norm": 472.93890380859375,
-      "learning_rate": 4.741757121890428e-05,
-      "loss": 67.7341,
-      "step": 57840
-    },
-    {
-      "epoch": 0.23372132015174715,
-      "grad_norm": 731.5870971679688,
-      "learning_rate": 4.7416025919630904e-05,
-      "loss": 96.6892,
-      "step": 57850
-    },
-    {
-      "epoch": 0.2337617214171148,
-      "grad_norm": 973.33935546875,
-      "learning_rate": 4.741448018334555e-05,
-      "loss": 106.4843,
-      "step": 57860
-    },
-    {
-      "epoch": 0.2338021226824824,
-      "grad_norm": 1150.518310546875,
-      "learning_rate": 4.741293401007837e-05,
-      "loss": 57.7009,
-      "step": 57870
-    },
-    {
-      "epoch": 0.23384252394785005,
-      "grad_norm": 879.5131225585938,
-      "learning_rate": 4.741138739985951e-05,
-      "loss": 75.2222,
-      "step": 57880
-    },
-    {
-      "epoch": 0.23388292521321769,
-      "grad_norm": 1559.7349853515625,
-      "learning_rate": 4.7409840352719106e-05,
-      "loss": 95.9505,
-      "step": 57890
-    },
-    {
-      "epoch": 0.2339233264785853,
-      "grad_norm": 1342.956298828125,
-      "learning_rate": 4.740829286868733e-05,
-      "loss": 74.7077,
-      "step": 57900
-    },
-    {
-      "epoch": 0.23396372774395294,
-      "grad_norm": 1618.971923828125,
-      "learning_rate": 4.740674494779435e-05,
-      "loss": 72.2942,
-      "step": 57910
-    },
-    {
-      "epoch": 0.23400412900932058,
-      "grad_norm": 598.3280029296875,
-      "learning_rate": 4.740519659007033e-05,
-      "loss": 71.8578,
-      "step": 57920
-    },
-    {
-      "epoch": 0.2340445302746882,
-      "grad_norm": 1528.6131591796875,
-      "learning_rate": 4.7403647795545484e-05,
-      "loss": 109.1537,
-      "step": 57930
-    },
-    {
-      "epoch": 0.23408493154005583,
-      "grad_norm": 1394.4090576171875,
-      "learning_rate": 4.7402098564249974e-05,
-      "loss": 58.2658,
-      "step": 57940
-    },
-    {
-      "epoch": 0.23412533280542347,
-      "grad_norm": 415.70263671875,
-      "learning_rate": 4.7400548896214024e-05,
-      "loss": 67.7115,
-      "step": 57950
-    },
-    {
-      "epoch": 0.2341657340707911,
-      "grad_norm": 928.4489135742188,
-      "learning_rate": 4.739899879146785e-05,
-      "loss": 54.671,
-      "step": 57960
-    },
-    {
-      "epoch": 0.23420613533615872,
-      "grad_norm": 995.4201049804688,
-      "learning_rate": 4.739744825004165e-05,
-      "loss": 85.7247,
-      "step": 57970
-    },
-    {
-      "epoch": 0.23424653660152636,
-      "grad_norm": 950.1658935546875,
-      "learning_rate": 4.739589727196568e-05,
-      "loss": 53.7052,
-      "step": 57980
-    },
-    {
-      "epoch": 0.234286937866894,
-      "grad_norm": 431.6142883300781,
-      "learning_rate": 4.739434585727015e-05,
-      "loss": 56.5389,
-      "step": 57990
-    },
-    {
-      "epoch": 0.23432733913226161,
-      "grad_norm": 1080.2391357421875,
-      "learning_rate": 4.7392794005985326e-05,
-      "loss": 82.0306,
-      "step": 58000
-    },
-    {
-      "epoch": 0.23436774039762925,
-      "grad_norm": 1104.7730712890625,
-      "learning_rate": 4.739124171814145e-05,
-      "loss": 64.9178,
-      "step": 58010
-    },
-    {
-      "epoch": 0.2344081416629969,
-      "grad_norm": 1969.99755859375,
-      "learning_rate": 4.7389688993768786e-05,
-      "loss": 97.3219,
-      "step": 58020
-    },
-    {
-      "epoch": 0.2344485429283645,
-      "grad_norm": 792.9730224609375,
-      "learning_rate": 4.738813583289762e-05,
-      "loss": 55.4297,
-      "step": 58030
-    },
-    {
-      "epoch": 0.23448894419373215,
-      "grad_norm": 1705.9310302734375,
-      "learning_rate": 4.7386582235558205e-05,
-      "loss": 105.6232,
-      "step": 58040
-    },
-    {
-      "epoch": 0.2345293454590998,
-      "grad_norm": 630.877197265625,
-      "learning_rate": 4.738502820178085e-05,
-      "loss": 72.6224,
-      "step": 58050
-    },
-    {
-      "epoch": 0.2345697467244674,
-      "grad_norm": 656.330322265625,
-      "learning_rate": 4.738347373159585e-05,
-      "loss": 106.0807,
-      "step": 58060
-    },
-    {
-      "epoch": 0.23461014798983504,
-      "grad_norm": 917.6051635742188,
-      "learning_rate": 4.73819188250335e-05,
-      "loss": 82.6257,
-      "step": 58070
-    },
-    {
-      "epoch": 0.23465054925520268,
-      "grad_norm": 574.9983520507812,
-      "learning_rate": 4.738036348212412e-05,
-      "loss": 86.4343,
-      "step": 58080
-    },
-    {
-      "epoch": 0.2346909505205703,
-      "grad_norm": 593.380859375,
-      "learning_rate": 4.737880770289803e-05,
-      "loss": 84.0278,
-      "step": 58090
-    },
-    {
-      "epoch": 0.23473135178593793,
-      "grad_norm": 342.4287109375,
-      "learning_rate": 4.737725148738557e-05,
-      "loss": 48.0205,
-      "step": 58100
-    },
-    {
-      "epoch": 0.23477175305130557,
-      "grad_norm": 520.4551391601562,
-      "learning_rate": 4.737569483561707e-05,
-      "loss": 48.0131,
-      "step": 58110
-    },
-    {
-      "epoch": 0.2348121543166732,
-      "grad_norm": 431.37115478515625,
-      "learning_rate": 4.737413774762287e-05,
-      "loss": 49.1139,
-      "step": 58120
-    },
-    {
-      "epoch": 0.23485255558204082,
-      "grad_norm": 1225.8043212890625,
-      "learning_rate": 4.737258022343335e-05,
-      "loss": 55.6529,
-      "step": 58130
-    },
-    {
-      "epoch": 0.23489295684740846,
-      "grad_norm": 475.2739562988281,
-      "learning_rate": 4.737102226307884e-05,
-      "loss": 72.1621,
-      "step": 58140
-    },
-    {
-      "epoch": 0.2349333581127761,
-      "grad_norm": 5526.28271484375,
-      "learning_rate": 4.736946386658976e-05,
-      "loss": 81.7198,
-      "step": 58150
-    },
-    {
-      "epoch": 0.23497375937814372,
-      "grad_norm": 836.2157592773438,
-      "learning_rate": 4.7367905033996445e-05,
-      "loss": 64.377,
-      "step": 58160
-    },
-    {
-      "epoch": 0.23501416064351136,
-      "grad_norm": 880.4158935546875,
-      "learning_rate": 4.736634576532931e-05,
-      "loss": 92.249,
-      "step": 58170
-    },
-    {
-      "epoch": 0.235054561908879,
-      "grad_norm": 582.840576171875,
-      "learning_rate": 4.736478606061875e-05,
-      "loss": 52.3027,
-      "step": 58180
-    },
-    {
-      "epoch": 0.2350949631742466,
-      "grad_norm": 587.2545776367188,
-      "learning_rate": 4.7363225919895185e-05,
-      "loss": 57.8036,
-      "step": 58190
-    },
-    {
-      "epoch": 0.23513536443961425,
-      "grad_norm": 975.4487915039062,
-      "learning_rate": 4.7361665343189e-05,
-      "loss": 74.3185,
-      "step": 58200
-    },
-    {
-      "epoch": 0.2351757657049819,
-      "grad_norm": 1377.298583984375,
-      "learning_rate": 4.736010433053064e-05,
-      "loss": 80.2555,
-      "step": 58210
-    },
-    {
-      "epoch": 0.2352161669703495,
-      "grad_norm": 806.4180297851562,
-      "learning_rate": 4.735854288195054e-05,
-      "loss": 63.3192,
-      "step": 58220
-    },
-    {
-      "epoch": 0.23525656823571714,
-      "grad_norm": 1805.80810546875,
-      "learning_rate": 4.735698099747913e-05,
-      "loss": 73.7252,
-      "step": 58230
-    },
-    {
-      "epoch": 0.23529696950108478,
-      "grad_norm": 1094.813232421875,
-      "learning_rate": 4.735541867714687e-05,
-      "loss": 114.6661,
-      "step": 58240
-    },
-    {
-      "epoch": 0.2353373707664524,
-      "grad_norm": 3966.76123046875,
-      "learning_rate": 4.73538559209842e-05,
-      "loss": 168.1956,
-      "step": 58250
-    },
-    {
-      "epoch": 0.23537777203182003,
-      "grad_norm": 496.0440673828125,
-      "learning_rate": 4.735229272902162e-05,
-      "loss": 90.3166,
-      "step": 58260
-    },
-    {
-      "epoch": 0.23541817329718767,
-      "grad_norm": 639.0549926757812,
-      "learning_rate": 4.735072910128957e-05,
-      "loss": 101.5113,
-      "step": 58270
-    },
-    {
-      "epoch": 0.2354585745625553,
-      "grad_norm": 643.226318359375,
-      "learning_rate": 4.734916503781856e-05,
-      "loss": 46.3206,
-      "step": 58280
-    },
-    {
-      "epoch": 0.23549897582792292,
-      "grad_norm": 1588.71142578125,
-      "learning_rate": 4.7347600538639067e-05,
-      "loss": 82.4462,
-      "step": 58290
-    },
-    {
-      "epoch": 0.23553937709329056,
-      "grad_norm": 306.2060852050781,
-      "learning_rate": 4.73460356037816e-05,
-      "loss": 86.551,
-      "step": 58300
-    },
-    {
-      "epoch": 0.2355797783586582,
-      "grad_norm": 838.1617431640625,
-      "learning_rate": 4.734447023327666e-05,
-      "loss": 62.018,
-      "step": 58310
-    },
-    {
-      "epoch": 0.23562017962402582,
-      "grad_norm": 839.5139770507812,
-      "learning_rate": 4.7342904427154766e-05,
-      "loss": 68.918,
-      "step": 58320
-    },
-    {
-      "epoch": 0.23566058088939346,
-      "grad_norm": 2884.371826171875,
-      "learning_rate": 4.734133818544645e-05,
-      "loss": 93.363,
-      "step": 58330
-    },
-    {
-      "epoch": 0.2357009821547611,
-      "grad_norm": 1337.994873046875,
-      "learning_rate": 4.733977150818225e-05,
-      "loss": 80.4189,
-      "step": 58340
-    },
-    {
-      "epoch": 0.2357413834201287,
-      "grad_norm": 916.3220825195312,
-      "learning_rate": 4.7338204395392694e-05,
-      "loss": 64.318,
-      "step": 58350
-    },
-    {
-      "epoch": 0.23578178468549635,
-      "grad_norm": 4077.7041015625,
-      "learning_rate": 4.733663684710835e-05,
-      "loss": 106.2886,
-      "step": 58360
-    },
-    {
-      "epoch": 0.235822185950864,
-      "grad_norm": 471.49566650390625,
-      "learning_rate": 4.7335068863359764e-05,
-      "loss": 76.7145,
-      "step": 58370
-    },
-    {
-      "epoch": 0.2358625872162316,
-      "grad_norm": 821.7672119140625,
-      "learning_rate": 4.733350044417752e-05,
-      "loss": 59.775,
-      "step": 58380
-    },
-    {
-      "epoch": 0.23590298848159924,
-      "grad_norm": 628.004638671875,
-      "learning_rate": 4.733193158959218e-05,
-      "loss": 65.8933,
-      "step": 58390
-    },
-    {
-      "epoch": 0.23594338974696688,
-      "grad_norm": 487.54107666015625,
-      "learning_rate": 4.733036229963435e-05,
-      "loss": 78.1683,
-      "step": 58400
-    },
-    {
-      "epoch": 0.2359837910123345,
-      "grad_norm": 967.7349243164062,
-      "learning_rate": 4.732879257433459e-05,
-      "loss": 87.9057,
-      "step": 58410
-    },
-    {
-      "epoch": 0.23602419227770213,
-      "grad_norm": 834.74755859375,
-      "learning_rate": 4.7327222413723536e-05,
-      "loss": 74.7745,
-      "step": 58420
-    },
-    {
-      "epoch": 0.23606459354306977,
-      "grad_norm": 1843.8294677734375,
-      "learning_rate": 4.7325651817831784e-05,
-      "loss": 69.399,
-      "step": 58430
-    },
-    {
-      "epoch": 0.2361049948084374,
-      "grad_norm": 900.3565063476562,
-      "learning_rate": 4.732408078668995e-05,
-      "loss": 70.5233,
-      "step": 58440
-    },
-    {
-      "epoch": 0.23614539607380503,
-      "grad_norm": 420.2508544921875,
-      "learning_rate": 4.7322509320328675e-05,
-      "loss": 67.6304,
-      "step": 58450
-    },
-    {
-      "epoch": 0.23618579733917267,
-      "grad_norm": 1126.9072265625,
-      "learning_rate": 4.732093741877859e-05,
-      "loss": 95.9313,
-      "step": 58460
-    },
-    {
-      "epoch": 0.2362261986045403,
-      "grad_norm": 496.6501770019531,
-      "learning_rate": 4.731936508207033e-05,
-      "loss": 65.5002,
-      "step": 58470
-    },
-    {
-      "epoch": 0.23626659986990792,
-      "grad_norm": 511.6619567871094,
-      "learning_rate": 4.731779231023456e-05,
-      "loss": 68.3283,
-      "step": 58480
-    },
-    {
-      "epoch": 0.23630700113527556,
-      "grad_norm": 728.8594970703125,
-      "learning_rate": 4.731621910330194e-05,
-      "loss": 71.2413,
-      "step": 58490
-    },
-    {
-      "epoch": 0.2363474024006432,
-      "grad_norm": 721.872802734375,
-      "learning_rate": 4.731464546130314e-05,
-      "loss": 64.1135,
-      "step": 58500
-    },
-    {
-      "epoch": 0.2363878036660108,
-      "grad_norm": 557.9456787109375,
-      "learning_rate": 4.7313071384268836e-05,
-      "loss": 71.9643,
-      "step": 58510
-    },
-    {
-      "epoch": 0.23642820493137845,
-      "grad_norm": 952.9915771484375,
-      "learning_rate": 4.731149687222972e-05,
-      "loss": 72.1631,
-      "step": 58520
-    },
-    {
-      "epoch": 0.2364686061967461,
-      "grad_norm": 406.0745544433594,
-      "learning_rate": 4.7309921925216484e-05,
-      "loss": 87.8844,
-      "step": 58530
-    },
-    {
-      "epoch": 0.2365090074621137,
-      "grad_norm": 6206.943359375,
-      "learning_rate": 4.730834654325984e-05,
-      "loss": 117.5255,
-      "step": 58540
-    },
-    {
-      "epoch": 0.23654940872748134,
-      "grad_norm": 0.0,
-      "learning_rate": 4.7306770726390496e-05,
-      "loss": 52.3261,
-      "step": 58550
-    },
-    {
-      "epoch": 0.23658980999284898,
-      "grad_norm": 1959.0308837890625,
-      "learning_rate": 4.730519447463916e-05,
-      "loss": 87.7869,
-      "step": 58560
-    },
-    {
-      "epoch": 0.2366302112582166,
-      "grad_norm": 1353.8363037109375,
-      "learning_rate": 4.730361778803658e-05,
-      "loss": 92.4961,
-      "step": 58570
-    },
-    {
-      "epoch": 0.23667061252358423,
-      "grad_norm": 1127.2203369140625,
-      "learning_rate": 4.730204066661349e-05,
-      "loss": 101.3056,
-      "step": 58580
-    },
-    {
-      "epoch": 0.23671101378895187,
-      "grad_norm": 1524.7484130859375,
-      "learning_rate": 4.730046311040064e-05,
-      "loss": 144.1944,
-      "step": 58590
-    },
-    {
-      "epoch": 0.23675141505431951,
-      "grad_norm": 831.2205200195312,
-      "learning_rate": 4.7298885119428773e-05,
-      "loss": 70.5365,
-      "step": 58600
-    },
-    {
-      "epoch": 0.23679181631968713,
-      "grad_norm": 913.74853515625,
-      "learning_rate": 4.729730669372866e-05,
-      "loss": 68.5036,
-      "step": 58610
-    },
-    {
-      "epoch": 0.23683221758505477,
-      "grad_norm": 2814.275146484375,
-      "learning_rate": 4.729572783333108e-05,
-      "loss": 90.047,
-      "step": 58620
-    },
-    {
-      "epoch": 0.2368726188504224,
-      "grad_norm": 1022.5120849609375,
-      "learning_rate": 4.72941485382668e-05,
-      "loss": 84.8219,
-      "step": 58630
-    },
-    {
-      "epoch": 0.23691302011579002,
-      "grad_norm": 665.8732299804688,
-      "learning_rate": 4.729256880856662e-05,
-      "loss": 63.3056,
-      "step": 58640
-    },
-    {
-      "epoch": 0.23695342138115766,
-      "grad_norm": 2222.1650390625,
-      "learning_rate": 4.7290988644261336e-05,
-      "loss": 77.9888,
-      "step": 58650
-    },
-    {
-      "epoch": 0.2369938226465253,
-      "grad_norm": 780.3485717773438,
-      "learning_rate": 4.728940804538176e-05,
-      "loss": 58.1334,
-      "step": 58660
-    },
-    {
-      "epoch": 0.2370342239118929,
-      "grad_norm": 568.1907958984375,
-      "learning_rate": 4.728782701195869e-05,
-      "loss": 53.3181,
-      "step": 58670
-    },
-    {
-      "epoch": 0.23707462517726055,
-      "grad_norm": 662.433349609375,
-      "learning_rate": 4.728624554402295e-05,
-      "loss": 67.1239,
-      "step": 58680
-    },
-    {
-      "epoch": 0.2371150264426282,
-      "grad_norm": 742.7056274414062,
-      "learning_rate": 4.7284663641605384e-05,
-      "loss": 56.0102,
-      "step": 58690
-    },
-    {
-      "epoch": 0.2371554277079958,
-      "grad_norm": 1382.75439453125,
-      "learning_rate": 4.728308130473683e-05,
-      "loss": 52.9646,
-      "step": 58700
-    },
-    {
-      "epoch": 0.23719582897336344,
-      "grad_norm": 1125.7822265625,
-      "learning_rate": 4.7281498533448136e-05,
-      "loss": 69.9835,
-      "step": 58710
-    },
-    {
-      "epoch": 0.23723623023873108,
-      "grad_norm": 446.69677734375,
-      "learning_rate": 4.7279915327770155e-05,
-      "loss": 54.0501,
-      "step": 58720
-    },
-    {
-      "epoch": 0.2372766315040987,
-      "grad_norm": 2016.18310546875,
-      "learning_rate": 4.7278331687733754e-05,
-      "loss": 111.3491,
-      "step": 58730
-    },
-    {
-      "epoch": 0.23731703276946634,
-      "grad_norm": 1517.979736328125,
-      "learning_rate": 4.727674761336981e-05,
-      "loss": 95.3152,
-      "step": 58740
-    },
-    {
-      "epoch": 0.23735743403483398,
-      "grad_norm": 965.1806030273438,
-      "learning_rate": 4.72751631047092e-05,
-      "loss": 77.6278,
-      "step": 58750
-    },
-    {
-      "epoch": 0.23739783530020162,
-      "grad_norm": 697.4866333007812,
-      "learning_rate": 4.727357816178282e-05,
-      "loss": 86.2706,
-      "step": 58760
-    },
-    {
-      "epoch": 0.23743823656556923,
-      "grad_norm": 521.2189331054688,
-      "learning_rate": 4.727199278462156e-05,
-      "loss": 59.1101,
-      "step": 58770
-    },
-    {
-      "epoch": 0.23747863783093687,
-      "grad_norm": 456.9306945800781,
-      "learning_rate": 4.727040697325634e-05,
-      "loss": 57.2227,
-      "step": 58780
-    },
-    {
-      "epoch": 0.2375190390963045,
-      "grad_norm": 1488.9251708984375,
-      "learning_rate": 4.726882072771807e-05,
-      "loss": 83.9987,
-      "step": 58790
-    },
-    {
-      "epoch": 0.23755944036167212,
-      "grad_norm": 816.0877075195312,
-      "learning_rate": 4.7267234048037664e-05,
-      "loss": 75.2146,
-      "step": 58800
-    },
-    {
-      "epoch": 0.23759984162703976,
-      "grad_norm": 1049.84375,
-      "learning_rate": 4.726564693424608e-05,
-      "loss": 85.2812,
-      "step": 58810
-    },
-    {
-      "epoch": 0.2376402428924074,
-      "grad_norm": 660.714599609375,
-      "learning_rate": 4.7264059386374236e-05,
-      "loss": 50.8552,
-      "step": 58820
-    },
-    {
-      "epoch": 0.237680644157775,
-      "grad_norm": 835.5247192382812,
-      "learning_rate": 4.72624714044531e-05,
-      "loss": 85.9735,
-      "step": 58830
-    },
-    {
-      "epoch": 0.23772104542314265,
-      "grad_norm": 934.1461181640625,
-      "learning_rate": 4.7260882988513624e-05,
-      "loss": 82.1574,
-      "step": 58840
-    },
-    {
-      "epoch": 0.2377614466885103,
-      "grad_norm": 253.60675048828125,
-      "learning_rate": 4.725929413858677e-05,
-      "loss": 86.5755,
-      "step": 58850
-    },
-    {
-      "epoch": 0.2378018479538779,
-      "grad_norm": 476.08544921875,
-      "learning_rate": 4.725770485470351e-05,
-      "loss": 79.0762,
-      "step": 58860
-    },
-    {
-      "epoch": 0.23784224921924554,
-      "grad_norm": 1652.21484375,
-      "learning_rate": 4.725611513689485e-05,
-      "loss": 76.6932,
-      "step": 58870
-    },
-    {
-      "epoch": 0.23788265048461318,
-      "grad_norm": 504.9270324707031,
-      "learning_rate": 4.725452498519175e-05,
-      "loss": 76.0411,
-      "step": 58880
-    },
-    {
-      "epoch": 0.2379230517499808,
-      "grad_norm": 1309.692626953125,
-      "learning_rate": 4.7252934399625234e-05,
-      "loss": 52.6607,
-      "step": 58890
-    },
-    {
-      "epoch": 0.23796345301534844,
-      "grad_norm": 1309.8441162109375,
-      "learning_rate": 4.725134338022631e-05,
-      "loss": 88.8201,
-      "step": 58900
-    },
-    {
-      "epoch": 0.23800385428071608,
-      "grad_norm": 1632.8326416015625,
-      "learning_rate": 4.7249751927025996e-05,
-      "loss": 88.7722,
-      "step": 58910
-    },
-    {
-      "epoch": 0.23804425554608372,
-      "grad_norm": 535.5170288085938,
-      "learning_rate": 4.7248160040055304e-05,
-      "loss": 47.7584,
-      "step": 58920
-    },
-    {
-      "epoch": 0.23808465681145133,
-      "grad_norm": 708.9509887695312,
-      "learning_rate": 4.724656771934528e-05,
-      "loss": 83.726,
-      "step": 58930
-    },
-    {
-      "epoch": 0.23812505807681897,
-      "grad_norm": 579.57275390625,
-      "learning_rate": 4.7244974964926965e-05,
-      "loss": 62.5475,
-      "step": 58940
-    },
-    {
-      "epoch": 0.2381654593421866,
-      "grad_norm": 651.062744140625,
-      "learning_rate": 4.724338177683141e-05,
-      "loss": 63.3703,
-      "step": 58950
-    },
-    {
-      "epoch": 0.23820586060755422,
-      "grad_norm": 1036.9290771484375,
-      "learning_rate": 4.724178815508967e-05,
-      "loss": 59.7281,
-      "step": 58960
-    },
-    {
-      "epoch": 0.23824626187292186,
-      "grad_norm": 406.64044189453125,
-      "learning_rate": 4.724019409973283e-05,
-      "loss": 75.1738,
-      "step": 58970
-    },
-    {
-      "epoch": 0.2382866631382895,
-      "grad_norm": 533.7338256835938,
-      "learning_rate": 4.723859961079195e-05,
-      "loss": 76.1549,
-      "step": 58980
-    },
-    {
-      "epoch": 0.2383270644036571,
-      "grad_norm": 595.7689819335938,
-      "learning_rate": 4.7237004688298125e-05,
-      "loss": 134.3011,
-      "step": 58990
-    },
-    {
-      "epoch": 0.23836746566902475,
-      "grad_norm": 1294.3282470703125,
-      "learning_rate": 4.723540933228244e-05,
-      "loss": 100.5167,
-      "step": 59000
-    },
-    {
-      "epoch": 0.2384078669343924,
-      "grad_norm": 612.8972778320312,
-      "learning_rate": 4.7233813542776006e-05,
-      "loss": 75.6335,
-      "step": 59010
-    },
-    {
-      "epoch": 0.23844826819976,
-      "grad_norm": 1156.35498046875,
-      "learning_rate": 4.723221731980993e-05,
-      "loss": 70.4552,
-      "step": 59020
-    },
-    {
-      "epoch": 0.23848866946512765,
-      "grad_norm": 1819.1259765625,
-      "learning_rate": 4.723062066341533e-05,
-      "loss": 64.2368,
-      "step": 59030
-    },
-    {
-      "epoch": 0.23852907073049529,
-      "grad_norm": 765.05615234375,
-      "learning_rate": 4.722902357362333e-05,
-      "loss": 80.1383,
-      "step": 59040
-    },
-    {
-      "epoch": 0.2385694719958629,
-      "grad_norm": 623.2485961914062,
-      "learning_rate": 4.7227426050465084e-05,
-      "loss": 73.9087,
-      "step": 59050
-    },
-    {
-      "epoch": 0.23860987326123054,
-      "grad_norm": 719.6350708007812,
-      "learning_rate": 4.722582809397171e-05,
-      "loss": 59.0293,
-      "step": 59060
-    },
-    {
-      "epoch": 0.23865027452659818,
-      "grad_norm": 410.1199645996094,
-      "learning_rate": 4.722422970417438e-05,
-      "loss": 58.2241,
-      "step": 59070
-    },
-    {
-      "epoch": 0.23869067579196582,
-      "grad_norm": 598.96240234375,
-      "learning_rate": 4.722263088110426e-05,
-      "loss": 79.149,
-      "step": 59080
-    },
-    {
-      "epoch": 0.23873107705733343,
-      "grad_norm": 504.0932922363281,
-      "learning_rate": 4.72210316247925e-05,
-      "loss": 67.0265,
-      "step": 59090
-    },
-    {
-      "epoch": 0.23877147832270107,
-      "grad_norm": 1157.12060546875,
-      "learning_rate": 4.721943193527029e-05,
-      "loss": 89.9269,
-      "step": 59100
-    },
-    {
-      "epoch": 0.2388118795880687,
-      "grad_norm": 1023.6410522460938,
-      "learning_rate": 4.7217831812568815e-05,
-      "loss": 59.8878,
-      "step": 59110
-    },
-    {
-      "epoch": 0.23885228085343632,
-      "grad_norm": 299.01300048828125,
-      "learning_rate": 4.721623125671927e-05,
-      "loss": 106.7344,
-      "step": 59120
-    },
-    {
-      "epoch": 0.23889268211880396,
-      "grad_norm": 490.2772521972656,
-      "learning_rate": 4.7214630267752856e-05,
-      "loss": 84.1761,
-      "step": 59130
-    },
-    {
-      "epoch": 0.2389330833841716,
-      "grad_norm": 689.8060913085938,
-      "learning_rate": 4.721302884570079e-05,
-      "loss": 61.8113,
-      "step": 59140
-    },
-    {
-      "epoch": 0.23897348464953921,
-      "grad_norm": 573.6192016601562,
-      "learning_rate": 4.7211426990594296e-05,
-      "loss": 84.5191,
-      "step": 59150
-    },
-    {
-      "epoch": 0.23901388591490685,
-      "grad_norm": 554.744384765625,
-      "learning_rate": 4.720982470246459e-05,
-      "loss": 68.6441,
-      "step": 59160
-    },
-    {
-      "epoch": 0.2390542871802745,
-      "grad_norm": 279.83795166015625,
-      "learning_rate": 4.720822198134293e-05,
-      "loss": 70.0726,
-      "step": 59170
-    },
-    {
-      "epoch": 0.2390946884456421,
-      "grad_norm": 1090.882568359375,
-      "learning_rate": 4.7206618827260534e-05,
-      "loss": 41.3233,
-      "step": 59180
-    },
-    {
-      "epoch": 0.23913508971100975,
-      "grad_norm": 2077.9443359375,
-      "learning_rate": 4.720501524024867e-05,
-      "loss": 81.099,
-      "step": 59190
-    },
-    {
-      "epoch": 0.2391754909763774,
-      "grad_norm": 537.9066772460938,
-      "learning_rate": 4.720341122033862e-05,
-      "loss": 71.1957,
-      "step": 59200
-    },
-    {
-      "epoch": 0.239215892241745,
-      "grad_norm": 1677.8251953125,
-      "learning_rate": 4.720180676756162e-05,
-      "loss": 70.0094,
-      "step": 59210
-    },
-    {
-      "epoch": 0.23925629350711264,
-      "grad_norm": 618.0469970703125,
-      "learning_rate": 4.720020188194897e-05,
-      "loss": 60.1108,
-      "step": 59220
-    },
-    {
-      "epoch": 0.23929669477248028,
-      "grad_norm": 843.9923095703125,
-      "learning_rate": 4.719859656353196e-05,
-      "loss": 54.7157,
-      "step": 59230
-    },
-    {
-      "epoch": 0.23933709603784792,
-      "grad_norm": 1101.4031982421875,
-      "learning_rate": 4.719699081234188e-05,
-      "loss": 101.9431,
-      "step": 59240
-    },
-    {
-      "epoch": 0.23937749730321553,
-      "grad_norm": 541.9828491210938,
-      "learning_rate": 4.719538462841003e-05,
-      "loss": 85.9788,
-      "step": 59250
-    },
-    {
-      "epoch": 0.23941789856858317,
-      "grad_norm": 470.3389587402344,
-      "learning_rate": 4.719377801176774e-05,
-      "loss": 73.3981,
-      "step": 59260
-    },
-    {
-      "epoch": 0.2394582998339508,
-      "grad_norm": 867.8907470703125,
-      "learning_rate": 4.719217096244631e-05,
-      "loss": 56.2378,
-      "step": 59270
-    },
-    {
-      "epoch": 0.23949870109931842,
-      "grad_norm": 1353.7008056640625,
-      "learning_rate": 4.7190563480477095e-05,
-      "loss": 57.5405,
-      "step": 59280
-    },
-    {
-      "epoch": 0.23953910236468606,
-      "grad_norm": 968.8933715820312,
-      "learning_rate": 4.718895556589141e-05,
-      "loss": 55.8982,
-      "step": 59290
-    },
-    {
-      "epoch": 0.2395795036300537,
-      "grad_norm": 1031.3507080078125,
-      "learning_rate": 4.718734721872062e-05,
-      "loss": 72.3141,
-      "step": 59300
-    },
-    {
-      "epoch": 0.23961990489542132,
-      "grad_norm": 531.7722778320312,
-      "learning_rate": 4.718573843899607e-05,
-      "loss": 79.3897,
-      "step": 59310
-    },
-    {
-      "epoch": 0.23966030616078896,
-      "grad_norm": 1232.718994140625,
-      "learning_rate": 4.718412922674913e-05,
-      "loss": 79.8532,
-      "step": 59320
-    },
-    {
-      "epoch": 0.2397007074261566,
-      "grad_norm": 3018.275390625,
-      "learning_rate": 4.718251958201117e-05,
-      "loss": 88.7309,
-      "step": 59330
-    },
-    {
-      "epoch": 0.2397411086915242,
-      "grad_norm": 1827.9061279296875,
-      "learning_rate": 4.718090950481356e-05,
-      "loss": 101.574,
-      "step": 59340
-    },
-    {
-      "epoch": 0.23978150995689185,
-      "grad_norm": 715.8681640625,
-      "learning_rate": 4.71792989951877e-05,
-      "loss": 72.9536,
-      "step": 59350
-    },
-    {
-      "epoch": 0.2398219112222595,
-      "grad_norm": 1006.0107421875,
-      "learning_rate": 4.717768805316501e-05,
-      "loss": 67.8689,
-      "step": 59360
-    },
-    {
-      "epoch": 0.2398623124876271,
-      "grad_norm": 1562.623779296875,
-      "learning_rate": 4.717607667877685e-05,
-      "loss": 94.7435,
-      "step": 59370
-    },
-    {
-      "epoch": 0.23990271375299474,
-      "grad_norm": 822.7528686523438,
-      "learning_rate": 4.717446487205466e-05,
-      "loss": 71.5322,
-      "step": 59380
-    },
-    {
-      "epoch": 0.23994311501836238,
-      "grad_norm": 733.3274536132812,
-      "learning_rate": 4.717285263302987e-05,
-      "loss": 63.2148,
-      "step": 59390
-    },
-    {
-      "epoch": 0.23998351628373002,
-      "grad_norm": 1009.9617919921875,
-      "learning_rate": 4.71712399617339e-05,
-      "loss": 92.269,
-      "step": 59400
-    },
-    {
-      "epoch": 0.24002391754909763,
-      "grad_norm": 768.0454711914062,
-      "learning_rate": 4.716962685819819e-05,
-      "loss": 46.8281,
-      "step": 59410
-    },
-    {
-      "epoch": 0.24006431881446527,
-      "grad_norm": 392.7835693359375,
-      "learning_rate": 4.716801332245419e-05,
-      "loss": 57.8384,
-      "step": 59420
-    },
-    {
-      "epoch": 0.2401047200798329,
-      "grad_norm": 1200.4879150390625,
-      "learning_rate": 4.7166399354533365e-05,
-      "loss": 63.8987,
-      "step": 59430
-    },
-    {
-      "epoch": 0.24014512134520052,
-      "grad_norm": 540.96142578125,
-      "learning_rate": 4.7164784954467166e-05,
-      "loss": 86.5465,
-      "step": 59440
-    },
-    {
-      "epoch": 0.24018552261056816,
-      "grad_norm": 679.489501953125,
-      "learning_rate": 4.716317012228707e-05,
-      "loss": 52.6366,
-      "step": 59450
-    },
-    {
-      "epoch": 0.2402259238759358,
-      "grad_norm": 1019.4818115234375,
-      "learning_rate": 4.716155485802457e-05,
-      "loss": 55.8305,
-      "step": 59460
-    },
-    {
-      "epoch": 0.24026632514130342,
-      "grad_norm": 330.5364990234375,
-      "learning_rate": 4.715993916171114e-05,
-      "loss": 42.9955,
-      "step": 59470
-    },
-    {
-      "epoch": 0.24030672640667106,
-      "grad_norm": 1147.410888671875,
-      "learning_rate": 4.715832303337829e-05,
-      "loss": 93.0537,
-      "step": 59480
-    },
-    {
-      "epoch": 0.2403471276720387,
-      "grad_norm": 1502.9691162109375,
-      "learning_rate": 4.715670647305753e-05,
-      "loss": 66.4446,
-      "step": 59490
-    },
-    {
-      "epoch": 0.2403875289374063,
-      "grad_norm": 1007.5733032226562,
-      "learning_rate": 4.715508948078037e-05,
-      "loss": 68.1062,
-      "step": 59500
-    },
-    {
-      "epoch": 0.24042793020277395,
-      "grad_norm": 1082.7093505859375,
-      "learning_rate": 4.715347205657833e-05,
-      "loss": 85.7707,
-      "step": 59510
-    },
-    {
-      "epoch": 0.2404683314681416,
-      "grad_norm": 591.158447265625,
-      "learning_rate": 4.715185420048295e-05,
-      "loss": 53.6973,
-      "step": 59520
-    },
-    {
-      "epoch": 0.2405087327335092,
-      "grad_norm": 357.9864196777344,
-      "learning_rate": 4.715023591252576e-05,
-      "loss": 70.244,
-      "step": 59530
-    },
-    {
-      "epoch": 0.24054913399887684,
-      "grad_norm": 975.7410888671875,
-      "learning_rate": 4.714861719273833e-05,
-      "loss": 84.0754,
-      "step": 59540
-    },
-    {
-      "epoch": 0.24058953526424448,
-      "grad_norm": 986.4381103515625,
-      "learning_rate": 4.7146998041152204e-05,
-      "loss": 64.928,
-      "step": 59550
-    },
-    {
-      "epoch": 0.24062993652961212,
-      "grad_norm": 996.6936645507812,
-      "learning_rate": 4.714537845779894e-05,
-      "loss": 47.7855,
-      "step": 59560
-    },
-    {
-      "epoch": 0.24067033779497973,
-      "grad_norm": 864.1629638671875,
-      "learning_rate": 4.7143758442710124e-05,
-      "loss": 65.1874,
-      "step": 59570
-    },
-    {
-      "epoch": 0.24071073906034737,
-      "grad_norm": 291.1817321777344,
-      "learning_rate": 4.7142137995917336e-05,
-      "loss": 57.6154,
-      "step": 59580
-    },
-    {
-      "epoch": 0.240751140325715,
-      "grad_norm": 353.8094177246094,
-      "learning_rate": 4.714051711745217e-05,
-      "loss": 62.8998,
-      "step": 59590
-    },
-    {
-      "epoch": 0.24079154159108263,
-      "grad_norm": 667.8562622070312,
-      "learning_rate": 4.713889580734623e-05,
-      "loss": 61.3621,
-      "step": 59600
-    },
-    {
-      "epoch": 0.24083194285645027,
-      "grad_norm": 600.4000854492188,
-      "learning_rate": 4.713727406563111e-05,
-      "loss": 101.2895,
-      "step": 59610
-    },
-    {
-      "epoch": 0.2408723441218179,
-      "grad_norm": 1322.2716064453125,
-      "learning_rate": 4.713565189233844e-05,
-      "loss": 73.4781,
-      "step": 59620
-    },
-    {
-      "epoch": 0.24091274538718552,
-      "grad_norm": 378.68841552734375,
-      "learning_rate": 4.7134029287499834e-05,
-      "loss": 68.5723,
-      "step": 59630
-    },
-    {
-      "epoch": 0.24095314665255316,
-      "grad_norm": 1208.138916015625,
-      "learning_rate": 4.7132406251146935e-05,
-      "loss": 66.9278,
-      "step": 59640
-    },
-    {
-      "epoch": 0.2409935479179208,
-      "grad_norm": 402.3450927734375,
-      "learning_rate": 4.713078278331138e-05,
-      "loss": 84.2316,
-      "step": 59650
-    },
-    {
-      "epoch": 0.2410339491832884,
-      "grad_norm": 785.4451293945312,
-      "learning_rate": 4.712915888402483e-05,
-      "loss": 82.4133,
-      "step": 59660
-    },
-    {
-      "epoch": 0.24107435044865605,
-      "grad_norm": 485.10504150390625,
-      "learning_rate": 4.7127534553318925e-05,
-      "loss": 62.9998,
-      "step": 59670
-    },
-    {
-      "epoch": 0.2411147517140237,
-      "grad_norm": 791.6461791992188,
-      "learning_rate": 4.712590979122534e-05,
-      "loss": 59.8595,
-      "step": 59680
-    },
-    {
-      "epoch": 0.2411551529793913,
-      "grad_norm": 417.98040771484375,
-      "learning_rate": 4.712428459777576e-05,
-      "loss": 71.6051,
-      "step": 59690
-    },
-    {
-      "epoch": 0.24119555424475894,
-      "grad_norm": 866.4073486328125,
-      "learning_rate": 4.712265897300186e-05,
-      "loss": 50.0575,
-      "step": 59700
-    },
-    {
-      "epoch": 0.24123595551012658,
-      "grad_norm": 1018.1526489257812,
-      "learning_rate": 4.712103291693533e-05,
-      "loss": 88.681,
-      "step": 59710
-    },
-    {
-      "epoch": 0.24127635677549422,
-      "grad_norm": 2852.052978515625,
-      "learning_rate": 4.7119406429607885e-05,
-      "loss": 85.578,
-      "step": 59720
-    },
-    {
-      "epoch": 0.24131675804086183,
-      "grad_norm": 682.5413208007812,
-      "learning_rate": 4.711777951105121e-05,
-      "loss": 73.2862,
-      "step": 59730
-    },
-    {
-      "epoch": 0.24135715930622947,
-      "grad_norm": 1790.8092041015625,
-      "learning_rate": 4.7116152161297045e-05,
-      "loss": 67.5014,
-      "step": 59740
-    },
-    {
-      "epoch": 0.24139756057159711,
-      "grad_norm": 671.518310546875,
-      "learning_rate": 4.71145243803771e-05,
-      "loss": 67.0482,
-      "step": 59750
-    },
-    {
-      "epoch": 0.24143796183696473,
-      "grad_norm": 248.55958557128906,
-      "learning_rate": 4.711289616832312e-05,
-      "loss": 52.1543,
-      "step": 59760
-    },
-    {
-      "epoch": 0.24147836310233237,
-      "grad_norm": 676.2057495117188,
-      "learning_rate": 4.7111267525166845e-05,
-      "loss": 89.2593,
-      "step": 59770
-    },
-    {
-      "epoch": 0.2415187643677,
-      "grad_norm": 846.4784545898438,
-      "learning_rate": 4.710963845094003e-05,
-      "loss": 94.6082,
-      "step": 59780
-    },
-    {
-      "epoch": 0.24155916563306762,
-      "grad_norm": 735.9619750976562,
-      "learning_rate": 4.710800894567443e-05,
-      "loss": 85.0373,
-      "step": 59790
-    },
-    {
-      "epoch": 0.24159956689843526,
-      "grad_norm": 0.0,
-      "learning_rate": 4.710637900940181e-05,
-      "loss": 56.9569,
-      "step": 59800
-    },
-    {
-      "epoch": 0.2416399681638029,
-      "grad_norm": 1803.2908935546875,
-      "learning_rate": 4.7104748642153954e-05,
-      "loss": 76.5792,
-      "step": 59810
-    },
-    {
-      "epoch": 0.2416803694291705,
-      "grad_norm": 886.1052856445312,
-      "learning_rate": 4.710311784396264e-05,
-      "loss": 44.95,
-      "step": 59820
-    },
-    {
-      "epoch": 0.24172077069453815,
-      "grad_norm": 2984.022705078125,
-      "learning_rate": 4.710148661485966e-05,
-      "loss": 104.791,
-      "step": 59830
-    },
-    {
-      "epoch": 0.2417611719599058,
-      "grad_norm": 1253.9434814453125,
-      "learning_rate": 4.709985495487682e-05,
-      "loss": 71.7768,
-      "step": 59840
-    },
-    {
-      "epoch": 0.2418015732252734,
-      "grad_norm": 785.1651000976562,
-      "learning_rate": 4.7098222864045945e-05,
-      "loss": 81.442,
-      "step": 59850
-    },
-    {
-      "epoch": 0.24184197449064104,
-      "grad_norm": 1157.2420654296875,
-      "learning_rate": 4.709659034239883e-05,
-      "loss": 91.4549,
-      "step": 59860
-    },
-    {
-      "epoch": 0.24188237575600868,
-      "grad_norm": 2114.29833984375,
-      "learning_rate": 4.7094957389967306e-05,
-      "loss": 82.0255,
-      "step": 59870
-    },
-    {
-      "epoch": 0.24192277702137632,
-      "grad_norm": 1235.954345703125,
-      "learning_rate": 4.7093324006783214e-05,
-      "loss": 95.1471,
-      "step": 59880
-    },
-    {
-      "epoch": 0.24196317828674394,
-      "grad_norm": 990.7355346679688,
-      "learning_rate": 4.709169019287839e-05,
-      "loss": 83.1672,
-      "step": 59890
-    },
-    {
-      "epoch": 0.24200357955211158,
-      "grad_norm": 790.0713500976562,
-      "learning_rate": 4.7090055948284706e-05,
-      "loss": 75.6788,
-      "step": 59900
-    },
-    {
-      "epoch": 0.24204398081747922,
-      "grad_norm": 613.9937133789062,
-      "learning_rate": 4.7088421273034e-05,
-      "loss": 74.3221,
-      "step": 59910
-    },
-    {
-      "epoch": 0.24208438208284683,
-      "grad_norm": 1069.763427734375,
-      "learning_rate": 4.708678616715815e-05,
-      "loss": 94.9927,
-      "step": 59920
-    },
-    {
-      "epoch": 0.24212478334821447,
-      "grad_norm": 1685.7093505859375,
-      "learning_rate": 4.7085150630689034e-05,
-      "loss": 79.8483,
-      "step": 59930
-    },
-    {
-      "epoch": 0.2421651846135821,
-      "grad_norm": 1826.313720703125,
-      "learning_rate": 4.7083514663658536e-05,
-      "loss": 110.4784,
-      "step": 59940
-    },
-    {
-      "epoch": 0.24220558587894972,
-      "grad_norm": 2508.780029296875,
-      "learning_rate": 4.7081878266098545e-05,
-      "loss": 95.3383,
-      "step": 59950
-    },
-    {
-      "epoch": 0.24224598714431736,
-      "grad_norm": 659.6832275390625,
-      "learning_rate": 4.708024143804097e-05,
-      "loss": 79.6695,
-      "step": 59960
-    },
-    {
-      "epoch": 0.242286388409685,
-      "grad_norm": 570.7511596679688,
-      "learning_rate": 4.707860417951773e-05,
-      "loss": 68.3188,
-      "step": 59970
-    },
-    {
-      "epoch": 0.2423267896750526,
-      "grad_norm": 1179.564208984375,
-      "learning_rate": 4.707696649056073e-05,
-      "loss": 109.7361,
-      "step": 59980
-    },
-    {
-      "epoch": 0.24236719094042025,
-      "grad_norm": 632.1577758789062,
-      "learning_rate": 4.70753283712019e-05,
-      "loss": 86.0636,
-      "step": 59990
-    },
-    {
-      "epoch": 0.2424075922057879,
-      "grad_norm": 975.7955322265625,
-      "learning_rate": 4.707368982147318e-05,
-      "loss": 87.0066,
-      "step": 60000
-    },
-    {
-      "epoch": 0.2424479934711555,
-      "grad_norm": 3105.386962890625,
-      "learning_rate": 4.707205084140651e-05,
-      "loss": 77.5594,
-      "step": 60010
-    },
-    {
-      "epoch": 0.24248839473652314,
-      "grad_norm": 834.739013671875,
-      "learning_rate": 4.707041143103384e-05,
-      "loss": 72.7049,
-      "step": 60020
-    },
-    {
-      "epoch": 0.24252879600189078,
-      "grad_norm": 481.8167419433594,
-      "learning_rate": 4.706877159038715e-05,
-      "loss": 91.9634,
-      "step": 60030
-    },
-    {
-      "epoch": 0.24256919726725842,
-      "grad_norm": 651.197021484375,
-      "learning_rate": 4.706713131949839e-05,
-      "loss": 57.6539,
-      "step": 60040
-    },
-    {
-      "epoch": 0.24260959853262604,
-      "grad_norm": 821.2056884765625,
-      "learning_rate": 4.706549061839954e-05,
-      "loss": 67.2985,
-      "step": 60050
-    },
-    {
-      "epoch": 0.24264999979799368,
-      "grad_norm": 445.24578857421875,
-      "learning_rate": 4.70638494871226e-05,
-      "loss": 70.1479,
-      "step": 60060
-    },
-    {
-      "epoch": 0.24269040106336132,
-      "grad_norm": 902.0319213867188,
-      "learning_rate": 4.7062207925699544e-05,
-      "loss": 85.1113,
-      "step": 60070
-    },
-    {
-      "epoch": 0.24273080232872893,
-      "grad_norm": 845.628662109375,
-      "learning_rate": 4.7060565934162394e-05,
-      "loss": 86.7892,
-      "step": 60080
-    },
-    {
-      "epoch": 0.24277120359409657,
-      "grad_norm": 936.5902709960938,
-      "learning_rate": 4.7058923512543154e-05,
-      "loss": 84.2543,
-      "step": 60090
-    },
-    {
-      "epoch": 0.2428116048594642,
-      "grad_norm": 770.0076904296875,
-      "learning_rate": 4.7057280660873835e-05,
-      "loss": 77.7724,
-      "step": 60100
-    },
-    {
-      "epoch": 0.24285200612483182,
-      "grad_norm": 1131.8125,
-      "learning_rate": 4.705563737918648e-05,
-      "loss": 69.4776,
-      "step": 60110
-    },
-    {
-      "epoch": 0.24289240739019946,
-      "grad_norm": 940.4691772460938,
-      "learning_rate": 4.705399366751312e-05,
-      "loss": 88.3803,
-      "step": 60120
-    },
-    {
-      "epoch": 0.2429328086555671,
-      "grad_norm": 763.4826049804688,
-      "learning_rate": 4.705234952588579e-05,
-      "loss": 79.9678,
-      "step": 60130
-    },
-    {
-      "epoch": 0.2429732099209347,
-      "grad_norm": 712.3192749023438,
-      "learning_rate": 4.705070495433657e-05,
-      "loss": 71.1181,
-      "step": 60140
-    },
-    {
-      "epoch": 0.24301361118630235,
-      "grad_norm": 1741.696044921875,
-      "learning_rate": 4.704905995289749e-05,
-      "loss": 70.7245,
-      "step": 60150
-    },
-    {
-      "epoch": 0.24305401245167,
-      "grad_norm": 588.7029418945312,
-      "learning_rate": 4.7047414521600644e-05,
-      "loss": 57.2675,
-      "step": 60160
-    },
-    {
-      "epoch": 0.2430944137170376,
-      "grad_norm": 2938.370361328125,
-      "learning_rate": 4.704576866047808e-05,
-      "loss": 61.0899,
-      "step": 60170
-    },
-    {
-      "epoch": 0.24313481498240525,
-      "grad_norm": 936.1781005859375,
-      "learning_rate": 4.704412236956193e-05,
-      "loss": 62.6005,
-      "step": 60180
-    },
-    {
-      "epoch": 0.24317521624777289,
-      "grad_norm": 804.415771484375,
-      "learning_rate": 4.7042475648884254e-05,
-      "loss": 79.235,
-      "step": 60190
-    },
-    {
-      "epoch": 0.24321561751314053,
-      "grad_norm": 729.1420288085938,
-      "learning_rate": 4.704082849847718e-05,
-      "loss": 92.9888,
-      "step": 60200
-    },
-    {
-      "epoch": 0.24325601877850814,
-      "grad_norm": 604.2557983398438,
-      "learning_rate": 4.703918091837279e-05,
-      "loss": 55.6201,
-      "step": 60210
-    },
-    {
-      "epoch": 0.24329642004387578,
-      "grad_norm": 580.6284790039062,
-      "learning_rate": 4.703753290860323e-05,
-      "loss": 45.9183,
-      "step": 60220
-    },
-    {
-      "epoch": 0.24333682130924342,
-      "grad_norm": 361.0988464355469,
-      "learning_rate": 4.703588446920062e-05,
-      "loss": 96.7986,
-      "step": 60230
-    },
-    {
-      "epoch": 0.24337722257461103,
-      "grad_norm": 5288.78271484375,
-      "learning_rate": 4.70342356001971e-05,
-      "loss": 62.3638,
-      "step": 60240
-    },
-    {
-      "epoch": 0.24341762383997867,
-      "grad_norm": 981.030517578125,
-      "learning_rate": 4.70325863016248e-05,
-      "loss": 72.3735,
-      "step": 60250
-    },
-    {
-      "epoch": 0.2434580251053463,
-      "grad_norm": 1044.311279296875,
-      "learning_rate": 4.703093657351591e-05,
-      "loss": 101.13,
-      "step": 60260
-    },
-    {
-      "epoch": 0.24349842637071392,
-      "grad_norm": 695.3463134765625,
-      "learning_rate": 4.702928641590255e-05,
-      "loss": 61.2876,
-      "step": 60270
-    },
-    {
-      "epoch": 0.24353882763608156,
-      "grad_norm": 825.5418701171875,
-      "learning_rate": 4.702763582881692e-05,
-      "loss": 77.4805,
-      "step": 60280
-    },
-    {
-      "epoch": 0.2435792289014492,
-      "grad_norm": 568.7557983398438,
-      "learning_rate": 4.702598481229118e-05,
-      "loss": 52.6858,
-      "step": 60290
-    },
-    {
-      "epoch": 0.24361963016681681,
-      "grad_norm": 1038.3533935546875,
-      "learning_rate": 4.702433336635753e-05,
-      "loss": 72.8244,
-      "step": 60300
-    },
-    {
-      "epoch": 0.24366003143218445,
-      "grad_norm": 534.8966674804688,
-      "learning_rate": 4.702268149104816e-05,
-      "loss": 54.4095,
-      "step": 60310
-    },
-    {
-      "epoch": 0.2437004326975521,
-      "grad_norm": 683.1052856445312,
-      "learning_rate": 4.702102918639528e-05,
-      "loss": 55.7349,
-      "step": 60320
-    },
-    {
-      "epoch": 0.2437408339629197,
-      "grad_norm": 483.41552734375,
-      "learning_rate": 4.70193764524311e-05,
-      "loss": 74.4342,
-      "step": 60330
-    },
-    {
-      "epoch": 0.24378123522828735,
-      "grad_norm": 688.1597900390625,
-      "learning_rate": 4.701772328918784e-05,
-      "loss": 56.6066,
-      "step": 60340
-    },
-    {
-      "epoch": 0.243821636493655,
-      "grad_norm": 874.982177734375,
-      "learning_rate": 4.701606969669773e-05,
-      "loss": 62.7406,
-      "step": 60350
-    },
-    {
-      "epoch": 0.24386203775902263,
-      "grad_norm": 1608.3681640625,
-      "learning_rate": 4.7014415674993e-05,
-      "loss": 87.2219,
-      "step": 60360
-    },
-    {
-      "epoch": 0.24390243902439024,
-      "grad_norm": 852.6019287109375,
-      "learning_rate": 4.701276122410591e-05,
-      "loss": 65.5959,
-      "step": 60370
-    },
-    {
-      "epoch": 0.24394284028975788,
-      "grad_norm": 796.290771484375,
-      "learning_rate": 4.70111063440687e-05,
-      "loss": 64.5564,
-      "step": 60380
-    },
-    {
-      "epoch": 0.24398324155512552,
-      "grad_norm": 1192.7855224609375,
-      "learning_rate": 4.7009451034913645e-05,
-      "loss": 56.2537,
-      "step": 60390
-    },
-    {
-      "epoch": 0.24402364282049313,
-      "grad_norm": 691.683837890625,
-      "learning_rate": 4.7007795296673006e-05,
-      "loss": 58.2113,
-      "step": 60400
-    },
-    {
-      "epoch": 0.24406404408586077,
-      "grad_norm": 586.52685546875,
-      "learning_rate": 4.700613912937907e-05,
-      "loss": 47.3722,
-      "step": 60410
-    },
-    {
-      "epoch": 0.2441044453512284,
-      "grad_norm": 1319.3663330078125,
-      "learning_rate": 4.700448253306412e-05,
-      "loss": 75.1673,
-      "step": 60420
-    },
-    {
-      "epoch": 0.24414484661659602,
-      "grad_norm": 1594.365234375,
-      "learning_rate": 4.7002825507760465e-05,
-      "loss": 120.8496,
-      "step": 60430
-    },
-    {
-      "epoch": 0.24418524788196366,
-      "grad_norm": 789.883544921875,
-      "learning_rate": 4.700116805350039e-05,
-      "loss": 65.3508,
-      "step": 60440
-    },
-    {
-      "epoch": 0.2442256491473313,
-      "grad_norm": 627.7055053710938,
-      "learning_rate": 4.699951017031621e-05,
-      "loss": 80.6547,
-      "step": 60450
-    },
-    {
-      "epoch": 0.24426605041269892,
-      "grad_norm": 564.2537841796875,
-      "learning_rate": 4.699785185824026e-05,
-      "loss": 73.5011,
-      "step": 60460
-    },
-    {
-      "epoch": 0.24430645167806656,
-      "grad_norm": 589.3305053710938,
-      "learning_rate": 4.6996193117304864e-05,
-      "loss": 50.1976,
-      "step": 60470
-    },
-    {
-      "epoch": 0.2443468529434342,
-      "grad_norm": 2720.202392578125,
-      "learning_rate": 4.699453394754236e-05,
-      "loss": 65.688,
-      "step": 60480
-    },
-    {
-      "epoch": 0.2443872542088018,
-      "grad_norm": 926.1817016601562,
-      "learning_rate": 4.6992874348985093e-05,
-      "loss": 99.6495,
-      "step": 60490
-    },
-    {
-      "epoch": 0.24442765547416945,
-      "grad_norm": 845.654541015625,
-      "learning_rate": 4.6991214321665414e-05,
-      "loss": 63.7764,
-      "step": 60500
-    },
-    {
-      "epoch": 0.2444680567395371,
-      "grad_norm": 458.4206237792969,
-      "learning_rate": 4.698955386561569e-05,
-      "loss": 53.1651,
-      "step": 60510
-    },
-    {
-      "epoch": 0.24450845800490473,
-      "grad_norm": 1341.775146484375,
-      "learning_rate": 4.6987892980868296e-05,
-      "loss": 85.8457,
-      "step": 60520
-    },
-    {
-      "epoch": 0.24454885927027234,
-      "grad_norm": 1214.4681396484375,
-      "learning_rate": 4.6986231667455605e-05,
-      "loss": 59.1112,
-      "step": 60530
-    },
-    {
-      "epoch": 0.24458926053563998,
-      "grad_norm": 791.0181884765625,
-      "learning_rate": 4.6984569925410016e-05,
-      "loss": 113.4298,
-      "step": 60540
-    },
-    {
-      "epoch": 0.24462966180100762,
-      "grad_norm": 1435.430908203125,
-      "learning_rate": 4.6982907754763906e-05,
-      "loss": 106.5237,
-      "step": 60550
-    },
-    {
-      "epoch": 0.24467006306637523,
-      "grad_norm": 671.5120239257812,
-      "learning_rate": 4.69812451555497e-05,
-      "loss": 75.014,
-      "step": 60560
-    },
-    {
-      "epoch": 0.24471046433174287,
-      "grad_norm": 419.3539733886719,
-      "learning_rate": 4.697958212779981e-05,
-      "loss": 40.2964,
-      "step": 60570
-    },
-    {
-      "epoch": 0.2447508655971105,
-      "grad_norm": 1304.94189453125,
-      "learning_rate": 4.697791867154663e-05,
-      "loss": 52.5812,
-      "step": 60580
-    },
-    {
-      "epoch": 0.24479126686247812,
-      "grad_norm": 577.18505859375,
-      "learning_rate": 4.697625478682263e-05,
-      "loss": 82.3342,
-      "step": 60590
-    },
-    {
-      "epoch": 0.24483166812784576,
-      "grad_norm": 455.48480224609375,
-      "learning_rate": 4.6974590473660216e-05,
-      "loss": 55.9024,
-      "step": 60600
-    },
-    {
-      "epoch": 0.2448720693932134,
-      "grad_norm": 592.9940185546875,
-      "learning_rate": 4.697292573209185e-05,
-      "loss": 65.5786,
-      "step": 60610
-    },
-    {
-      "epoch": 0.24491247065858102,
-      "grad_norm": 368.27581787109375,
-      "learning_rate": 4.697126056214999e-05,
-      "loss": 34.5167,
-      "step": 60620
-    },
-    {
-      "epoch": 0.24495287192394866,
-      "grad_norm": 400.2873840332031,
-      "learning_rate": 4.6969594963867084e-05,
-      "loss": 104.06,
-      "step": 60630
-    },
-    {
-      "epoch": 0.2449932731893163,
-      "grad_norm": 744.5037231445312,
-      "learning_rate": 4.696792893727562e-05,
-      "loss": 87.9777,
-      "step": 60640
-    },
-    {
-      "epoch": 0.2450336744546839,
-      "grad_norm": 654.1832885742188,
-      "learning_rate": 4.696626248240807e-05,
-      "loss": 54.0729,
-      "step": 60650
-    },
-    {
-      "epoch": 0.24507407572005155,
-      "grad_norm": 339.2250671386719,
-      "learning_rate": 4.6964595599296926e-05,
-      "loss": 40.1769,
-      "step": 60660
-    },
-    {
-      "epoch": 0.2451144769854192,
-      "grad_norm": 632.5075073242188,
-      "learning_rate": 4.696292828797468e-05,
-      "loss": 79.759,
-      "step": 60670
-    },
-    {
-      "epoch": 0.24515487825078683,
-      "grad_norm": 884.93310546875,
-      "learning_rate": 4.696126054847385e-05,
-      "loss": 73.9001,
-      "step": 60680
-    },
-    {
-      "epoch": 0.24519527951615444,
-      "grad_norm": 1115.2449951171875,
-      "learning_rate": 4.695959238082692e-05,
-      "loss": 70.6572,
-      "step": 60690
-    },
-    {
-      "epoch": 0.24523568078152208,
-      "grad_norm": 1520.8741455078125,
-      "learning_rate": 4.6957923785066445e-05,
-      "loss": 90.2716,
-      "step": 60700
-    },
-    {
-      "epoch": 0.24527608204688972,
-      "grad_norm": 1127.73876953125,
-      "learning_rate": 4.6956254761224936e-05,
-      "loss": 70.0457,
-      "step": 60710
-    },
-    {
-      "epoch": 0.24531648331225733,
-      "grad_norm": 1073.5108642578125,
-      "learning_rate": 4.695458530933494e-05,
-      "loss": 100.9174,
-      "step": 60720
-    },
-    {
-      "epoch": 0.24535688457762497,
-      "grad_norm": 773.4358520507812,
-      "learning_rate": 4.6952915429429e-05,
-      "loss": 82.9172,
-      "step": 60730
-    },
-    {
-      "epoch": 0.2453972858429926,
-      "grad_norm": 1714.5184326171875,
-      "learning_rate": 4.6951245121539675e-05,
-      "loss": 74.2328,
-      "step": 60740
-    },
-    {
-      "epoch": 0.24543768710836023,
-      "grad_norm": 1113.5059814453125,
-      "learning_rate": 4.694957438569951e-05,
-      "loss": 68.1501,
-      "step": 60750
-    },
-    {
-      "epoch": 0.24547808837372787,
-      "grad_norm": 1123.2255859375,
-      "learning_rate": 4.694790322194111e-05,
-      "loss": 57.4253,
-      "step": 60760
-    },
-    {
-      "epoch": 0.2455184896390955,
-      "grad_norm": 1276.2410888671875,
-      "learning_rate": 4.6946231630297036e-05,
-      "loss": 110.3862,
-      "step": 60770
-    },
-    {
-      "epoch": 0.24555889090446312,
-      "grad_norm": 695.6715087890625,
-      "learning_rate": 4.694455961079987e-05,
-      "loss": 70.8457,
-      "step": 60780
-    },
-    {
-      "epoch": 0.24559929216983076,
-      "grad_norm": 1895.229736328125,
-      "learning_rate": 4.694288716348221e-05,
-      "loss": 62.3386,
-      "step": 60790
-    },
-    {
-      "epoch": 0.2456396934351984,
-      "grad_norm": 894.3408813476562,
-      "learning_rate": 4.694121428837668e-05,
-      "loss": 73.6542,
-      "step": 60800
-    },
-    {
-      "epoch": 0.245680094700566,
-      "grad_norm": 773.7852172851562,
-      "learning_rate": 4.693954098551587e-05,
-      "loss": 57.4668,
-      "step": 60810
-    },
-    {
-      "epoch": 0.24572049596593365,
-      "grad_norm": 873.3363037109375,
-      "learning_rate": 4.693786725493242e-05,
-      "loss": 88.7946,
-      "step": 60820
-    },
-    {
-      "epoch": 0.2457608972313013,
-      "grad_norm": 787.7183227539062,
-      "learning_rate": 4.6936193096658955e-05,
-      "loss": 95.2631,
-      "step": 60830
-    },
-    {
-      "epoch": 0.2458012984966689,
-      "grad_norm": 1632.471923828125,
-      "learning_rate": 4.693451851072811e-05,
-      "loss": 76.3312,
-      "step": 60840
-    },
-    {
-      "epoch": 0.24584169976203654,
-      "grad_norm": 624.8790283203125,
-      "learning_rate": 4.693284349717254e-05,
-      "loss": 60.229,
-      "step": 60850
-    },
-    {
-      "epoch": 0.24588210102740418,
-      "grad_norm": 776.8783569335938,
-      "learning_rate": 4.693116805602489e-05,
-      "loss": 67.0748,
-      "step": 60860
-    },
-    {
-      "epoch": 0.24592250229277182,
-      "grad_norm": 357.4325866699219,
-      "learning_rate": 4.692949218731782e-05,
-      "loss": 75.164,
-      "step": 60870
-    },
-    {
-      "epoch": 0.24596290355813943,
-      "grad_norm": 934.4722290039062,
-      "learning_rate": 4.692781589108402e-05,
-      "loss": 104.851,
-      "step": 60880
-    },
-    {
-      "epoch": 0.24600330482350707,
-      "grad_norm": 788.913330078125,
-      "learning_rate": 4.692613916735615e-05,
-      "loss": 76.9564,
-      "step": 60890
-    },
-    {
-      "epoch": 0.24604370608887471,
-      "grad_norm": 1184.293701171875,
-      "learning_rate": 4.692446201616692e-05,
-      "loss": 104.7225,
-      "step": 60900
-    },
-    {
-      "epoch": 0.24608410735424233,
-      "grad_norm": 768.3632202148438,
-      "learning_rate": 4.692278443754901e-05,
-      "loss": 96.9056,
-      "step": 60910
-    },
-    {
-      "epoch": 0.24612450861960997,
-      "grad_norm": 1289.7335205078125,
-      "learning_rate": 4.6921106431535135e-05,
-      "loss": 51.5294,
-      "step": 60920
-    },
-    {
-      "epoch": 0.2461649098849776,
-      "grad_norm": 537.193359375,
-      "learning_rate": 4.6919427998158e-05,
-      "loss": 87.4695,
-      "step": 60930
-    },
-    {
-      "epoch": 0.24620531115034522,
-      "grad_norm": 3655.863525390625,
-      "learning_rate": 4.691774913745033e-05,
-      "loss": 92.1189,
-      "step": 60940
-    },
-    {
-      "epoch": 0.24624571241571286,
-      "grad_norm": 465.6102600097656,
-      "learning_rate": 4.691606984944486e-05,
-      "loss": 79.8114,
-      "step": 60950
-    },
-    {
-      "epoch": 0.2462861136810805,
-      "grad_norm": 531.8803100585938,
-      "learning_rate": 4.691439013417433e-05,
-      "loss": 62.0595,
-      "step": 60960
-    },
-    {
-      "epoch": 0.2463265149464481,
-      "grad_norm": 1052.3104248046875,
-      "learning_rate": 4.691270999167147e-05,
-      "loss": 67.6164,
-      "step": 60970
-    },
-    {
-      "epoch": 0.24636691621181575,
-      "grad_norm": 624.75390625,
-      "learning_rate": 4.691102942196906e-05,
-      "loss": 67.1735,
-      "step": 60980
-    },
-    {
-      "epoch": 0.2464073174771834,
-      "grad_norm": 626.8027954101562,
-      "learning_rate": 4.6909348425099835e-05,
-      "loss": 64.8553,
-      "step": 60990
-    },
-    {
-      "epoch": 0.246447718742551,
-      "grad_norm": 250.1268768310547,
-      "learning_rate": 4.690766700109659e-05,
-      "loss": 63.7221,
-      "step": 61000
-    },
-    {
-      "epoch": 0.24648812000791864,
-      "grad_norm": 978.392333984375,
-      "learning_rate": 4.6905985149992107e-05,
-      "loss": 138.7521,
-      "step": 61010
-    },
-    {
-      "epoch": 0.24652852127328628,
-      "grad_norm": 770.6021728515625,
-      "learning_rate": 4.690430287181915e-05,
-      "loss": 104.2015,
-      "step": 61020
-    },
-    {
-      "epoch": 0.24656892253865392,
-      "grad_norm": 1836.369873046875,
-      "learning_rate": 4.690262016661054e-05,
-      "loss": 64.2337,
-      "step": 61030
-    },
-    {
-      "epoch": 0.24660932380402154,
-      "grad_norm": 1369.9937744140625,
-      "learning_rate": 4.690093703439907e-05,
-      "loss": 114.4457,
-      "step": 61040
-    },
-    {
-      "epoch": 0.24664972506938918,
-      "grad_norm": 756.3161010742188,
-      "learning_rate": 4.689925347521757e-05,
-      "loss": 49.0843,
-      "step": 61050
-    },
-    {
-      "epoch": 0.24669012633475682,
-      "grad_norm": 446.1201171875,
-      "learning_rate": 4.689756948909884e-05,
-      "loss": 59.4647,
-      "step": 61060
-    },
-    {
-      "epoch": 0.24673052760012443,
-      "grad_norm": 1100.18994140625,
-      "learning_rate": 4.689588507607572e-05,
-      "loss": 55.1963,
-      "step": 61070
-    },
-    {
-      "epoch": 0.24677092886549207,
-      "grad_norm": 863.7999877929688,
-      "learning_rate": 4.689420023618104e-05,
-      "loss": 67.0081,
-      "step": 61080
-    },
-    {
-      "epoch": 0.2468113301308597,
-      "grad_norm": 696.3552856445312,
-      "learning_rate": 4.6892514969447664e-05,
-      "loss": 80.1907,
-      "step": 61090
-    },
-    {
-      "epoch": 0.24685173139622732,
-      "grad_norm": 695.3787231445312,
-      "learning_rate": 4.6890829275908434e-05,
-      "loss": 70.8243,
-      "step": 61100
-    },
-    {
-      "epoch": 0.24689213266159496,
-      "grad_norm": 893.7193603515625,
-      "learning_rate": 4.6889143155596214e-05,
-      "loss": 99.1561,
-      "step": 61110
-    },
-    {
-      "epoch": 0.2469325339269626,
-      "grad_norm": 1360.45263671875,
-      "learning_rate": 4.688745660854388e-05,
-      "loss": 103.6161,
-      "step": 61120
-    },
-    {
-      "epoch": 0.2469729351923302,
-      "grad_norm": 726.0472412109375,
-      "learning_rate": 4.688576963478432e-05,
-      "loss": 77.2796,
-      "step": 61130
-    },
-    {
-      "epoch": 0.24701333645769785,
-      "grad_norm": 120.09856414794922,
-      "learning_rate": 4.68840822343504e-05,
-      "loss": 77.299,
-      "step": 61140
-    },
-    {
-      "epoch": 0.2470537377230655,
-      "grad_norm": 1099.0858154296875,
-      "learning_rate": 4.6882394407275044e-05,
-      "loss": 103.8713,
-      "step": 61150
-    },
-    {
-      "epoch": 0.2470941389884331,
-      "grad_norm": 704.1102294921875,
-      "learning_rate": 4.688070615359114e-05,
-      "loss": 100.1313,
-      "step": 61160
-    },
-    {
-      "epoch": 0.24713454025380074,
-      "grad_norm": 429.2733459472656,
-      "learning_rate": 4.6879017473331595e-05,
-      "loss": 105.0034,
-      "step": 61170
-    },
-    {
-      "epoch": 0.24717494151916838,
-      "grad_norm": 715.7286376953125,
-      "learning_rate": 4.6877328366529346e-05,
-      "loss": 57.8755,
-      "step": 61180
-    },
-    {
-      "epoch": 0.24721534278453602,
-      "grad_norm": 708.1912841796875,
-      "learning_rate": 4.687563883321732e-05,
-      "loss": 67.7243,
-      "step": 61190
-    },
-    {
-      "epoch": 0.24725574404990364,
-      "grad_norm": 885.6137084960938,
-      "learning_rate": 4.687394887342845e-05,
-      "loss": 63.6241,
-      "step": 61200
-    },
-    {
-      "epoch": 0.24729614531527128,
-      "grad_norm": 549.2717895507812,
-      "learning_rate": 4.687225848719568e-05,
-      "loss": 66.4841,
-      "step": 61210
-    },
-    {
-      "epoch": 0.24733654658063892,
-      "grad_norm": 390.741455078125,
-      "learning_rate": 4.687056767455198e-05,
-      "loss": 59.3939,
-      "step": 61220
-    },
-    {
-      "epoch": 0.24737694784600653,
-      "grad_norm": 770.025390625,
-      "learning_rate": 4.6868876435530296e-05,
-      "loss": 78.1586,
-      "step": 61230
-    },
-    {
-      "epoch": 0.24741734911137417,
-      "grad_norm": 565.1214599609375,
-      "learning_rate": 4.686718477016361e-05,
-      "loss": 68.8117,
-      "step": 61240
-    },
-    {
-      "epoch": 0.2474577503767418,
-      "grad_norm": 944.383544921875,
-      "learning_rate": 4.6865492678484895e-05,
-      "loss": 72.7522,
-      "step": 61250
-    },
-    {
-      "epoch": 0.24749815164210942,
-      "grad_norm": 2328.01513671875,
-      "learning_rate": 4.6863800160527147e-05,
-      "loss": 86.5814,
-      "step": 61260
-    },
-    {
-      "epoch": 0.24753855290747706,
-      "grad_norm": 648.8901977539062,
-      "learning_rate": 4.686210721632336e-05,
-      "loss": 85.4312,
-      "step": 61270
-    },
-    {
-      "epoch": 0.2475789541728447,
-      "grad_norm": 1613.290283203125,
-      "learning_rate": 4.6860413845906534e-05,
-      "loss": 80.4138,
-      "step": 61280
-    },
-    {
-      "epoch": 0.2476193554382123,
-      "grad_norm": 515.4049682617188,
-      "learning_rate": 4.685872004930969e-05,
-      "loss": 48.0804,
-      "step": 61290
-    },
-    {
-      "epoch": 0.24765975670357995,
-      "grad_norm": 258.1792297363281,
-      "learning_rate": 4.685702582656584e-05,
-      "loss": 48.0407,
-      "step": 61300
-    },
-    {
-      "epoch": 0.2477001579689476,
-      "grad_norm": 1516.9779052734375,
-      "learning_rate": 4.685533117770803e-05,
-      "loss": 76.3524,
-      "step": 61310
-    },
-    {
-      "epoch": 0.2477405592343152,
-      "grad_norm": 1197.6373291015625,
-      "learning_rate": 4.6853636102769274e-05,
-      "loss": 95.2485,
-      "step": 61320
-    },
-    {
-      "epoch": 0.24778096049968285,
-      "grad_norm": 433.1604309082031,
-      "learning_rate": 4.6851940601782635e-05,
-      "loss": 83.6786,
-      "step": 61330
-    },
-    {
-      "epoch": 0.24782136176505049,
-      "grad_norm": 622.607666015625,
-      "learning_rate": 4.685024467478116e-05,
-      "loss": 75.2844,
-      "step": 61340
-    },
-    {
-      "epoch": 0.24786176303041813,
-      "grad_norm": 710.9769287109375,
-      "learning_rate": 4.684854832179792e-05,
-      "loss": 106.4003,
-      "step": 61350
-    },
-    {
-      "epoch": 0.24790216429578574,
-      "grad_norm": 1580.35009765625,
-      "learning_rate": 4.684685154286599e-05,
-      "loss": 58.8193,
-      "step": 61360
-    },
-    {
-      "epoch": 0.24794256556115338,
-      "grad_norm": 917.3915405273438,
-      "learning_rate": 4.684515433801843e-05,
-      "loss": 78.9511,
-      "step": 61370
-    },
-    {
-      "epoch": 0.24798296682652102,
-      "grad_norm": 1568.43408203125,
-      "learning_rate": 4.684345670728834e-05,
-      "loss": 82.2572,
-      "step": 61380
-    },
-    {
-      "epoch": 0.24802336809188863,
-      "grad_norm": 768.3124389648438,
-      "learning_rate": 4.6841758650708824e-05,
-      "loss": 83.1386,
-      "step": 61390
-    },
-    {
-      "epoch": 0.24806376935725627,
-      "grad_norm": 972.3724975585938,
-      "learning_rate": 4.684006016831297e-05,
-      "loss": 111.0157,
-      "step": 61400
-    },
-    {
-      "epoch": 0.2481041706226239,
-      "grad_norm": 710.788330078125,
-      "learning_rate": 4.68383612601339e-05,
-      "loss": 70.9548,
-      "step": 61410
-    },
-    {
-      "epoch": 0.24814457188799152,
-      "grad_norm": 1372.3570556640625,
-      "learning_rate": 4.6836661926204736e-05,
-      "loss": 57.5385,
-      "step": 61420
-    },
-    {
-      "epoch": 0.24818497315335916,
-      "grad_norm": 2235.90380859375,
-      "learning_rate": 4.6834962166558605e-05,
-      "loss": 86.6557,
-      "step": 61430
-    },
-    {
-      "epoch": 0.2482253744187268,
-      "grad_norm": 1524.7860107421875,
-      "learning_rate": 4.6833261981228646e-05,
-      "loss": 86.687,
-      "step": 61440
-    },
-    {
-      "epoch": 0.24826577568409441,
-      "grad_norm": 514.3629760742188,
-      "learning_rate": 4.683156137024801e-05,
-      "loss": 73.8575,
-      "step": 61450
-    },
-    {
-      "epoch": 0.24830617694946205,
-      "grad_norm": 830.9341430664062,
-      "learning_rate": 4.6829860333649836e-05,
-      "loss": 63.1829,
-      "step": 61460
-    },
-    {
-      "epoch": 0.2483465782148297,
-      "grad_norm": 0.0,
-      "learning_rate": 4.68281588714673e-05,
-      "loss": 49.3014,
-      "step": 61470
-    },
-    {
-      "epoch": 0.2483869794801973,
-      "grad_norm": 1122.3907470703125,
-      "learning_rate": 4.682645698373357e-05,
-      "loss": 56.4632,
-      "step": 61480
-    },
-    {
-      "epoch": 0.24842738074556495,
-      "grad_norm": 838.9703369140625,
-      "learning_rate": 4.682475467048182e-05,
-      "loss": 51.4873,
-      "step": 61490
-    },
-    {
-      "epoch": 0.2484677820109326,
-      "grad_norm": 966.208251953125,
-      "learning_rate": 4.682305193174524e-05,
-      "loss": 92.0763,
-      "step": 61500
-    },
-    {
-      "epoch": 0.24850818327630023,
-      "grad_norm": 623.809326171875,
-      "learning_rate": 4.682134876755704e-05,
-      "loss": 87.1039,
-      "step": 61510
-    },
-    {
-      "epoch": 0.24854858454166784,
-      "grad_norm": 769.28759765625,
-      "learning_rate": 4.68196451779504e-05,
-      "loss": 98.8148,
-      "step": 61520
-    },
-    {
-      "epoch": 0.24858898580703548,
-      "grad_norm": 437.263671875,
-      "learning_rate": 4.6817941162958544e-05,
-      "loss": 78.6662,
-      "step": 61530
-    },
-    {
-      "epoch": 0.24862938707240312,
-      "grad_norm": 566.79638671875,
-      "learning_rate": 4.681623672261469e-05,
-      "loss": 67.6416,
-      "step": 61540
-    },
-    {
-      "epoch": 0.24866978833777073,
-      "grad_norm": 378.28082275390625,
-      "learning_rate": 4.6814531856952084e-05,
-      "loss": 51.0772,
-      "step": 61550
-    },
-    {
-      "epoch": 0.24871018960313837,
-      "grad_norm": 2680.483154296875,
-      "learning_rate": 4.6812826566003934e-05,
-      "loss": 102.0762,
-      "step": 61560
-    },
-    {
-      "epoch": 0.248750590868506,
-      "grad_norm": 787.8114624023438,
-      "learning_rate": 4.68111208498035e-05,
-      "loss": 40.7656,
-      "step": 61570
-    },
-    {
-      "epoch": 0.24879099213387362,
-      "grad_norm": 701.6543579101562,
-      "learning_rate": 4.6809414708384046e-05,
-      "loss": 71.9813,
-      "step": 61580
-    },
-    {
-      "epoch": 0.24883139339924126,
-      "grad_norm": 643.1692504882812,
-      "learning_rate": 4.680770814177882e-05,
-      "loss": 84.7958,
-      "step": 61590
-    },
-    {
-      "epoch": 0.2488717946646089,
-      "grad_norm": 857.9775390625,
-      "learning_rate": 4.68060011500211e-05,
-      "loss": 78.1685,
-      "step": 61600
-    },
-    {
-      "epoch": 0.24891219592997652,
-      "grad_norm": 897.5280151367188,
-      "learning_rate": 4.680429373314415e-05,
-      "loss": 57.9258,
-      "step": 61610
-    },
-    {
-      "epoch": 0.24895259719534416,
-      "grad_norm": 756.8377685546875,
-      "learning_rate": 4.680258589118128e-05,
-      "loss": 90.2208,
-      "step": 61620
-    },
-    {
-      "epoch": 0.2489929984607118,
-      "grad_norm": 650.5374145507812,
-      "learning_rate": 4.680087762416576e-05,
-      "loss": 81.8518,
-      "step": 61630
-    },
-    {
-      "epoch": 0.2490333997260794,
-      "grad_norm": 675.606201171875,
-      "learning_rate": 4.6799168932130915e-05,
-      "loss": 72.3285,
-      "step": 61640
-    },
-    {
-      "epoch": 0.24907380099144705,
-      "grad_norm": 932.6596069335938,
-      "learning_rate": 4.679745981511005e-05,
-      "loss": 51.2554,
-      "step": 61650
-    },
-    {
-      "epoch": 0.2491142022568147,
-      "grad_norm": 891.9270629882812,
-      "learning_rate": 4.679575027313649e-05,
-      "loss": 54.2277,
-      "step": 61660
-    },
-    {
-      "epoch": 0.24915460352218233,
-      "grad_norm": 927.309814453125,
-      "learning_rate": 4.6794040306243545e-05,
-      "loss": 68.7457,
-      "step": 61670
-    },
-    {
-      "epoch": 0.24919500478754994,
-      "grad_norm": 2662.16845703125,
-      "learning_rate": 4.679232991446456e-05,
-      "loss": 63.2462,
-      "step": 61680
-    },
-    {
-      "epoch": 0.24923540605291758,
-      "grad_norm": 604.02978515625,
-      "learning_rate": 4.67906190978329e-05,
-      "loss": 37.8021,
-      "step": 61690
-    },
-    {
-      "epoch": 0.24927580731828522,
-      "grad_norm": 291.47430419921875,
-      "learning_rate": 4.6788907856381895e-05,
-      "loss": 131.591,
-      "step": 61700
-    },
-    {
-      "epoch": 0.24931620858365283,
-      "grad_norm": 640.17529296875,
-      "learning_rate": 4.678719619014491e-05,
-      "loss": 86.4894,
-      "step": 61710
-    },
-    {
-      "epoch": 0.24935660984902047,
-      "grad_norm": 745.0997924804688,
-      "learning_rate": 4.678548409915532e-05,
-      "loss": 62.8037,
-      "step": 61720
-    },
-    {
-      "epoch": 0.2493970111143881,
-      "grad_norm": 576.8112182617188,
-      "learning_rate": 4.67837715834465e-05,
-      "loss": 96.3769,
-      "step": 61730
-    },
-    {
-      "epoch": 0.24943741237975572,
-      "grad_norm": 687.8386840820312,
-      "learning_rate": 4.678205864305184e-05,
-      "loss": 78.1074,
-      "step": 61740
-    },
-    {
-      "epoch": 0.24947781364512336,
-      "grad_norm": 542.1369018554688,
-      "learning_rate": 4.678034527800474e-05,
-      "loss": 70.0832,
-      "step": 61750
-    },
-    {
-      "epoch": 0.249518214910491,
-      "grad_norm": 945.9921875,
-      "learning_rate": 4.677863148833859e-05,
-      "loss": 87.9454,
-      "step": 61760
-    },
-    {
-      "epoch": 0.24955861617585862,
-      "grad_norm": 443.54583740234375,
-      "learning_rate": 4.6776917274086806e-05,
-      "loss": 63.2674,
-      "step": 61770
-    },
-    {
-      "epoch": 0.24959901744122626,
-      "grad_norm": 485.80865478515625,
-      "learning_rate": 4.67752026352828e-05,
-      "loss": 86.9183,
-      "step": 61780
-    },
-    {
-      "epoch": 0.2496394187065939,
-      "grad_norm": 514.6878051757812,
-      "learning_rate": 4.677348757196002e-05,
-      "loss": 95.4372,
-      "step": 61790
-    },
-    {
-      "epoch": 0.2496798199719615,
-      "grad_norm": 1996.046142578125,
-      "learning_rate": 4.6771772084151885e-05,
-      "loss": 93.6957,
-      "step": 61800
-    },
-    {
-      "epoch": 0.24972022123732915,
-      "grad_norm": 893.337646484375,
-      "learning_rate": 4.6770056171891846e-05,
-      "loss": 75.4777,
-      "step": 61810
-    },
-    {
-      "epoch": 0.2497606225026968,
-      "grad_norm": 999.063232421875,
-      "learning_rate": 4.676833983521335e-05,
-      "loss": 133.2376,
-      "step": 61820
-    },
-    {
-      "epoch": 0.24980102376806443,
-      "grad_norm": 1203.1588134765625,
-      "learning_rate": 4.676662307414987e-05,
-      "loss": 92.8217,
-      "step": 61830
-    },
-    {
-      "epoch": 0.24984142503343204,
-      "grad_norm": 1015.3056030273438,
-      "learning_rate": 4.676490588873486e-05,
-      "loss": 71.3694,
-      "step": 61840
-    },
-    {
-      "epoch": 0.24988182629879968,
-      "grad_norm": 504.0675354003906,
-      "learning_rate": 4.6763188279001804e-05,
-      "loss": 37.9027,
-      "step": 61850
-    },
-    {
-      "epoch": 0.24992222756416732,
-      "grad_norm": 1208.43310546875,
-      "learning_rate": 4.6761470244984196e-05,
-      "loss": 79.1104,
-      "step": 61860
-    },
-    {
-      "epoch": 0.24996262882953493,
-      "grad_norm": 1380.7022705078125,
-      "learning_rate": 4.675975178671551e-05,
-      "loss": 59.1448,
-      "step": 61870
-    },
-    {
-      "epoch": 0.25000303009490255,
-      "grad_norm": 348.9476623535156,
-      "learning_rate": 4.675803290422927e-05,
-      "loss": 85.426,
-      "step": 61880
-    },
-    {
-      "epoch": 0.2500434313602702,
-      "grad_norm": 709.4434204101562,
-      "learning_rate": 4.6756313597558977e-05,
-      "loss": 52.583,
-      "step": 61890
-    },
-    {
-      "epoch": 0.2500838326256378,
-      "grad_norm": 490.9687805175781,
-      "learning_rate": 4.675459386673815e-05,
-      "loss": 84.7173,
-      "step": 61900
-    },
-    {
-      "epoch": 0.25012423389100547,
-      "grad_norm": 935.8824462890625,
-      "learning_rate": 4.6752873711800306e-05,
-      "loss": 61.2777,
-      "step": 61910
-    },
-    {
-      "epoch": 0.2501646351563731,
-      "grad_norm": 1041.913818359375,
-      "learning_rate": 4.6751153132779e-05,
-      "loss": 90.0399,
-      "step": 61920
-    },
-    {
-      "epoch": 0.25020503642174075,
-      "grad_norm": 516.0934448242188,
-      "learning_rate": 4.674943212970776e-05,
-      "loss": 53.4762,
-      "step": 61930
-    },
-    {
-      "epoch": 0.2502454376871084,
-      "grad_norm": 1382.7197265625,
-      "learning_rate": 4.674771070262014e-05,
-      "loss": 128.2385,
-      "step": 61940
-    },
-    {
-      "epoch": 0.25028583895247597,
-      "grad_norm": 759.5361938476562,
-      "learning_rate": 4.67459888515497e-05,
-      "loss": 56.0513,
-      "step": 61950
-    },
-    {
-      "epoch": 0.2503262402178436,
-      "grad_norm": 1789.28466796875,
-      "learning_rate": 4.674426657653003e-05,
-      "loss": 59.7396,
-      "step": 61960
-    },
-    {
-      "epoch": 0.25036664148321125,
-      "grad_norm": 456.0893859863281,
-      "learning_rate": 4.6742543877594675e-05,
-      "loss": 42.1051,
-      "step": 61970
-    },
-    {
-      "epoch": 0.2504070427485789,
-      "grad_norm": 1111.8485107421875,
-      "learning_rate": 4.6740820754777235e-05,
-      "loss": 62.1872,
-      "step": 61980
-    },
-    {
-      "epoch": 0.25044744401394653,
-      "grad_norm": 927.1618041992188,
-      "learning_rate": 4.6739097208111306e-05,
-      "loss": 67.374,
-      "step": 61990
-    },
-    {
-      "epoch": 0.25048784527931417,
-      "grad_norm": 898.595458984375,
-      "learning_rate": 4.6737373237630476e-05,
-      "loss": 97.9774,
-      "step": 62000
-    },
-    {
-      "epoch": 0.25052824654468175,
-      "grad_norm": 0.0,
-      "learning_rate": 4.6735648843368376e-05,
-      "loss": 87.7824,
-      "step": 62010
-    },
-    {
-      "epoch": 0.2505686478100494,
-      "grad_norm": 850.822265625,
-      "learning_rate": 4.6733924025358597e-05,
-      "loss": 58.0294,
-      "step": 62020
-    },
-    {
-      "epoch": 0.25060904907541703,
-      "grad_norm": 1727.8907470703125,
-      "learning_rate": 4.673219878363479e-05,
-      "loss": 71.4364,
-      "step": 62030
-    },
-    {
-      "epoch": 0.2506494503407847,
-      "grad_norm": 878.0247802734375,
-      "learning_rate": 4.6730473118230575e-05,
-      "loss": 90.9596,
-      "step": 62040
-    },
-    {
-      "epoch": 0.2506898516061523,
-      "grad_norm": 1600.9791259765625,
-      "learning_rate": 4.67287470291796e-05,
-      "loss": 64.3531,
-      "step": 62050
-    },
-    {
-      "epoch": 0.25073025287151995,
-      "grad_norm": 617.0396118164062,
-      "learning_rate": 4.672702051651552e-05,
-      "loss": 60.0307,
-      "step": 62060
-    },
-    {
-      "epoch": 0.2507706541368876,
-      "grad_norm": 583.3132934570312,
-      "learning_rate": 4.672529358027198e-05,
-      "loss": 82.0429,
-      "step": 62070
-    },
-    {
-      "epoch": 0.2508110554022552,
-      "grad_norm": 579.264892578125,
-      "learning_rate": 4.6723566220482664e-05,
-      "loss": 71.0199,
-      "step": 62080
-    },
-    {
-      "epoch": 0.2508514566676228,
-      "grad_norm": 723.0690307617188,
-      "learning_rate": 4.672183843718123e-05,
-      "loss": 72.1286,
-      "step": 62090
-    },
-    {
-      "epoch": 0.25089185793299046,
-      "grad_norm": 543.1493530273438,
-      "learning_rate": 4.672011023040138e-05,
-      "loss": 64.0717,
-      "step": 62100
-    },
-    {
-      "epoch": 0.2509322591983581,
-      "grad_norm": 724.9541625976562,
-      "learning_rate": 4.671838160017681e-05,
-      "loss": 63.9457,
-      "step": 62110
-    },
-    {
-      "epoch": 0.25097266046372574,
-      "grad_norm": 565.6871948242188,
-      "learning_rate": 4.6716652546541194e-05,
-      "loss": 52.1139,
-      "step": 62120
-    },
-    {
-      "epoch": 0.2510130617290934,
-      "grad_norm": 1354.7672119140625,
-      "learning_rate": 4.671492306952826e-05,
-      "loss": 76.9495,
-      "step": 62130
-    },
-    {
-      "epoch": 0.25105346299446096,
-      "grad_norm": 636.530029296875,
-      "learning_rate": 4.6713193169171724e-05,
-      "loss": 55.6664,
-      "step": 62140
-    },
-    {
-      "epoch": 0.2510938642598286,
-      "grad_norm": 1726.2562255859375,
-      "learning_rate": 4.6711462845505304e-05,
-      "loss": 81.8684,
-      "step": 62150
-    },
-    {
-      "epoch": 0.25113426552519624,
-      "grad_norm": 742.2718505859375,
-      "learning_rate": 4.6709732098562745e-05,
-      "loss": 68.1843,
-      "step": 62160
-    },
-    {
-      "epoch": 0.2511746667905639,
-      "grad_norm": 725.1959228515625,
-      "learning_rate": 4.670800092837777e-05,
-      "loss": 62.3066,
-      "step": 62170
-    },
-    {
-      "epoch": 0.2512150680559315,
-      "grad_norm": 1269.651123046875,
-      "learning_rate": 4.670626933498415e-05,
-      "loss": 75.0705,
-      "step": 62180
-    },
-    {
-      "epoch": 0.25125546932129916,
-      "grad_norm": 1048.4793701171875,
-      "learning_rate": 4.670453731841563e-05,
-      "loss": 67.2275,
-      "step": 62190
-    },
-    {
-      "epoch": 0.25129587058666675,
-      "grad_norm": 1658.6903076171875,
-      "learning_rate": 4.670280487870598e-05,
-      "loss": 87.6034,
-      "step": 62200
-    },
-    {
-      "epoch": 0.2513362718520344,
-      "grad_norm": 706.1924438476562,
-      "learning_rate": 4.670107201588898e-05,
-      "loss": 66.6542,
-      "step": 62210
-    },
-    {
-      "epoch": 0.251376673117402,
-      "grad_norm": 1182.62939453125,
-      "learning_rate": 4.669933872999841e-05,
-      "loss": 61.4396,
-      "step": 62220
-    },
-    {
-      "epoch": 0.25141707438276967,
-      "grad_norm": 973.4239501953125,
-      "learning_rate": 4.669760502106805e-05,
-      "loss": 49.4658,
-      "step": 62230
-    },
-    {
-      "epoch": 0.2514574756481373,
-      "grad_norm": 954.9755859375,
-      "learning_rate": 4.6695870889131724e-05,
-      "loss": 58.9961,
-      "step": 62240
-    },
-    {
-      "epoch": 0.25149787691350495,
-      "grad_norm": 983.7535400390625,
-      "learning_rate": 4.669413633422322e-05,
-      "loss": 67.7966,
-      "step": 62250
-    },
-    {
-      "epoch": 0.2515382781788726,
-      "grad_norm": 934.3507690429688,
-      "learning_rate": 4.669240135637635e-05,
-      "loss": 64.6385,
-      "step": 62260
-    },
-    {
-      "epoch": 0.25157867944424017,
-      "grad_norm": 535.4703369140625,
-      "learning_rate": 4.669066595562496e-05,
-      "loss": 99.6696,
-      "step": 62270
-    },
-    {
-      "epoch": 0.2516190807096078,
-      "grad_norm": 887.987060546875,
-      "learning_rate": 4.668893013200286e-05,
-      "loss": 66.7841,
-      "step": 62280
-    },
-    {
-      "epoch": 0.25165948197497545,
-      "grad_norm": 1214.275146484375,
-      "learning_rate": 4.66871938855439e-05,
-      "loss": 75.537,
-      "step": 62290
-    },
-    {
-      "epoch": 0.2516998832403431,
-      "grad_norm": 572.87158203125,
-      "learning_rate": 4.6685457216281936e-05,
-      "loss": 67.1349,
-      "step": 62300
-    },
-    {
-      "epoch": 0.25174028450571073,
-      "grad_norm": 507.9439697265625,
-      "learning_rate": 4.668372012425082e-05,
-      "loss": 79.3798,
-      "step": 62310
-    },
-    {
-      "epoch": 0.25178068577107837,
-      "grad_norm": 265.6028747558594,
-      "learning_rate": 4.6681982609484416e-05,
-      "loss": 57.9518,
-      "step": 62320
-    },
-    {
-      "epoch": 0.25182108703644596,
-      "grad_norm": 2266.7421875,
-      "learning_rate": 4.6680244672016595e-05,
-      "loss": 74.0739,
-      "step": 62330
-    },
-    {
-      "epoch": 0.2518614883018136,
-      "grad_norm": 727.9891967773438,
-      "learning_rate": 4.6678506311881245e-05,
-      "loss": 107.9434,
-      "step": 62340
-    },
-    {
-      "epoch": 0.25190188956718124,
-      "grad_norm": 769.7603759765625,
-      "learning_rate": 4.667676752911225e-05,
-      "loss": 62.9798,
-      "step": 62350
-    },
-    {
-      "epoch": 0.2519422908325489,
-      "grad_norm": 367.056396484375,
-      "learning_rate": 4.667502832374352e-05,
-      "loss": 76.6132,
-      "step": 62360
-    },
-    {
-      "epoch": 0.2519826920979165,
-      "grad_norm": 1385.6729736328125,
-      "learning_rate": 4.667328869580895e-05,
-      "loss": 56.0849,
-      "step": 62370
-    },
-    {
-      "epoch": 0.25202309336328416,
-      "grad_norm": 509.8416442871094,
-      "learning_rate": 4.6671548645342456e-05,
-      "loss": 59.8157,
-      "step": 62380
-    },
-    {
-      "epoch": 0.2520634946286518,
-      "grad_norm": 1469.747314453125,
-      "learning_rate": 4.666980817237797e-05,
-      "loss": 89.1393,
-      "step": 62390
-    },
-    {
-      "epoch": 0.2521038958940194,
-      "grad_norm": 1213.3016357421875,
-      "learning_rate": 4.6668067276949414e-05,
-      "loss": 59.1252,
-      "step": 62400
-    },
-    {
-      "epoch": 0.252144297159387,
-      "grad_norm": 990.9741821289062,
-      "learning_rate": 4.666632595909072e-05,
-      "loss": 68.1975,
-      "step": 62410
-    },
-    {
-      "epoch": 0.25218469842475466,
-      "grad_norm": 916.1857299804688,
-      "learning_rate": 4.666458421883586e-05,
-      "loss": 66.8495,
-      "step": 62420
-    },
-    {
-      "epoch": 0.2522250996901223,
-      "grad_norm": 381.66485595703125,
-      "learning_rate": 4.666284205621877e-05,
-      "loss": 91.8923,
-      "step": 62430
-    },
-    {
-      "epoch": 0.25226550095548994,
-      "grad_norm": 894.6453247070312,
-      "learning_rate": 4.666109947127343e-05,
-      "loss": 76.0347,
-      "step": 62440
-    },
-    {
-      "epoch": 0.2523059022208576,
-      "grad_norm": 455.5608215332031,
-      "learning_rate": 4.6659356464033795e-05,
-      "loss": 54.6976,
-      "step": 62450
-    },
-    {
-      "epoch": 0.25234630348622517,
-      "grad_norm": 581.934814453125,
-      "learning_rate": 4.6657613034533866e-05,
-      "loss": 60.0454,
-      "step": 62460
-    },
-    {
-      "epoch": 0.2523867047515928,
-      "grad_norm": 911.0252685546875,
-      "learning_rate": 4.665586918280761e-05,
-      "loss": 102.2392,
-      "step": 62470
-    },
-    {
-      "epoch": 0.25242710601696045,
-      "grad_norm": 1507.02685546875,
-      "learning_rate": 4.665412490888904e-05,
-      "loss": 90.9116,
-      "step": 62480
-    },
-    {
-      "epoch": 0.2524675072823281,
-      "grad_norm": 2237.4755859375,
-      "learning_rate": 4.6652380212812155e-05,
-      "loss": 93.2901,
-      "step": 62490
-    },
-    {
-      "epoch": 0.2525079085476957,
-      "grad_norm": 816.4346923828125,
-      "learning_rate": 4.665063509461097e-05,
-      "loss": 80.2981,
-      "step": 62500
-    },
-    {
-      "epoch": 0.25254830981306337,
-      "grad_norm": 769.7588500976562,
-      "learning_rate": 4.6648889554319506e-05,
-      "loss": 55.1148,
-      "step": 62510
-    },
-    {
-      "epoch": 0.25258871107843095,
-      "grad_norm": 401.32989501953125,
-      "learning_rate": 4.66471435919718e-05,
-      "loss": 67.0569,
-      "step": 62520
-    },
-    {
-      "epoch": 0.2526291123437986,
-      "grad_norm": 888.2326049804688,
-      "learning_rate": 4.6645397207601884e-05,
-      "loss": 78.4589,
-      "step": 62530
-    },
-    {
-      "epoch": 0.25266951360916623,
-      "grad_norm": 310.5882568359375,
-      "learning_rate": 4.66436504012438e-05,
-      "loss": 98.2279,
-      "step": 62540
-    },
-    {
-      "epoch": 0.25270991487453387,
-      "grad_norm": 984.7472534179688,
-      "learning_rate": 4.664190317293161e-05,
-      "loss": 97.2269,
-      "step": 62550
-    },
-    {
-      "epoch": 0.2527503161399015,
-      "grad_norm": 793.189697265625,
-      "learning_rate": 4.6640155522699374e-05,
-      "loss": 127.4911,
-      "step": 62560
-    },
-    {
-      "epoch": 0.25279071740526915,
-      "grad_norm": 2629.53466796875,
-      "learning_rate": 4.6638407450581165e-05,
-      "loss": 95.0568,
-      "step": 62570
-    },
-    {
-      "epoch": 0.2528311186706368,
-      "grad_norm": 380.42779541015625,
-      "learning_rate": 4.663665895661107e-05,
-      "loss": 81.795,
-      "step": 62580
-    },
-    {
-      "epoch": 0.2528715199360044,
-      "grad_norm": 690.9551391601562,
-      "learning_rate": 4.663491004082316e-05,
-      "loss": 50.0963,
-      "step": 62590
-    },
-    {
-      "epoch": 0.252911921201372,
-      "grad_norm": 1449.210693359375,
-      "learning_rate": 4.6633160703251554e-05,
-      "loss": 65.8343,
-      "step": 62600
-    },
-    {
-      "epoch": 0.25295232246673965,
-      "grad_norm": 1021.6344604492188,
-      "learning_rate": 4.6631410943930334e-05,
-      "loss": 107.6034,
-      "step": 62610
-    },
-    {
-      "epoch": 0.2529927237321073,
-      "grad_norm": 869.527099609375,
-      "learning_rate": 4.662966076289362e-05,
-      "loss": 102.0092,
-      "step": 62620
-    },
-    {
-      "epoch": 0.25303312499747493,
-      "grad_norm": 584.1009521484375,
-      "learning_rate": 4.662791016017554e-05,
-      "loss": 69.7825,
-      "step": 62630
-    },
-    {
-      "epoch": 0.2530735262628426,
-      "grad_norm": 802.0068359375,
-      "learning_rate": 4.6626159135810205e-05,
-      "loss": 72.057,
-      "step": 62640
-    },
-    {
-      "epoch": 0.25311392752821016,
-      "grad_norm": 2002.407958984375,
-      "learning_rate": 4.662440768983177e-05,
-      "loss": 76.6827,
-      "step": 62650
-    },
-    {
-      "epoch": 0.2531543287935778,
-      "grad_norm": 573.6917724609375,
-      "learning_rate": 4.662265582227438e-05,
-      "loss": 116.2082,
-      "step": 62660
-    },
-    {
-      "epoch": 0.25319473005894544,
-      "grad_norm": 1771.079345703125,
-      "learning_rate": 4.662090353317217e-05,
-      "loss": 81.9311,
-      "step": 62670
-    },
-    {
-      "epoch": 0.2532351313243131,
-      "grad_norm": 1918.1868896484375,
-      "learning_rate": 4.661915082255932e-05,
-      "loss": 109.1544,
-      "step": 62680
-    },
-    {
-      "epoch": 0.2532755325896807,
-      "grad_norm": 1021.0386962890625,
-      "learning_rate": 4.661739769047e-05,
-      "loss": 71.0243,
-      "step": 62690
-    },
-    {
-      "epoch": 0.25331593385504836,
-      "grad_norm": 1025.5999755859375,
-      "learning_rate": 4.6615644136938375e-05,
-      "loss": 69.5765,
-      "step": 62700
-    },
-    {
-      "epoch": 0.253356335120416,
-      "grad_norm": 441.26617431640625,
-      "learning_rate": 4.661389016199864e-05,
-      "loss": 48.4432,
-      "step": 62710
-    },
-    {
-      "epoch": 0.2533967363857836,
-      "grad_norm": 2817.1181640625,
-      "learning_rate": 4.6612135765685e-05,
-      "loss": 43.8955,
-      "step": 62720
-    },
-    {
-      "epoch": 0.2534371376511512,
-      "grad_norm": 539.3783569335938,
-      "learning_rate": 4.6610380948031627e-05,
-      "loss": 63.5467,
-      "step": 62730
-    },
-    {
-      "epoch": 0.25347753891651886,
-      "grad_norm": 735.4268188476562,
-      "learning_rate": 4.660862570907277e-05,
-      "loss": 107.1801,
-      "step": 62740
-    },
-    {
-      "epoch": 0.2535179401818865,
-      "grad_norm": 1085.6962890625,
-      "learning_rate": 4.6606870048842624e-05,
-      "loss": 88.6567,
-      "step": 62750
-    },
-    {
-      "epoch": 0.25355834144725414,
-      "grad_norm": 488.97369384765625,
-      "learning_rate": 4.660511396737541e-05,
-      "loss": 74.2245,
-      "step": 62760
-    },
-    {
-      "epoch": 0.2535987427126218,
-      "grad_norm": 3399.495361328125,
-      "learning_rate": 4.660335746470539e-05,
-      "loss": 188.06,
-      "step": 62770
-    },
-    {
-      "epoch": 0.25363914397798937,
-      "grad_norm": 515.9884033203125,
-      "learning_rate": 4.6601600540866794e-05,
-      "loss": 61.2756,
-      "step": 62780
-    },
-    {
-      "epoch": 0.253679545243357,
-      "grad_norm": 328.14276123046875,
-      "learning_rate": 4.659984319589387e-05,
-      "loss": 66.6262,
-      "step": 62790
-    },
-    {
-      "epoch": 0.25371994650872465,
-      "grad_norm": 1215.5234375,
-      "learning_rate": 4.659808542982088e-05,
-      "loss": 107.4222,
-      "step": 62800
-    },
-    {
-      "epoch": 0.2537603477740923,
-      "grad_norm": 1078.8668212890625,
-      "learning_rate": 4.65963272426821e-05,
-      "loss": 101.8976,
-      "step": 62810
-    },
-    {
-      "epoch": 0.2538007490394599,
-      "grad_norm": 620.238037109375,
-      "learning_rate": 4.659456863451181e-05,
-      "loss": 49.1838,
-      "step": 62820
-    },
-    {
-      "epoch": 0.25384115030482757,
-      "grad_norm": 357.8902587890625,
-      "learning_rate": 4.6592809605344276e-05,
-      "loss": 76.9179,
-      "step": 62830
-    },
-    {
-      "epoch": 0.25388155157019515,
-      "grad_norm": 763.6502075195312,
-      "learning_rate": 4.65910501552138e-05,
-      "loss": 68.4214,
-      "step": 62840
-    },
-    {
-      "epoch": 0.2539219528355628,
-      "grad_norm": 650.4541015625,
-      "learning_rate": 4.658929028415468e-05,
-      "loss": 67.5361,
-      "step": 62850
-    },
-    {
-      "epoch": 0.25396235410093043,
-      "grad_norm": 1105.059814453125,
-      "learning_rate": 4.658752999220125e-05,
-      "loss": 107.3979,
-      "step": 62860
-    },
-    {
-      "epoch": 0.25400275536629807,
-      "grad_norm": 1753.42822265625,
-      "learning_rate": 4.65857692793878e-05,
-      "loss": 59.3314,
-      "step": 62870
-    },
-    {
-      "epoch": 0.2540431566316657,
-      "grad_norm": 610.6260375976562,
-      "learning_rate": 4.6584008145748656e-05,
-      "loss": 75.3817,
-      "step": 62880
-    },
-    {
-      "epoch": 0.25408355789703335,
-      "grad_norm": 815.94482421875,
-      "learning_rate": 4.6582246591318175e-05,
-      "loss": 49.2156,
-      "step": 62890
-    },
-    {
-      "epoch": 0.254123959162401,
-      "grad_norm": 802.24072265625,
-      "learning_rate": 4.658048461613068e-05,
-      "loss": 84.062,
-      "step": 62900
-    },
-    {
-      "epoch": 0.2541643604277686,
-      "grad_norm": 517.5693969726562,
-      "learning_rate": 4.6578722220220525e-05,
-      "loss": 101.054,
-      "step": 62910
-    },
-    {
-      "epoch": 0.2542047616931362,
-      "grad_norm": 948.2103271484375,
-      "learning_rate": 4.657695940362207e-05,
-      "loss": 78.7945,
-      "step": 62920
-    },
-    {
-      "epoch": 0.25424516295850386,
-      "grad_norm": 751.4588623046875,
-      "learning_rate": 4.657519616636968e-05,
-      "loss": 68.9439,
-      "step": 62930
-    },
-    {
-      "epoch": 0.2542855642238715,
-      "grad_norm": 209.2197723388672,
-      "learning_rate": 4.6573432508497735e-05,
-      "loss": 68.4884,
-      "step": 62940
-    },
-    {
-      "epoch": 0.25432596548923914,
-      "grad_norm": 793.4098510742188,
-      "learning_rate": 4.6571668430040625e-05,
-      "loss": 97.053,
-      "step": 62950
-    },
-    {
-      "epoch": 0.2543663667546068,
-      "grad_norm": 606.1543579101562,
-      "learning_rate": 4.6569903931032735e-05,
-      "loss": 69.4006,
-      "step": 62960
-    },
-    {
-      "epoch": 0.25440676801997436,
-      "grad_norm": 757.6336059570312,
-      "learning_rate": 4.656813901150845e-05,
-      "loss": 79.5474,
-      "step": 62970
-    },
-    {
-      "epoch": 0.254447169285342,
-      "grad_norm": 866.605712890625,
-      "learning_rate": 4.6566373671502196e-05,
-      "loss": 56.2643,
-      "step": 62980
-    },
-    {
-      "epoch": 0.25448757055070964,
-      "grad_norm": 121.89437866210938,
-      "learning_rate": 4.656460791104839e-05,
-      "loss": 48.1909,
-      "step": 62990
-    },
-    {
-      "epoch": 0.2545279718160773,
-      "grad_norm": 1390.025390625,
-      "learning_rate": 4.656284173018144e-05,
-      "loss": 107.6447,
-      "step": 63000
-    },
-    {
-      "epoch": 0.2545683730814449,
-      "grad_norm": 1153.6959228515625,
-      "learning_rate": 4.656107512893579e-05,
-      "loss": 75.1273,
-      "step": 63010
-    },
-    {
-      "epoch": 0.25460877434681256,
-      "grad_norm": 441.3359069824219,
-      "learning_rate": 4.655930810734589e-05,
-      "loss": 46.4667,
-      "step": 63020
-    },
-    {
-      "epoch": 0.2546491756121802,
-      "grad_norm": 914.875732421875,
-      "learning_rate": 4.655754066544617e-05,
-      "loss": 58.5283,
-      "step": 63030
-    },
-    {
-      "epoch": 0.2546895768775478,
-      "grad_norm": 1101.917236328125,
-      "learning_rate": 4.65557728032711e-05,
-      "loss": 84.4799,
-      "step": 63040
-    },
-    {
-      "epoch": 0.2547299781429154,
-      "grad_norm": 985.5469360351562,
-      "learning_rate": 4.655400452085514e-05,
-      "loss": 63.2999,
-      "step": 63050
-    },
-    {
-      "epoch": 0.25477037940828307,
-      "grad_norm": 596.2723999023438,
-      "learning_rate": 4.6552235818232764e-05,
-      "loss": 72.8403,
-      "step": 63060
-    },
-    {
-      "epoch": 0.2548107806736507,
-      "grad_norm": 654.357666015625,
-      "learning_rate": 4.655046669543845e-05,
-      "loss": 57.4079,
-      "step": 63070
-    },
-    {
-      "epoch": 0.25485118193901835,
-      "grad_norm": 634.196044921875,
-      "learning_rate": 4.6548697152506705e-05,
-      "loss": 80.194,
-      "step": 63080
-    },
-    {
-      "epoch": 0.254891583204386,
-      "grad_norm": 701.0132446289062,
-      "learning_rate": 4.6546927189472014e-05,
-      "loss": 46.6557,
-      "step": 63090
-    },
-    {
-      "epoch": 0.25493198446975357,
-      "grad_norm": 1064.72412109375,
-      "learning_rate": 4.654515680636888e-05,
-      "loss": 73.7024,
-      "step": 63100
-    },
-    {
-      "epoch": 0.2549723857351212,
-      "grad_norm": 967.7996826171875,
-      "learning_rate": 4.654338600323182e-05,
-      "loss": 62.9957,
-      "step": 63110
-    },
-    {
-      "epoch": 0.25501278700048885,
-      "grad_norm": 533.4601440429688,
-      "learning_rate": 4.654161478009536e-05,
-      "loss": 63.6593,
-      "step": 63120
-    },
-    {
-      "epoch": 0.2550531882658565,
-      "grad_norm": 588.319580078125,
-      "learning_rate": 4.6539843136994036e-05,
-      "loss": 60.8479,
-      "step": 63130
-    },
-    {
-      "epoch": 0.25509358953122413,
-      "grad_norm": 1351.3424072265625,
-      "learning_rate": 4.653807107396237e-05,
-      "loss": 87.2913,
-      "step": 63140
-    },
-    {
-      "epoch": 0.25513399079659177,
-      "grad_norm": 780.9151000976562,
-      "learning_rate": 4.653629859103492e-05,
-      "loss": 64.7784,
-      "step": 63150
-    },
-    {
-      "epoch": 0.25517439206195935,
-      "grad_norm": 618.2050170898438,
-      "learning_rate": 4.653452568824625e-05,
-      "loss": 44.7988,
-      "step": 63160
-    },
-    {
-      "epoch": 0.255214793327327,
-      "grad_norm": 295.4721984863281,
-      "learning_rate": 4.653275236563091e-05,
-      "loss": 83.7096,
-      "step": 63170
-    },
-    {
-      "epoch": 0.25525519459269463,
-      "grad_norm": 567.467041015625,
-      "learning_rate": 4.653097862322348e-05,
-      "loss": 47.7101,
-      "step": 63180
-    },
-    {
-      "epoch": 0.2552955958580623,
-      "grad_norm": 1230.712646484375,
-      "learning_rate": 4.652920446105853e-05,
-      "loss": 82.6227,
-      "step": 63190
-    },
-    {
-      "epoch": 0.2553359971234299,
-      "grad_norm": 1003.6646118164062,
-      "learning_rate": 4.652742987917066e-05,
-      "loss": 82.4499,
-      "step": 63200
-    },
-    {
-      "epoch": 0.25537639838879755,
-      "grad_norm": 488.0013122558594,
-      "learning_rate": 4.652565487759446e-05,
-      "loss": 69.6767,
-      "step": 63210
-    },
-    {
-      "epoch": 0.2554167996541652,
-      "grad_norm": 423.2750244140625,
-      "learning_rate": 4.652387945636454e-05,
-      "loss": 99.7746,
-      "step": 63220
-    },
-    {
-      "epoch": 0.2554572009195328,
-      "grad_norm": 531.3796997070312,
-      "learning_rate": 4.65221036155155e-05,
-      "loss": 65.6895,
-      "step": 63230
-    },
-    {
-      "epoch": 0.2554976021849004,
-      "grad_norm": 664.51513671875,
-      "learning_rate": 4.652032735508198e-05,
-      "loss": 76.6833,
-      "step": 63240
-    },
-    {
-      "epoch": 0.25553800345026806,
-      "grad_norm": 1186.4224853515625,
-      "learning_rate": 4.65185506750986e-05,
-      "loss": 75.191,
-      "step": 63250
-    },
-    {
-      "epoch": 0.2555784047156357,
-      "grad_norm": 1025.0635986328125,
-      "learning_rate": 4.651677357559998e-05,
-      "loss": 79.5044,
-      "step": 63260
-    },
-    {
-      "epoch": 0.25561880598100334,
-      "grad_norm": 1971.0733642578125,
-      "learning_rate": 4.65149960566208e-05,
-      "loss": 78.158,
-      "step": 63270
-    },
-    {
-      "epoch": 0.255659207246371,
-      "grad_norm": 601.2109375,
-      "learning_rate": 4.651321811819568e-05,
-      "loss": 109.8638,
-      "step": 63280
-    },
-    {
-      "epoch": 0.25569960851173856,
-      "grad_norm": 926.903076171875,
-      "learning_rate": 4.65114397603593e-05,
-      "loss": 67.0008,
-      "step": 63290
-    },
-    {
-      "epoch": 0.2557400097771062,
-      "grad_norm": 543.5036010742188,
-      "learning_rate": 4.6509660983146334e-05,
-      "loss": 90.5155,
-      "step": 63300
-    },
-    {
-      "epoch": 0.25578041104247384,
-      "grad_norm": 1474.077392578125,
-      "learning_rate": 4.650788178659146e-05,
-      "loss": 76.4021,
-      "step": 63310
-    },
-    {
-      "epoch": 0.2558208123078415,
-      "grad_norm": 1020.8681030273438,
-      "learning_rate": 4.650610217072934e-05,
-      "loss": 78.0826,
-      "step": 63320
-    },
-    {
-      "epoch": 0.2558612135732091,
-      "grad_norm": 963.8372192382812,
-      "learning_rate": 4.650432213559469e-05,
-      "loss": 63.825,
-      "step": 63330
-    },
-    {
-      "epoch": 0.25590161483857676,
-      "grad_norm": 1577.4927978515625,
-      "learning_rate": 4.650254168122222e-05,
-      "loss": 76.4572,
-      "step": 63340
-    },
-    {
-      "epoch": 0.2559420161039444,
-      "grad_norm": 985.75439453125,
-      "learning_rate": 4.650076080764662e-05,
-      "loss": 45.7849,
-      "step": 63350
-    },
-    {
-      "epoch": 0.255982417369312,
-      "grad_norm": 651.5304565429688,
-      "learning_rate": 4.649897951490262e-05,
-      "loss": 86.52,
-      "step": 63360
-    },
-    {
-      "epoch": 0.2560228186346796,
-      "grad_norm": 1333.999267578125,
-      "learning_rate": 4.649719780302495e-05,
-      "loss": 59.2886,
-      "step": 63370
-    },
-    {
-      "epoch": 0.25606321990004727,
-      "grad_norm": 926.7659912109375,
-      "learning_rate": 4.649541567204834e-05,
-      "loss": 91.8786,
-      "step": 63380
-    },
-    {
-      "epoch": 0.2561036211654149,
-      "grad_norm": 720.6255493164062,
-      "learning_rate": 4.649363312200753e-05,
-      "loss": 76.3271,
-      "step": 63390
-    },
-    {
-      "epoch": 0.25614402243078255,
-      "grad_norm": 0.0,
-      "learning_rate": 4.649185015293728e-05,
-      "loss": 68.2173,
-      "step": 63400
-    },
-    {
-      "epoch": 0.2561844236961502,
-      "grad_norm": 905.1610717773438,
-      "learning_rate": 4.649006676487234e-05,
-      "loss": 72.3964,
-      "step": 63410
-    },
-    {
-      "epoch": 0.25622482496151777,
-      "grad_norm": 1136.3037109375,
-      "learning_rate": 4.6488282957847494e-05,
-      "loss": 102.3837,
-      "step": 63420
-    },
-    {
-      "epoch": 0.2562652262268854,
-      "grad_norm": 1587.19921875,
-      "learning_rate": 4.648649873189751e-05,
-      "loss": 70.8869,
-      "step": 63430
-    },
-    {
-      "epoch": 0.25630562749225305,
-      "grad_norm": 683.1766967773438,
-      "learning_rate": 4.648471408705717e-05,
-      "loss": 78.3768,
-      "step": 63440
-    },
-    {
-      "epoch": 0.2563460287576207,
-      "grad_norm": 719.8806762695312,
-      "learning_rate": 4.648292902336126e-05,
-      "loss": 81.2503,
-      "step": 63450
-    },
-    {
-      "epoch": 0.25638643002298833,
-      "grad_norm": 1979.6317138671875,
-      "learning_rate": 4.648114354084459e-05,
-      "loss": 129.6083,
-      "step": 63460
-    },
-    {
-      "epoch": 0.25642683128835597,
-      "grad_norm": 565.4490356445312,
-      "learning_rate": 4.647935763954198e-05,
-      "loss": 78.8221,
-      "step": 63470
-    },
-    {
-      "epoch": 0.25646723255372356,
-      "grad_norm": 446.4573669433594,
-      "learning_rate": 4.647757131948822e-05,
-      "loss": 103.6806,
-      "step": 63480
-    },
-    {
-      "epoch": 0.2565076338190912,
-      "grad_norm": 1135.3199462890625,
-      "learning_rate": 4.6475784580718155e-05,
-      "loss": 97.5097,
-      "step": 63490
-    },
-    {
-      "epoch": 0.25654803508445884,
-      "grad_norm": 725.8003540039062,
-      "learning_rate": 4.6473997423266614e-05,
-      "loss": 65.7066,
-      "step": 63500
-    },
-    {
-      "epoch": 0.2565884363498265,
-      "grad_norm": 844.005859375,
-      "learning_rate": 4.6472209847168435e-05,
-      "loss": 75.9603,
-      "step": 63510
-    },
-    {
-      "epoch": 0.2566288376151941,
-      "grad_norm": 555.097412109375,
-      "learning_rate": 4.647042185245847e-05,
-      "loss": 68.8893,
-      "step": 63520
-    },
-    {
-      "epoch": 0.25666923888056176,
-      "grad_norm": 434.0204772949219,
-      "learning_rate": 4.646863343917158e-05,
-      "loss": 82.4939,
-      "step": 63530
-    },
-    {
-      "epoch": 0.2567096401459294,
-      "grad_norm": 918.038818359375,
-      "learning_rate": 4.646684460734263e-05,
-      "loss": 61.4218,
-      "step": 63540
-    },
-    {
-      "epoch": 0.256750041411297,
-      "grad_norm": 617.525390625,
-      "learning_rate": 4.646505535700649e-05,
-      "loss": 55.3601,
-      "step": 63550
-    },
-    {
-      "epoch": 0.2567904426766646,
-      "grad_norm": 864.6348266601562,
-      "learning_rate": 4.6463265688198044e-05,
-      "loss": 76.1844,
-      "step": 63560
-    },
-    {
-      "epoch": 0.25683084394203226,
-      "grad_norm": 2091.900390625,
-      "learning_rate": 4.6461475600952184e-05,
-      "loss": 76.1381,
-      "step": 63570
-    },
-    {
-      "epoch": 0.2568712452073999,
-      "grad_norm": 1115.5269775390625,
-      "learning_rate": 4.645968509530381e-05,
-      "loss": 69.8563,
-      "step": 63580
-    },
-    {
-      "epoch": 0.25691164647276754,
-      "grad_norm": 808.4495849609375,
-      "learning_rate": 4.645789417128783e-05,
-      "loss": 65.0377,
-      "step": 63590
-    },
-    {
-      "epoch": 0.2569520477381352,
-      "grad_norm": 867.1176147460938,
-      "learning_rate": 4.645610282893915e-05,
-      "loss": 78.228,
-      "step": 63600
-    },
-    {
-      "epoch": 0.25699244900350277,
-      "grad_norm": 452.4241943359375,
-      "learning_rate": 4.64543110682927e-05,
-      "loss": 80.4785,
-      "step": 63610
-    },
-    {
-      "epoch": 0.2570328502688704,
-      "grad_norm": 626.6029663085938,
-      "learning_rate": 4.6452518889383414e-05,
-      "loss": 68.1861,
-      "step": 63620
-    },
-    {
-      "epoch": 0.25707325153423805,
-      "grad_norm": 561.9072265625,
-      "learning_rate": 4.645072629224622e-05,
-      "loss": 68.4941,
-      "step": 63630
-    },
-    {
-      "epoch": 0.2571136527996057,
-      "grad_norm": 495.2870788574219,
-      "learning_rate": 4.6448933276916076e-05,
-      "loss": 78.5616,
-      "step": 63640
-    },
-    {
-      "epoch": 0.2571540540649733,
-      "grad_norm": 757.0482788085938,
-      "learning_rate": 4.644713984342793e-05,
-      "loss": 61.9383,
-      "step": 63650
-    },
-    {
-      "epoch": 0.25719445533034097,
-      "grad_norm": 666.8660278320312,
-      "learning_rate": 4.644534599181677e-05,
-      "loss": 60.4049,
-      "step": 63660
-    },
-    {
-      "epoch": 0.2572348565957086,
-      "grad_norm": 487.714111328125,
-      "learning_rate": 4.644355172211753e-05,
-      "loss": 65.7735,
-      "step": 63670
-    },
-    {
-      "epoch": 0.2572752578610762,
-      "grad_norm": 403.6894836425781,
-      "learning_rate": 4.644175703436522e-05,
-      "loss": 74.5089,
-      "step": 63680
-    },
-    {
-      "epoch": 0.25731565912644383,
-      "grad_norm": 806.9922485351562,
-      "learning_rate": 4.643996192859481e-05,
-      "loss": 47.5238,
-      "step": 63690
-    },
-    {
-      "epoch": 0.25735606039181147,
-      "grad_norm": 1357.157470703125,
-      "learning_rate": 4.643816640484131e-05,
-      "loss": 86.1968,
-      "step": 63700
-    },
-    {
-      "epoch": 0.2573964616571791,
-      "grad_norm": 1357.3590087890625,
-      "learning_rate": 4.643637046313972e-05,
-      "loss": 82.4154,
-      "step": 63710
-    },
-    {
-      "epoch": 0.25743686292254675,
-      "grad_norm": 1421.4974365234375,
-      "learning_rate": 4.6434574103525044e-05,
-      "loss": 91.1663,
-      "step": 63720
-    },
-    {
-      "epoch": 0.2574772641879144,
-      "grad_norm": 1042.117431640625,
-      "learning_rate": 4.6432777326032316e-05,
-      "loss": 86.4203,
-      "step": 63730
-    },
-    {
-      "epoch": 0.257517665453282,
-      "grad_norm": 1068.4251708984375,
-      "learning_rate": 4.6430980130696555e-05,
-      "loss": 79.4296,
-      "step": 63740
-    },
-    {
-      "epoch": 0.2575580667186496,
-      "grad_norm": 709.8405151367188,
-      "learning_rate": 4.642918251755281e-05,
-      "loss": 59.4794,
-      "step": 63750
-    },
-    {
-      "epoch": 0.25759846798401725,
-      "grad_norm": 752.4713745117188,
-      "learning_rate": 4.6427384486636113e-05,
-      "loss": 80.5621,
-      "step": 63760
-    },
-    {
-      "epoch": 0.2576388692493849,
-      "grad_norm": 500.284423828125,
-      "learning_rate": 4.6425586037981526e-05,
-      "loss": 70.1939,
-      "step": 63770
-    },
-    {
-      "epoch": 0.25767927051475253,
-      "grad_norm": 1230.3875732421875,
-      "learning_rate": 4.6423787171624114e-05,
-      "loss": 93.3144,
-      "step": 63780
-    },
-    {
-      "epoch": 0.2577196717801202,
-      "grad_norm": 375.65838623046875,
-      "learning_rate": 4.642198788759894e-05,
-      "loss": 83.3093,
-      "step": 63790
-    },
-    {
-      "epoch": 0.25776007304548776,
-      "grad_norm": 1165.2003173828125,
-      "learning_rate": 4.642018818594107e-05,
-      "loss": 67.3937,
-      "step": 63800
-    },
-    {
-      "epoch": 0.2578004743108554,
-      "grad_norm": 1046.146728515625,
-      "learning_rate": 4.641838806668562e-05,
-      "loss": 72.0059,
-      "step": 63810
-    },
-    {
-      "epoch": 0.25784087557622304,
-      "grad_norm": 695.942138671875,
-      "learning_rate": 4.6416587529867664e-05,
-      "loss": 54.245,
-      "step": 63820
-    },
-    {
-      "epoch": 0.2578812768415907,
-      "grad_norm": 1145.5577392578125,
-      "learning_rate": 4.6414786575522306e-05,
-      "loss": 88.9208,
-      "step": 63830
-    },
-    {
-      "epoch": 0.2579216781069583,
-      "grad_norm": 453.61181640625,
-      "learning_rate": 4.6412985203684654e-05,
-      "loss": 66.1485,
-      "step": 63840
-    },
-    {
-      "epoch": 0.25796207937232596,
-      "grad_norm": 765.479248046875,
-      "learning_rate": 4.6411183414389837e-05,
-      "loss": 74.9149,
-      "step": 63850
-    },
-    {
-      "epoch": 0.2580024806376936,
-      "grad_norm": 794.6134643554688,
-      "learning_rate": 4.6409381207672974e-05,
-      "loss": 58.8936,
-      "step": 63860
-    },
-    {
-      "epoch": 0.2580428819030612,
-      "grad_norm": 582.8801879882812,
-      "learning_rate": 4.64075785835692e-05,
-      "loss": 95.4306,
-      "step": 63870
-    },
-    {
-      "epoch": 0.2580832831684288,
-      "grad_norm": 1164.59619140625,
-      "learning_rate": 4.640577554211366e-05,
-      "loss": 64.1848,
-      "step": 63880
-    },
-    {
-      "epoch": 0.25812368443379646,
-      "grad_norm": 1310.0181884765625,
-      "learning_rate": 4.640397208334151e-05,
-      "loss": 63.126,
-      "step": 63890
-    },
-    {
-      "epoch": 0.2581640856991641,
-      "grad_norm": 571.4081420898438,
-      "learning_rate": 4.64021682072879e-05,
-      "loss": 42.7484,
-      "step": 63900
-    },
-    {
-      "epoch": 0.25820448696453174,
-      "grad_norm": 726.766845703125,
-      "learning_rate": 4.640036391398801e-05,
-      "loss": 43.3208,
-      "step": 63910
-    },
-    {
-      "epoch": 0.2582448882298994,
-      "grad_norm": 754.6783447265625,
-      "learning_rate": 4.639855920347701e-05,
-      "loss": 110.9717,
-      "step": 63920
-    },
-    {
-      "epoch": 0.25828528949526697,
-      "grad_norm": 1137.88818359375,
-      "learning_rate": 4.639675407579007e-05,
-      "loss": 70.8244,
-      "step": 63930
-    },
-    {
-      "epoch": 0.2583256907606346,
-      "grad_norm": 508.732421875,
-      "learning_rate": 4.6394948530962396e-05,
-      "loss": 55.6226,
-      "step": 63940
-    },
-    {
-      "epoch": 0.25836609202600225,
-      "grad_norm": 405.8502197265625,
-      "learning_rate": 4.639314256902919e-05,
-      "loss": 51.9442,
-      "step": 63950
-    },
-    {
-      "epoch": 0.2584064932913699,
-      "grad_norm": 941.8868408203125,
-      "learning_rate": 4.6391336190025644e-05,
-      "loss": 97.1299,
-      "step": 63960
-    },
-    {
-      "epoch": 0.2584468945567375,
-      "grad_norm": 368.6508483886719,
-      "learning_rate": 4.6389529393987e-05,
-      "loss": 52.9645,
-      "step": 63970
-    },
-    {
-      "epoch": 0.25848729582210517,
-      "grad_norm": 846.4256591796875,
-      "learning_rate": 4.638772218094847e-05,
-      "loss": 76.5249,
-      "step": 63980
-    },
-    {
-      "epoch": 0.25852769708747275,
-      "grad_norm": 552.8169555664062,
-      "learning_rate": 4.638591455094527e-05,
-      "loss": 58.1772,
-      "step": 63990
-    },
-    {
-      "epoch": 0.2585680983528404,
-      "grad_norm": 1205.1046142578125,
-      "learning_rate": 4.638410650401267e-05,
-      "loss": 59.0313,
-      "step": 64000
-    },
-    {
-      "epoch": 0.25860849961820803,
-      "grad_norm": 979.0299682617188,
-      "learning_rate": 4.63822980401859e-05,
-      "loss": 75.0273,
-      "step": 64010
-    },
-    {
-      "epoch": 0.25864890088357567,
-      "grad_norm": 1404.583740234375,
-      "learning_rate": 4.638048915950022e-05,
-      "loss": 77.1647,
-      "step": 64020
-    },
-    {
-      "epoch": 0.2586893021489433,
-      "grad_norm": 1869.6007080078125,
-      "learning_rate": 4.637867986199089e-05,
-      "loss": 57.802,
-      "step": 64030
-    },
-    {
-      "epoch": 0.25872970341431095,
-      "grad_norm": 502.42138671875,
-      "learning_rate": 4.6376870147693196e-05,
-      "loss": 62.1255,
-      "step": 64040
-    },
-    {
-      "epoch": 0.2587701046796786,
-      "grad_norm": 700.3970947265625,
-      "learning_rate": 4.6375060016642415e-05,
-      "loss": 72.0223,
-      "step": 64050
-    },
-    {
-      "epoch": 0.2588105059450462,
-      "grad_norm": 1100.4659423828125,
-      "learning_rate": 4.6373249468873833e-05,
-      "loss": 84.0306,
-      "step": 64060
-    },
-    {
-      "epoch": 0.2588509072104138,
-      "grad_norm": 818.85595703125,
-      "learning_rate": 4.637143850442275e-05,
-      "loss": 82.6929,
-      "step": 64070
-    },
-    {
-      "epoch": 0.25889130847578146,
-      "grad_norm": 373.8831787109375,
-      "learning_rate": 4.6369627123324465e-05,
-      "loss": 65.1804,
-      "step": 64080
-    },
-    {
-      "epoch": 0.2589317097411491,
-      "grad_norm": 932.3283081054688,
-      "learning_rate": 4.6367815325614306e-05,
-      "loss": 72.5139,
-      "step": 64090
-    },
-    {
-      "epoch": 0.25897211100651674,
-      "grad_norm": 1563.8875732421875,
-      "learning_rate": 4.636600311132758e-05,
-      "loss": 100.7779,
-      "step": 64100
-    },
-    {
-      "epoch": 0.2590125122718844,
-      "grad_norm": 1865.01220703125,
-      "learning_rate": 4.6364190480499624e-05,
-      "loss": 105.2925,
-      "step": 64110
-    },
-    {
-      "epoch": 0.25905291353725196,
-      "grad_norm": 947.768798828125,
-      "learning_rate": 4.636237743316578e-05,
-      "loss": 58.8598,
-      "step": 64120
-    },
-    {
-      "epoch": 0.2590933148026196,
-      "grad_norm": 663.779052734375,
-      "learning_rate": 4.636056396936138e-05,
-      "loss": 97.7154,
-      "step": 64130
-    },
-    {
-      "epoch": 0.25913371606798724,
-      "grad_norm": 814.3782958984375,
-      "learning_rate": 4.6358750089121795e-05,
-      "loss": 72.49,
-      "step": 64140
-    },
-    {
-      "epoch": 0.2591741173333549,
-      "grad_norm": 634.1887817382812,
-      "learning_rate": 4.635693579248238e-05,
-      "loss": 64.081,
-      "step": 64150
-    },
-    {
-      "epoch": 0.2592145185987225,
-      "grad_norm": 1665.0477294921875,
-      "learning_rate": 4.635512107947851e-05,
-      "loss": 60.826,
-      "step": 64160
-    },
-    {
-      "epoch": 0.25925491986409016,
-      "grad_norm": 587.009765625,
-      "learning_rate": 4.635330595014555e-05,
-      "loss": 62.3891,
-      "step": 64170
-    },
-    {
-      "epoch": 0.2592953211294578,
-      "grad_norm": 1239.5069580078125,
-      "learning_rate": 4.635149040451891e-05,
-      "loss": 83.9093,
-      "step": 64180
-    },
-    {
-      "epoch": 0.2593357223948254,
-      "grad_norm": 743.6141357421875,
-      "learning_rate": 4.634967444263397e-05,
-      "loss": 111.2367,
-      "step": 64190
-    },
-    {
-      "epoch": 0.259376123660193,
-      "grad_norm": 848.8331909179688,
-      "learning_rate": 4.6347858064526125e-05,
-      "loss": 56.9608,
-      "step": 64200
-    },
-    {
-      "epoch": 0.25941652492556067,
-      "grad_norm": 628.3904418945312,
-      "learning_rate": 4.6346041270230804e-05,
-      "loss": 72.7587,
-      "step": 64210
-    },
-    {
-      "epoch": 0.2594569261909283,
-      "grad_norm": 849.9680786132812,
-      "learning_rate": 4.634422405978342e-05,
-      "loss": 47.3868,
-      "step": 64220
-    },
-    {
-      "epoch": 0.25949732745629595,
-      "grad_norm": 474.90264892578125,
-      "learning_rate": 4.6342406433219394e-05,
-      "loss": 102.6614,
-      "step": 64230
-    },
-    {
-      "epoch": 0.2595377287216636,
-      "grad_norm": 2507.705322265625,
-      "learning_rate": 4.634058839057417e-05,
-      "loss": 70.4811,
-      "step": 64240
-    },
-    {
-      "epoch": 0.25957812998703117,
-      "grad_norm": 1063.095458984375,
-      "learning_rate": 4.6338769931883185e-05,
-      "loss": 78.3297,
-      "step": 64250
-    },
-    {
-      "epoch": 0.2596185312523988,
-      "grad_norm": 622.4473266601562,
-      "learning_rate": 4.63369510571819e-05,
-      "loss": 77.0825,
-      "step": 64260
-    },
-    {
-      "epoch": 0.25965893251776645,
-      "grad_norm": 674.91455078125,
-      "learning_rate": 4.633513176650577e-05,
-      "loss": 60.1044,
-      "step": 64270
-    },
-    {
-      "epoch": 0.2596993337831341,
-      "grad_norm": 857.1925659179688,
-      "learning_rate": 4.6333312059890256e-05,
-      "loss": 75.2282,
-      "step": 64280
-    },
-    {
-      "epoch": 0.25973973504850173,
-      "grad_norm": 1186.2684326171875,
-      "learning_rate": 4.633149193737084e-05,
-      "loss": 90.6651,
-      "step": 64290
-    },
-    {
-      "epoch": 0.25978013631386937,
-      "grad_norm": 503.4192810058594,
-      "learning_rate": 4.632967139898301e-05,
-      "loss": 52.2867,
-      "step": 64300
-    },
-    {
-      "epoch": 0.25982053757923695,
-      "grad_norm": 0.0,
-      "learning_rate": 4.632785044476225e-05,
-      "loss": 54.2169,
-      "step": 64310
-    },
-    {
-      "epoch": 0.2598609388446046,
-      "grad_norm": 808.451904296875,
-      "learning_rate": 4.6326029074744074e-05,
-      "loss": 89.7694,
-      "step": 64320
-    },
-    {
-      "epoch": 0.25990134010997223,
-      "grad_norm": 3421.571533203125,
-      "learning_rate": 4.6324207288963974e-05,
-      "loss": 94.0056,
-      "step": 64330
-    },
-    {
-      "epoch": 0.2599417413753399,
-      "grad_norm": 300.9707336425781,
-      "learning_rate": 4.632238508745748e-05,
-      "loss": 72.748,
-      "step": 64340
-    },
-    {
-      "epoch": 0.2599821426407075,
-      "grad_norm": 1028.27197265625,
-      "learning_rate": 4.632056247026011e-05,
-      "loss": 66.6853,
-      "step": 64350
-    },
-    {
-      "epoch": 0.26002254390607515,
-      "grad_norm": 730.0326538085938,
-      "learning_rate": 4.63187394374074e-05,
-      "loss": 90.4567,
-      "step": 64360
-    },
-    {
-      "epoch": 0.2600629451714428,
-      "grad_norm": 859.37890625,
-      "learning_rate": 4.6316915988934884e-05,
-      "loss": 76.2773,
-      "step": 64370
-    },
-    {
-      "epoch": 0.2601033464368104,
-      "grad_norm": 956.5285034179688,
-      "learning_rate": 4.631509212487811e-05,
-      "loss": 70.2754,
-      "step": 64380
-    },
-    {
-      "epoch": 0.260143747702178,
-      "grad_norm": 627.6688232421875,
-      "learning_rate": 4.6313267845272656e-05,
-      "loss": 61.4442,
-      "step": 64390
-    },
-    {
-      "epoch": 0.26018414896754566,
-      "grad_norm": 716.159423828125,
-      "learning_rate": 4.631144315015407e-05,
-      "loss": 72.7093,
-      "step": 64400
-    },
-    {
-      "epoch": 0.2602245502329133,
-      "grad_norm": 447.9820861816406,
-      "learning_rate": 4.630961803955792e-05,
-      "loss": 66.9453,
-      "step": 64410
-    },
-    {
-      "epoch": 0.26026495149828094,
-      "grad_norm": 588.1676025390625,
-      "learning_rate": 4.63077925135198e-05,
-      "loss": 44.5305,
-      "step": 64420
-    },
-    {
-      "epoch": 0.2603053527636486,
-      "grad_norm": 594.3401489257812,
-      "learning_rate": 4.6305966572075286e-05,
-      "loss": 60.6098,
-      "step": 64430
-    },
-    {
-      "epoch": 0.26034575402901616,
-      "grad_norm": 1224.4241943359375,
-      "learning_rate": 4.630414021525999e-05,
-      "loss": 95.3115,
-      "step": 64440
-    },
-    {
-      "epoch": 0.2603861552943838,
-      "grad_norm": 1307.3175048828125,
-      "learning_rate": 4.6302313443109526e-05,
-      "loss": 59.6623,
-      "step": 64450
-    },
-    {
-      "epoch": 0.26042655655975144,
-      "grad_norm": 667.8453979492188,
-      "learning_rate": 4.6300486255659484e-05,
-      "loss": 68.9579,
-      "step": 64460
-    },
-    {
-      "epoch": 0.2604669578251191,
-      "grad_norm": 648.3145141601562,
-      "learning_rate": 4.6298658652945494e-05,
-      "loss": 72.2152,
-      "step": 64470
-    },
-    {
-      "epoch": 0.2605073590904867,
-      "grad_norm": 1166.9095458984375,
-      "learning_rate": 4.629683063500319e-05,
-      "loss": 56.1489,
-      "step": 64480
-    },
-    {
-      "epoch": 0.26054776035585436,
-      "grad_norm": 856.700927734375,
-      "learning_rate": 4.629500220186821e-05,
-      "loss": 92.3313,
-      "step": 64490
-    },
-    {
-      "epoch": 0.260588161621222,
-      "grad_norm": 1207.9443359375,
-      "learning_rate": 4.629317335357619e-05,
-      "loss": 62.1099,
-      "step": 64500
-    },
-    {
-      "epoch": 0.2606285628865896,
-      "grad_norm": 618.7341918945312,
-      "learning_rate": 4.6291344090162804e-05,
-      "loss": 89.1145,
-      "step": 64510
-    },
-    {
-      "epoch": 0.2606689641519572,
-      "grad_norm": 1234.6435546875,
-      "learning_rate": 4.62895144116637e-05,
-      "loss": 81.7492,
-      "step": 64520
-    },
-    {
-      "epoch": 0.26070936541732487,
-      "grad_norm": 1199.2274169921875,
-      "learning_rate": 4.628768431811455e-05,
-      "loss": 93.2701,
-      "step": 64530
-    },
-    {
-      "epoch": 0.2607497666826925,
-      "grad_norm": 787.503662109375,
-      "learning_rate": 4.6285853809551036e-05,
-      "loss": 67.4107,
-      "step": 64540
-    },
-    {
-      "epoch": 0.26079016794806015,
-      "grad_norm": 1185.4876708984375,
-      "learning_rate": 4.6284022886008836e-05,
-      "loss": 137.017,
-      "step": 64550
-    },
-    {
-      "epoch": 0.2608305692134278,
-      "grad_norm": 1550.175048828125,
-      "learning_rate": 4.628219154752367e-05,
-      "loss": 105.3924,
-      "step": 64560
-    },
-    {
-      "epoch": 0.26087097047879537,
-      "grad_norm": 474.1405334472656,
-      "learning_rate": 4.628035979413121e-05,
-      "loss": 63.6596,
-      "step": 64570
-    },
-    {
-      "epoch": 0.260911371744163,
-      "grad_norm": 631.3251953125,
-      "learning_rate": 4.627852762586718e-05,
-      "loss": 88.0774,
-      "step": 64580
-    },
-    {
-      "epoch": 0.26095177300953065,
-      "grad_norm": 932.6629638671875,
-      "learning_rate": 4.627669504276731e-05,
-      "loss": 69.9022,
-      "step": 64590
-    },
-    {
-      "epoch": 0.2609921742748983,
-      "grad_norm": 462.3409423828125,
-      "learning_rate": 4.6274862044867304e-05,
-      "loss": 105.5615,
-      "step": 64600
-    },
-    {
-      "epoch": 0.26103257554026593,
-      "grad_norm": 421.7193908691406,
-      "learning_rate": 4.627302863220291e-05,
-      "loss": 83.5629,
-      "step": 64610
-    },
-    {
-      "epoch": 0.26107297680563357,
-      "grad_norm": 849.374267578125,
-      "learning_rate": 4.627119480480987e-05,
-      "loss": 108.5556,
-      "step": 64620
-    },
-    {
-      "epoch": 0.26111337807100116,
-      "grad_norm": 851.120849609375,
-      "learning_rate": 4.626936056272394e-05,
-      "loss": 76.4113,
-      "step": 64630
-    },
-    {
-      "epoch": 0.2611537793363688,
-      "grad_norm": 587.6618041992188,
-      "learning_rate": 4.626752590598088e-05,
-      "loss": 97.3091,
-      "step": 64640
-    },
-    {
-      "epoch": 0.26119418060173644,
-      "grad_norm": 1624.475341796875,
-      "learning_rate": 4.6265690834616446e-05,
-      "loss": 66.4535,
-      "step": 64650
-    },
-    {
-      "epoch": 0.2612345818671041,
-      "grad_norm": 194.85707092285156,
-      "learning_rate": 4.626385534866642e-05,
-      "loss": 51.6759,
-      "step": 64660
-    },
-    {
-      "epoch": 0.2612749831324717,
-      "grad_norm": 1180.03369140625,
-      "learning_rate": 4.626201944816659e-05,
-      "loss": 51.8923,
-      "step": 64670
-    },
-    {
-      "epoch": 0.26131538439783936,
-      "grad_norm": 951.8189697265625,
-      "learning_rate": 4.626018313315275e-05,
-      "loss": 76.8008,
-      "step": 64680
-    },
-    {
-      "epoch": 0.261355785663207,
-      "grad_norm": 894.181396484375,
-      "learning_rate": 4.625834640366068e-05,
-      "loss": 93.1069,
-      "step": 64690
-    },
-    {
-      "epoch": 0.2613961869285746,
-      "grad_norm": 575.2242431640625,
-      "learning_rate": 4.625650925972622e-05,
-      "loss": 85.9263,
-      "step": 64700
-    },
-    {
-      "epoch": 0.2614365881939422,
-      "grad_norm": 925.6211547851562,
-      "learning_rate": 4.6254671701385154e-05,
-      "loss": 69.3511,
-      "step": 64710
-    },
-    {
-      "epoch": 0.26147698945930986,
-      "grad_norm": 588.945068359375,
-      "learning_rate": 4.625283372867333e-05,
-      "loss": 36.42,
-      "step": 64720
-    },
-    {
-      "epoch": 0.2615173907246775,
-      "grad_norm": 1043.0775146484375,
-      "learning_rate": 4.625099534162656e-05,
-      "loss": 81.0917,
-      "step": 64730
-    },
-    {
-      "epoch": 0.26155779199004514,
-      "grad_norm": 1658.7032470703125,
-      "learning_rate": 4.62491565402807e-05,
-      "loss": 62.2384,
-      "step": 64740
-    },
-    {
-      "epoch": 0.2615981932554128,
-      "grad_norm": 1211.529052734375,
-      "learning_rate": 4.6247317324671605e-05,
-      "loss": 74.7915,
-      "step": 64750
-    },
-    {
-      "epoch": 0.26163859452078037,
-      "grad_norm": 1145.8216552734375,
-      "learning_rate": 4.6245477694835106e-05,
-      "loss": 52.6177,
-      "step": 64760
-    },
-    {
-      "epoch": 0.261678995786148,
-      "grad_norm": 758.6449584960938,
-      "learning_rate": 4.6243637650807086e-05,
-      "loss": 85.0178,
-      "step": 64770
-    },
-    {
-      "epoch": 0.26171939705151565,
-      "grad_norm": 427.8953857421875,
-      "learning_rate": 4.624179719262342e-05,
-      "loss": 76.1282,
-      "step": 64780
-    },
-    {
-      "epoch": 0.2617597983168833,
-      "grad_norm": 750.1713256835938,
-      "learning_rate": 4.623995632031997e-05,
-      "loss": 74.4636,
-      "step": 64790
-    },
-    {
-      "epoch": 0.2618001995822509,
-      "grad_norm": 421.75958251953125,
-      "learning_rate": 4.6238115033932636e-05,
-      "loss": 56.7678,
-      "step": 64800
-    },
-    {
-      "epoch": 0.26184060084761857,
-      "grad_norm": 1215.94482421875,
-      "learning_rate": 4.623627333349732e-05,
-      "loss": 72.2328,
-      "step": 64810
-    },
-    {
-      "epoch": 0.2618810021129862,
-      "grad_norm": 0.0,
-      "learning_rate": 4.623443121904992e-05,
-      "loss": 68.4072,
-      "step": 64820
-    },
-    {
-      "epoch": 0.2619214033783538,
-      "grad_norm": 739.1293334960938,
-      "learning_rate": 4.623258869062636e-05,
-      "loss": 58.8318,
-      "step": 64830
-    },
-    {
-      "epoch": 0.26196180464372143,
-      "grad_norm": 760.8939819335938,
-      "learning_rate": 4.623074574826254e-05,
-      "loss": 85.5972,
-      "step": 64840
-    },
-    {
-      "epoch": 0.26200220590908907,
-      "grad_norm": 902.6166381835938,
-      "learning_rate": 4.622890239199441e-05,
-      "loss": 77.9647,
-      "step": 64850
-    },
-    {
-      "epoch": 0.2620426071744567,
-      "grad_norm": 504.2374572753906,
-      "learning_rate": 4.622705862185789e-05,
-      "loss": 61.3637,
-      "step": 64860
-    },
-    {
-      "epoch": 0.26208300843982435,
-      "grad_norm": 550.212890625,
-      "learning_rate": 4.622521443788894e-05,
-      "loss": 58.025,
-      "step": 64870
-    },
-    {
-      "epoch": 0.262123409705192,
-      "grad_norm": 1491.5367431640625,
-      "learning_rate": 4.622336984012351e-05,
-      "loss": 119.6012,
-      "step": 64880
-    },
-    {
-      "epoch": 0.2621638109705596,
-      "grad_norm": 1074.4013671875,
-      "learning_rate": 4.622152482859755e-05,
-      "loss": 53.4368,
-      "step": 64890
-    },
-    {
-      "epoch": 0.2622042122359272,
-      "grad_norm": 1384.9111328125,
-      "learning_rate": 4.621967940334705e-05,
-      "loss": 73.9773,
-      "step": 64900
-    },
-    {
-      "epoch": 0.26224461350129485,
-      "grad_norm": 594.1865234375,
-      "learning_rate": 4.621783356440796e-05,
-      "loss": 59.9528,
-      "step": 64910
-    },
-    {
-      "epoch": 0.2622850147666625,
-      "grad_norm": 1251.0433349609375,
-      "learning_rate": 4.621598731181629e-05,
-      "loss": 69.5674,
-      "step": 64920
-    },
-    {
-      "epoch": 0.26232541603203013,
-      "grad_norm": 799.28564453125,
-      "learning_rate": 4.621414064560803e-05,
-      "loss": 90.669,
-      "step": 64930
-    },
-    {
-      "epoch": 0.2623658172973978,
-      "grad_norm": 766.1452026367188,
-      "learning_rate": 4.6212293565819166e-05,
-      "loss": 65.2333,
-      "step": 64940
-    },
-    {
-      "epoch": 0.26240621856276536,
-      "grad_norm": 2778.8935546875,
-      "learning_rate": 4.6210446072485725e-05,
-      "loss": 89.6359,
-      "step": 64950
-    },
-    {
-      "epoch": 0.262446619828133,
-      "grad_norm": 1777.0819091796875,
-      "learning_rate": 4.6208598165643715e-05,
-      "loss": 106.0751,
-      "step": 64960
-    },
-    {
-      "epoch": 0.26248702109350064,
-      "grad_norm": 607.8473510742188,
-      "learning_rate": 4.6206749845329164e-05,
-      "loss": 65.859,
-      "step": 64970
-    },
-    {
-      "epoch": 0.2625274223588683,
-      "grad_norm": 640.949951171875,
-      "learning_rate": 4.62049011115781e-05,
-      "loss": 93.8094,
-      "step": 64980
-    },
-    {
-      "epoch": 0.2625678236242359,
-      "grad_norm": 449.35650634765625,
-      "learning_rate": 4.620305196442659e-05,
-      "loss": 52.2666,
-      "step": 64990
-    },
-    {
-      "epoch": 0.26260822488960356,
-      "grad_norm": 1173.4410400390625,
-      "learning_rate": 4.620120240391065e-05,
-      "loss": 93.0513,
-      "step": 65000
-    },
-    {
-      "epoch": 0.2626486261549712,
-      "grad_norm": 489.27716064453125,
-      "learning_rate": 4.619935243006636e-05,
-      "loss": 64.3064,
-      "step": 65010
-    },
-    {
-      "epoch": 0.2626890274203388,
-      "grad_norm": 265.78076171875,
-      "learning_rate": 4.619750204292978e-05,
-      "loss": 80.1383,
-      "step": 65020
-    },
-    {
-      "epoch": 0.2627294286857064,
-      "grad_norm": 581.3447265625,
-      "learning_rate": 4.619565124253698e-05,
-      "loss": 41.3353,
-      "step": 65030
-    },
-    {
-      "epoch": 0.26276982995107406,
-      "grad_norm": 3157.60595703125,
-      "learning_rate": 4.619380002892406e-05,
-      "loss": 86.7737,
-      "step": 65040
-    },
-    {
-      "epoch": 0.2628102312164417,
-      "grad_norm": 679.6066284179688,
-      "learning_rate": 4.619194840212708e-05,
-      "loss": 73.4557,
-      "step": 65050
-    },
-    {
-      "epoch": 0.26285063248180934,
-      "grad_norm": 352.4366760253906,
-      "learning_rate": 4.6190096362182167e-05,
-      "loss": 84.3029,
-      "step": 65060
-    },
-    {
-      "epoch": 0.262891033747177,
-      "grad_norm": 1345.6533203125,
-      "learning_rate": 4.618824390912541e-05,
-      "loss": 76.0789,
-      "step": 65070
-    },
-    {
-      "epoch": 0.26293143501254457,
-      "grad_norm": 606.6847534179688,
-      "learning_rate": 4.618639104299294e-05,
-      "loss": 84.0097,
-      "step": 65080
-    },
-    {
-      "epoch": 0.2629718362779122,
-      "grad_norm": 777.6514892578125,
-      "learning_rate": 4.618453776382086e-05,
-      "loss": 54.7092,
-      "step": 65090
-    },
-    {
-      "epoch": 0.26301223754327985,
-      "grad_norm": 686.7308959960938,
-      "learning_rate": 4.61826840716453e-05,
-      "loss": 61.8434,
-      "step": 65100
-    },
-    {
-      "epoch": 0.2630526388086475,
-      "grad_norm": 538.2279052734375,
-      "learning_rate": 4.618082996650243e-05,
-      "loss": 49.7606,
-      "step": 65110
-    },
-    {
-      "epoch": 0.2630930400740151,
-      "grad_norm": 537.077392578125,
-      "learning_rate": 4.617897544842836e-05,
-      "loss": 66.0555,
-      "step": 65120
-    },
-    {
-      "epoch": 0.26313344133938277,
-      "grad_norm": 676.6751098632812,
-      "learning_rate": 4.617712051745927e-05,
-      "loss": 64.1107,
-      "step": 65130
-    },
-    {
-      "epoch": 0.2631738426047504,
-      "grad_norm": 587.270263671875,
-      "learning_rate": 4.61752651736313e-05,
-      "loss": 63.3736,
-      "step": 65140
-    },
-    {
-      "epoch": 0.263214243870118,
-      "grad_norm": 1263.456787109375,
-      "learning_rate": 4.617340941698064e-05,
-      "loss": 59.6258,
-      "step": 65150
-    },
-    {
-      "epoch": 0.26325464513548563,
-      "grad_norm": 427.99737548828125,
-      "learning_rate": 4.617155324754346e-05,
-      "loss": 52.6464,
-      "step": 65160
-    },
-    {
-      "epoch": 0.26329504640085327,
-      "grad_norm": 0.0,
-      "learning_rate": 4.616969666535596e-05,
-      "loss": 73.0998,
-      "step": 65170
-    },
-    {
-      "epoch": 0.2633354476662209,
-      "grad_norm": 617.71337890625,
-      "learning_rate": 4.6167839670454315e-05,
-      "loss": 44.5023,
-      "step": 65180
-    },
-    {
-      "epoch": 0.26337584893158855,
-      "grad_norm": 517.5345458984375,
-      "learning_rate": 4.616598226287474e-05,
-      "loss": 95.0356,
-      "step": 65190
-    },
-    {
-      "epoch": 0.2634162501969562,
-      "grad_norm": 1026.675537109375,
-      "learning_rate": 4.616412444265345e-05,
-      "loss": 67.167,
-      "step": 65200
-    },
-    {
-      "epoch": 0.2634566514623238,
-      "grad_norm": 1128.9427490234375,
-      "learning_rate": 4.616226620982665e-05,
-      "loss": 89.6982,
-      "step": 65210
-    },
-    {
-      "epoch": 0.2634970527276914,
-      "grad_norm": 809.5255737304688,
-      "learning_rate": 4.6160407564430574e-05,
-      "loss": 65.6736,
-      "step": 65220
-    },
-    {
-      "epoch": 0.26353745399305906,
-      "grad_norm": 1231.8623046875,
-      "learning_rate": 4.6158548506501464e-05,
-      "loss": 86.9045,
-      "step": 65230
-    },
-    {
-      "epoch": 0.2635778552584267,
-      "grad_norm": 1171.7540283203125,
-      "learning_rate": 4.6156689036075555e-05,
-      "loss": 77.3384,
-      "step": 65240
-    },
-    {
-      "epoch": 0.26361825652379434,
-      "grad_norm": 619.1737060546875,
-      "learning_rate": 4.615482915318911e-05,
-      "loss": 65.3949,
-      "step": 65250
-    },
-    {
-      "epoch": 0.263658657789162,
-      "grad_norm": 518.4265747070312,
-      "learning_rate": 4.6152968857878366e-05,
-      "loss": 55.1953,
-      "step": 65260
-    },
-    {
-      "epoch": 0.26369905905452956,
-      "grad_norm": 1298.9517822265625,
-      "learning_rate": 4.615110815017961e-05,
-      "loss": 56.1695,
-      "step": 65270
-    },
-    {
-      "epoch": 0.2637394603198972,
-      "grad_norm": 452.3489074707031,
-      "learning_rate": 4.614924703012911e-05,
-      "loss": 58.1403,
-      "step": 65280
-    },
-    {
-      "epoch": 0.26377986158526484,
-      "grad_norm": 809.9728393554688,
-      "learning_rate": 4.614738549776315e-05,
-      "loss": 49.3975,
-      "step": 65290
-    },
-    {
-      "epoch": 0.2638202628506325,
-      "grad_norm": 703.8427124023438,
-      "learning_rate": 4.614552355311802e-05,
-      "loss": 54.3999,
-      "step": 65300
-    },
-    {
-      "epoch": 0.2638606641160001,
-      "grad_norm": 518.8963623046875,
-      "learning_rate": 4.6143661196230026e-05,
-      "loss": 59.6565,
-      "step": 65310
-    },
-    {
-      "epoch": 0.26390106538136776,
-      "grad_norm": 780.0380249023438,
-      "learning_rate": 4.614179842713547e-05,
-      "loss": 85.9944,
-      "step": 65320
-    },
-    {
-      "epoch": 0.2639414666467354,
-      "grad_norm": 855.755126953125,
-      "learning_rate": 4.613993524587067e-05,
-      "loss": 57.039,
-      "step": 65330
-    },
-    {
-      "epoch": 0.263981867912103,
-      "grad_norm": 722.5667114257812,
-      "learning_rate": 4.613807165247195e-05,
-      "loss": 77.862,
-      "step": 65340
-    },
-    {
-      "epoch": 0.2640222691774706,
-      "grad_norm": 833.8828125,
-      "learning_rate": 4.613620764697564e-05,
-      "loss": 70.7339,
-      "step": 65350
-    },
-    {
-      "epoch": 0.26406267044283827,
-      "grad_norm": 827.6605224609375,
-      "learning_rate": 4.6134343229418075e-05,
-      "loss": 91.2179,
-      "step": 65360
-    },
-    {
-      "epoch": 0.2641030717082059,
-      "grad_norm": 398.13287353515625,
-      "learning_rate": 4.613247839983561e-05,
-      "loss": 85.9632,
-      "step": 65370
-    },
-    {
-      "epoch": 0.26414347297357355,
-      "grad_norm": 661.8615112304688,
-      "learning_rate": 4.613061315826461e-05,
-      "loss": 89.6971,
-      "step": 65380
-    },
-    {
-      "epoch": 0.2641838742389412,
-      "grad_norm": 425.14483642578125,
-      "learning_rate": 4.612874750474142e-05,
-      "loss": 46.4946,
-      "step": 65390
-    },
-    {
-      "epoch": 0.26422427550430877,
-      "grad_norm": 709.1620483398438,
-      "learning_rate": 4.612688143930242e-05,
-      "loss": 97.9462,
-      "step": 65400
-    },
-    {
-      "epoch": 0.2642646767696764,
-      "grad_norm": 888.79248046875,
-      "learning_rate": 4.612501496198398e-05,
-      "loss": 94.5246,
-      "step": 65410
-    },
-    {
-      "epoch": 0.26430507803504405,
-      "grad_norm": 629.85791015625,
-      "learning_rate": 4.612314807282251e-05,
-      "loss": 47.7306,
-      "step": 65420
-    },
-    {
-      "epoch": 0.2643454793004117,
-      "grad_norm": 76.55754089355469,
-      "learning_rate": 4.612128077185439e-05,
-      "loss": 79.3505,
-      "step": 65430
-    },
-    {
-      "epoch": 0.26438588056577933,
-      "grad_norm": 1617.7518310546875,
-      "learning_rate": 4.611941305911602e-05,
-      "loss": 100.8537,
-      "step": 65440
-    },
-    {
-      "epoch": 0.26442628183114697,
-      "grad_norm": 653.6577758789062,
-      "learning_rate": 4.611754493464383e-05,
-      "loss": 61.9429,
-      "step": 65450
-    },
-    {
-      "epoch": 0.2644666830965146,
-      "grad_norm": 1721.381103515625,
-      "learning_rate": 4.611567639847422e-05,
-      "loss": 72.449,
-      "step": 65460
-    },
-    {
-      "epoch": 0.2645070843618822,
-      "grad_norm": 2042.4002685546875,
-      "learning_rate": 4.611380745064363e-05,
-      "loss": 66.475,
-      "step": 65470
-    },
-    {
-      "epoch": 0.26454748562724983,
-      "grad_norm": 603.5230712890625,
-      "learning_rate": 4.61119380911885e-05,
-      "loss": 81.908,
-      "step": 65480
-    },
-    {
-      "epoch": 0.2645878868926175,
-      "grad_norm": 507.7789001464844,
-      "learning_rate": 4.611006832014526e-05,
-      "loss": 57.1852,
-      "step": 65490
-    },
-    {
-      "epoch": 0.2646282881579851,
-      "grad_norm": 646.4906005859375,
-      "learning_rate": 4.610819813755038e-05,
-      "loss": 48.6094,
-      "step": 65500
-    },
-    {
-      "epoch": 0.26466868942335275,
-      "grad_norm": 783.2797241210938,
-      "learning_rate": 4.61063275434403e-05,
-      "loss": 67.8832,
-      "step": 65510
-    },
-    {
-      "epoch": 0.2647090906887204,
-      "grad_norm": 698.7806396484375,
-      "learning_rate": 4.610445653785151e-05,
-      "loss": 53.7803,
-      "step": 65520
-    },
-    {
-      "epoch": 0.264749491954088,
-      "grad_norm": 563.870361328125,
-      "learning_rate": 4.610258512082046e-05,
-      "loss": 54.008,
-      "step": 65530
-    },
-    {
-      "epoch": 0.2647898932194556,
-      "grad_norm": 734.525146484375,
-      "learning_rate": 4.610071329238366e-05,
-      "loss": 59.3586,
-      "step": 65540
-    },
-    {
-      "epoch": 0.26483029448482326,
-      "grad_norm": 568.80419921875,
-      "learning_rate": 4.6098841052577583e-05,
-      "loss": 52.5479,
-      "step": 65550
-    },
-    {
-      "epoch": 0.2648706957501909,
-      "grad_norm": 473.8629150390625,
-      "learning_rate": 4.6096968401438745e-05,
-      "loss": 88.4755,
-      "step": 65560
-    },
-    {
-      "epoch": 0.26491109701555854,
-      "grad_norm": 804.3765258789062,
-      "learning_rate": 4.609509533900364e-05,
-      "loss": 82.4179,
-      "step": 65570
-    },
-    {
-      "epoch": 0.2649514982809262,
-      "grad_norm": 556.0540161132812,
-      "learning_rate": 4.6093221865308786e-05,
-      "loss": 108.8011,
-      "step": 65580
-    },
-    {
-      "epoch": 0.26499189954629376,
-      "grad_norm": 761.5044555664062,
-      "learning_rate": 4.609134798039073e-05,
-      "loss": 61.9983,
-      "step": 65590
-    },
-    {
-      "epoch": 0.2650323008116614,
-      "grad_norm": 1213.9462890625,
-      "learning_rate": 4.6089473684285974e-05,
-      "loss": 93.0749,
-      "step": 65600
-    },
-    {
-      "epoch": 0.26507270207702904,
-      "grad_norm": 691.1778564453125,
-      "learning_rate": 4.608759897703107e-05,
-      "loss": 55.7611,
-      "step": 65610
-    },
-    {
-      "epoch": 0.2651131033423967,
-      "grad_norm": 1190.127685546875,
-      "learning_rate": 4.608572385866257e-05,
-      "loss": 56.0981,
-      "step": 65620
-    },
-    {
-      "epoch": 0.2651535046077643,
-      "grad_norm": 1123.38916015625,
-      "learning_rate": 4.6083848329217026e-05,
-      "loss": 88.4509,
-      "step": 65630
-    },
-    {
-      "epoch": 0.26519390587313196,
-      "grad_norm": 358.7076110839844,
-      "learning_rate": 4.608197238873101e-05,
-      "loss": 70.7137,
-      "step": 65640
-    },
-    {
-      "epoch": 0.2652343071384996,
-      "grad_norm": 1214.915283203125,
-      "learning_rate": 4.6080096037241086e-05,
-      "loss": 89.7655,
-      "step": 65650
-    },
-    {
-      "epoch": 0.2652747084038672,
-      "grad_norm": 890.3026123046875,
-      "learning_rate": 4.607821927478383e-05,
-      "loss": 86.5065,
-      "step": 65660
-    },
-    {
-      "epoch": 0.2653151096692348,
-      "grad_norm": 835.5069580078125,
-      "learning_rate": 4.607634210139584e-05,
-      "loss": 84.1165,
-      "step": 65670
-    },
-    {
-      "epoch": 0.26535551093460247,
-      "grad_norm": 350.78350830078125,
-      "learning_rate": 4.607446451711372e-05,
-      "loss": 67.6091,
-      "step": 65680
-    },
-    {
-      "epoch": 0.2653959121999701,
-      "grad_norm": 1582.5438232421875,
-      "learning_rate": 4.607258652197406e-05,
-      "loss": 86.531,
-      "step": 65690
-    },
-    {
-      "epoch": 0.26543631346533775,
-      "grad_norm": 1221.990234375,
-      "learning_rate": 4.6070708116013476e-05,
-      "loss": 70.8549,
-      "step": 65700
-    },
-    {
-      "epoch": 0.2654767147307054,
-      "grad_norm": 931.9036865234375,
-      "learning_rate": 4.606882929926858e-05,
-      "loss": 47.7265,
-      "step": 65710
-    },
-    {
-      "epoch": 0.26551711599607297,
-      "grad_norm": 787.3499755859375,
-      "learning_rate": 4.6066950071776015e-05,
-      "loss": 82.7044,
-      "step": 65720
-    },
-    {
-      "epoch": 0.2655575172614406,
-      "grad_norm": 3718.5693359375,
-      "learning_rate": 4.606507043357242e-05,
-      "loss": 120.9925,
-      "step": 65730
-    },
-    {
-      "epoch": 0.26559791852680825,
-      "grad_norm": 836.9452514648438,
-      "learning_rate": 4.606319038469443e-05,
-      "loss": 59.5213,
-      "step": 65740
-    },
-    {
-      "epoch": 0.2656383197921759,
-      "grad_norm": 694.1288452148438,
-      "learning_rate": 4.606130992517869e-05,
-      "loss": 70.1149,
-      "step": 65750
-    },
-    {
-      "epoch": 0.26567872105754353,
-      "grad_norm": 1005.6154174804688,
-      "learning_rate": 4.605942905506188e-05,
-      "loss": 49.102,
-      "step": 65760
-    },
-    {
-      "epoch": 0.26571912232291117,
-      "grad_norm": 684.3242797851562,
-      "learning_rate": 4.605754777438065e-05,
-      "loss": 79.9185,
-      "step": 65770
-    },
-    {
-      "epoch": 0.2657595235882788,
-      "grad_norm": 1068.396240234375,
-      "learning_rate": 4.605566608317169e-05,
-      "loss": 109.3487,
-      "step": 65780
-    },
-    {
-      "epoch": 0.2657999248536464,
-      "grad_norm": 2943.5751953125,
-      "learning_rate": 4.6053783981471675e-05,
-      "loss": 78.831,
-      "step": 65790
-    },
-    {
-      "epoch": 0.26584032611901404,
-      "grad_norm": 386.4954833984375,
-      "learning_rate": 4.605190146931731e-05,
-      "loss": 91.7346,
-      "step": 65800
-    },
-    {
-      "epoch": 0.2658807273843817,
-      "grad_norm": 1254.3067626953125,
-      "learning_rate": 4.605001854674529e-05,
-      "loss": 88.2726,
-      "step": 65810
-    },
-    {
-      "epoch": 0.2659211286497493,
-      "grad_norm": 553.072509765625,
-      "learning_rate": 4.604813521379231e-05,
-      "loss": 71.1216,
-      "step": 65820
-    },
-    {
-      "epoch": 0.26596152991511696,
-      "grad_norm": 780.3787841796875,
-      "learning_rate": 4.60462514704951e-05,
-      "loss": 71.7296,
-      "step": 65830
-    },
-    {
-      "epoch": 0.2660019311804846,
-      "grad_norm": 1093.623046875,
-      "learning_rate": 4.6044367316890386e-05,
-      "loss": 53.4292,
-      "step": 65840
-    },
-    {
-      "epoch": 0.2660423324458522,
-      "grad_norm": 986.8058471679688,
-      "learning_rate": 4.6042482753014895e-05,
-      "loss": 70.1599,
-      "step": 65850
-    },
-    {
-      "epoch": 0.2660827337112198,
-      "grad_norm": 614.8383178710938,
-      "learning_rate": 4.604059777890537e-05,
-      "loss": 68.2766,
-      "step": 65860
-    },
-    {
-      "epoch": 0.26612313497658746,
-      "grad_norm": 608.2391357421875,
-      "learning_rate": 4.603871239459856e-05,
-      "loss": 51.4506,
-      "step": 65870
-    },
-    {
-      "epoch": 0.2661635362419551,
-      "grad_norm": 1213.140869140625,
-      "learning_rate": 4.6036826600131216e-05,
-      "loss": 66.3176,
-      "step": 65880
-    },
-    {
-      "epoch": 0.26620393750732274,
-      "grad_norm": 1492.8607177734375,
-      "learning_rate": 4.603494039554011e-05,
-      "loss": 83.3855,
-      "step": 65890
-    },
-    {
-      "epoch": 0.2662443387726904,
-      "grad_norm": 2795.365478515625,
-      "learning_rate": 4.603305378086201e-05,
-      "loss": 86.1449,
-      "step": 65900
-    },
-    {
-      "epoch": 0.26628474003805797,
-      "grad_norm": 429.2622985839844,
-      "learning_rate": 4.60311667561337e-05,
-      "loss": 93.9114,
-      "step": 65910
-    },
-    {
-      "epoch": 0.2663251413034256,
-      "grad_norm": 1739.33544921875,
-      "learning_rate": 4.602927932139197e-05,
-      "loss": 72.7662,
-      "step": 65920
-    },
-    {
-      "epoch": 0.26636554256879325,
-      "grad_norm": 850.324951171875,
-      "learning_rate": 4.6027391476673606e-05,
-      "loss": 87.2089,
-      "step": 65930
-    },
-    {
-      "epoch": 0.2664059438341609,
-      "grad_norm": 248.5250244140625,
-      "learning_rate": 4.602550322201542e-05,
-      "loss": 73.9269,
-      "step": 65940
-    },
-    {
-      "epoch": 0.2664463450995285,
-      "grad_norm": 414.97100830078125,
-      "learning_rate": 4.602361455745423e-05,
-      "loss": 38.258,
-      "step": 65950
-    },
-    {
-      "epoch": 0.26648674636489617,
-      "grad_norm": 821.9193725585938,
-      "learning_rate": 4.602172548302684e-05,
-      "loss": 51.6796,
-      "step": 65960
-    },
-    {
-      "epoch": 0.2665271476302638,
-      "grad_norm": 972.9696044921875,
-      "learning_rate": 4.60198359987701e-05,
-      "loss": 75.2604,
-      "step": 65970
-    },
-    {
-      "epoch": 0.2665675488956314,
-      "grad_norm": 589.0814819335938,
-      "learning_rate": 4.6017946104720836e-05,
-      "loss": 86.6934,
-      "step": 65980
-    },
-    {
-      "epoch": 0.26660795016099903,
-      "grad_norm": 812.1279296875,
-      "learning_rate": 4.6016055800915884e-05,
-      "loss": 77.9313,
-      "step": 65990
-    },
-    {
-      "epoch": 0.26664835142636667,
-      "grad_norm": 456.3207702636719,
-      "learning_rate": 4.601416508739211e-05,
-      "loss": 49.5027,
-      "step": 66000
-    },
-    {
-      "epoch": 0.2666887526917343,
-      "grad_norm": 1623.6546630859375,
-      "learning_rate": 4.6012273964186365e-05,
-      "loss": 64.7836,
-      "step": 66010
-    },
-    {
-      "epoch": 0.26672915395710195,
-      "grad_norm": 779.8026733398438,
-      "learning_rate": 4.601038243133552e-05,
-      "loss": 108.0493,
-      "step": 66020
-    },
-    {
-      "epoch": 0.2667695552224696,
-      "grad_norm": 685.0618286132812,
-      "learning_rate": 4.600849048887646e-05,
-      "loss": 79.6456,
-      "step": 66030
-    },
-    {
-      "epoch": 0.2668099564878372,
-      "grad_norm": 1468.7335205078125,
-      "learning_rate": 4.6006598136846056e-05,
-      "loss": 63.8021,
-      "step": 66040
-    },
-    {
-      "epoch": 0.2668503577532048,
-      "grad_norm": 932.735107421875,
-      "learning_rate": 4.600470537528121e-05,
-      "loss": 51.6305,
-      "step": 66050
-    },
-    {
-      "epoch": 0.26689075901857245,
-      "grad_norm": 815.7344970703125,
-      "learning_rate": 4.6002812204218816e-05,
-      "loss": 66.5829,
-      "step": 66060
-    },
-    {
-      "epoch": 0.2669311602839401,
-      "grad_norm": 606.532958984375,
-      "learning_rate": 4.600091862369579e-05,
-      "loss": 77.9135,
-      "step": 66070
-    },
-    {
-      "epoch": 0.26697156154930773,
-      "grad_norm": 760.3806762695312,
-      "learning_rate": 4.599902463374903e-05,
-      "loss": 64.1426,
-      "step": 66080
-    },
-    {
-      "epoch": 0.2670119628146754,
-      "grad_norm": 1549.1610107421875,
-      "learning_rate": 4.599713023441549e-05,
-      "loss": 67.7165,
-      "step": 66090
-    },
-    {
-      "epoch": 0.267052364080043,
-      "grad_norm": 500.93182373046875,
-      "learning_rate": 4.599523542573207e-05,
-      "loss": 65.0257,
-      "step": 66100
-    },
-    {
-      "epoch": 0.2670927653454106,
-      "grad_norm": 301.22589111328125,
-      "learning_rate": 4.599334020773574e-05,
-      "loss": 69.3244,
-      "step": 66110
-    },
-    {
-      "epoch": 0.26713316661077824,
-      "grad_norm": 374.9650573730469,
-      "learning_rate": 4.599144458046343e-05,
-      "loss": 57.7532,
-      "step": 66120
-    },
-    {
-      "epoch": 0.2671735678761459,
-      "grad_norm": 521.96337890625,
-      "learning_rate": 4.59895485439521e-05,
-      "loss": 80.4386,
-      "step": 66130
-    },
-    {
-      "epoch": 0.2672139691415135,
-      "grad_norm": 1903.363525390625,
-      "learning_rate": 4.5987652098238714e-05,
-      "loss": 68.9257,
-      "step": 66140
-    },
-    {
-      "epoch": 0.26725437040688116,
-      "grad_norm": 561.215087890625,
-      "learning_rate": 4.598575524336025e-05,
-      "loss": 56.7605,
-      "step": 66150
-    },
-    {
-      "epoch": 0.2672947716722488,
-      "grad_norm": 502.40240478515625,
-      "learning_rate": 4.598385797935368e-05,
-      "loss": 61.1385,
-      "step": 66160
-    },
-    {
-      "epoch": 0.2673351729376164,
-      "grad_norm": 948.8161010742188,
-      "learning_rate": 4.5981960306255996e-05,
-      "loss": 71.2674,
-      "step": 66170
-    },
-    {
-      "epoch": 0.267375574202984,
-      "grad_norm": 683.00244140625,
-      "learning_rate": 4.598006222410419e-05,
-      "loss": 56.7152,
-      "step": 66180
-    },
-    {
-      "epoch": 0.26741597546835166,
-      "grad_norm": 916.79833984375,
-      "learning_rate": 4.597816373293528e-05,
-      "loss": 75.4204,
-      "step": 66190
-    },
-    {
-      "epoch": 0.2674563767337193,
-      "grad_norm": 406.5462341308594,
-      "learning_rate": 4.597626483278625e-05,
-      "loss": 63.3937,
-      "step": 66200
-    },
-    {
-      "epoch": 0.26749677799908694,
-      "grad_norm": 953.0973510742188,
-      "learning_rate": 4.5974365523694155e-05,
-      "loss": 55.9855,
-      "step": 66210
-    },
-    {
-      "epoch": 0.2675371792644546,
-      "grad_norm": 664.46923828125,
-      "learning_rate": 4.5972465805695996e-05,
-      "loss": 100.492,
-      "step": 66220
-    },
-    {
-      "epoch": 0.26757758052982217,
-      "grad_norm": 810.0478515625,
-      "learning_rate": 4.597056567882883e-05,
-      "loss": 68.5419,
-      "step": 66230
-    },
-    {
-      "epoch": 0.2676179817951898,
-      "grad_norm": 821.2228393554688,
-      "learning_rate": 4.596866514312967e-05,
-      "loss": 71.8012,
-      "step": 66240
-    },
-    {
-      "epoch": 0.26765838306055745,
-      "grad_norm": 1174.8955078125,
-      "learning_rate": 4.5966764198635606e-05,
-      "loss": 92.9574,
-      "step": 66250
-    },
-    {
-      "epoch": 0.2676987843259251,
-      "grad_norm": 1086.9854736328125,
-      "learning_rate": 4.596486284538367e-05,
-      "loss": 57.3918,
-      "step": 66260
-    },
-    {
-      "epoch": 0.2677391855912927,
-      "grad_norm": 690.481689453125,
-      "learning_rate": 4.5962961083410946e-05,
-      "loss": 71.6719,
-      "step": 66270
-    },
-    {
-      "epoch": 0.26777958685666037,
-      "grad_norm": 1423.809814453125,
-      "learning_rate": 4.596105891275449e-05,
-      "loss": 66.3168,
-      "step": 66280
-    },
-    {
-      "epoch": 0.267819988122028,
-      "grad_norm": 1662.720947265625,
-      "learning_rate": 4.595915633345141e-05,
-      "loss": 94.1194,
-      "step": 66290
-    },
-    {
-      "epoch": 0.2678603893873956,
-      "grad_norm": 687.7420654296875,
-      "learning_rate": 4.595725334553879e-05,
-      "loss": 125.3564,
-      "step": 66300
-    },
-    {
-      "epoch": 0.26790079065276323,
-      "grad_norm": 580.8500366210938,
-      "learning_rate": 4.595534994905372e-05,
-      "loss": 45.3163,
-      "step": 66310
-    },
-    {
-      "epoch": 0.26794119191813087,
-      "grad_norm": 0.0,
-      "learning_rate": 4.5953446144033316e-05,
-      "loss": 59.1748,
-      "step": 66320
-    },
-    {
-      "epoch": 0.2679815931834985,
-      "grad_norm": 292.1687927246094,
-      "learning_rate": 4.595154193051469e-05,
-      "loss": 62.6477,
-      "step": 66330
-    },
-    {
-      "epoch": 0.26802199444886615,
-      "grad_norm": 379.4153747558594,
-      "learning_rate": 4.594963730853497e-05,
-      "loss": 90.7049,
-      "step": 66340
-    },
-    {
-      "epoch": 0.2680623957142338,
-      "grad_norm": 2087.564208984375,
-      "learning_rate": 4.5947732278131286e-05,
-      "loss": 75.824,
-      "step": 66350
-    },
-    {
-      "epoch": 0.2681027969796014,
-      "grad_norm": 1189.827880859375,
-      "learning_rate": 4.594582683934078e-05,
-      "loss": 92.8753,
-      "step": 66360
-    },
-    {
-      "epoch": 0.268143198244969,
-      "grad_norm": 524.31005859375,
-      "learning_rate": 4.5943920992200585e-05,
-      "loss": 57.8241,
-      "step": 66370
-    },
-    {
-      "epoch": 0.26818359951033666,
-      "grad_norm": 1099.4833984375,
-      "learning_rate": 4.5942014736747875e-05,
-      "loss": 65.1616,
-      "step": 66380
-    },
-    {
-      "epoch": 0.2682240007757043,
-      "grad_norm": 1009.6073608398438,
-      "learning_rate": 4.59401080730198e-05,
-      "loss": 68.1652,
-      "step": 66390
-    },
-    {
-      "epoch": 0.26826440204107194,
-      "grad_norm": 4793.3828125,
-      "learning_rate": 4.593820100105355e-05,
-      "loss": 111.5321,
-      "step": 66400
-    },
-    {
-      "epoch": 0.2683048033064396,
-      "grad_norm": 940.0222778320312,
-      "learning_rate": 4.5936293520886275e-05,
-      "loss": 83.8232,
-      "step": 66410
-    },
-    {
-      "epoch": 0.2683452045718072,
-      "grad_norm": 1047.490478515625,
-      "learning_rate": 4.59343856325552e-05,
-      "loss": 58.4591,
-      "step": 66420
-    },
-    {
-      "epoch": 0.2683856058371748,
-      "grad_norm": 874.4302978515625,
-      "learning_rate": 4.593247733609748e-05,
-      "loss": 104.9535,
-      "step": 66430
-    },
-    {
-      "epoch": 0.26842600710254244,
-      "grad_norm": 742.1166381835938,
-      "learning_rate": 4.593056863155034e-05,
-      "loss": 84.062,
-      "step": 66440
-    },
-    {
-      "epoch": 0.2684664083679101,
-      "grad_norm": 313.023681640625,
-      "learning_rate": 4.5928659518951e-05,
-      "loss": 44.992,
-      "step": 66450
-    },
-    {
-      "epoch": 0.2685068096332777,
-      "grad_norm": 1163.264892578125,
-      "learning_rate": 4.592674999833666e-05,
-      "loss": 71.2603,
-      "step": 66460
-    },
-    {
-      "epoch": 0.26854721089864536,
-      "grad_norm": 1168.4986572265625,
-      "learning_rate": 4.592484006974456e-05,
-      "loss": 46.842,
-      "step": 66470
-    },
-    {
-      "epoch": 0.268587612164013,
-      "grad_norm": 2281.073974609375,
-      "learning_rate": 4.5922929733211926e-05,
-      "loss": 54.4491,
-      "step": 66480
-    },
-    {
-      "epoch": 0.2686280134293806,
-      "grad_norm": 656.7672729492188,
-      "learning_rate": 4.5921018988776e-05,
-      "loss": 88.2362,
-      "step": 66490
-    },
-    {
-      "epoch": 0.2686684146947482,
-      "grad_norm": 787.104736328125,
-      "learning_rate": 4.591910783647404e-05,
-      "loss": 102.6616,
-      "step": 66500
-    },
-    {
-      "epoch": 0.26870881596011587,
-      "grad_norm": 602.447265625,
-      "learning_rate": 4.591719627634331e-05,
-      "loss": 70.4162,
-      "step": 66510
-    },
-    {
-      "epoch": 0.2687492172254835,
-      "grad_norm": 607.5802612304688,
-      "learning_rate": 4.591528430842107e-05,
-      "loss": 67.1125,
-      "step": 66520
-    },
-    {
-      "epoch": 0.26878961849085115,
-      "grad_norm": 1654.757080078125,
-      "learning_rate": 4.5913371932744584e-05,
-      "loss": 82.985,
-      "step": 66530
-    },
-    {
-      "epoch": 0.2688300197562188,
-      "grad_norm": 711.9852905273438,
-      "learning_rate": 4.591145914935116e-05,
-      "loss": 66.6757,
-      "step": 66540
-    },
-    {
-      "epoch": 0.26887042102158637,
-      "grad_norm": 956.3167114257812,
-      "learning_rate": 4.590954595827806e-05,
-      "loss": 78.8319,
-      "step": 66550
-    },
-    {
-      "epoch": 0.268910822286954,
-      "grad_norm": 1775.7945556640625,
-      "learning_rate": 4.59076323595626e-05,
-      "loss": 88.6002,
-      "step": 66560
-    },
-    {
-      "epoch": 0.26895122355232165,
-      "grad_norm": 776.41015625,
-      "learning_rate": 4.5905718353242086e-05,
-      "loss": 70.2173,
-      "step": 66570
-    },
-    {
-      "epoch": 0.2689916248176893,
-      "grad_norm": 934.8427124023438,
-      "learning_rate": 4.590380393935383e-05,
-      "loss": 71.2888,
-      "step": 66580
-    },
-    {
-      "epoch": 0.26903202608305693,
-      "grad_norm": 1015.8541259765625,
-      "learning_rate": 4.5901889117935153e-05,
-      "loss": 94.4522,
-      "step": 66590
-    },
-    {
-      "epoch": 0.26907242734842457,
-      "grad_norm": 0.0,
-      "learning_rate": 4.589997388902338e-05,
-      "loss": 66.415,
-      "step": 66600
-    },
-    {
-      "epoch": 0.2691128286137922,
-      "grad_norm": 1408.8917236328125,
-      "learning_rate": 4.589805825265587e-05,
-      "loss": 83.8427,
-      "step": 66610
-    },
-    {
-      "epoch": 0.2691532298791598,
-      "grad_norm": 654.0909423828125,
-      "learning_rate": 4.5896142208869954e-05,
-      "loss": 59.3255,
-      "step": 66620
-    },
-    {
-      "epoch": 0.26919363114452743,
-      "grad_norm": 644.368896484375,
-      "learning_rate": 4.589422575770298e-05,
-      "loss": 51.3406,
-      "step": 66630
-    },
-    {
-      "epoch": 0.2692340324098951,
-      "grad_norm": 2795.22998046875,
-      "learning_rate": 4.589230889919232e-05,
-      "loss": 64.9199,
-      "step": 66640
-    },
-    {
-      "epoch": 0.2692744336752627,
-      "grad_norm": 940.733642578125,
-      "learning_rate": 4.589039163337534e-05,
-      "loss": 102.8163,
-      "step": 66650
-    },
-    {
-      "epoch": 0.26931483494063035,
-      "grad_norm": 847.7637939453125,
-      "learning_rate": 4.588847396028942e-05,
-      "loss": 77.151,
-      "step": 66660
-    },
-    {
-      "epoch": 0.269355236205998,
-      "grad_norm": 710.346435546875,
-      "learning_rate": 4.588655587997195e-05,
-      "loss": 57.7109,
-      "step": 66670
-    },
-    {
-      "epoch": 0.2693956374713656,
-      "grad_norm": 607.9664306640625,
-      "learning_rate": 4.5884637392460314e-05,
-      "loss": 71.083,
-      "step": 66680
-    },
-    {
-      "epoch": 0.2694360387367332,
-      "grad_norm": 615.1112060546875,
-      "learning_rate": 4.588271849779192e-05,
-      "loss": 63.7689,
-      "step": 66690
-    },
-    {
-      "epoch": 0.26947644000210086,
-      "grad_norm": 664.7433471679688,
-      "learning_rate": 4.588079919600419e-05,
-      "loss": 80.6283,
-      "step": 66700
-    },
-    {
-      "epoch": 0.2695168412674685,
-      "grad_norm": 407.6193542480469,
-      "learning_rate": 4.587887948713452e-05,
-      "loss": 49.1637,
-      "step": 66710
-    },
-    {
-      "epoch": 0.26955724253283614,
-      "grad_norm": 785.1792602539062,
-      "learning_rate": 4.5876959371220344e-05,
-      "loss": 64.5263,
-      "step": 66720
-    },
-    {
-      "epoch": 0.2695976437982038,
-      "grad_norm": 565.7954711914062,
-      "learning_rate": 4.587503884829909e-05,
-      "loss": 86.9049,
-      "step": 66730
-    },
-    {
-      "epoch": 0.2696380450635714,
-      "grad_norm": 1209.4056396484375,
-      "learning_rate": 4.587311791840822e-05,
-      "loss": 103.0452,
-      "step": 66740
-    },
-    {
-      "epoch": 0.269678446328939,
-      "grad_norm": 384.4944152832031,
-      "learning_rate": 4.5871196581585166e-05,
-      "loss": 46.5085,
-      "step": 66750
-    },
-    {
-      "epoch": 0.26971884759430664,
-      "grad_norm": 818.3886108398438,
-      "learning_rate": 4.5869274837867394e-05,
-      "loss": 100.6729,
-      "step": 66760
-    },
-    {
-      "epoch": 0.2697592488596743,
-      "grad_norm": 964.607421875,
-      "learning_rate": 4.5867352687292355e-05,
-      "loss": 64.0361,
-      "step": 66770
-    },
-    {
-      "epoch": 0.2697996501250419,
-      "grad_norm": 641.611328125,
-      "learning_rate": 4.5865430129897536e-05,
-      "loss": 44.5035,
-      "step": 66780
-    },
-    {
-      "epoch": 0.26984005139040956,
-      "grad_norm": 822.2906494140625,
-      "learning_rate": 4.5863507165720415e-05,
-      "loss": 63.5622,
-      "step": 66790
-    },
-    {
-      "epoch": 0.2698804526557772,
-      "grad_norm": 370.7065124511719,
-      "learning_rate": 4.586158379479848e-05,
-      "loss": 57.749,
-      "step": 66800
-    },
-    {
-      "epoch": 0.2699208539211448,
-      "grad_norm": 565.5477905273438,
-      "learning_rate": 4.5859660017169224e-05,
-      "loss": 80.5675,
-      "step": 66810
-    },
-    {
-      "epoch": 0.2699612551865124,
-      "grad_norm": 869.1512451171875,
-      "learning_rate": 4.5857735832870166e-05,
-      "loss": 76.6965,
-      "step": 66820
-    },
-    {
-      "epoch": 0.27000165645188007,
-      "grad_norm": 390.1742858886719,
-      "learning_rate": 4.5855811241938806e-05,
-      "loss": 87.9975,
-      "step": 66830
-    },
-    {
-      "epoch": 0.2700420577172477,
-      "grad_norm": 815.9105834960938,
-      "learning_rate": 4.585388624441267e-05,
-      "loss": 60.8064,
-      "step": 66840
-    },
-    {
-      "epoch": 0.27008245898261535,
-      "grad_norm": 777.6511840820312,
-      "learning_rate": 4.585196084032928e-05,
-      "loss": 50.49,
-      "step": 66850
-    },
-    {
-      "epoch": 0.270122860247983,
-      "grad_norm": 975.7098999023438,
-      "learning_rate": 4.585003502972618e-05,
-      "loss": 82.6168,
-      "step": 66860
-    },
-    {
-      "epoch": 0.27016326151335057,
-      "grad_norm": 773.058349609375,
-      "learning_rate": 4.584810881264092e-05,
-      "loss": 58.9796,
-      "step": 66870
-    },
-    {
-      "epoch": 0.2702036627787182,
-      "grad_norm": 342.78265380859375,
-      "learning_rate": 4.5846182189111035e-05,
-      "loss": 77.3388,
-      "step": 66880
-    },
-    {
-      "epoch": 0.27024406404408585,
-      "grad_norm": 759.47314453125,
-      "learning_rate": 4.584425515917411e-05,
-      "loss": 53.6399,
-      "step": 66890
-    },
-    {
-      "epoch": 0.2702844653094535,
-      "grad_norm": 392.1265869140625,
-      "learning_rate": 4.584232772286768e-05,
-      "loss": 48.1632,
-      "step": 66900
-    },
-    {
-      "epoch": 0.27032486657482113,
-      "grad_norm": 704.0643920898438,
-      "learning_rate": 4.5840399880229354e-05,
-      "loss": 73.1397,
-      "step": 66910
-    },
-    {
-      "epoch": 0.27036526784018877,
-      "grad_norm": 828.9876098632812,
-      "learning_rate": 4.58384716312967e-05,
-      "loss": 60.1899,
-      "step": 66920
-    },
-    {
-      "epoch": 0.2704056691055564,
-      "grad_norm": 529.6195678710938,
-      "learning_rate": 4.583654297610731e-05,
-      "loss": 59.9641,
-      "step": 66930
-    },
-    {
-      "epoch": 0.270446070370924,
-      "grad_norm": 733.8816528320312,
-      "learning_rate": 4.583461391469879e-05,
-      "loss": 90.0057,
-      "step": 66940
-    },
-    {
-      "epoch": 0.27048647163629164,
-      "grad_norm": 1088.7060546875,
-      "learning_rate": 4.583268444710875e-05,
-      "loss": 103.6154,
-      "step": 66950
-    },
-    {
-      "epoch": 0.2705268729016593,
-      "grad_norm": 952.4298095703125,
-      "learning_rate": 4.583075457337479e-05,
-      "loss": 78.6486,
-      "step": 66960
-    },
-    {
-      "epoch": 0.2705672741670269,
-      "grad_norm": 490.1788024902344,
-      "learning_rate": 4.5828824293534555e-05,
-      "loss": 64.2271,
-      "step": 66970
-    },
-    {
-      "epoch": 0.27060767543239456,
-      "grad_norm": 1667.7532958984375,
-      "learning_rate": 4.5826893607625665e-05,
-      "loss": 64.5124,
-      "step": 66980
-    },
-    {
-      "epoch": 0.2706480766977622,
-      "grad_norm": 823.8807373046875,
-      "learning_rate": 4.582496251568576e-05,
-      "loss": 71.0763,
-      "step": 66990
-    },
-    {
-      "epoch": 0.2706884779631298,
-      "grad_norm": 0.0,
-      "learning_rate": 4.5823031017752485e-05,
-      "loss": 65.76,
-      "step": 67000
-    },
-    {
-      "epoch": 0.2707288792284974,
-      "grad_norm": 962.5149536132812,
-      "learning_rate": 4.5821099113863506e-05,
-      "loss": 87.7109,
-      "step": 67010
-    },
-    {
-      "epoch": 0.27076928049386506,
-      "grad_norm": 1748.1005859375,
-      "learning_rate": 4.581916680405648e-05,
-      "loss": 62.5524,
-      "step": 67020
-    },
-    {
-      "epoch": 0.2708096817592327,
-      "grad_norm": 1395.0408935546875,
-      "learning_rate": 4.581723408836908e-05,
-      "loss": 70.1472,
-      "step": 67030
-    },
-    {
-      "epoch": 0.27085008302460034,
-      "grad_norm": 1065.451904296875,
-      "learning_rate": 4.581530096683898e-05,
-      "loss": 65.4421,
-      "step": 67040
-    },
-    {
-      "epoch": 0.270890484289968,
-      "grad_norm": 1557.5567626953125,
-      "learning_rate": 4.5813367439503875e-05,
-      "loss": 50.4025,
-      "step": 67050
-    },
-    {
-      "epoch": 0.27093088555533557,
-      "grad_norm": 990.9954223632812,
-      "learning_rate": 4.5811433506401456e-05,
-      "loss": 60.8415,
-      "step": 67060
-    },
-    {
-      "epoch": 0.2709712868207032,
-      "grad_norm": 535.752685546875,
-      "learning_rate": 4.580949916756942e-05,
-      "loss": 56.8241,
-      "step": 67070
-    },
-    {
-      "epoch": 0.27101168808607085,
-      "grad_norm": 589.4478149414062,
-      "learning_rate": 4.580756442304549e-05,
-      "loss": 43.5019,
-      "step": 67080
-    },
-    {
-      "epoch": 0.2710520893514385,
-      "grad_norm": 471.8979187011719,
-      "learning_rate": 4.580562927286738e-05,
-      "loss": 83.2732,
-      "step": 67090
-    },
-    {
-      "epoch": 0.2710924906168061,
-      "grad_norm": 1266.64794921875,
-      "learning_rate": 4.5803693717072815e-05,
-      "loss": 97.8621,
-      "step": 67100
-    },
-    {
-      "epoch": 0.27113289188217377,
-      "grad_norm": 939.3637084960938,
-      "learning_rate": 4.5801757755699534e-05,
-      "loss": 73.3667,
-      "step": 67110
-    },
-    {
-      "epoch": 0.2711732931475414,
-      "grad_norm": 516.1859741210938,
-      "learning_rate": 4.579982138878527e-05,
-      "loss": 64.6031,
-      "step": 67120
-    },
-    {
-      "epoch": 0.271213694412909,
-      "grad_norm": 688.78125,
-      "learning_rate": 4.579788461636778e-05,
-      "loss": 86.9311,
-      "step": 67130
-    },
-    {
-      "epoch": 0.27125409567827663,
-      "grad_norm": 726.7952270507812,
-      "learning_rate": 4.579594743848482e-05,
-      "loss": 97.9171,
-      "step": 67140
-    },
-    {
-      "epoch": 0.27129449694364427,
-      "grad_norm": 1185.85498046875,
-      "learning_rate": 4.579400985517416e-05,
-      "loss": 92.5374,
-      "step": 67150
-    },
-    {
-      "epoch": 0.2713348982090119,
-      "grad_norm": 928.0086059570312,
-      "learning_rate": 4.579207186647357e-05,
-      "loss": 46.1306,
-      "step": 67160
-    },
-    {
-      "epoch": 0.27137529947437955,
-      "grad_norm": 5591.84228515625,
-      "learning_rate": 4.579013347242085e-05,
-      "loss": 74.2452,
-      "step": 67170
-    },
-    {
-      "epoch": 0.2714157007397472,
-      "grad_norm": 886.691650390625,
-      "learning_rate": 4.5788194673053756e-05,
-      "loss": 55.0841,
-      "step": 67180
-    },
-    {
-      "epoch": 0.2714561020051148,
-      "grad_norm": 1341.62109375,
-      "learning_rate": 4.578625546841011e-05,
-      "loss": 62.2266,
-      "step": 67190
-    },
-    {
-      "epoch": 0.2714965032704824,
-      "grad_norm": 839.7449951171875,
-      "learning_rate": 4.5784315858527715e-05,
-      "loss": 73.6595,
-      "step": 67200
-    },
-    {
-      "epoch": 0.27153690453585005,
-      "grad_norm": 570.7716674804688,
-      "learning_rate": 4.578237584344438e-05,
-      "loss": 68.3076,
-      "step": 67210
-    },
-    {
-      "epoch": 0.2715773058012177,
-      "grad_norm": 673.6726684570312,
-      "learning_rate": 4.578043542319793e-05,
-      "loss": 59.9907,
-      "step": 67220
-    },
-    {
-      "epoch": 0.27161770706658533,
-      "grad_norm": 966.4290161132812,
-      "learning_rate": 4.577849459782619e-05,
-      "loss": 56.2375,
-      "step": 67230
-    },
-    {
-      "epoch": 0.271658108331953,
-      "grad_norm": 516.3497924804688,
-      "learning_rate": 4.5776553367367e-05,
-      "loss": 86.4351,
-      "step": 67240
-    },
-    {
-      "epoch": 0.2716985095973206,
-      "grad_norm": 2003.4329833984375,
-      "learning_rate": 4.577461173185821e-05,
-      "loss": 76.1077,
-      "step": 67250
-    },
-    {
-      "epoch": 0.2717389108626882,
-      "grad_norm": 829.5421752929688,
-      "learning_rate": 4.5772669691337665e-05,
-      "loss": 67.1411,
-      "step": 67260
-    },
-    {
-      "epoch": 0.27177931212805584,
-      "grad_norm": 783.5784912109375,
-      "learning_rate": 4.577072724584323e-05,
-      "loss": 72.7227,
-      "step": 67270
-    },
-    {
-      "epoch": 0.2718197133934235,
-      "grad_norm": 875.1643676757812,
-      "learning_rate": 4.576878439541278e-05,
-      "loss": 80.1554,
-      "step": 67280
-    },
-    {
-      "epoch": 0.2718601146587911,
-      "grad_norm": 432.68487548828125,
-      "learning_rate": 4.576684114008418e-05,
-      "loss": 44.3341,
-      "step": 67290
-    },
-    {
-      "epoch": 0.27190051592415876,
-      "grad_norm": 427.4791259765625,
-      "learning_rate": 4.5764897479895317e-05,
-      "loss": 74.892,
-      "step": 67300
-    },
-    {
-      "epoch": 0.2719409171895264,
-      "grad_norm": 1177.6551513671875,
-      "learning_rate": 4.576295341488409e-05,
-      "loss": 68.8277,
-      "step": 67310
-    },
-    {
-      "epoch": 0.271981318454894,
-      "grad_norm": 739.357421875,
-      "learning_rate": 4.57610089450884e-05,
-      "loss": 83.5637,
-      "step": 67320
-    },
-    {
-      "epoch": 0.2720217197202616,
-      "grad_norm": 464.8116760253906,
-      "learning_rate": 4.575906407054615e-05,
-      "loss": 69.9499,
-      "step": 67330
-    },
-    {
-      "epoch": 0.27206212098562926,
-      "grad_norm": 439.9012145996094,
-      "learning_rate": 4.5757118791295264e-05,
-      "loss": 54.5363,
-      "step": 67340
-    },
-    {
-      "epoch": 0.2721025222509969,
-      "grad_norm": 415.9515380859375,
-      "learning_rate": 4.575517310737365e-05,
-      "loss": 73.2503,
-      "step": 67350
-    },
-    {
-      "epoch": 0.27214292351636454,
-      "grad_norm": 784.4149780273438,
-      "learning_rate": 4.575322701881926e-05,
-      "loss": 64.8702,
-      "step": 67360
-    },
-    {
-      "epoch": 0.2721833247817322,
-      "grad_norm": 389.4613342285156,
-      "learning_rate": 4.575128052567002e-05,
-      "loss": 76.2548,
-      "step": 67370
-    },
-    {
-      "epoch": 0.27222372604709977,
-      "grad_norm": 942.818603515625,
-      "learning_rate": 4.5749333627963884e-05,
-      "loss": 76.8624,
-      "step": 67380
-    },
-    {
-      "epoch": 0.2722641273124674,
-      "grad_norm": 895.9558715820312,
-      "learning_rate": 4.574738632573881e-05,
-      "loss": 69.907,
-      "step": 67390
-    },
-    {
-      "epoch": 0.27230452857783505,
-      "grad_norm": 585.9174194335938,
-      "learning_rate": 4.574543861903274e-05,
-      "loss": 88.2039,
-      "step": 67400
-    },
-    {
-      "epoch": 0.2723449298432027,
-      "grad_norm": 556.5941162109375,
-      "learning_rate": 4.5743490507883685e-05,
-      "loss": 128.176,
-      "step": 67410
-    },
-    {
-      "epoch": 0.2723853311085703,
-      "grad_norm": 730.61865234375,
-      "learning_rate": 4.574154199232959e-05,
-      "loss": 69.3492,
-      "step": 67420
-    },
-    {
-      "epoch": 0.27242573237393797,
-      "grad_norm": 887.3568725585938,
-      "learning_rate": 4.573959307240847e-05,
-      "loss": 69.9397,
-      "step": 67430
-    },
-    {
-      "epoch": 0.2724661336393056,
-      "grad_norm": 549.9888916015625,
-      "learning_rate": 4.5737643748158295e-05,
-      "loss": 76.4168,
-      "step": 67440
-    },
-    {
-      "epoch": 0.2725065349046732,
-      "grad_norm": 1629.8140869140625,
-      "learning_rate": 4.573569401961708e-05,
-      "loss": 59.1505,
-      "step": 67450
-    },
-    {
-      "epoch": 0.27254693617004083,
-      "grad_norm": 982.314453125,
-      "learning_rate": 4.573374388682283e-05,
-      "loss": 70.0269,
-      "step": 67460
-    },
-    {
-      "epoch": 0.27258733743540847,
-      "grad_norm": 499.3196716308594,
-      "learning_rate": 4.573179334981358e-05,
-      "loss": 41.5002,
-      "step": 67470
-    },
-    {
-      "epoch": 0.2726277387007761,
-      "grad_norm": 764.8738403320312,
-      "learning_rate": 4.5729842408627334e-05,
-      "loss": 65.5099,
-      "step": 67480
-    },
-    {
-      "epoch": 0.27266813996614375,
-      "grad_norm": 813.9560546875,
-      "learning_rate": 4.572789106330214e-05,
-      "loss": 67.2205,
-      "step": 67490
-    },
-    {
-      "epoch": 0.2727085412315114,
-      "grad_norm": 813.4544677734375,
-      "learning_rate": 4.572593931387604e-05,
-      "loss": 58.4749,
-      "step": 67500
-    },
-    {
-      "epoch": 0.272748942496879,
-      "grad_norm": 1321.0914306640625,
-      "learning_rate": 4.572398716038709e-05,
-      "loss": 76.9033,
-      "step": 67510
-    },
-    {
-      "epoch": 0.2727893437622466,
-      "grad_norm": 809.7509155273438,
-      "learning_rate": 4.572203460287333e-05,
-      "loss": 44.7534,
-      "step": 67520
-    },
-    {
-      "epoch": 0.27282974502761426,
-      "grad_norm": 502.35369873046875,
-      "learning_rate": 4.5720081641372844e-05,
-      "loss": 75.6196,
-      "step": 67530
-    },
-    {
-      "epoch": 0.2728701462929819,
-      "grad_norm": 285.2467956542969,
-      "learning_rate": 4.57181282759237e-05,
-      "loss": 49.662,
-      "step": 67540
-    },
-    {
-      "epoch": 0.27291054755834954,
-      "grad_norm": 1159.55419921875,
-      "learning_rate": 4.571617450656397e-05,
-      "loss": 68.6516,
-      "step": 67550
-    },
-    {
-      "epoch": 0.2729509488237172,
-      "grad_norm": 883.5384521484375,
-      "learning_rate": 4.5714220333331756e-05,
-      "loss": 48.4618,
-      "step": 67560
-    },
-    {
-      "epoch": 0.2729913500890848,
-      "grad_norm": 714.6936645507812,
-      "learning_rate": 4.571226575626516e-05,
-      "loss": 70.2103,
-      "step": 67570
-    },
-    {
-      "epoch": 0.2730317513544524,
-      "grad_norm": 803.3701782226562,
-      "learning_rate": 4.5710310775402274e-05,
-      "loss": 65.889,
-      "step": 67580
-    },
-    {
-      "epoch": 0.27307215261982004,
-      "grad_norm": 463.9601745605469,
-      "learning_rate": 4.570835539078121e-05,
-      "loss": 65.3356,
-      "step": 67590
-    },
-    {
-      "epoch": 0.2731125538851877,
-      "grad_norm": 5190.55419921875,
-      "learning_rate": 4.5706399602440106e-05,
-      "loss": 79.4295,
-      "step": 67600
-    },
-    {
-      "epoch": 0.2731529551505553,
-      "grad_norm": 1560.6190185546875,
-      "learning_rate": 4.5704443410417075e-05,
-      "loss": 60.054,
-      "step": 67610
-    },
-    {
-      "epoch": 0.27319335641592296,
-      "grad_norm": 566.1893310546875,
-      "learning_rate": 4.5702486814750265e-05,
-      "loss": 65.9659,
-      "step": 67620
-    },
-    {
-      "epoch": 0.2732337576812906,
-      "grad_norm": 497.7099914550781,
-      "learning_rate": 4.570052981547782e-05,
-      "loss": 60.542,
-      "step": 67630
-    },
-    {
-      "epoch": 0.2732741589466582,
-      "grad_norm": 595.028076171875,
-      "learning_rate": 4.569857241263788e-05,
-      "loss": 113.0741,
-      "step": 67640
-    },
-    {
-      "epoch": 0.2733145602120258,
-      "grad_norm": 562.2115478515625,
-      "learning_rate": 4.569661460626862e-05,
-      "loss": 73.3425,
-      "step": 67650
-    },
-    {
-      "epoch": 0.27335496147739347,
-      "grad_norm": 3808.25244140625,
-      "learning_rate": 4.5694656396408195e-05,
-      "loss": 101.9904,
-      "step": 67660
-    },
-    {
-      "epoch": 0.2733953627427611,
-      "grad_norm": 1519.3927001953125,
-      "learning_rate": 4.56926977830948e-05,
-      "loss": 89.2418,
-      "step": 67670
-    },
-    {
-      "epoch": 0.27343576400812875,
-      "grad_norm": 1025.804931640625,
-      "learning_rate": 4.56907387663666e-05,
-      "loss": 69.6141,
-      "step": 67680
-    },
-    {
-      "epoch": 0.2734761652734964,
-      "grad_norm": 852.5979614257812,
-      "learning_rate": 4.56887793462618e-05,
-      "loss": 88.1962,
-      "step": 67690
-    },
-    {
-      "epoch": 0.27351656653886397,
-      "grad_norm": 589.6854858398438,
-      "learning_rate": 4.5686819522818594e-05,
-      "loss": 73.4965,
-      "step": 67700
-    },
-    {
-      "epoch": 0.2735569678042316,
-      "grad_norm": 1321.8143310546875,
-      "learning_rate": 4.56848592960752e-05,
-      "loss": 79.9866,
-      "step": 67710
-    },
-    {
-      "epoch": 0.27359736906959925,
-      "grad_norm": 1166.6270751953125,
-      "learning_rate": 4.568289866606981e-05,
-      "loss": 55.3361,
-      "step": 67720
-    },
-    {
-      "epoch": 0.2736377703349669,
-      "grad_norm": 440.9592590332031,
-      "learning_rate": 4.568093763284067e-05,
-      "loss": 41.0626,
-      "step": 67730
-    },
-    {
-      "epoch": 0.27367817160033453,
-      "grad_norm": 598.6520385742188,
-      "learning_rate": 4.567897619642601e-05,
-      "loss": 80.7803,
-      "step": 67740
-    },
-    {
-      "epoch": 0.27371857286570217,
-      "grad_norm": 897.6016235351562,
-      "learning_rate": 4.567701435686404e-05,
-      "loss": 94.469,
-      "step": 67750
-    },
-    {
-      "epoch": 0.2737589741310698,
-      "grad_norm": 926.674560546875,
-      "learning_rate": 4.567505211419305e-05,
-      "loss": 68.1472,
-      "step": 67760
-    },
-    {
-      "epoch": 0.2737993753964374,
-      "grad_norm": 497.29339599609375,
-      "learning_rate": 4.567308946845127e-05,
-      "loss": 47.2474,
-      "step": 67770
-    },
-    {
-      "epoch": 0.27383977666180503,
-      "grad_norm": 891.9127807617188,
-      "learning_rate": 4.567112641967697e-05,
-      "loss": 88.2951,
-      "step": 67780
-    },
-    {
-      "epoch": 0.2738801779271727,
-      "grad_norm": 655.0907592773438,
-      "learning_rate": 4.566916296790842e-05,
-      "loss": 61.0948,
-      "step": 67790
-    },
-    {
-      "epoch": 0.2739205791925403,
-      "grad_norm": 577.0123901367188,
-      "learning_rate": 4.566719911318389e-05,
-      "loss": 89.3447,
-      "step": 67800
-    },
-    {
-      "epoch": 0.27396098045790795,
-      "grad_norm": 598.5209350585938,
-      "learning_rate": 4.5665234855541675e-05,
-      "loss": 42.0639,
-      "step": 67810
-    },
-    {
-      "epoch": 0.2740013817232756,
-      "grad_norm": 578.243408203125,
-      "learning_rate": 4.566327019502007e-05,
-      "loss": 50.7425,
-      "step": 67820
-    },
-    {
-      "epoch": 0.2740417829886432,
-      "grad_norm": 849.9849853515625,
-      "learning_rate": 4.566130513165737e-05,
-      "loss": 58.8878,
-      "step": 67830
-    },
-    {
-      "epoch": 0.2740821842540108,
-      "grad_norm": 585.877685546875,
-      "learning_rate": 4.565933966549189e-05,
-      "loss": 64.4063,
-      "step": 67840
-    },
-    {
-      "epoch": 0.27412258551937846,
-      "grad_norm": 600.2329711914062,
-      "learning_rate": 4.565737379656195e-05,
-      "loss": 62.7049,
-      "step": 67850
-    },
-    {
-      "epoch": 0.2741629867847461,
-      "grad_norm": 1287.9954833984375,
-      "learning_rate": 4.5655407524905866e-05,
-      "loss": 55.8217,
-      "step": 67860
-    },
-    {
-      "epoch": 0.27420338805011374,
-      "grad_norm": 640.1063842773438,
-      "learning_rate": 4.5653440850561986e-05,
-      "loss": 107.2745,
-      "step": 67870
-    },
-    {
-      "epoch": 0.2742437893154814,
-      "grad_norm": 1059.8924560546875,
-      "learning_rate": 4.565147377356864e-05,
-      "loss": 64.9931,
-      "step": 67880
-    },
-    {
-      "epoch": 0.274284190580849,
-      "grad_norm": 1559.3463134765625,
-      "learning_rate": 4.564950629396418e-05,
-      "loss": 117.1461,
-      "step": 67890
-    },
-    {
-      "epoch": 0.2743245918462166,
-      "grad_norm": 367.62322998046875,
-      "learning_rate": 4.564753841178697e-05,
-      "loss": 45.144,
-      "step": 67900
-    },
-    {
-      "epoch": 0.27436499311158424,
-      "grad_norm": 2093.43505859375,
-      "learning_rate": 4.564557012707536e-05,
-      "loss": 118.0313,
-      "step": 67910
-    },
-    {
-      "epoch": 0.2744053943769519,
-      "grad_norm": 1412.019287109375,
-      "learning_rate": 4.5643601439867734e-05,
-      "loss": 85.6378,
-      "step": 67920
-    },
-    {
-      "epoch": 0.2744457956423195,
-      "grad_norm": 1107.6322021484375,
-      "learning_rate": 4.564163235020247e-05,
-      "loss": 77.5563,
-      "step": 67930
-    },
-    {
-      "epoch": 0.27448619690768716,
-      "grad_norm": 938.7551879882812,
-      "learning_rate": 4.563966285811796e-05,
-      "loss": 53.3907,
-      "step": 67940
-    },
-    {
-      "epoch": 0.2745265981730548,
-      "grad_norm": 898.0419921875,
-      "learning_rate": 4.5637692963652596e-05,
-      "loss": 56.9066,
-      "step": 67950
-    },
-    {
-      "epoch": 0.2745669994384224,
-      "grad_norm": 654.0993041992188,
-      "learning_rate": 4.5635722666844775e-05,
-      "loss": 71.2898,
-      "step": 67960
-    },
-    {
-      "epoch": 0.27460740070379,
-      "grad_norm": 762.6791381835938,
-      "learning_rate": 4.563375196773293e-05,
-      "loss": 111.2552,
-      "step": 67970
-    },
-    {
-      "epoch": 0.27464780196915767,
-      "grad_norm": 683.8074951171875,
-      "learning_rate": 4.5631780866355454e-05,
-      "loss": 79.4953,
-      "step": 67980
-    },
-    {
-      "epoch": 0.2746882032345253,
-      "grad_norm": 0.0,
-      "learning_rate": 4.56298093627508e-05,
-      "loss": 79.2197,
-      "step": 67990
-    },
-    {
-      "epoch": 0.27472860449989295,
-      "grad_norm": 525.245361328125,
-      "learning_rate": 4.562783745695738e-05,
-      "loss": 53.7112,
-      "step": 68000
-    },
-    {
-      "epoch": 0.2747690057652606,
-      "grad_norm": 588.7833251953125,
-      "learning_rate": 4.562586514901366e-05,
-      "loss": 50.6445,
-      "step": 68010
-    },
-    {
-      "epoch": 0.27480940703062817,
-      "grad_norm": 515.099853515625,
-      "learning_rate": 4.5623892438958074e-05,
-      "loss": 48.6461,
-      "step": 68020
-    },
-    {
-      "epoch": 0.2748498082959958,
-      "grad_norm": 835.2371826171875,
-      "learning_rate": 4.562191932682908e-05,
-      "loss": 49.5645,
-      "step": 68030
-    },
-    {
-      "epoch": 0.27489020956136345,
-      "grad_norm": 1118.2117919921875,
-      "learning_rate": 4.561994581266516e-05,
-      "loss": 97.7809,
-      "step": 68040
-    },
-    {
-      "epoch": 0.2749306108267311,
-      "grad_norm": 514.0653076171875,
-      "learning_rate": 4.561797189650478e-05,
-      "loss": 76.3595,
-      "step": 68050
-    },
-    {
-      "epoch": 0.27497101209209873,
-      "grad_norm": 461.3988342285156,
-      "learning_rate": 4.561599757838643e-05,
-      "loss": 66.2146,
-      "step": 68060
-    },
-    {
-      "epoch": 0.27501141335746637,
-      "grad_norm": 710.6083374023438,
-      "learning_rate": 4.561402285834858e-05,
-      "loss": 65.2828,
-      "step": 68070
-    },
-    {
-      "epoch": 0.275051814622834,
-      "grad_norm": 1474.451416015625,
-      "learning_rate": 4.561204773642974e-05,
-      "loss": 54.9727,
-      "step": 68080
-    },
-    {
-      "epoch": 0.2750922158882016,
-      "grad_norm": 606.7708129882812,
-      "learning_rate": 4.5610072212668434e-05,
-      "loss": 57.3314,
-      "step": 68090
-    },
-    {
-      "epoch": 0.27513261715356924,
-      "grad_norm": 706.001953125,
-      "learning_rate": 4.560809628710315e-05,
-      "loss": 72.5477,
-      "step": 68100
-    },
-    {
-      "epoch": 0.2751730184189369,
-      "grad_norm": 805.1146850585938,
-      "learning_rate": 4.560611995977242e-05,
-      "loss": 62.7448,
-      "step": 68110
-    },
-    {
-      "epoch": 0.2752134196843045,
-      "grad_norm": 502.4281311035156,
-      "learning_rate": 4.560414323071477e-05,
-      "loss": 40.9372,
-      "step": 68120
-    },
-    {
-      "epoch": 0.27525382094967216,
-      "grad_norm": 468.7350158691406,
-      "learning_rate": 4.560216609996874e-05,
-      "loss": 71.8095,
-      "step": 68130
-    },
-    {
-      "epoch": 0.2752942222150398,
-      "grad_norm": 557.2088623046875,
-      "learning_rate": 4.5600188567572876e-05,
-      "loss": 64.4124,
-      "step": 68140
-    },
-    {
-      "epoch": 0.2753346234804074,
-      "grad_norm": 1210.1270751953125,
-      "learning_rate": 4.559821063356574e-05,
-      "loss": 85.0767,
-      "step": 68150
-    },
-    {
-      "epoch": 0.275375024745775,
-      "grad_norm": 1038.1043701171875,
-      "learning_rate": 4.559623229798587e-05,
-      "loss": 71.8373,
-      "step": 68160
-    },
-    {
-      "epoch": 0.27541542601114266,
-      "grad_norm": 929.40576171875,
-      "learning_rate": 4.5594253560871854e-05,
-      "loss": 72.4507,
-      "step": 68170
-    },
-    {
-      "epoch": 0.2754558272765103,
-      "grad_norm": 957.2470092773438,
-      "learning_rate": 4.559227442226226e-05,
-      "loss": 97.403,
-      "step": 68180
-    },
-    {
-      "epoch": 0.27549622854187794,
-      "grad_norm": 706.5764770507812,
-      "learning_rate": 4.559029488219567e-05,
-      "loss": 64.6209,
-      "step": 68190
-    },
-    {
-      "epoch": 0.2755366298072456,
-      "grad_norm": 434.8182373046875,
-      "learning_rate": 4.558831494071069e-05,
-      "loss": 87.0814,
-      "step": 68200
-    },
-    {
-      "epoch": 0.2755770310726132,
-      "grad_norm": 760.851318359375,
-      "learning_rate": 4.5586334597845904e-05,
-      "loss": 80.2775,
-      "step": 68210
-    },
-    {
-      "epoch": 0.2756174323379808,
-      "grad_norm": 256.12982177734375,
-      "learning_rate": 4.558435385363993e-05,
-      "loss": 53.6615,
-      "step": 68220
-    },
-    {
-      "epoch": 0.27565783360334845,
-      "grad_norm": 421.9570617675781,
-      "learning_rate": 4.5582372708131385e-05,
-      "loss": 75.5364,
-      "step": 68230
-    },
-    {
-      "epoch": 0.2756982348687161,
-      "grad_norm": 812.7416381835938,
-      "learning_rate": 4.558039116135887e-05,
-      "loss": 65.1539,
-      "step": 68240
-    },
-    {
-      "epoch": 0.2757386361340837,
-      "grad_norm": 639.4966430664062,
-      "learning_rate": 4.557840921336105e-05,
-      "loss": 54.6641,
-      "step": 68250
-    },
-    {
-      "epoch": 0.27577903739945137,
-      "grad_norm": 370.4179992675781,
-      "learning_rate": 4.557642686417654e-05,
-      "loss": 98.5805,
-      "step": 68260
-    },
-    {
-      "epoch": 0.275819438664819,
-      "grad_norm": 516.9441528320312,
-      "learning_rate": 4.5574444113844e-05,
-      "loss": 76.0973,
-      "step": 68270
-    },
-    {
-      "epoch": 0.2758598399301866,
-      "grad_norm": 434.20660400390625,
-      "learning_rate": 4.5572460962402075e-05,
-      "loss": 59.694,
-      "step": 68280
-    },
-    {
-      "epoch": 0.27590024119555423,
-      "grad_norm": 1096.5814208984375,
-      "learning_rate": 4.557047740988944e-05,
-      "loss": 83.0501,
-      "step": 68290
-    },
-    {
-      "epoch": 0.27594064246092187,
-      "grad_norm": 679.775634765625,
-      "learning_rate": 4.556849345634475e-05,
-      "loss": 71.1283,
-      "step": 68300
-    },
-    {
-      "epoch": 0.2759810437262895,
-      "grad_norm": 1577.58251953125,
-      "learning_rate": 4.5566509101806695e-05,
-      "loss": 99.9247,
-      "step": 68310
-    },
-    {
-      "epoch": 0.27602144499165715,
-      "grad_norm": 1189.281982421875,
-      "learning_rate": 4.556452434631395e-05,
-      "loss": 80.3163,
-      "step": 68320
-    },
-    {
-      "epoch": 0.2760618462570248,
-      "grad_norm": 809.91015625,
-      "learning_rate": 4.5562539189905223e-05,
-      "loss": 104.2898,
-      "step": 68330
-    },
-    {
-      "epoch": 0.2761022475223924,
-      "grad_norm": 1438.5238037109375,
-      "learning_rate": 4.5560553632619205e-05,
-      "loss": 99.692,
-      "step": 68340
-    },
-    {
-      "epoch": 0.27614264878776,
-      "grad_norm": 330.85479736328125,
-      "learning_rate": 4.555856767449461e-05,
-      "loss": 67.1174,
-      "step": 68350
-    },
-    {
-      "epoch": 0.27618305005312765,
-      "grad_norm": 803.5668334960938,
-      "learning_rate": 4.555658131557015e-05,
-      "loss": 99.5837,
-      "step": 68360
-    },
-    {
-      "epoch": 0.2762234513184953,
-      "grad_norm": 674.5006713867188,
-      "learning_rate": 4.555459455588456e-05,
-      "loss": 65.4563,
-      "step": 68370
-    },
-    {
-      "epoch": 0.27626385258386293,
-      "grad_norm": 773.7620849609375,
-      "learning_rate": 4.555260739547657e-05,
-      "loss": 43.1016,
-      "step": 68380
-    },
-    {
-      "epoch": 0.2763042538492306,
-      "grad_norm": 639.3969116210938,
-      "learning_rate": 4.55506198343849e-05,
-      "loss": 78.3155,
-      "step": 68390
-    },
-    {
-      "epoch": 0.2763446551145982,
-      "grad_norm": 373.80694580078125,
-      "learning_rate": 4.5548631872648326e-05,
-      "loss": 83.402,
-      "step": 68400
-    },
-    {
-      "epoch": 0.2763850563799658,
-      "grad_norm": 420.725830078125,
-      "learning_rate": 4.55466435103056e-05,
-      "loss": 61.3818,
-      "step": 68410
-    },
-    {
-      "epoch": 0.27642545764533344,
-      "grad_norm": 877.37890625,
-      "learning_rate": 4.554465474739548e-05,
-      "loss": 59.281,
-      "step": 68420
-    },
-    {
-      "epoch": 0.2764658589107011,
-      "grad_norm": 295.0977478027344,
-      "learning_rate": 4.5542665583956736e-05,
-      "loss": 36.8095,
-      "step": 68430
-    },
-    {
-      "epoch": 0.2765062601760687,
-      "grad_norm": 288.376953125,
-      "learning_rate": 4.5540676020028145e-05,
-      "loss": 68.784,
-      "step": 68440
-    },
-    {
-      "epoch": 0.27654666144143636,
-      "grad_norm": 1751.3602294921875,
-      "learning_rate": 4.5538686055648506e-05,
-      "loss": 93.4555,
-      "step": 68450
-    },
-    {
-      "epoch": 0.276587062706804,
-      "grad_norm": 857.876953125,
-      "learning_rate": 4.5536695690856606e-05,
-      "loss": 53.5522,
-      "step": 68460
-    },
-    {
-      "epoch": 0.2766274639721716,
-      "grad_norm": 1057.3465576171875,
-      "learning_rate": 4.553470492569125e-05,
-      "loss": 58.7135,
-      "step": 68470
-    },
-    {
-      "epoch": 0.2766678652375392,
-      "grad_norm": 602.7147216796875,
-      "learning_rate": 4.553271376019125e-05,
-      "loss": 73.2039,
-      "step": 68480
-    },
-    {
-      "epoch": 0.27670826650290686,
-      "grad_norm": 799.6278686523438,
-      "learning_rate": 4.5530722194395425e-05,
-      "loss": 66.5906,
-      "step": 68490
-    },
-    {
-      "epoch": 0.2767486677682745,
-      "grad_norm": 539.0422973632812,
-      "learning_rate": 4.5528730228342605e-05,
-      "loss": 79.3498,
-      "step": 68500
-    },
-    {
-      "epoch": 0.27678906903364214,
-      "grad_norm": 838.4744873046875,
-      "learning_rate": 4.552673786207161e-05,
-      "loss": 46.1416,
-      "step": 68510
-    },
-    {
-      "epoch": 0.2768294702990098,
-      "grad_norm": 1674.3841552734375,
-      "learning_rate": 4.55247450956213e-05,
-      "loss": 72.0151,
-      "step": 68520
-    },
-    {
-      "epoch": 0.2768698715643774,
-      "grad_norm": 5940.70263671875,
-      "learning_rate": 4.552275192903052e-05,
-      "loss": 75.9517,
-      "step": 68530
-    },
-    {
-      "epoch": 0.276910272829745,
-      "grad_norm": 1089.8138427734375,
-      "learning_rate": 4.552075836233812e-05,
-      "loss": 97.0178,
-      "step": 68540
-    },
-    {
-      "epoch": 0.27695067409511265,
-      "grad_norm": 608.082275390625,
-      "learning_rate": 4.551876439558298e-05,
-      "loss": 85.0739,
-      "step": 68550
-    },
-    {
-      "epoch": 0.2769910753604803,
-      "grad_norm": 1392.0042724609375,
-      "learning_rate": 4.5516770028803954e-05,
-      "loss": 60.7242,
-      "step": 68560
-    },
-    {
-      "epoch": 0.27703147662584793,
-      "grad_norm": 758.804931640625,
-      "learning_rate": 4.5514775262039934e-05,
-      "loss": 61.6169,
-      "step": 68570
-    },
-    {
-      "epoch": 0.27707187789121557,
-      "grad_norm": 3674.3056640625,
-      "learning_rate": 4.551278009532981e-05,
-      "loss": 77.6101,
-      "step": 68580
-    },
-    {
-      "epoch": 0.2771122791565832,
-      "grad_norm": 1095.73876953125,
-      "learning_rate": 4.551078452871248e-05,
-      "loss": 54.9264,
-      "step": 68590
-    },
-    {
-      "epoch": 0.2771526804219508,
-      "grad_norm": 1469.385009765625,
-      "learning_rate": 4.550878856222685e-05,
-      "loss": 70.6401,
-      "step": 68600
-    },
-    {
-      "epoch": 0.27719308168731843,
-      "grad_norm": 866.8749389648438,
-      "learning_rate": 4.5506792195911817e-05,
-      "loss": 50.9549,
-      "step": 68610
-    },
-    {
-      "epoch": 0.27723348295268607,
-      "grad_norm": 735.7921142578125,
-      "learning_rate": 4.550479542980632e-05,
-      "loss": 85.6192,
-      "step": 68620
-    },
-    {
-      "epoch": 0.2772738842180537,
-      "grad_norm": 400.9674377441406,
-      "learning_rate": 4.550279826394928e-05,
-      "loss": 97.7626,
-      "step": 68630
-    },
-    {
-      "epoch": 0.27731428548342135,
-      "grad_norm": 700.01318359375,
-      "learning_rate": 4.5500800698379624e-05,
-      "loss": 66.2745,
-      "step": 68640
-    },
-    {
-      "epoch": 0.277354686748789,
-      "grad_norm": 1081.5074462890625,
-      "learning_rate": 4.549880273313631e-05,
-      "loss": 73.2586,
-      "step": 68650
-    },
-    {
-      "epoch": 0.2773950880141566,
-      "grad_norm": 770.979736328125,
-      "learning_rate": 4.5496804368258286e-05,
-      "loss": 56.8278,
-      "step": 68660
-    },
-    {
-      "epoch": 0.2774354892795242,
-      "grad_norm": 733.8268432617188,
-      "learning_rate": 4.549480560378451e-05,
-      "loss": 57.6247,
-      "step": 68670
-    },
-    {
-      "epoch": 0.27747589054489186,
-      "grad_norm": 744.9702758789062,
-      "learning_rate": 4.5492806439753935e-05,
-      "loss": 59.9518,
-      "step": 68680
-    },
-    {
-      "epoch": 0.2775162918102595,
-      "grad_norm": 701.0172729492188,
-      "learning_rate": 4.549080687620555e-05,
-      "loss": 86.6771,
-      "step": 68690
-    },
-    {
-      "epoch": 0.27755669307562714,
-      "grad_norm": 387.11041259765625,
-      "learning_rate": 4.548880691317835e-05,
-      "loss": 79.3791,
-      "step": 68700
-    },
-    {
-      "epoch": 0.2775970943409948,
-      "grad_norm": 1299.3056640625,
-      "learning_rate": 4.54868065507113e-05,
-      "loss": 64.44,
-      "step": 68710
-    },
-    {
-      "epoch": 0.2776374956063624,
-      "grad_norm": 795.0128173828125,
-      "learning_rate": 4.548480578884341e-05,
-      "loss": 60.6718,
-      "step": 68720
-    },
-    {
-      "epoch": 0.27767789687173,
-      "grad_norm": 332.22100830078125,
-      "learning_rate": 4.5482804627613686e-05,
-      "loss": 61.0125,
-      "step": 68730
-    },
-    {
-      "epoch": 0.27771829813709764,
-      "grad_norm": 525.1163940429688,
-      "learning_rate": 4.548080306706114e-05,
-      "loss": 74.0871,
-      "step": 68740
-    },
-    {
-      "epoch": 0.2777586994024653,
-      "grad_norm": 1393.9156494140625,
-      "learning_rate": 4.54788011072248e-05,
-      "loss": 72.4176,
-      "step": 68750
-    },
-    {
-      "epoch": 0.2777991006678329,
-      "grad_norm": 950.6893920898438,
-      "learning_rate": 4.547679874814368e-05,
-      "loss": 62.7376,
-      "step": 68760
-    },
-    {
-      "epoch": 0.27783950193320056,
-      "grad_norm": 1314.624267578125,
-      "learning_rate": 4.547479598985683e-05,
-      "loss": 95.1351,
-      "step": 68770
-    },
-    {
-      "epoch": 0.2778799031985682,
-      "grad_norm": 360.00762939453125,
-      "learning_rate": 4.547279283240329e-05,
-      "loss": 52.8572,
-      "step": 68780
-    },
-    {
-      "epoch": 0.2779203044639358,
-      "grad_norm": 789.6798706054688,
-      "learning_rate": 4.547078927582212e-05,
-      "loss": 79.6554,
-      "step": 68790
-    },
-    {
-      "epoch": 0.2779607057293034,
-      "grad_norm": 973.8438110351562,
-      "learning_rate": 4.5468785320152365e-05,
-      "loss": 66.0115,
-      "step": 68800
-    },
-    {
-      "epoch": 0.27800110699467107,
-      "grad_norm": 351.57525634765625,
-      "learning_rate": 4.546678096543311e-05,
-      "loss": 62.568,
-      "step": 68810
-    },
-    {
-      "epoch": 0.2780415082600387,
-      "grad_norm": 1510.792724609375,
-      "learning_rate": 4.546477621170342e-05,
-      "loss": 82.8039,
-      "step": 68820
-    },
-    {
-      "epoch": 0.27808190952540635,
-      "grad_norm": 1050.9688720703125,
-      "learning_rate": 4.546277105900237e-05,
-      "loss": 71.6794,
-      "step": 68830
-    },
-    {
-      "epoch": 0.278122310790774,
-      "grad_norm": 1624.618408203125,
-      "learning_rate": 4.5460765507369084e-05,
-      "loss": 64.593,
-      "step": 68840
-    },
-    {
-      "epoch": 0.2781627120561416,
-      "grad_norm": 1015.446533203125,
-      "learning_rate": 4.5458759556842624e-05,
-      "loss": 73.7418,
-      "step": 68850
-    },
-    {
-      "epoch": 0.2782031133215092,
-      "grad_norm": 1055.7169189453125,
-      "learning_rate": 4.545675320746212e-05,
-      "loss": 89.4522,
-      "step": 68860
-    },
-    {
-      "epoch": 0.27824351458687685,
-      "grad_norm": 939.6693725585938,
-      "learning_rate": 4.545474645926668e-05,
-      "loss": 55.4124,
-      "step": 68870
-    },
-    {
-      "epoch": 0.2782839158522445,
-      "grad_norm": 524.6759033203125,
-      "learning_rate": 4.5452739312295436e-05,
-      "loss": 67.2065,
-      "step": 68880
-    },
-    {
-      "epoch": 0.27832431711761213,
-      "grad_norm": 636.6887817382812,
-      "learning_rate": 4.54507317665875e-05,
-      "loss": 54.9611,
-      "step": 68890
-    },
-    {
-      "epoch": 0.27836471838297977,
-      "grad_norm": 1205.8494873046875,
-      "learning_rate": 4.544872382218202e-05,
-      "loss": 55.5338,
-      "step": 68900
-    },
-    {
-      "epoch": 0.2784051196483474,
-      "grad_norm": 400.2951965332031,
-      "learning_rate": 4.544671547911814e-05,
-      "loss": 72.6024,
-      "step": 68910
-    },
-    {
-      "epoch": 0.278445520913715,
-      "grad_norm": 671.6885986328125,
-      "learning_rate": 4.5444706737435014e-05,
-      "loss": 88.0325,
-      "step": 68920
-    },
-    {
-      "epoch": 0.27848592217908263,
-      "grad_norm": 905.302490234375,
-      "learning_rate": 4.544269759717181e-05,
-      "loss": 90.4285,
-      "step": 68930
-    },
-    {
-      "epoch": 0.2785263234444503,
-      "grad_norm": 713.8814086914062,
-      "learning_rate": 4.5440688058367686e-05,
-      "loss": 70.4518,
-      "step": 68940
-    },
-    {
-      "epoch": 0.2785667247098179,
-      "grad_norm": 986.9356689453125,
-      "learning_rate": 4.543867812106183e-05,
-      "loss": 61.4856,
-      "step": 68950
-    },
-    {
-      "epoch": 0.27860712597518555,
-      "grad_norm": 852.2457885742188,
-      "learning_rate": 4.543666778529342e-05,
-      "loss": 67.6646,
-      "step": 68960
-    },
-    {
-      "epoch": 0.2786475272405532,
-      "grad_norm": 1037.552490234375,
-      "learning_rate": 4.543465705110165e-05,
-      "loss": 50.1012,
-      "step": 68970
-    },
-    {
-      "epoch": 0.2786879285059208,
-      "grad_norm": 327.4298095703125,
-      "learning_rate": 4.543264591852572e-05,
-      "loss": 61.3465,
-      "step": 68980
-    },
-    {
-      "epoch": 0.2787283297712884,
-      "grad_norm": 866.1886596679688,
-      "learning_rate": 4.543063438760483e-05,
-      "loss": 48.84,
-      "step": 68990
-    },
-    {
-      "epoch": 0.27876873103665606,
-      "grad_norm": 708.6840209960938,
-      "learning_rate": 4.542862245837821e-05,
-      "loss": 66.3793,
-      "step": 69000
-    },
-    {
-      "epoch": 0.2788091323020237,
-      "grad_norm": 1597.6319580078125,
-      "learning_rate": 4.5426610130885087e-05,
-      "loss": 80.3279,
-      "step": 69010
-    },
-    {
-      "epoch": 0.27884953356739134,
-      "grad_norm": 906.7168579101562,
-      "learning_rate": 4.542459740516467e-05,
-      "loss": 73.8522,
-      "step": 69020
-    },
-    {
-      "epoch": 0.278889934832759,
-      "grad_norm": 890.2028198242188,
-      "learning_rate": 4.542258428125622e-05,
-      "loss": 90.2994,
-      "step": 69030
-    },
-    {
-      "epoch": 0.2789303360981266,
-      "grad_norm": 612.5167846679688,
-      "learning_rate": 4.542057075919897e-05,
-      "loss": 87.7907,
-      "step": 69040
-    },
-    {
-      "epoch": 0.2789707373634942,
-      "grad_norm": 983.6898193359375,
-      "learning_rate": 4.541855683903219e-05,
-      "loss": 65.7596,
-      "step": 69050
-    },
-    {
-      "epoch": 0.27901113862886184,
-      "grad_norm": 366.17645263671875,
-      "learning_rate": 4.541654252079513e-05,
-      "loss": 63.9307,
-      "step": 69060
-    },
-    {
-      "epoch": 0.2790515398942295,
-      "grad_norm": 489.0401306152344,
-      "learning_rate": 4.541452780452705e-05,
-      "loss": 41.8797,
-      "step": 69070
-    },
-    {
-      "epoch": 0.2790919411595971,
-      "grad_norm": 983.3743896484375,
-      "learning_rate": 4.5412512690267246e-05,
-      "loss": 111.5006,
-      "step": 69080
-    },
-    {
-      "epoch": 0.27913234242496476,
-      "grad_norm": 1284.0035400390625,
-      "learning_rate": 4.5410497178055e-05,
-      "loss": 58.7547,
-      "step": 69090
-    },
-    {
-      "epoch": 0.2791727436903324,
-      "grad_norm": 731.1244506835938,
-      "learning_rate": 4.5408481267929605e-05,
-      "loss": 81.0387,
-      "step": 69100
-    },
-    {
-      "epoch": 0.2792131449557,
-      "grad_norm": 717.0902099609375,
-      "learning_rate": 4.540646495993036e-05,
-      "loss": 130.641,
-      "step": 69110
-    },
-    {
-      "epoch": 0.27925354622106763,
-      "grad_norm": 844.9999389648438,
-      "learning_rate": 4.540444825409657e-05,
-      "loss": 67.0038,
-      "step": 69120
-    },
-    {
-      "epoch": 0.27929394748643527,
-      "grad_norm": 399.6801452636719,
-      "learning_rate": 4.540243115046756e-05,
-      "loss": 55.7817,
-      "step": 69130
-    },
-    {
-      "epoch": 0.2793343487518029,
-      "grad_norm": 438.31207275390625,
-      "learning_rate": 4.540041364908265e-05,
-      "loss": 63.9536,
-      "step": 69140
-    },
-    {
-      "epoch": 0.27937475001717055,
-      "grad_norm": 1663.7740478515625,
-      "learning_rate": 4.539839574998117e-05,
-      "loss": 80.7982,
-      "step": 69150
-    },
-    {
-      "epoch": 0.2794151512825382,
-      "grad_norm": 1818.9991455078125,
-      "learning_rate": 4.5396377453202466e-05,
-      "loss": 113.5125,
-      "step": 69160
-    },
-    {
-      "epoch": 0.27945555254790583,
-      "grad_norm": 736.188720703125,
-      "learning_rate": 4.539435875878588e-05,
-      "loss": 81.4038,
-      "step": 69170
-    },
-    {
-      "epoch": 0.2794959538132734,
-      "grad_norm": 879.2432250976562,
-      "learning_rate": 4.539233966677078e-05,
-      "loss": 87.2281,
-      "step": 69180
-    },
-    {
-      "epoch": 0.27953635507864105,
-      "grad_norm": 560.682373046875,
-      "learning_rate": 4.539032017719651e-05,
-      "loss": 82.0745,
-      "step": 69190
-    },
-    {
-      "epoch": 0.2795767563440087,
-      "grad_norm": 943.4149780273438,
-      "learning_rate": 4.5388300290102456e-05,
-      "loss": 117.8961,
-      "step": 69200
-    },
-    {
-      "epoch": 0.27961715760937633,
-      "grad_norm": 854.2255249023438,
-      "learning_rate": 4.538628000552799e-05,
-      "loss": 46.8455,
-      "step": 69210
-    },
-    {
-      "epoch": 0.27965755887474397,
-      "grad_norm": 768.5675659179688,
-      "learning_rate": 4.5384259323512504e-05,
-      "loss": 77.2018,
-      "step": 69220
-    },
-    {
-      "epoch": 0.2796979601401116,
-      "grad_norm": 790.1107177734375,
-      "learning_rate": 4.538223824409538e-05,
-      "loss": 62.6797,
-      "step": 69230
-    },
-    {
-      "epoch": 0.2797383614054792,
-      "grad_norm": 0.0,
-      "learning_rate": 4.538021676731603e-05,
-      "loss": 50.2399,
-      "step": 69240
-    },
-    {
-      "epoch": 0.27977876267084684,
-      "grad_norm": 2715.3037109375,
-      "learning_rate": 4.537819489321386e-05,
-      "loss": 107.9521,
-      "step": 69250
-    },
-    {
-      "epoch": 0.2798191639362145,
-      "grad_norm": 2060.965576171875,
-      "learning_rate": 4.537617262182829e-05,
-      "loss": 65.6806,
-      "step": 69260
-    },
-    {
-      "epoch": 0.2798595652015821,
-      "grad_norm": 1369.305419921875,
-      "learning_rate": 4.5374149953198746e-05,
-      "loss": 92.2145,
-      "step": 69270
-    },
-    {
-      "epoch": 0.27989996646694976,
-      "grad_norm": 830.0508422851562,
-      "learning_rate": 4.5372126887364655e-05,
-      "loss": 106.5613,
-      "step": 69280
-    },
-    {
-      "epoch": 0.2799403677323174,
-      "grad_norm": 942.8829345703125,
-      "learning_rate": 4.5370103424365474e-05,
-      "loss": 76.8458,
-      "step": 69290
-    },
-    {
-      "epoch": 0.279980768997685,
-      "grad_norm": 842.5079345703125,
-      "learning_rate": 4.536807956424063e-05,
-      "loss": 53.2971,
-      "step": 69300
-    },
-    {
-      "epoch": 0.2800211702630526,
-      "grad_norm": 1059.9853515625,
-      "learning_rate": 4.5366055307029585e-05,
-      "loss": 80.9336,
-      "step": 69310
-    },
-    {
-      "epoch": 0.28006157152842026,
-      "grad_norm": 326.0080261230469,
-      "learning_rate": 4.536403065277182e-05,
-      "loss": 78.7748,
-      "step": 69320
-    },
-    {
-      "epoch": 0.2801019727937879,
-      "grad_norm": 275.553955078125,
-      "learning_rate": 4.536200560150678e-05,
-      "loss": 69.185,
-      "step": 69330
-    },
-    {
-      "epoch": 0.28014237405915554,
-      "grad_norm": 686.8462524414062,
-      "learning_rate": 4.5359980153273964e-05,
-      "loss": 68.1509,
-      "step": 69340
-    },
-    {
-      "epoch": 0.2801827753245232,
-      "grad_norm": 630.9837646484375,
-      "learning_rate": 4.535795430811285e-05,
-      "loss": 107.0847,
-      "step": 69350
-    },
-    {
-      "epoch": 0.2802231765898908,
-      "grad_norm": 2537.67138671875,
-      "learning_rate": 4.535592806606294e-05,
-      "loss": 64.2041,
-      "step": 69360
-    },
-    {
-      "epoch": 0.2802635778552584,
-      "grad_norm": 871.0073852539062,
-      "learning_rate": 4.5353901427163725e-05,
-      "loss": 56.6685,
-      "step": 69370
-    },
-    {
-      "epoch": 0.28030397912062605,
-      "grad_norm": 805.7468872070312,
-      "learning_rate": 4.535187439145473e-05,
-      "loss": 56.4696,
-      "step": 69380
-    },
-    {
-      "epoch": 0.2803443803859937,
-      "grad_norm": 638.688720703125,
-      "learning_rate": 4.534984695897546e-05,
-      "loss": 61.3559,
-      "step": 69390
-    },
-    {
-      "epoch": 0.2803847816513613,
-      "grad_norm": 916.9631958007812,
-      "learning_rate": 4.534781912976546e-05,
-      "loss": 81.307,
-      "step": 69400
-    },
-    {
-      "epoch": 0.28042518291672897,
-      "grad_norm": 695.0220947265625,
-      "learning_rate": 4.534579090386424e-05,
-      "loss": 69.4657,
-      "step": 69410
-    },
-    {
-      "epoch": 0.2804655841820966,
-      "grad_norm": 2208.393310546875,
-      "learning_rate": 4.5343762281311345e-05,
-      "loss": 78.4579,
-      "step": 69420
-    },
-    {
-      "epoch": 0.2805059854474642,
-      "grad_norm": 903.3010864257812,
-      "learning_rate": 4.534173326214634e-05,
-      "loss": 85.2893,
-      "step": 69430
-    },
-    {
-      "epoch": 0.28054638671283183,
-      "grad_norm": 1086.91455078125,
-      "learning_rate": 4.533970384640877e-05,
-      "loss": 56.4757,
-      "step": 69440
-    },
-    {
-      "epoch": 0.28058678797819947,
-      "grad_norm": 841.8659057617188,
-      "learning_rate": 4.53376740341382e-05,
-      "loss": 71.8159,
-      "step": 69450
-    },
-    {
-      "epoch": 0.2806271892435671,
-      "grad_norm": 1415.8238525390625,
-      "learning_rate": 4.533564382537421e-05,
-      "loss": 62.5434,
-      "step": 69460
-    },
-    {
-      "epoch": 0.28066759050893475,
-      "grad_norm": 974.2418212890625,
-      "learning_rate": 4.533361322015637e-05,
-      "loss": 69.978,
-      "step": 69470
-    },
-    {
-      "epoch": 0.2807079917743024,
-      "grad_norm": 2103.853515625,
-      "learning_rate": 4.533158221852427e-05,
-      "loss": 81.6764,
-      "step": 69480
-    },
-    {
-      "epoch": 0.28074839303967003,
-      "grad_norm": 644.2200317382812,
-      "learning_rate": 4.532955082051751e-05,
-      "loss": 49.908,
-      "step": 69490
-    },
-    {
-      "epoch": 0.2807887943050376,
-      "grad_norm": 603.8206176757812,
-      "learning_rate": 4.532751902617569e-05,
-      "loss": 112.3736,
-      "step": 69500
-    },
-    {
-      "epoch": 0.28082919557040525,
-      "grad_norm": 700.8045654296875,
-      "learning_rate": 4.532548683553842e-05,
-      "loss": 65.1945,
-      "step": 69510
-    },
-    {
-      "epoch": 0.2808695968357729,
-      "grad_norm": 565.0137329101562,
-      "learning_rate": 4.5323454248645324e-05,
-      "loss": 89.2239,
-      "step": 69520
-    },
-    {
-      "epoch": 0.28090999810114053,
-      "grad_norm": 550.55859375,
-      "learning_rate": 4.532142126553603e-05,
-      "loss": 62.9486,
-      "step": 69530
-    },
-    {
-      "epoch": 0.2809503993665082,
-      "grad_norm": 819.1644287109375,
-      "learning_rate": 4.5319387886250156e-05,
-      "loss": 79.9527,
-      "step": 69540
-    },
-    {
-      "epoch": 0.2809908006318758,
-      "grad_norm": 1997.134033203125,
-      "learning_rate": 4.531735411082735e-05,
-      "loss": 73.3801,
-      "step": 69550
-    },
-    {
-      "epoch": 0.2810312018972434,
-      "grad_norm": 233.41488647460938,
-      "learning_rate": 4.531531993930727e-05,
-      "loss": 79.5541,
-      "step": 69560
-    },
-    {
-      "epoch": 0.28107160316261104,
-      "grad_norm": 981.9132080078125,
-      "learning_rate": 4.5313285371729575e-05,
-      "loss": 66.0245,
-      "step": 69570
-    },
-    {
-      "epoch": 0.2811120044279787,
-      "grad_norm": 600.988037109375,
-      "learning_rate": 4.531125040813392e-05,
-      "loss": 71.3912,
-      "step": 69580
-    },
-    {
-      "epoch": 0.2811524056933463,
-      "grad_norm": 760.9716796875,
-      "learning_rate": 4.530921504855997e-05,
-      "loss": 52.4341,
-      "step": 69590
-    },
-    {
-      "epoch": 0.28119280695871396,
-      "grad_norm": 1831.3338623046875,
-      "learning_rate": 4.530717929304743e-05,
-      "loss": 109.2623,
-      "step": 69600
-    },
-    {
-      "epoch": 0.2812332082240816,
-      "grad_norm": 838.1921997070312,
-      "learning_rate": 4.5305143141635976e-05,
-      "loss": 68.4752,
-      "step": 69610
-    },
-    {
-      "epoch": 0.2812736094894492,
-      "grad_norm": 446.0285949707031,
-      "learning_rate": 4.5303106594365296e-05,
-      "loss": 53.7918,
-      "step": 69620
-    },
-    {
-      "epoch": 0.2813140107548168,
-      "grad_norm": 1072.6981201171875,
-      "learning_rate": 4.530106965127511e-05,
-      "loss": 71.5331,
-      "step": 69630
-    },
-    {
-      "epoch": 0.28135441202018446,
-      "grad_norm": 841.3654174804688,
-      "learning_rate": 4.529903231240511e-05,
-      "loss": 74.8359,
-      "step": 69640
-    },
-    {
-      "epoch": 0.2813948132855521,
-      "grad_norm": 685.6234741210938,
-      "learning_rate": 4.529699457779503e-05,
-      "loss": 77.722,
-      "step": 69650
-    },
-    {
-      "epoch": 0.28143521455091974,
-      "grad_norm": 1382.5294189453125,
-      "learning_rate": 4.5294956447484584e-05,
-      "loss": 72.9175,
-      "step": 69660
-    },
-    {
-      "epoch": 0.2814756158162874,
-      "grad_norm": 651.1674194335938,
-      "learning_rate": 4.529291792151351e-05,
-      "loss": 73.6339,
-      "step": 69670
-    },
-    {
-      "epoch": 0.281516017081655,
-      "grad_norm": 540.8804931640625,
-      "learning_rate": 4.529087899992156e-05,
-      "loss": 36.5959,
-      "step": 69680
-    },
-    {
-      "epoch": 0.2815564183470226,
-      "grad_norm": 648.30615234375,
-      "learning_rate": 4.528883968274848e-05,
-      "loss": 96.0729,
-      "step": 69690
-    },
-    {
-      "epoch": 0.28159681961239025,
-      "grad_norm": 1536.320556640625,
-      "learning_rate": 4.528679997003403e-05,
-      "loss": 62.9985,
-      "step": 69700
-    },
-    {
-      "epoch": 0.2816372208777579,
-      "grad_norm": 458.5219421386719,
-      "learning_rate": 4.528475986181796e-05,
-      "loss": 69.9463,
-      "step": 69710
-    },
-    {
-      "epoch": 0.28167762214312553,
-      "grad_norm": 342.8268127441406,
-      "learning_rate": 4.5282719358140056e-05,
-      "loss": 87.2414,
-      "step": 69720
-    },
-    {
-      "epoch": 0.28171802340849317,
-      "grad_norm": 564.786376953125,
-      "learning_rate": 4.5280678459040095e-05,
-      "loss": 61.6622,
-      "step": 69730
-    },
-    {
-      "epoch": 0.2817584246738608,
-      "grad_norm": 665.1903686523438,
-      "learning_rate": 4.5278637164557866e-05,
-      "loss": 116.5534,
-      "step": 69740
-    },
-    {
-      "epoch": 0.2817988259392284,
-      "grad_norm": 736.6189575195312,
-      "learning_rate": 4.527659547473317e-05,
-      "loss": 67.6608,
-      "step": 69750
-    },
-    {
-      "epoch": 0.28183922720459603,
-      "grad_norm": 660.351806640625,
-      "learning_rate": 4.52745533896058e-05,
-      "loss": 84.5275,
-      "step": 69760
-    },
-    {
-      "epoch": 0.2818796284699637,
-      "grad_norm": 641.83642578125,
-      "learning_rate": 4.527251090921558e-05,
-      "loss": 64.5504,
-      "step": 69770
-    },
-    {
-      "epoch": 0.2819200297353313,
-      "grad_norm": 966.1611328125,
-      "learning_rate": 4.527046803360232e-05,
-      "loss": 66.6436,
-      "step": 69780
-    },
-    {
-      "epoch": 0.28196043100069895,
-      "grad_norm": 1488.1563720703125,
-      "learning_rate": 4.526842476280585e-05,
-      "loss": 72.4152,
-      "step": 69790
-    },
-    {
-      "epoch": 0.2820008322660666,
-      "grad_norm": 809.4866333007812,
-      "learning_rate": 4.5266381096866e-05,
-      "loss": 56.8854,
-      "step": 69800
-    },
-    {
-      "epoch": 0.28204123353143423,
-      "grad_norm": 945.8799438476562,
-      "learning_rate": 4.526433703582262e-05,
-      "loss": 85.1573,
-      "step": 69810
-    },
-    {
-      "epoch": 0.2820816347968018,
-      "grad_norm": 500.84307861328125,
-      "learning_rate": 4.5262292579715556e-05,
-      "loss": 37.6372,
-      "step": 69820
-    },
-    {
-      "epoch": 0.28212203606216946,
-      "grad_norm": 909.1983642578125,
-      "learning_rate": 4.526024772858467e-05,
-      "loss": 69.4535,
-      "step": 69830
-    },
-    {
-      "epoch": 0.2821624373275371,
-      "grad_norm": 1079.6373291015625,
-      "learning_rate": 4.525820248246982e-05,
-      "loss": 111.5219,
-      "step": 69840
-    },
-    {
-      "epoch": 0.28220283859290474,
-      "grad_norm": 727.4838256835938,
-      "learning_rate": 4.5256156841410886e-05,
-      "loss": 66.2702,
-      "step": 69850
-    },
-    {
-      "epoch": 0.2822432398582724,
-      "grad_norm": 1162.87353515625,
-      "learning_rate": 4.525411080544775e-05,
-      "loss": 62.2597,
-      "step": 69860
-    },
-    {
-      "epoch": 0.28228364112364,
-      "grad_norm": 2795.06640625,
-      "learning_rate": 4.5252064374620285e-05,
-      "loss": 99.6256,
-      "step": 69870
-    },
-    {
-      "epoch": 0.2823240423890076,
-      "grad_norm": 672.3999633789062,
-      "learning_rate": 4.5250017548968404e-05,
-      "loss": 104.1018,
-      "step": 69880
-    },
-    {
-      "epoch": 0.28236444365437524,
-      "grad_norm": 1359.24609375,
-      "learning_rate": 4.524797032853201e-05,
-      "loss": 90.7683,
-      "step": 69890
-    },
-    {
-      "epoch": 0.2824048449197429,
-      "grad_norm": 744.1480102539062,
-      "learning_rate": 4.5245922713350996e-05,
-      "loss": 73.5604,
-      "step": 69900
-    },
-    {
-      "epoch": 0.2824452461851105,
-      "grad_norm": 629.0797729492188,
-      "learning_rate": 4.524387470346531e-05,
-      "loss": 79.3253,
-      "step": 69910
-    },
-    {
-      "epoch": 0.28248564745047816,
-      "grad_norm": 846.8230590820312,
-      "learning_rate": 4.524182629891486e-05,
-      "loss": 52.2285,
-      "step": 69920
-    },
-    {
-      "epoch": 0.2825260487158458,
-      "grad_norm": 916.1362915039062,
-      "learning_rate": 4.523977749973958e-05,
-      "loss": 47.4978,
-      "step": 69930
-    },
-    {
-      "epoch": 0.2825664499812134,
-      "grad_norm": 878.1392822265625,
-      "learning_rate": 4.523772830597942e-05,
-      "loss": 69.1207,
-      "step": 69940
-    },
-    {
-      "epoch": 0.282606851246581,
-      "grad_norm": 1516.745361328125,
-      "learning_rate": 4.523567871767434e-05,
-      "loss": 70.5018,
-      "step": 69950
-    },
-    {
-      "epoch": 0.28264725251194867,
-      "grad_norm": 1540.39501953125,
-      "learning_rate": 4.523362873486427e-05,
-      "loss": 61.0742,
-      "step": 69960
-    },
-    {
-      "epoch": 0.2826876537773163,
-      "grad_norm": 588.6622314453125,
-      "learning_rate": 4.52315783575892e-05,
-      "loss": 105.1381,
-      "step": 69970
-    },
-    {
-      "epoch": 0.28272805504268395,
-      "grad_norm": 792.640380859375,
-      "learning_rate": 4.522952758588909e-05,
-      "loss": 74.2773,
-      "step": 69980
-    },
-    {
-      "epoch": 0.2827684563080516,
-      "grad_norm": 381.4082336425781,
-      "learning_rate": 4.5227476419803916e-05,
-      "loss": 42.4223,
-      "step": 69990
-    },
-    {
-      "epoch": 0.2828088575734192,
-      "grad_norm": 2376.44775390625,
-      "learning_rate": 4.522542485937369e-05,
-      "loss": 73.0111,
-      "step": 70000
-    },
-    {
-      "epoch": 0.2828492588387868,
-      "grad_norm": 478.2873229980469,
-      "learning_rate": 4.5223372904638386e-05,
-      "loss": 89.4105,
-      "step": 70010
-    },
-    {
-      "epoch": 0.28288966010415445,
-      "grad_norm": 1394.88623046875,
-      "learning_rate": 4.5221320555638016e-05,
-      "loss": 57.0316,
-      "step": 70020
-    },
-    {
-      "epoch": 0.2829300613695221,
-      "grad_norm": 1728.24267578125,
-      "learning_rate": 4.521926781241259e-05,
-      "loss": 60.0436,
-      "step": 70030
-    },
-    {
-      "epoch": 0.28297046263488973,
-      "grad_norm": 313.620361328125,
-      "learning_rate": 4.521721467500213e-05,
-      "loss": 78.582,
-      "step": 70040
-    },
-    {
-      "epoch": 0.28301086390025737,
-      "grad_norm": 876.0878295898438,
-      "learning_rate": 4.521516114344666e-05,
-      "loss": 96.7629,
-      "step": 70050
-    },
-    {
-      "epoch": 0.283051265165625,
-      "grad_norm": 715.8422241210938,
-      "learning_rate": 4.521310721778622e-05,
-      "loss": 68.7341,
-      "step": 70060
-    },
-    {
-      "epoch": 0.2830916664309926,
-      "grad_norm": 984.6596069335938,
-      "learning_rate": 4.5211052898060855e-05,
-      "loss": 85.3589,
-      "step": 70070
-    },
-    {
-      "epoch": 0.28313206769636023,
-      "grad_norm": 440.2537536621094,
-      "learning_rate": 4.5208998184310596e-05,
-      "loss": 66.0765,
-      "step": 70080
-    },
-    {
-      "epoch": 0.2831724689617279,
-      "grad_norm": 958.81982421875,
-      "learning_rate": 4.520694307657551e-05,
-      "loss": 62.4521,
-      "step": 70090
-    },
-    {
-      "epoch": 0.2832128702270955,
-      "grad_norm": 592.9948120117188,
-      "learning_rate": 4.5204887574895684e-05,
-      "loss": 76.5842,
-      "step": 70100
-    },
-    {
-      "epoch": 0.28325327149246315,
-      "grad_norm": 810.19287109375,
-      "learning_rate": 4.520283167931115e-05,
-      "loss": 67.7459,
-      "step": 70110
-    },
-    {
-      "epoch": 0.2832936727578308,
-      "grad_norm": 486.7115173339844,
-      "learning_rate": 4.5200775389862026e-05,
-      "loss": 57.3887,
-      "step": 70120
-    },
-    {
-      "epoch": 0.2833340740231984,
-      "grad_norm": 1011.84033203125,
-      "learning_rate": 4.519871870658838e-05,
-      "loss": 77.9214,
-      "step": 70130
-    },
-    {
-      "epoch": 0.283374475288566,
-      "grad_norm": 1170.3331298828125,
-      "learning_rate": 4.519666162953032e-05,
-      "loss": 81.3161,
-      "step": 70140
-    },
-    {
-      "epoch": 0.28341487655393366,
-      "grad_norm": 885.1085815429688,
-      "learning_rate": 4.519460415872794e-05,
-      "loss": 73.1431,
-      "step": 70150
-    },
-    {
-      "epoch": 0.2834552778193013,
-      "grad_norm": 746.1101684570312,
-      "learning_rate": 4.519254629422136e-05,
-      "loss": 75.5062,
-      "step": 70160
-    },
-    {
-      "epoch": 0.28349567908466894,
-      "grad_norm": 805.8640747070312,
-      "learning_rate": 4.5190488036050685e-05,
-      "loss": 62.1489,
-      "step": 70170
-    },
-    {
-      "epoch": 0.2835360803500366,
-      "grad_norm": 1288.6954345703125,
-      "learning_rate": 4.518842938425605e-05,
-      "loss": 70.3223,
-      "step": 70180
-    },
-    {
-      "epoch": 0.2835764816154042,
-      "grad_norm": 820.5307006835938,
-      "learning_rate": 4.51863703388776e-05,
-      "loss": 49.1447,
-      "step": 70190
-    },
-    {
-      "epoch": 0.2836168828807718,
-      "grad_norm": 279.9125671386719,
-      "learning_rate": 4.5184310899955465e-05,
-      "loss": 41.5823,
-      "step": 70200
-    },
-    {
-      "epoch": 0.28365728414613944,
-      "grad_norm": 2208.78564453125,
-      "learning_rate": 4.518225106752979e-05,
-      "loss": 63.2972,
-      "step": 70210
-    },
-    {
-      "epoch": 0.2836976854115071,
-      "grad_norm": 609.0946655273438,
-      "learning_rate": 4.5180190841640747e-05,
-      "loss": 59.1739,
-      "step": 70220
-    },
-    {
-      "epoch": 0.2837380866768747,
-      "grad_norm": 761.7667236328125,
-      "learning_rate": 4.517813022232849e-05,
-      "loss": 76.7521,
-      "step": 70230
-    },
-    {
-      "epoch": 0.28377848794224236,
-      "grad_norm": 623.1316528320312,
-      "learning_rate": 4.51760692096332e-05,
-      "loss": 54.5164,
-      "step": 70240
-    },
-    {
-      "epoch": 0.28381888920761,
-      "grad_norm": 875.2012329101562,
-      "learning_rate": 4.5174007803595055e-05,
-      "loss": 69.7034,
-      "step": 70250
-    },
-    {
-      "epoch": 0.2838592904729776,
-      "grad_norm": 294.3750915527344,
-      "learning_rate": 4.517194600425423e-05,
-      "loss": 75.7382,
-      "step": 70260
-    },
-    {
-      "epoch": 0.28389969173834523,
-      "grad_norm": 793.992919921875,
-      "learning_rate": 4.516988381165095e-05,
-      "loss": 57.0891,
-      "step": 70270
-    },
-    {
-      "epoch": 0.28394009300371287,
-      "grad_norm": 265.2918395996094,
-      "learning_rate": 4.516782122582538e-05,
-      "loss": 53.4736,
-      "step": 70280
-    },
-    {
-      "epoch": 0.2839804942690805,
-      "grad_norm": 637.1311645507812,
-      "learning_rate": 4.516575824681777e-05,
-      "loss": 89.6522,
-      "step": 70290
-    },
-    {
-      "epoch": 0.28402089553444815,
-      "grad_norm": 2700.84521484375,
-      "learning_rate": 4.516369487466832e-05,
-      "loss": 78.3818,
-      "step": 70300
-    },
-    {
-      "epoch": 0.2840612967998158,
-      "grad_norm": 775.3411865234375,
-      "learning_rate": 4.5161631109417246e-05,
-      "loss": 75.4549,
-      "step": 70310
-    },
-    {
-      "epoch": 0.28410169806518343,
-      "grad_norm": 616.4786376953125,
-      "learning_rate": 4.5159566951104796e-05,
-      "loss": 78.6188,
-      "step": 70320
-    },
-    {
-      "epoch": 0.284142099330551,
-      "grad_norm": 265.2325744628906,
-      "learning_rate": 4.515750239977122e-05,
-      "loss": 58.4777,
-      "step": 70330
-    },
-    {
-      "epoch": 0.28418250059591865,
-      "grad_norm": 356.9336242675781,
-      "learning_rate": 4.5155437455456744e-05,
-      "loss": 88.5666,
-      "step": 70340
-    },
-    {
-      "epoch": 0.2842229018612863,
-      "grad_norm": 1054.8193359375,
-      "learning_rate": 4.515337211820165e-05,
-      "loss": 74.0532,
-      "step": 70350
-    },
-    {
-      "epoch": 0.28426330312665393,
-      "grad_norm": 1073.7655029296875,
-      "learning_rate": 4.5151306388046175e-05,
-      "loss": 57.4097,
-      "step": 70360
-    },
-    {
-      "epoch": 0.2843037043920216,
-      "grad_norm": 1117.8515625,
-      "learning_rate": 4.5149240265030627e-05,
-      "loss": 91.991,
-      "step": 70370
-    },
-    {
-      "epoch": 0.2843441056573892,
-      "grad_norm": 554.7203979492188,
-      "learning_rate": 4.5147173749195255e-05,
-      "loss": 51.3412,
-      "step": 70380
-    },
-    {
-      "epoch": 0.2843845069227568,
-      "grad_norm": 669.3203125,
-      "learning_rate": 4.514510684058036e-05,
-      "loss": 90.7202,
-      "step": 70390
-    },
-    {
-      "epoch": 0.28442490818812444,
-      "grad_norm": 861.2157592773438,
-      "learning_rate": 4.5143039539226234e-05,
-      "loss": 70.1871,
-      "step": 70400
-    },
-    {
-      "epoch": 0.2844653094534921,
-      "grad_norm": 345.44781494140625,
-      "learning_rate": 4.514097184517318e-05,
-      "loss": 49.2981,
-      "step": 70410
-    },
-    {
-      "epoch": 0.2845057107188597,
-      "grad_norm": 1595.496337890625,
-      "learning_rate": 4.5138903758461515e-05,
-      "loss": 47.9407,
-      "step": 70420
-    },
-    {
-      "epoch": 0.28454611198422736,
-      "grad_norm": 560.5866088867188,
-      "learning_rate": 4.5136835279131556e-05,
-      "loss": 67.7868,
-      "step": 70430
-    },
-    {
-      "epoch": 0.284586513249595,
-      "grad_norm": 215.32069396972656,
-      "learning_rate": 4.513476640722362e-05,
-      "loss": 59.5168,
-      "step": 70440
-    },
-    {
-      "epoch": 0.2846269145149626,
-      "grad_norm": 858.737548828125,
-      "learning_rate": 4.513269714277805e-05,
-      "loss": 53.7752,
-      "step": 70450
-    },
-    {
-      "epoch": 0.2846673157803302,
-      "grad_norm": 575.0614624023438,
-      "learning_rate": 4.5130627485835186e-05,
-      "loss": 95.9444,
-      "step": 70460
-    },
-    {
-      "epoch": 0.28470771704569786,
-      "grad_norm": 2945.451416015625,
-      "learning_rate": 4.512855743643537e-05,
-      "loss": 69.3364,
-      "step": 70470
-    },
-    {
-      "epoch": 0.2847481183110655,
-      "grad_norm": 578.4557495117188,
-      "learning_rate": 4.512648699461897e-05,
-      "loss": 53.4026,
-      "step": 70480
-    },
-    {
-      "epoch": 0.28478851957643314,
-      "grad_norm": 362.3124084472656,
-      "learning_rate": 4.512441616042634e-05,
-      "loss": 86.2023,
-      "step": 70490
-    },
-    {
-      "epoch": 0.2848289208418008,
-      "grad_norm": 521.9320068359375,
-      "learning_rate": 4.512234493389785e-05,
-      "loss": 69.5704,
-      "step": 70500
-    },
-    {
-      "epoch": 0.2848693221071684,
-      "grad_norm": 606.285888671875,
-      "learning_rate": 4.5120273315073897e-05,
-      "loss": 67.9938,
-      "step": 70510
-    },
-    {
-      "epoch": 0.284909723372536,
-      "grad_norm": 977.4638671875,
-      "learning_rate": 4.511820130399485e-05,
-      "loss": 90.854,
-      "step": 70520
-    },
-    {
-      "epoch": 0.28495012463790365,
-      "grad_norm": 402.9257507324219,
-      "learning_rate": 4.5116128900701114e-05,
-      "loss": 79.0524,
-      "step": 70530
-    },
-    {
-      "epoch": 0.2849905259032713,
-      "grad_norm": 1074.8834228515625,
-      "learning_rate": 4.511405610523309e-05,
-      "loss": 80.932,
-      "step": 70540
-    },
-    {
-      "epoch": 0.2850309271686389,
-      "grad_norm": 1405.5771484375,
-      "learning_rate": 4.5111982917631194e-05,
-      "loss": 103.0241,
-      "step": 70550
-    },
-    {
-      "epoch": 0.28507132843400657,
-      "grad_norm": 1106.503662109375,
-      "learning_rate": 4.510990933793583e-05,
-      "loss": 70.0878,
-      "step": 70560
-    },
-    {
-      "epoch": 0.2851117296993742,
-      "grad_norm": 788.6580810546875,
-      "learning_rate": 4.5107835366187425e-05,
-      "loss": 32.8911,
-      "step": 70570
-    },
-    {
-      "epoch": 0.2851521309647418,
-      "grad_norm": 1015.0027465820312,
-      "learning_rate": 4.5105761002426415e-05,
-      "loss": 55.8387,
-      "step": 70580
-    },
-    {
-      "epoch": 0.28519253223010943,
-      "grad_norm": 537.4610595703125,
-      "learning_rate": 4.510368624669325e-05,
-      "loss": 58.8079,
-      "step": 70590
-    },
-    {
-      "epoch": 0.28523293349547707,
-      "grad_norm": 1022.7601318359375,
-      "learning_rate": 4.510161109902837e-05,
-      "loss": 66.04,
-      "step": 70600
-    },
-    {
-      "epoch": 0.2852733347608447,
-      "grad_norm": 785.5269165039062,
-      "learning_rate": 4.5099535559472234e-05,
-      "loss": 87.9221,
-      "step": 70610
-    },
-    {
-      "epoch": 0.28531373602621235,
-      "grad_norm": 765.5206298828125,
-      "learning_rate": 4.50974596280653e-05,
-      "loss": 72.3749,
-      "step": 70620
-    },
-    {
-      "epoch": 0.28535413729158,
-      "grad_norm": 1824.7486572265625,
-      "learning_rate": 4.509538330484805e-05,
-      "loss": 70.8314,
-      "step": 70630
-    },
-    {
-      "epoch": 0.28539453855694763,
-      "grad_norm": 784.1112670898438,
-      "learning_rate": 4.509330658986095e-05,
-      "loss": 64.7922,
-      "step": 70640
-    },
-    {
-      "epoch": 0.2854349398223152,
-      "grad_norm": 621.5699462890625,
-      "learning_rate": 4.5091229483144495e-05,
-      "loss": 94.8581,
-      "step": 70650
-    },
-    {
-      "epoch": 0.28547534108768285,
-      "grad_norm": 724.716796875,
-      "learning_rate": 4.508915198473919e-05,
-      "loss": 54.7563,
-      "step": 70660
-    },
-    {
-      "epoch": 0.2855157423530505,
-      "grad_norm": 435.5616149902344,
-      "learning_rate": 4.50870740946855e-05,
-      "loss": 69.268,
-      "step": 70670
-    },
-    {
-      "epoch": 0.28555614361841813,
-      "grad_norm": 1086.2598876953125,
-      "learning_rate": 4.508499581302398e-05,
-      "loss": 56.1892,
-      "step": 70680
-    },
-    {
-      "epoch": 0.2855965448837858,
-      "grad_norm": 1214.41015625,
-      "learning_rate": 4.5082917139795125e-05,
-      "loss": 75.7297,
-      "step": 70690
-    },
-    {
-      "epoch": 0.2856369461491534,
-      "grad_norm": 665.9967041015625,
-      "learning_rate": 4.508083807503945e-05,
-      "loss": 56.6401,
-      "step": 70700
-    },
-    {
-      "epoch": 0.285677347414521,
-      "grad_norm": 1049.6688232421875,
-      "learning_rate": 4.50787586187975e-05,
-      "loss": 70.0839,
-      "step": 70710
-    },
-    {
-      "epoch": 0.28571774867988864,
-      "grad_norm": 680.3385009765625,
-      "learning_rate": 4.507667877110982e-05,
-      "loss": 65.669,
-      "step": 70720
-    },
-    {
-      "epoch": 0.2857581499452563,
-      "grad_norm": 351.40545654296875,
-      "learning_rate": 4.507459853201695e-05,
-      "loss": 52.2754,
-      "step": 70730
-    },
-    {
-      "epoch": 0.2857985512106239,
-      "grad_norm": 699.9556274414062,
-      "learning_rate": 4.507251790155944e-05,
-      "loss": 66.7143,
-      "step": 70740
-    },
-    {
-      "epoch": 0.28583895247599156,
-      "grad_norm": 483.2270202636719,
-      "learning_rate": 4.5070436879777865e-05,
-      "loss": 72.3885,
-      "step": 70750
-    },
-    {
-      "epoch": 0.2858793537413592,
-      "grad_norm": 13059.109375,
-      "learning_rate": 4.506835546671278e-05,
-      "loss": 100.1066,
-      "step": 70760
-    },
-    {
-      "epoch": 0.2859197550067268,
-      "grad_norm": 1270.271240234375,
-      "learning_rate": 4.506627366240479e-05,
-      "loss": 76.3044,
-      "step": 70770
-    },
-    {
-      "epoch": 0.2859601562720944,
-      "grad_norm": 373.1097106933594,
-      "learning_rate": 4.506419146689446e-05,
-      "loss": 138.1525,
-      "step": 70780
-    },
-    {
-      "epoch": 0.28600055753746206,
-      "grad_norm": 627.6198120117188,
-      "learning_rate": 4.506210888022239e-05,
-      "loss": 80.5968,
-      "step": 70790
-    },
-    {
-      "epoch": 0.2860409588028297,
-      "grad_norm": 396.57012939453125,
-      "learning_rate": 4.5060025902429174e-05,
-      "loss": 53.7545,
-      "step": 70800
-    },
-    {
-      "epoch": 0.28608136006819734,
-      "grad_norm": 534.5454711914062,
-      "learning_rate": 4.505794253355542e-05,
-      "loss": 87.9787,
-      "step": 70810
-    },
-    {
-      "epoch": 0.286121761333565,
-      "grad_norm": 1753.4989013671875,
-      "learning_rate": 4.505585877364175e-05,
-      "loss": 80.0071,
-      "step": 70820
-    },
-    {
-      "epoch": 0.2861621625989326,
-      "grad_norm": 940.0836791992188,
-      "learning_rate": 4.505377462272879e-05,
-      "loss": 64.5939,
-      "step": 70830
-    },
-    {
-      "epoch": 0.2862025638643002,
-      "grad_norm": 524.5255737304688,
-      "learning_rate": 4.5051690080857176e-05,
-      "loss": 54.3899,
-      "step": 70840
-    },
-    {
-      "epoch": 0.28624296512966785,
-      "grad_norm": 982.0265502929688,
-      "learning_rate": 4.504960514806753e-05,
-      "loss": 64.2541,
-      "step": 70850
-    },
-    {
-      "epoch": 0.2862833663950355,
-      "grad_norm": 1201.5341796875,
-      "learning_rate": 4.504751982440052e-05,
-      "loss": 88.8859,
-      "step": 70860
-    },
-    {
-      "epoch": 0.28632376766040313,
-      "grad_norm": 624.1221313476562,
-      "learning_rate": 4.5045434109896786e-05,
-      "loss": 59.5705,
-      "step": 70870
-    },
-    {
-      "epoch": 0.28636416892577077,
-      "grad_norm": 422.28778076171875,
-      "learning_rate": 4.504334800459699e-05,
-      "loss": 44.9323,
-      "step": 70880
-    },
-    {
-      "epoch": 0.2864045701911384,
-      "grad_norm": 665.96923828125,
-      "learning_rate": 4.504126150854181e-05,
-      "loss": 60.9275,
-      "step": 70890
-    },
-    {
-      "epoch": 0.286444971456506,
-      "grad_norm": 1277.694580078125,
-      "learning_rate": 4.503917462177192e-05,
-      "loss": 77.8568,
-      "step": 70900
-    },
-    {
-      "epoch": 0.28648537272187363,
-      "grad_norm": 1883.61962890625,
-      "learning_rate": 4.5037087344328e-05,
-      "loss": 62.4995,
-      "step": 70910
-    },
-    {
-      "epoch": 0.2865257739872413,
-      "grad_norm": 885.0050659179688,
-      "learning_rate": 4.5034999676250745e-05,
-      "loss": 62.4988,
-      "step": 70920
-    },
-    {
-      "epoch": 0.2865661752526089,
-      "grad_norm": 1359.623779296875,
-      "learning_rate": 4.503291161758087e-05,
-      "loss": 71.7273,
-      "step": 70930
-    },
-    {
-      "epoch": 0.28660657651797655,
-      "grad_norm": 497.2101745605469,
-      "learning_rate": 4.5030823168359046e-05,
-      "loss": 76.8819,
-      "step": 70940
-    },
-    {
-      "epoch": 0.2866469777833442,
-      "grad_norm": 891.3270874023438,
-      "learning_rate": 4.502873432862603e-05,
-      "loss": 69.6958,
-      "step": 70950
-    },
-    {
-      "epoch": 0.28668737904871183,
-      "grad_norm": 441.0372314453125,
-      "learning_rate": 4.5026645098422515e-05,
-      "loss": 102.5139,
-      "step": 70960
-    },
-    {
-      "epoch": 0.2867277803140794,
-      "grad_norm": 993.6549072265625,
-      "learning_rate": 4.5024555477789255e-05,
-      "loss": 79.794,
-      "step": 70970
-    },
-    {
-      "epoch": 0.28676818157944706,
-      "grad_norm": 3066.687255859375,
-      "learning_rate": 4.5022465466766974e-05,
-      "loss": 105.9858,
-      "step": 70980
-    },
-    {
-      "epoch": 0.2868085828448147,
-      "grad_norm": 1127.01025390625,
-      "learning_rate": 4.502037506539642e-05,
-      "loss": 69.0879,
-      "step": 70990
-    },
-    {
-      "epoch": 0.28684898411018234,
-      "grad_norm": 1199.5037841796875,
-      "learning_rate": 4.5018284273718336e-05,
-      "loss": 58.2869,
-      "step": 71000
-    },
-    {
-      "epoch": 0.28688938537555,
-      "grad_norm": 706.9274291992188,
-      "learning_rate": 4.5016193091773504e-05,
-      "loss": 71.1858,
-      "step": 71010
-    },
-    {
-      "epoch": 0.2869297866409176,
-      "grad_norm": 758.85498046875,
-      "learning_rate": 4.501410151960268e-05,
-      "loss": 75.0063,
-      "step": 71020
-    },
-    {
-      "epoch": 0.2869701879062852,
-      "grad_norm": 938.1792602539062,
-      "learning_rate": 4.5012009557246645e-05,
-      "loss": 56.8306,
-      "step": 71030
-    },
-    {
-      "epoch": 0.28701058917165284,
-      "grad_norm": 539.9905395507812,
-      "learning_rate": 4.5009917204746184e-05,
-      "loss": 49.6551,
-      "step": 71040
-    },
-    {
-      "epoch": 0.2870509904370205,
-      "grad_norm": 1365.8111572265625,
-      "learning_rate": 4.5007824462142076e-05,
-      "loss": 67.7948,
-      "step": 71050
-    },
-    {
-      "epoch": 0.2870913917023881,
-      "grad_norm": 1149.0391845703125,
-      "learning_rate": 4.500573132947514e-05,
-      "loss": 62.5273,
-      "step": 71060
-    },
-    {
-      "epoch": 0.28713179296775576,
-      "grad_norm": 833.6195678710938,
-      "learning_rate": 4.500363780678617e-05,
-      "loss": 46.2563,
-      "step": 71070
-    },
-    {
-      "epoch": 0.2871721942331234,
-      "grad_norm": 2005.719482421875,
-      "learning_rate": 4.5001543894115975e-05,
-      "loss": 65.34,
-      "step": 71080
-    },
-    {
-      "epoch": 0.287212595498491,
-      "grad_norm": 499.4972839355469,
-      "learning_rate": 4.4999449591505396e-05,
-      "loss": 46.8325,
-      "step": 71090
-    },
-    {
-      "epoch": 0.2872529967638586,
-      "grad_norm": 1662.39013671875,
-      "learning_rate": 4.499735489899524e-05,
-      "loss": 75.1804,
-      "step": 71100
-    },
-    {
-      "epoch": 0.28729339802922627,
-      "grad_norm": 297.105712890625,
-      "learning_rate": 4.4995259816626356e-05,
-      "loss": 59.1684,
-      "step": 71110
-    },
-    {
-      "epoch": 0.2873337992945939,
-      "grad_norm": 506.93670654296875,
-      "learning_rate": 4.499316434443959e-05,
-      "loss": 41.7648,
-      "step": 71120
-    },
-    {
-      "epoch": 0.28737420055996155,
-      "grad_norm": 666.0968627929688,
-      "learning_rate": 4.49910684824758e-05,
-      "loss": 110.1385,
-      "step": 71130
-    },
-    {
-      "epoch": 0.2874146018253292,
-      "grad_norm": 786.0115966796875,
-      "learning_rate": 4.498897223077582e-05,
-      "loss": 72.202,
-      "step": 71140
-    },
-    {
-      "epoch": 0.2874550030906968,
-      "grad_norm": 7043.43310546875,
-      "learning_rate": 4.498687558938055e-05,
-      "loss": 108.4283,
-      "step": 71150
-    },
-    {
-      "epoch": 0.2874954043560644,
-      "grad_norm": 857.16748046875,
-      "learning_rate": 4.4984778558330844e-05,
-      "loss": 96.0511,
-      "step": 71160
-    },
-    {
-      "epoch": 0.28753580562143205,
-      "grad_norm": 637.8214111328125,
-      "learning_rate": 4.4982681137667594e-05,
-      "loss": 44.1582,
-      "step": 71170
-    },
-    {
-      "epoch": 0.2875762068867997,
-      "grad_norm": 904.2376098632812,
-      "learning_rate": 4.498058332743168e-05,
-      "loss": 98.6093,
-      "step": 71180
-    },
-    {
-      "epoch": 0.28761660815216733,
-      "grad_norm": 1919.4237060546875,
-      "learning_rate": 4.4978485127664015e-05,
-      "loss": 103.2059,
-      "step": 71190
-    },
-    {
-      "epoch": 0.28765700941753497,
-      "grad_norm": 810.6397094726562,
-      "learning_rate": 4.4976386538405495e-05,
-      "loss": 67.5545,
-      "step": 71200
-    },
-    {
-      "epoch": 0.2876974106829026,
-      "grad_norm": 1610.69384765625,
-      "learning_rate": 4.4974287559697035e-05,
-      "loss": 60.4293,
-      "step": 71210
-    },
-    {
-      "epoch": 0.2877378119482702,
-      "grad_norm": 1313.538818359375,
-      "learning_rate": 4.497218819157956e-05,
-      "loss": 57.7007,
-      "step": 71220
-    },
-    {
-      "epoch": 0.28777821321363783,
-      "grad_norm": 872.7113037109375,
-      "learning_rate": 4.497008843409399e-05,
-      "loss": 51.1471,
-      "step": 71230
-    },
-    {
-      "epoch": 0.2878186144790055,
-      "grad_norm": 398.4093017578125,
-      "learning_rate": 4.496798828728126e-05,
-      "loss": 73.8484,
-      "step": 71240
-    },
-    {
-      "epoch": 0.2878590157443731,
-      "grad_norm": 3197.44970703125,
-      "learning_rate": 4.496588775118232e-05,
-      "loss": 97.3321,
-      "step": 71250
-    },
-    {
-      "epoch": 0.28789941700974075,
-      "grad_norm": 422.7282409667969,
-      "learning_rate": 4.496378682583813e-05,
-      "loss": 79.5314,
-      "step": 71260
-    },
-    {
-      "epoch": 0.2879398182751084,
-      "grad_norm": 409.5416564941406,
-      "learning_rate": 4.4961685511289625e-05,
-      "loss": 93.6449,
-      "step": 71270
-    },
-    {
-      "epoch": 0.28798021954047603,
-      "grad_norm": 714.4796752929688,
-      "learning_rate": 4.495958380757779e-05,
-      "loss": 105.323,
-      "step": 71280
-    },
-    {
-      "epoch": 0.2880206208058436,
-      "grad_norm": 1594.0919189453125,
-      "learning_rate": 4.4957481714743585e-05,
-      "loss": 68.0684,
-      "step": 71290
-    },
-    {
-      "epoch": 0.28806102207121126,
-      "grad_norm": 460.36334228515625,
-      "learning_rate": 4.495537923282801e-05,
-      "loss": 64.8978,
-      "step": 71300
-    },
-    {
-      "epoch": 0.2881014233365789,
-      "grad_norm": 1116.5792236328125,
-      "learning_rate": 4.4953276361872034e-05,
-      "loss": 61.4736,
-      "step": 71310
-    },
-    {
-      "epoch": 0.28814182460194654,
-      "grad_norm": 463.9264221191406,
-      "learning_rate": 4.4951173101916675e-05,
-      "loss": 59.6178,
-      "step": 71320
-    },
-    {
-      "epoch": 0.2881822258673142,
-      "grad_norm": 958.1243896484375,
-      "learning_rate": 4.494906945300291e-05,
-      "loss": 76.7924,
-      "step": 71330
-    },
-    {
-      "epoch": 0.2882226271326818,
-      "grad_norm": 479.6456604003906,
-      "learning_rate": 4.4946965415171775e-05,
-      "loss": 57.5769,
-      "step": 71340
-    },
-    {
-      "epoch": 0.2882630283980494,
-      "grad_norm": 595.579345703125,
-      "learning_rate": 4.4944860988464276e-05,
-      "loss": 72.5989,
-      "step": 71350
-    },
-    {
-      "epoch": 0.28830342966341704,
-      "grad_norm": 821.9955444335938,
-      "learning_rate": 4.494275617292144e-05,
-      "loss": 74.0442,
-      "step": 71360
-    },
-    {
-      "epoch": 0.2883438309287847,
-      "grad_norm": 890.1593627929688,
-      "learning_rate": 4.494065096858432e-05,
-      "loss": 67.1569,
-      "step": 71370
-    },
-    {
-      "epoch": 0.2883842321941523,
-      "grad_norm": 935.3182983398438,
-      "learning_rate": 4.4938545375493934e-05,
-      "loss": 57.2469,
-      "step": 71380
-    },
-    {
-      "epoch": 0.28842463345951996,
-      "grad_norm": 930.1670532226562,
-      "learning_rate": 4.493643939369134e-05,
-      "loss": 62.9051,
-      "step": 71390
-    },
-    {
-      "epoch": 0.2884650347248876,
-      "grad_norm": 923.1588745117188,
-      "learning_rate": 4.493433302321759e-05,
-      "loss": 72.7095,
-      "step": 71400
-    },
-    {
-      "epoch": 0.2885054359902552,
-      "grad_norm": 1056.5206298828125,
-      "learning_rate": 4.4932226264113764e-05,
-      "loss": 89.8381,
-      "step": 71410
-    },
-    {
-      "epoch": 0.28854583725562283,
-      "grad_norm": 556.2738647460938,
-      "learning_rate": 4.493011911642092e-05,
-      "loss": 64.1478,
-      "step": 71420
-    },
-    {
-      "epoch": 0.28858623852099047,
-      "grad_norm": 614.498046875,
-      "learning_rate": 4.4928011580180155e-05,
-      "loss": 71.601,
-      "step": 71430
-    },
-    {
-      "epoch": 0.2886266397863581,
-      "grad_norm": 1232.3363037109375,
-      "learning_rate": 4.492590365543253e-05,
-      "loss": 59.9572,
-      "step": 71440
-    },
-    {
-      "epoch": 0.28866704105172575,
-      "grad_norm": 1182.1708984375,
-      "learning_rate": 4.492379534221916e-05,
-      "loss": 87.4886,
-      "step": 71450
-    },
-    {
-      "epoch": 0.2887074423170934,
-      "grad_norm": 703.7147827148438,
-      "learning_rate": 4.492168664058114e-05,
-      "loss": 48.6093,
-      "step": 71460
-    },
-    {
-      "epoch": 0.28874784358246103,
-      "grad_norm": 1096.2550048828125,
-      "learning_rate": 4.491957755055959e-05,
-      "loss": 55.8847,
-      "step": 71470
-    },
-    {
-      "epoch": 0.2887882448478286,
-      "grad_norm": 824.9928588867188,
-      "learning_rate": 4.491746807219561e-05,
-      "loss": 69.6475,
-      "step": 71480
-    },
-    {
-      "epoch": 0.28882864611319625,
-      "grad_norm": 799.056640625,
-      "learning_rate": 4.491535820553034e-05,
-      "loss": 74.3196,
-      "step": 71490
-    },
-    {
-      "epoch": 0.2888690473785639,
-      "grad_norm": 1376.468017578125,
-      "learning_rate": 4.491324795060491e-05,
-      "loss": 56.1541,
-      "step": 71500
-    },
-    {
-      "epoch": 0.28890944864393153,
-      "grad_norm": 792.1198120117188,
-      "learning_rate": 4.491113730746046e-05,
-      "loss": 80.2355,
-      "step": 71510
-    },
-    {
-      "epoch": 0.2889498499092992,
-      "grad_norm": 1302.2103271484375,
-      "learning_rate": 4.490902627613813e-05,
-      "loss": 70.5781,
-      "step": 71520
-    },
-    {
-      "epoch": 0.2889902511746668,
-      "grad_norm": 1214.075927734375,
-      "learning_rate": 4.4906914856679094e-05,
-      "loss": 84.3187,
-      "step": 71530
-    },
-    {
-      "epoch": 0.2890306524400344,
-      "grad_norm": 401.3418273925781,
-      "learning_rate": 4.49048030491245e-05,
-      "loss": 53.576,
-      "step": 71540
-    },
-    {
-      "epoch": 0.28907105370540204,
-      "grad_norm": 885.4319458007812,
-      "learning_rate": 4.490269085351552e-05,
-      "loss": 46.2776,
-      "step": 71550
-    },
-    {
-      "epoch": 0.2891114549707697,
-      "grad_norm": 410.9359436035156,
-      "learning_rate": 4.4900578269893335e-05,
-      "loss": 75.6896,
-      "step": 71560
-    },
-    {
-      "epoch": 0.2891518562361373,
-      "grad_norm": 858.1013793945312,
-      "learning_rate": 4.4898465298299134e-05,
-      "loss": 78.359,
-      "step": 71570
-    },
-    {
-      "epoch": 0.28919225750150496,
-      "grad_norm": 1423.806640625,
-      "learning_rate": 4.489635193877411e-05,
-      "loss": 62.488,
-      "step": 71580
-    },
-    {
-      "epoch": 0.2892326587668726,
-      "grad_norm": 962.8115234375,
-      "learning_rate": 4.489423819135945e-05,
-      "loss": 68.6732,
-      "step": 71590
-    },
-    {
-      "epoch": 0.28927306003224024,
-      "grad_norm": 737.3809814453125,
-      "learning_rate": 4.4892124056096386e-05,
-      "loss": 85.0963,
-      "step": 71600
-    },
-    {
-      "epoch": 0.2893134612976078,
-      "grad_norm": 699.3038330078125,
-      "learning_rate": 4.489000953302612e-05,
-      "loss": 39.0792,
-      "step": 71610
-    },
-    {
-      "epoch": 0.28935386256297546,
-      "grad_norm": 526.4671020507812,
-      "learning_rate": 4.488789462218987e-05,
-      "loss": 73.1894,
-      "step": 71620
-    },
-    {
-      "epoch": 0.2893942638283431,
-      "grad_norm": 869.59130859375,
-      "learning_rate": 4.4885779323628886e-05,
-      "loss": 80.5256,
-      "step": 71630
-    },
-    {
-      "epoch": 0.28943466509371074,
-      "grad_norm": 2488.22216796875,
-      "learning_rate": 4.4883663637384396e-05,
-      "loss": 96.78,
-      "step": 71640
-    },
-    {
-      "epoch": 0.2894750663590784,
-      "grad_norm": 322.6717529296875,
-      "learning_rate": 4.488154756349764e-05,
-      "loss": 55.7131,
-      "step": 71650
-    },
-    {
-      "epoch": 0.289515467624446,
-      "grad_norm": 762.85498046875,
-      "learning_rate": 4.4879431102009886e-05,
-      "loss": 83.2775,
-      "step": 71660
-    },
-    {
-      "epoch": 0.2895558688898136,
-      "grad_norm": 312.4606018066406,
-      "learning_rate": 4.487731425296238e-05,
-      "loss": 73.5523,
-      "step": 71670
-    },
-    {
-      "epoch": 0.28959627015518125,
-      "grad_norm": 774.15478515625,
-      "learning_rate": 4.487519701639641e-05,
-      "loss": 58.399,
-      "step": 71680
-    },
-    {
-      "epoch": 0.2896366714205489,
-      "grad_norm": 939.1637573242188,
-      "learning_rate": 4.487307939235323e-05,
-      "loss": 103.4214,
-      "step": 71690
-    },
-    {
-      "epoch": 0.2896770726859165,
-      "grad_norm": 378.4061584472656,
-      "learning_rate": 4.487096138087414e-05,
-      "loss": 54.6242,
-      "step": 71700
-    },
-    {
-      "epoch": 0.28971747395128417,
-      "grad_norm": 841.2785034179688,
-      "learning_rate": 4.4868842982000425e-05,
-      "loss": 92.4388,
-      "step": 71710
-    },
-    {
-      "epoch": 0.2897578752166518,
-      "grad_norm": 1951.40673828125,
-      "learning_rate": 4.486672419577339e-05,
-      "loss": 99.056,
-      "step": 71720
-    },
-    {
-      "epoch": 0.2897982764820194,
-      "grad_norm": 870.7053833007812,
-      "learning_rate": 4.486460502223434e-05,
-      "loss": 82.6444,
-      "step": 71730
-    },
-    {
-      "epoch": 0.28983867774738703,
-      "grad_norm": 1036.0379638671875,
-      "learning_rate": 4.4862485461424585e-05,
-      "loss": 84.8484,
-      "step": 71740
-    },
-    {
-      "epoch": 0.28987907901275467,
-      "grad_norm": 1319.0260009765625,
-      "learning_rate": 4.4860365513385456e-05,
-      "loss": 65.2683,
-      "step": 71750
-    },
-    {
-      "epoch": 0.2899194802781223,
-      "grad_norm": 623.9202270507812,
-      "learning_rate": 4.4858245178158276e-05,
-      "loss": 51.1636,
-      "step": 71760
-    },
-    {
-      "epoch": 0.28995988154348995,
-      "grad_norm": 1197.3074951171875,
-      "learning_rate": 4.4856124455784375e-05,
-      "loss": 57.0653,
-      "step": 71770
-    },
-    {
-      "epoch": 0.2900002828088576,
-      "grad_norm": 951.6260986328125,
-      "learning_rate": 4.485400334630511e-05,
-      "loss": 79.302,
-      "step": 71780
-    },
-    {
-      "epoch": 0.29004068407422523,
-      "grad_norm": 714.1635131835938,
-      "learning_rate": 4.485188184976182e-05,
-      "loss": 64.435,
-      "step": 71790
-    },
-    {
-      "epoch": 0.2900810853395928,
-      "grad_norm": 1034.5103759765625,
-      "learning_rate": 4.484975996619589e-05,
-      "loss": 69.878,
-      "step": 71800
-    },
-    {
-      "epoch": 0.29012148660496045,
-      "grad_norm": 2745.012939453125,
-      "learning_rate": 4.484763769564866e-05,
-      "loss": 46.8337,
-      "step": 71810
-    },
-    {
-      "epoch": 0.2901618878703281,
-      "grad_norm": 905.9317626953125,
-      "learning_rate": 4.4845515038161515e-05,
-      "loss": 72.7523,
-      "step": 71820
-    },
-    {
-      "epoch": 0.29020228913569573,
-      "grad_norm": 596.106689453125,
-      "learning_rate": 4.484339199377583e-05,
-      "loss": 56.2039,
-      "step": 71830
-    },
-    {
-      "epoch": 0.2902426904010634,
-      "grad_norm": 770.546875,
-      "learning_rate": 4.484126856253301e-05,
-      "loss": 63.8412,
-      "step": 71840
-    },
-    {
-      "epoch": 0.290283091666431,
-      "grad_norm": 1505.4610595703125,
-      "learning_rate": 4.483914474447445e-05,
-      "loss": 84.4822,
-      "step": 71850
-    },
-    {
-      "epoch": 0.2903234929317986,
-      "grad_norm": 925.32421875,
-      "learning_rate": 4.483702053964154e-05,
-      "loss": 58.1121,
-      "step": 71860
-    },
-    {
-      "epoch": 0.29036389419716624,
-      "grad_norm": 754.6409912109375,
-      "learning_rate": 4.4834895948075704e-05,
-      "loss": 94.093,
-      "step": 71870
-    },
-    {
-      "epoch": 0.2904042954625339,
-      "grad_norm": 296.52349853515625,
-      "learning_rate": 4.483277096981836e-05,
-      "loss": 70.3271,
-      "step": 71880
-    },
-    {
-      "epoch": 0.2904446967279015,
-      "grad_norm": 387.88568115234375,
-      "learning_rate": 4.483064560491094e-05,
-      "loss": 46.6679,
-      "step": 71890
-    },
-    {
-      "epoch": 0.29048509799326916,
-      "grad_norm": 628.3577270507812,
-      "learning_rate": 4.482851985339487e-05,
-      "loss": 73.2846,
-      "step": 71900
-    },
-    {
-      "epoch": 0.2905254992586368,
-      "grad_norm": 429.7192077636719,
-      "learning_rate": 4.4826393715311595e-05,
-      "loss": 54.1423,
-      "step": 71910
-    },
-    {
-      "epoch": 0.29056590052400444,
-      "grad_norm": 933.8616943359375,
-      "learning_rate": 4.482426719070258e-05,
-      "loss": 82.952,
-      "step": 71920
-    },
-    {
-      "epoch": 0.290606301789372,
-      "grad_norm": 1550.8494873046875,
-      "learning_rate": 4.482214027960925e-05,
-      "loss": 60.8338,
-      "step": 71930
-    },
-    {
-      "epoch": 0.29064670305473966,
-      "grad_norm": 725.9008178710938,
-      "learning_rate": 4.48200129820731e-05,
-      "loss": 58.8863,
-      "step": 71940
-    },
-    {
-      "epoch": 0.2906871043201073,
-      "grad_norm": 426.5110168457031,
-      "learning_rate": 4.481788529813559e-05,
-      "loss": 84.5091,
-      "step": 71950
-    },
-    {
-      "epoch": 0.29072750558547494,
-      "grad_norm": 735.1624145507812,
-      "learning_rate": 4.481575722783821e-05,
-      "loss": 71.0047,
-      "step": 71960
-    },
-    {
-      "epoch": 0.2907679068508426,
-      "grad_norm": 1465.6466064453125,
-      "learning_rate": 4.481362877122243e-05,
-      "loss": 93.0289,
-      "step": 71970
-    },
-    {
-      "epoch": 0.2908083081162102,
-      "grad_norm": 448.85919189453125,
-      "learning_rate": 4.481149992832977e-05,
-      "loss": 49.5786,
-      "step": 71980
-    },
-    {
-      "epoch": 0.2908487093815778,
-      "grad_norm": 1719.1485595703125,
-      "learning_rate": 4.4809370699201706e-05,
-      "loss": 99.739,
-      "step": 71990
-    },
-    {
-      "epoch": 0.29088911064694545,
-      "grad_norm": 2338.972900390625,
-      "learning_rate": 4.480724108387977e-05,
-      "loss": 94.2316,
-      "step": 72000
-    },
-    {
-      "epoch": 0.2909295119123131,
-      "grad_norm": 495.1227722167969,
-      "learning_rate": 4.480511108240547e-05,
-      "loss": 57.7428,
-      "step": 72010
-    },
-    {
-      "epoch": 0.29096991317768073,
-      "grad_norm": 516.1085205078125,
-      "learning_rate": 4.480298069482033e-05,
-      "loss": 65.6927,
-      "step": 72020
-    },
-    {
-      "epoch": 0.29101031444304837,
-      "grad_norm": 592.4568481445312,
-      "learning_rate": 4.480084992116589e-05,
-      "loss": 54.2324,
-      "step": 72030
-    },
-    {
-      "epoch": 0.291050715708416,
-      "grad_norm": 661.2754516601562,
-      "learning_rate": 4.479871876148368e-05,
-      "loss": 101.6369,
-      "step": 72040
-    },
-    {
-      "epoch": 0.2910911169737836,
-      "grad_norm": 1018.503662109375,
-      "learning_rate": 4.479658721581527e-05,
-      "loss": 87.5494,
-      "step": 72050
-    },
-    {
-      "epoch": 0.29113151823915123,
-      "grad_norm": 1803.326416015625,
-      "learning_rate": 4.479445528420218e-05,
-      "loss": 121.1224,
-      "step": 72060
-    },
-    {
-      "epoch": 0.2911719195045189,
-      "grad_norm": 773.8688354492188,
-      "learning_rate": 4.479232296668601e-05,
-      "loss": 40.3872,
-      "step": 72070
-    },
-    {
-      "epoch": 0.2912123207698865,
-      "grad_norm": 598.6536865234375,
-      "learning_rate": 4.4790190263308306e-05,
-      "loss": 61.5503,
-      "step": 72080
-    },
-    {
-      "epoch": 0.29125272203525415,
-      "grad_norm": 1390.3765869140625,
-      "learning_rate": 4.478805717411066e-05,
-      "loss": 75.8073,
-      "step": 72090
-    },
-    {
-      "epoch": 0.2912931233006218,
-      "grad_norm": 1810.283447265625,
-      "learning_rate": 4.478592369913465e-05,
-      "loss": 66.2582,
-      "step": 72100
-    },
-    {
-      "epoch": 0.29133352456598943,
-      "grad_norm": 585.9959106445312,
-      "learning_rate": 4.478378983842186e-05,
-      "loss": 60.1694,
-      "step": 72110
-    },
-    {
-      "epoch": 0.291373925831357,
-      "grad_norm": 589.87109375,
-      "learning_rate": 4.4781655592013914e-05,
-      "loss": 69.5412,
-      "step": 72120
-    },
-    {
-      "epoch": 0.29141432709672466,
-      "grad_norm": 829.7304077148438,
-      "learning_rate": 4.47795209599524e-05,
-      "loss": 57.3538,
-      "step": 72130
-    },
-    {
-      "epoch": 0.2914547283620923,
-      "grad_norm": 331.1748962402344,
-      "learning_rate": 4.477738594227895e-05,
-      "loss": 54.7226,
-      "step": 72140
-    },
-    {
-      "epoch": 0.29149512962745994,
-      "grad_norm": 1150.1204833984375,
-      "learning_rate": 4.4775250539035174e-05,
-      "loss": 101.6825,
-      "step": 72150
-    },
-    {
-      "epoch": 0.2915355308928276,
-      "grad_norm": 1445.37255859375,
-      "learning_rate": 4.477311475026271e-05,
-      "loss": 73.3006,
-      "step": 72160
-    },
-    {
-      "epoch": 0.2915759321581952,
-      "grad_norm": 1057.3179931640625,
-      "learning_rate": 4.4770978576003196e-05,
-      "loss": 87.7836,
-      "step": 72170
-    },
-    {
-      "epoch": 0.2916163334235628,
-      "grad_norm": 258.7021484375,
-      "learning_rate": 4.4768842016298275e-05,
-      "loss": 47.9227,
-      "step": 72180
-    },
-    {
-      "epoch": 0.29165673468893044,
-      "grad_norm": 572.38623046875,
-      "learning_rate": 4.4766705071189595e-05,
-      "loss": 94.038,
-      "step": 72190
-    },
-    {
-      "epoch": 0.2916971359542981,
-      "grad_norm": 1339.6978759765625,
-      "learning_rate": 4.4764567740718825e-05,
-      "loss": 55.9968,
-      "step": 72200
-    },
-    {
-      "epoch": 0.2917375372196657,
-      "grad_norm": 671.230712890625,
-      "learning_rate": 4.4762430024927636e-05,
-      "loss": 83.76,
-      "step": 72210
-    },
-    {
-      "epoch": 0.29177793848503336,
-      "grad_norm": 459.9236145019531,
-      "learning_rate": 4.476029192385769e-05,
-      "loss": 52.4123,
-      "step": 72220
-    },
-    {
-      "epoch": 0.291818339750401,
-      "grad_norm": 1072.2008056640625,
-      "learning_rate": 4.4758153437550684e-05,
-      "loss": 61.8481,
-      "step": 72230
-    },
-    {
-      "epoch": 0.29185874101576864,
-      "grad_norm": 1238.638671875,
-      "learning_rate": 4.475601456604831e-05,
-      "loss": 101.6763,
-      "step": 72240
-    },
-    {
-      "epoch": 0.2918991422811362,
-      "grad_norm": 1013.4456787109375,
-      "learning_rate": 4.4753875309392266e-05,
-      "loss": 83.2611,
-      "step": 72250
-    },
-    {
-      "epoch": 0.29193954354650387,
-      "grad_norm": 609.5685424804688,
-      "learning_rate": 4.4751735667624237e-05,
-      "loss": 83.8193,
-      "step": 72260
-    },
-    {
-      "epoch": 0.2919799448118715,
-      "grad_norm": 683.1033325195312,
-      "learning_rate": 4.474959564078596e-05,
-      "loss": 51.5649,
-      "step": 72270
-    },
-    {
-      "epoch": 0.29202034607723915,
-      "grad_norm": 799.1218872070312,
-      "learning_rate": 4.4747455228919146e-05,
-      "loss": 48.9159,
-      "step": 72280
-    },
-    {
-      "epoch": 0.2920607473426068,
-      "grad_norm": 316.8408508300781,
-      "learning_rate": 4.4745314432065535e-05,
-      "loss": 80.174,
-      "step": 72290
-    },
-    {
-      "epoch": 0.2921011486079744,
-      "grad_norm": 549.0940551757812,
-      "learning_rate": 4.474317325026684e-05,
-      "loss": 56.1946,
-      "step": 72300
-    },
-    {
-      "epoch": 0.292141549873342,
-      "grad_norm": 591.0458984375,
-      "learning_rate": 4.474103168356483e-05,
-      "loss": 101.0572,
-      "step": 72310
-    },
-    {
-      "epoch": 0.29218195113870965,
-      "grad_norm": 3971.22802734375,
-      "learning_rate": 4.4738889732001234e-05,
-      "loss": 65.9609,
-      "step": 72320
-    },
-    {
-      "epoch": 0.2922223524040773,
-      "grad_norm": 947.2832641601562,
-      "learning_rate": 4.473674739561783e-05,
-      "loss": 108.9696,
-      "step": 72330
-    },
-    {
-      "epoch": 0.29226275366944493,
-      "grad_norm": 745.0115356445312,
-      "learning_rate": 4.473460467445637e-05,
-      "loss": 53.7305,
-      "step": 72340
-    },
-    {
-      "epoch": 0.29230315493481257,
-      "grad_norm": 1212.1033935546875,
-      "learning_rate": 4.473246156855863e-05,
-      "loss": 85.5423,
-      "step": 72350
-    },
-    {
-      "epoch": 0.2923435562001802,
-      "grad_norm": 1858.6473388671875,
-      "learning_rate": 4.473031807796639e-05,
-      "loss": 84.1671,
-      "step": 72360
-    },
-    {
-      "epoch": 0.2923839574655478,
-      "grad_norm": 1183.3797607421875,
-      "learning_rate": 4.4728174202721444e-05,
-      "loss": 85.8308,
-      "step": 72370
-    },
-    {
-      "epoch": 0.29242435873091543,
-      "grad_norm": 564.0057373046875,
-      "learning_rate": 4.472602994286559e-05,
-      "loss": 62.2768,
-      "step": 72380
-    },
-    {
-      "epoch": 0.2924647599962831,
-      "grad_norm": 740.6309204101562,
-      "learning_rate": 4.472388529844062e-05,
-      "loss": 46.9198,
-      "step": 72390
-    },
-    {
-      "epoch": 0.2925051612616507,
-      "grad_norm": 1595.9669189453125,
-      "learning_rate": 4.4721740269488355e-05,
-      "loss": 61.2028,
-      "step": 72400
-    },
-    {
-      "epoch": 0.29254556252701835,
-      "grad_norm": 1151.5087890625,
-      "learning_rate": 4.4719594856050604e-05,
-      "loss": 62.4049,
-      "step": 72410
-    },
-    {
-      "epoch": 0.292585963792386,
-      "grad_norm": 475.585205078125,
-      "learning_rate": 4.4717449058169216e-05,
-      "loss": 50.4637,
-      "step": 72420
-    },
-    {
-      "epoch": 0.29262636505775363,
-      "grad_norm": 519.3822631835938,
-      "learning_rate": 4.471530287588599e-05,
-      "loss": 77.0086,
-      "step": 72430
-    },
-    {
-      "epoch": 0.2926667663231212,
-      "grad_norm": 993.6683959960938,
-      "learning_rate": 4.471315630924279e-05,
-      "loss": 76.9779,
-      "step": 72440
-    },
-    {
-      "epoch": 0.29270716758848886,
-      "grad_norm": 1110.651611328125,
-      "learning_rate": 4.4711009358281456e-05,
-      "loss": 73.6586,
-      "step": 72450
-    },
-    {
-      "epoch": 0.2927475688538565,
-      "grad_norm": 703.5302124023438,
-      "learning_rate": 4.4708862023043854e-05,
-      "loss": 68.1201,
-      "step": 72460
-    },
-    {
-      "epoch": 0.29278797011922414,
-      "grad_norm": 773.6651611328125,
-      "learning_rate": 4.470671430357183e-05,
-      "loss": 62.8462,
-      "step": 72470
-    },
-    {
-      "epoch": 0.2928283713845918,
-      "grad_norm": 806.9835815429688,
-      "learning_rate": 4.470456619990727e-05,
-      "loss": 71.4623,
-      "step": 72480
-    },
-    {
-      "epoch": 0.2928687726499594,
-      "grad_norm": 805.1903686523438,
-      "learning_rate": 4.470241771209205e-05,
-      "loss": 75.0459,
-      "step": 72490
-    },
-    {
-      "epoch": 0.292909173915327,
-      "grad_norm": 408.85089111328125,
-      "learning_rate": 4.4700268840168045e-05,
-      "loss": 58.6624,
-      "step": 72500
-    },
-    {
-      "epoch": 0.29294957518069464,
-      "grad_norm": 701.9678344726562,
-      "learning_rate": 4.469811958417717e-05,
-      "loss": 55.822,
-      "step": 72510
-    },
-    {
-      "epoch": 0.2929899764460623,
-      "grad_norm": 563.7380981445312,
-      "learning_rate": 4.46959699441613e-05,
-      "loss": 49.1962,
-      "step": 72520
-    },
-    {
-      "epoch": 0.2930303777114299,
-      "grad_norm": 736.0975952148438,
-      "learning_rate": 4.469381992016236e-05,
-      "loss": 42.205,
-      "step": 72530
-    },
-    {
-      "epoch": 0.29307077897679756,
-      "grad_norm": 375.850830078125,
-      "learning_rate": 4.469166951222227e-05,
-      "loss": 58.5806,
-      "step": 72540
-    },
-    {
-      "epoch": 0.2931111802421652,
-      "grad_norm": 828.135498046875,
-      "learning_rate": 4.4689518720382937e-05,
-      "loss": 69.4176,
-      "step": 72550
-    },
-    {
-      "epoch": 0.29315158150753284,
-      "grad_norm": 955.2371215820312,
-      "learning_rate": 4.46873675446863e-05,
-      "loss": 100.2446,
-      "step": 72560
-    },
-    {
-      "epoch": 0.29319198277290043,
-      "grad_norm": 808.4039916992188,
-      "learning_rate": 4.468521598517429e-05,
-      "loss": 79.4082,
-      "step": 72570
-    },
-    {
-      "epoch": 0.29323238403826807,
-      "grad_norm": 563.0283813476562,
-      "learning_rate": 4.468306404188887e-05,
-      "loss": 33.0342,
-      "step": 72580
-    },
-    {
-      "epoch": 0.2932727853036357,
-      "grad_norm": 2005.7269287109375,
-      "learning_rate": 4.468091171487197e-05,
-      "loss": 87.8131,
-      "step": 72590
-    },
-    {
-      "epoch": 0.29331318656900335,
-      "grad_norm": 1091.245849609375,
-      "learning_rate": 4.4678759004165584e-05,
-      "loss": 60.274,
-      "step": 72600
-    },
-    {
-      "epoch": 0.293353587834371,
-      "grad_norm": 773.4925537109375,
-      "learning_rate": 4.467660590981165e-05,
-      "loss": 43.7714,
-      "step": 72610
-    },
-    {
-      "epoch": 0.29339398909973863,
-      "grad_norm": 414.9910583496094,
-      "learning_rate": 4.4674452431852155e-05,
-      "loss": 43.9692,
-      "step": 72620
-    },
-    {
-      "epoch": 0.2934343903651062,
-      "grad_norm": 974.5853881835938,
-      "learning_rate": 4.467229857032907e-05,
-      "loss": 73.6409,
-      "step": 72630
-    },
-    {
-      "epoch": 0.29347479163047385,
-      "grad_norm": 879.56494140625,
-      "learning_rate": 4.4670144325284414e-05,
-      "loss": 103.894,
-      "step": 72640
-    },
-    {
-      "epoch": 0.2935151928958415,
-      "grad_norm": 522.8005981445312,
-      "learning_rate": 4.466798969676015e-05,
-      "loss": 108.3693,
-      "step": 72650
-    },
-    {
-      "epoch": 0.29355559416120913,
-      "grad_norm": 680.3546142578125,
-      "learning_rate": 4.4665834684798316e-05,
-      "loss": 58.7339,
-      "step": 72660
-    },
-    {
-      "epoch": 0.2935959954265768,
-      "grad_norm": 639.8063354492188,
-      "learning_rate": 4.4663679289440895e-05,
-      "loss": 69.8376,
-      "step": 72670
-    },
-    {
-      "epoch": 0.2936363966919444,
-      "grad_norm": 279.01385498046875,
-      "learning_rate": 4.466152351072994e-05,
-      "loss": 60.8839,
-      "step": 72680
-    },
-    {
-      "epoch": 0.293676797957312,
-      "grad_norm": 744.683349609375,
-      "learning_rate": 4.465936734870745e-05,
-      "loss": 59.2348,
-      "step": 72690
-    },
-    {
-      "epoch": 0.29371719922267964,
-      "grad_norm": 884.4476928710938,
-      "learning_rate": 4.465721080341547e-05,
-      "loss": 55.9587,
-      "step": 72700
-    },
-    {
-      "epoch": 0.2937576004880473,
-      "grad_norm": 736.0975341796875,
-      "learning_rate": 4.465505387489606e-05,
-      "loss": 85.0535,
-      "step": 72710
-    },
-    {
-      "epoch": 0.2937980017534149,
-      "grad_norm": 1129.6541748046875,
-      "learning_rate": 4.465289656319124e-05,
-      "loss": 166.6917,
-      "step": 72720
-    },
-    {
-      "epoch": 0.29383840301878256,
-      "grad_norm": 918.02587890625,
-      "learning_rate": 4.465073886834309e-05,
-      "loss": 80.1093,
-      "step": 72730
-    },
-    {
-      "epoch": 0.2938788042841502,
-      "grad_norm": 741.1550903320312,
-      "learning_rate": 4.464858079039367e-05,
-      "loss": 62.3389,
-      "step": 72740
-    },
-    {
-      "epoch": 0.29391920554951784,
-      "grad_norm": 1366.1309814453125,
-      "learning_rate": 4.464642232938505e-05,
-      "loss": 127.8192,
-      "step": 72750
-    },
-    {
-      "epoch": 0.2939596068148854,
-      "grad_norm": 691.9977416992188,
-      "learning_rate": 4.464426348535931e-05,
-      "loss": 71.4388,
-      "step": 72760
-    },
-    {
-      "epoch": 0.29400000808025306,
-      "grad_norm": 743.5264282226562,
-      "learning_rate": 4.464210425835854e-05,
-      "loss": 77.9983,
-      "step": 72770
-    },
-    {
-      "epoch": 0.2940404093456207,
-      "grad_norm": 570.9989624023438,
-      "learning_rate": 4.463994464842484e-05,
-      "loss": 66.5061,
-      "step": 72780
-    },
-    {
-      "epoch": 0.29408081061098834,
-      "grad_norm": 419.94793701171875,
-      "learning_rate": 4.46377846556003e-05,
-      "loss": 80.6933,
-      "step": 72790
-    },
-    {
-      "epoch": 0.294121211876356,
-      "grad_norm": 399.4372253417969,
-      "learning_rate": 4.4635624279927044e-05,
-      "loss": 82.7092,
-      "step": 72800
-    },
-    {
-      "epoch": 0.2941616131417236,
-      "grad_norm": 1092.372802734375,
-      "learning_rate": 4.463346352144718e-05,
-      "loss": 61.0069,
-      "step": 72810
-    },
-    {
-      "epoch": 0.2942020144070912,
-      "grad_norm": 1244.2633056640625,
-      "learning_rate": 4.463130238020285e-05,
-      "loss": 69.0925,
-      "step": 72820
-    },
-    {
-      "epoch": 0.29424241567245885,
-      "grad_norm": 3014.16796875,
-      "learning_rate": 4.4629140856236155e-05,
-      "loss": 95.4817,
-      "step": 72830
-    },
-    {
-      "epoch": 0.2942828169378265,
-      "grad_norm": 855.9567260742188,
-      "learning_rate": 4.462697894958926e-05,
-      "loss": 79.4187,
-      "step": 72840
-    },
-    {
-      "epoch": 0.2943232182031941,
-      "grad_norm": 588.0103149414062,
-      "learning_rate": 4.4624816660304314e-05,
-      "loss": 46.0367,
-      "step": 72850
-    },
-    {
-      "epoch": 0.29436361946856177,
-      "grad_norm": 2154.02734375,
-      "learning_rate": 4.4622653988423455e-05,
-      "loss": 116.9557,
-      "step": 72860
-    },
-    {
-      "epoch": 0.2944040207339294,
-      "grad_norm": 743.9976196289062,
-      "learning_rate": 4.462049093398885e-05,
-      "loss": 52.3771,
-      "step": 72870
-    },
-    {
-      "epoch": 0.29444442199929705,
-      "grad_norm": 1056.8822021484375,
-      "learning_rate": 4.461832749704268e-05,
-      "loss": 70.0739,
-      "step": 72880
-    },
-    {
-      "epoch": 0.29448482326466463,
-      "grad_norm": 860.94580078125,
-      "learning_rate": 4.461616367762711e-05,
-      "loss": 53.0582,
-      "step": 72890
-    },
-    {
-      "epoch": 0.29452522453003227,
-      "grad_norm": 319.7754821777344,
-      "learning_rate": 4.4613999475784336e-05,
-      "loss": 50.5095,
-      "step": 72900
-    },
-    {
-      "epoch": 0.2945656257953999,
-      "grad_norm": 1478.31982421875,
-      "learning_rate": 4.4611834891556534e-05,
-      "loss": 94.7775,
-      "step": 72910
-    },
-    {
-      "epoch": 0.29460602706076755,
-      "grad_norm": 548.7947998046875,
-      "learning_rate": 4.460966992498593e-05,
-      "loss": 96.3625,
-      "step": 72920
-    },
-    {
-      "epoch": 0.2946464283261352,
-      "grad_norm": 567.1472778320312,
-      "learning_rate": 4.46075045761147e-05,
-      "loss": 66.8417,
-      "step": 72930
-    },
-    {
-      "epoch": 0.29468682959150283,
-      "grad_norm": 510.6759033203125,
-      "learning_rate": 4.460533884498509e-05,
-      "loss": 43.7635,
-      "step": 72940
-    },
-    {
-      "epoch": 0.2947272308568704,
-      "grad_norm": 1200.787353515625,
-      "learning_rate": 4.460317273163929e-05,
-      "loss": 79.7278,
-      "step": 72950
-    },
-    {
-      "epoch": 0.29476763212223805,
-      "grad_norm": 1466.562255859375,
-      "learning_rate": 4.460100623611955e-05,
-      "loss": 80.256,
-      "step": 72960
-    },
-    {
-      "epoch": 0.2948080333876057,
-      "grad_norm": 1230.124267578125,
-      "learning_rate": 4.45988393584681e-05,
-      "loss": 94.8147,
-      "step": 72970
-    },
-    {
-      "epoch": 0.29484843465297333,
-      "grad_norm": 1354.7569580078125,
-      "learning_rate": 4.4596672098727195e-05,
-      "loss": 79.9409,
-      "step": 72980
-    },
-    {
-      "epoch": 0.294888835918341,
-      "grad_norm": 486.84246826171875,
-      "learning_rate": 4.459450445693907e-05,
-      "loss": 48.9059,
-      "step": 72990
-    },
-    {
-      "epoch": 0.2949292371837086,
-      "grad_norm": 606.8703002929688,
-      "learning_rate": 4.4592336433146e-05,
-      "loss": 56.8745,
-      "step": 73000
-    },
-    {
-      "epoch": 0.2949696384490762,
-      "grad_norm": 1210.473388671875,
-      "learning_rate": 4.459016802739023e-05,
-      "loss": 81.2079,
-      "step": 73010
-    },
-    {
-      "epoch": 0.29501003971444384,
-      "grad_norm": 467.654296875,
-      "learning_rate": 4.458799923971406e-05,
-      "loss": 71.5637,
-      "step": 73020
-    },
-    {
-      "epoch": 0.2950504409798115,
-      "grad_norm": 1085.084228515625,
-      "learning_rate": 4.4585830070159764e-05,
-      "loss": 59.4551,
-      "step": 73030
-    },
-    {
-      "epoch": 0.2950908422451791,
-      "grad_norm": 1020.708984375,
-      "learning_rate": 4.458366051876962e-05,
-      "loss": 67.9901,
-      "step": 73040
-    },
-    {
-      "epoch": 0.29513124351054676,
-      "grad_norm": 413.11859130859375,
-      "learning_rate": 4.458149058558594e-05,
-      "loss": 94.3448,
-      "step": 73050
-    },
-    {
-      "epoch": 0.2951716447759144,
-      "grad_norm": 344.1728210449219,
-      "learning_rate": 4.457932027065102e-05,
-      "loss": 42.2916,
-      "step": 73060
-    },
-    {
-      "epoch": 0.29521204604128204,
-      "grad_norm": 0.0,
-      "learning_rate": 4.457714957400716e-05,
-      "loss": 72.4169,
-      "step": 73070
-    },
-    {
-      "epoch": 0.2952524473066496,
-      "grad_norm": 1693.0472412109375,
-      "learning_rate": 4.45749784956967e-05,
-      "loss": 81.7885,
-      "step": 73080
-    },
-    {
-      "epoch": 0.29529284857201726,
-      "grad_norm": 676.3005981445312,
-      "learning_rate": 4.457280703576194e-05,
-      "loss": 69.0328,
-      "step": 73090
-    },
-    {
-      "epoch": 0.2953332498373849,
-      "grad_norm": 723.7576293945312,
-      "learning_rate": 4.457063519424525e-05,
-      "loss": 74.3711,
-      "step": 73100
-    },
-    {
-      "epoch": 0.29537365110275254,
-      "grad_norm": 397.1194152832031,
-      "learning_rate": 4.456846297118894e-05,
-      "loss": 65.241,
-      "step": 73110
-    },
-    {
-      "epoch": 0.2954140523681202,
-      "grad_norm": 686.8461303710938,
-      "learning_rate": 4.456629036663537e-05,
-      "loss": 66.4102,
-      "step": 73120
-    },
-    {
-      "epoch": 0.2954544536334878,
-      "grad_norm": 991.1904296875,
-      "learning_rate": 4.45641173806269e-05,
-      "loss": 79.4375,
-      "step": 73130
-    },
-    {
-      "epoch": 0.2954948548988554,
-      "grad_norm": 1777.7161865234375,
-      "learning_rate": 4.4561944013205885e-05,
-      "loss": 74.0132,
-      "step": 73140
-    },
-    {
-      "epoch": 0.29553525616422305,
-      "grad_norm": 1228.3016357421875,
-      "learning_rate": 4.45597702644147e-05,
-      "loss": 55.363,
-      "step": 73150
-    },
-    {
-      "epoch": 0.2955756574295907,
-      "grad_norm": 846.4840087890625,
-      "learning_rate": 4.455759613429573e-05,
-      "loss": 76.3519,
-      "step": 73160
-    },
-    {
-      "epoch": 0.29561605869495833,
-      "grad_norm": 639.7553100585938,
-      "learning_rate": 4.455542162289136e-05,
-      "loss": 83.1235,
-      "step": 73170
-    },
-    {
-      "epoch": 0.29565645996032597,
-      "grad_norm": 485.4346618652344,
-      "learning_rate": 4.455324673024396e-05,
-      "loss": 106.5092,
-      "step": 73180
-    },
-    {
-      "epoch": 0.2956968612256936,
-      "grad_norm": 1039.0308837890625,
-      "learning_rate": 4.4551071456395957e-05,
-      "loss": 55.8722,
-      "step": 73190
-    },
-    {
-      "epoch": 0.2957372624910612,
-      "grad_norm": 4481.669921875,
-      "learning_rate": 4.454889580138975e-05,
-      "loss": 92.0912,
-      "step": 73200
-    },
-    {
-      "epoch": 0.29577766375642883,
-      "grad_norm": 1369.180419921875,
-      "learning_rate": 4.454671976526776e-05,
-      "loss": 81.6197,
-      "step": 73210
-    },
-    {
-      "epoch": 0.2958180650217965,
-      "grad_norm": 1476.4033203125,
-      "learning_rate": 4.45445433480724e-05,
-      "loss": 60.5057,
-      "step": 73220
-    },
-    {
-      "epoch": 0.2958584662871641,
-      "grad_norm": 817.6678466796875,
-      "learning_rate": 4.45423665498461e-05,
-      "loss": 107.7844,
-      "step": 73230
-    },
-    {
-      "epoch": 0.29589886755253175,
-      "grad_norm": 1333.8826904296875,
-      "learning_rate": 4.4540189370631315e-05,
-      "loss": 100.9942,
-      "step": 73240
-    },
-    {
-      "epoch": 0.2959392688178994,
-      "grad_norm": 465.54510498046875,
-      "learning_rate": 4.453801181047047e-05,
-      "loss": 77.737,
-      "step": 73250
-    },
-    {
-      "epoch": 0.29597967008326703,
-      "grad_norm": 1762.44287109375,
-      "learning_rate": 4.4535833869406027e-05,
-      "loss": 58.7377,
-      "step": 73260
-    },
-    {
-      "epoch": 0.2960200713486346,
-      "grad_norm": 568.57958984375,
-      "learning_rate": 4.4533655547480444e-05,
-      "loss": 56.8368,
-      "step": 73270
-    },
-    {
-      "epoch": 0.29606047261400226,
-      "grad_norm": 1187.254638671875,
-      "learning_rate": 4.45314768447362e-05,
-      "loss": 88.9359,
-      "step": 73280
-    },
-    {
-      "epoch": 0.2961008738793699,
-      "grad_norm": 902.0029296875,
-      "learning_rate": 4.452929776121575e-05,
-      "loss": 69.4659,
-      "step": 73290
-    },
-    {
-      "epoch": 0.29614127514473754,
-      "grad_norm": 0.0,
-      "learning_rate": 4.452711829696158e-05,
-      "loss": 47.5744,
-      "step": 73300
-    },
-    {
-      "epoch": 0.2961816764101052,
-      "grad_norm": 407.5782775878906,
-      "learning_rate": 4.452493845201619e-05,
-      "loss": 63.7359,
-      "step": 73310
-    },
-    {
-      "epoch": 0.2962220776754728,
-      "grad_norm": 674.528076171875,
-      "learning_rate": 4.4522758226422076e-05,
-      "loss": 43.8548,
-      "step": 73320
-    },
-    {
-      "epoch": 0.2962624789408404,
-      "grad_norm": 679.5468139648438,
-      "learning_rate": 4.452057762022174e-05,
-      "loss": 56.5893,
-      "step": 73330
-    },
-    {
-      "epoch": 0.29630288020620804,
-      "grad_norm": 982.80517578125,
-      "learning_rate": 4.4518396633457696e-05,
-      "loss": 103.2029,
-      "step": 73340
-    },
-    {
-      "epoch": 0.2963432814715757,
-      "grad_norm": 529.539794921875,
-      "learning_rate": 4.4516215266172453e-05,
-      "loss": 36.5715,
-      "step": 73350
-    },
-    {
-      "epoch": 0.2963836827369433,
-      "grad_norm": 703.641357421875,
-      "learning_rate": 4.451403351840855e-05,
-      "loss": 68.3904,
-      "step": 73360
-    },
-    {
-      "epoch": 0.29642408400231096,
-      "grad_norm": 865.1505126953125,
-      "learning_rate": 4.451185139020852e-05,
-      "loss": 64.2055,
-      "step": 73370
-    },
-    {
-      "epoch": 0.2964644852676786,
-      "grad_norm": 718.0322265625,
-      "learning_rate": 4.4509668881614894e-05,
-      "loss": 102.1239,
-      "step": 73380
-    },
-    {
-      "epoch": 0.29650488653304624,
-      "grad_norm": 1524.7041015625,
-      "learning_rate": 4.450748599267024e-05,
-      "loss": 81.8768,
-      "step": 73390
-    },
-    {
-      "epoch": 0.2965452877984138,
-      "grad_norm": 378.5061950683594,
-      "learning_rate": 4.450530272341709e-05,
-      "loss": 56.3704,
-      "step": 73400
-    },
-    {
-      "epoch": 0.29658568906378147,
-      "grad_norm": 1193.3980712890625,
-      "learning_rate": 4.4503119073898024e-05,
-      "loss": 62.8788,
-      "step": 73410
-    },
-    {
-      "epoch": 0.2966260903291491,
-      "grad_norm": 972.7481079101562,
-      "learning_rate": 4.4500935044155626e-05,
-      "loss": 55.7444,
-      "step": 73420
-    },
-    {
-      "epoch": 0.29666649159451675,
-      "grad_norm": 496.94207763671875,
-      "learning_rate": 4.4498750634232445e-05,
-      "loss": 55.0117,
-      "step": 73430
-    },
-    {
-      "epoch": 0.2967068928598844,
-      "grad_norm": 690.003173828125,
-      "learning_rate": 4.449656584417108e-05,
-      "loss": 63.6232,
-      "step": 73440
-    },
-    {
-      "epoch": 0.296747294125252,
-      "grad_norm": 1239.76123046875,
-      "learning_rate": 4.449438067401413e-05,
-      "loss": 56.0802,
-      "step": 73450
-    },
-    {
-      "epoch": 0.2967876953906196,
-      "grad_norm": 1073.9417724609375,
-      "learning_rate": 4.44921951238042e-05,
-      "loss": 74.7659,
-      "step": 73460
-    },
-    {
-      "epoch": 0.29682809665598725,
-      "grad_norm": 612.8600463867188,
-      "learning_rate": 4.449000919358388e-05,
-      "loss": 49.4253,
-      "step": 73470
-    },
-    {
-      "epoch": 0.2968684979213549,
-      "grad_norm": 803.32763671875,
-      "learning_rate": 4.4487822883395805e-05,
-      "loss": 70.9369,
-      "step": 73480
-    },
-    {
-      "epoch": 0.29690889918672253,
-      "grad_norm": 999.6846313476562,
-      "learning_rate": 4.448563619328259e-05,
-      "loss": 77.0513,
-      "step": 73490
-    },
-    {
-      "epoch": 0.29694930045209017,
-      "grad_norm": 260.62261962890625,
-      "learning_rate": 4.448344912328686e-05,
-      "loss": 40.874,
-      "step": 73500
-    },
-    {
-      "epoch": 0.2969897017174578,
-      "grad_norm": 1337.423828125,
-      "learning_rate": 4.4481261673451255e-05,
-      "loss": 63.2176,
-      "step": 73510
-    },
-    {
-      "epoch": 0.2970301029828254,
-      "grad_norm": 328.9664306640625,
-      "learning_rate": 4.447907384381843e-05,
-      "loss": 58.7723,
-      "step": 73520
-    },
-    {
-      "epoch": 0.29707050424819303,
-      "grad_norm": 767.3320922851562,
-      "learning_rate": 4.447688563443103e-05,
-      "loss": 100.0262,
-      "step": 73530
-    },
-    {
-      "epoch": 0.2971109055135607,
-      "grad_norm": 251.83306884765625,
-      "learning_rate": 4.447469704533172e-05,
-      "loss": 58.1344,
-      "step": 73540
-    },
-    {
-      "epoch": 0.2971513067789283,
-      "grad_norm": 696.3643798828125,
-      "learning_rate": 4.4472508076563166e-05,
-      "loss": 80.2865,
-      "step": 73550
-    },
-    {
-      "epoch": 0.29719170804429595,
-      "grad_norm": 578.1966552734375,
-      "learning_rate": 4.447031872816804e-05,
-      "loss": 63.5288,
-      "step": 73560
-    },
-    {
-      "epoch": 0.2972321093096636,
-      "grad_norm": 1685.81982421875,
-      "learning_rate": 4.446812900018902e-05,
-      "loss": 98.5778,
-      "step": 73570
-    },
-    {
-      "epoch": 0.29727251057503123,
-      "grad_norm": 605.77978515625,
-      "learning_rate": 4.4465938892668814e-05,
-      "loss": 108.6232,
-      "step": 73580
-    },
-    {
-      "epoch": 0.2973129118403988,
-      "grad_norm": 425.3101501464844,
-      "learning_rate": 4.44637484056501e-05,
-      "loss": 81.9932,
-      "step": 73590
-    },
-    {
-      "epoch": 0.29735331310576646,
-      "grad_norm": 664.3129272460938,
-      "learning_rate": 4.4461557539175594e-05,
-      "loss": 80.7212,
-      "step": 73600
-    },
-    {
-      "epoch": 0.2973937143711341,
-      "grad_norm": 932.316650390625,
-      "learning_rate": 4.4459366293287994e-05,
-      "loss": 83.0597,
-      "step": 73610
-    },
-    {
-      "epoch": 0.29743411563650174,
-      "grad_norm": 1423.0401611328125,
-      "learning_rate": 4.445717466803004e-05,
-      "loss": 64.9128,
-      "step": 73620
-    },
-    {
-      "epoch": 0.2974745169018694,
-      "grad_norm": 572.2749633789062,
-      "learning_rate": 4.445498266344444e-05,
-      "loss": 49.8087,
-      "step": 73630
-    },
-    {
-      "epoch": 0.297514918167237,
-      "grad_norm": 1016.2572021484375,
-      "learning_rate": 4.445279027957395e-05,
-      "loss": 64.3123,
-      "step": 73640
-    },
-    {
-      "epoch": 0.2975553194326046,
-      "grad_norm": 1676.9608154296875,
-      "learning_rate": 4.4450597516461287e-05,
-      "loss": 79.8256,
-      "step": 73650
-    },
-    {
-      "epoch": 0.29759572069797224,
-      "grad_norm": 501.0017395019531,
-      "learning_rate": 4.444840437414922e-05,
-      "loss": 63.8892,
-      "step": 73660
-    },
-    {
-      "epoch": 0.2976361219633399,
-      "grad_norm": 1168.860107421875,
-      "learning_rate": 4.444621085268049e-05,
-      "loss": 59.1573,
-      "step": 73670
-    },
-    {
-      "epoch": 0.2976765232287075,
-      "grad_norm": 664.9098510742188,
-      "learning_rate": 4.444401695209788e-05,
-      "loss": 57.3835,
-      "step": 73680
-    },
-    {
-      "epoch": 0.29771692449407516,
-      "grad_norm": 1014.5247192382812,
-      "learning_rate": 4.4441822672444134e-05,
-      "loss": 89.5824,
-      "step": 73690
-    },
-    {
-      "epoch": 0.2977573257594428,
-      "grad_norm": 765.1936645507812,
-      "learning_rate": 4.443962801376205e-05,
-      "loss": 38.8063,
-      "step": 73700
-    },
-    {
-      "epoch": 0.29779772702481044,
-      "grad_norm": 928.1898193359375,
-      "learning_rate": 4.443743297609442e-05,
-      "loss": 62.8045,
-      "step": 73710
-    },
-    {
-      "epoch": 0.29783812829017803,
-      "grad_norm": 685.7420654296875,
-      "learning_rate": 4.443523755948401e-05,
-      "loss": 66.0851,
-      "step": 73720
-    },
-    {
-      "epoch": 0.29787852955554567,
-      "grad_norm": 881.26416015625,
-      "learning_rate": 4.443304176397365e-05,
-      "loss": 71.4241,
-      "step": 73730
-    },
-    {
-      "epoch": 0.2979189308209133,
-      "grad_norm": 1430.6566162109375,
-      "learning_rate": 4.443084558960613e-05,
-      "loss": 110.0562,
-      "step": 73740
-    },
-    {
-      "epoch": 0.29795933208628095,
-      "grad_norm": 476.3144226074219,
-      "learning_rate": 4.442864903642428e-05,
-      "loss": 63.0536,
-      "step": 73750
-    },
-    {
-      "epoch": 0.2979997333516486,
-      "grad_norm": 597.589111328125,
-      "learning_rate": 4.4426452104470903e-05,
-      "loss": 49.1393,
-      "step": 73760
-    },
-    {
-      "epoch": 0.29804013461701623,
-      "grad_norm": 673.9664306640625,
-      "learning_rate": 4.4424254793788844e-05,
-      "loss": 49.0798,
-      "step": 73770
-    },
-    {
-      "epoch": 0.2980805358823838,
-      "grad_norm": 1329.530029296875,
-      "learning_rate": 4.4422057104420946e-05,
-      "loss": 71.3935,
-      "step": 73780
-    },
-    {
-      "epoch": 0.29812093714775145,
-      "grad_norm": 690.584228515625,
-      "learning_rate": 4.4419859036410036e-05,
-      "loss": 73.6381,
-      "step": 73790
-    },
-    {
-      "epoch": 0.2981613384131191,
-      "grad_norm": 627.472900390625,
-      "learning_rate": 4.441766058979898e-05,
-      "loss": 82.9257,
-      "step": 73800
-    },
-    {
-      "epoch": 0.29820173967848673,
-      "grad_norm": 638.86279296875,
-      "learning_rate": 4.441546176463063e-05,
-      "loss": 57.4086,
-      "step": 73810
-    },
-    {
-      "epoch": 0.2982421409438544,
-      "grad_norm": 844.6746215820312,
-      "learning_rate": 4.441326256094787e-05,
-      "loss": 74.4755,
-      "step": 73820
-    },
-    {
-      "epoch": 0.298282542209222,
-      "grad_norm": 1176.2183837890625,
-      "learning_rate": 4.4411062978793545e-05,
-      "loss": 48.4431,
-      "step": 73830
-    },
-    {
-      "epoch": 0.2983229434745896,
-      "grad_norm": 762.2763671875,
-      "learning_rate": 4.4408863018210564e-05,
-      "loss": 54.5466,
-      "step": 73840
-    },
-    {
-      "epoch": 0.29836334473995724,
-      "grad_norm": 2829.393310546875,
-      "learning_rate": 4.44066626792418e-05,
-      "loss": 88.412,
-      "step": 73850
-    },
-    {
-      "epoch": 0.2984037460053249,
-      "grad_norm": 2878.249267578125,
-      "learning_rate": 4.440446196193016e-05,
-      "loss": 129.003,
-      "step": 73860
-    },
-    {
-      "epoch": 0.2984441472706925,
-      "grad_norm": 631.333984375,
-      "learning_rate": 4.440226086631854e-05,
-      "loss": 59.6416,
-      "step": 73870
-    },
-    {
-      "epoch": 0.29848454853606016,
-      "grad_norm": 858.2770385742188,
-      "learning_rate": 4.440005939244986e-05,
-      "loss": 54.7916,
-      "step": 73880
-    },
-    {
-      "epoch": 0.2985249498014278,
-      "grad_norm": 945.3146362304688,
-      "learning_rate": 4.439785754036703e-05,
-      "loss": 70.1292,
-      "step": 73890
-    },
-    {
-      "epoch": 0.29856535106679544,
-      "grad_norm": 915.803955078125,
-      "learning_rate": 4.439565531011299e-05,
-      "loss": 66.0254,
-      "step": 73900
-    },
-    {
-      "epoch": 0.298605752332163,
-      "grad_norm": 683.1613159179688,
-      "learning_rate": 4.4393452701730655e-05,
-      "loss": 55.0487,
-      "step": 73910
-    },
-    {
-      "epoch": 0.29864615359753066,
-      "grad_norm": 1310.735107421875,
-      "learning_rate": 4.439124971526297e-05,
-      "loss": 99.916,
-      "step": 73920
-    },
-    {
-      "epoch": 0.2986865548628983,
-      "grad_norm": 276.0152893066406,
-      "learning_rate": 4.4389046350752905e-05,
-      "loss": 46.5066,
-      "step": 73930
-    },
-    {
-      "epoch": 0.29872695612826594,
-      "grad_norm": 1337.242919921875,
-      "learning_rate": 4.438684260824339e-05,
-      "loss": 112.974,
-      "step": 73940
-    },
-    {
-      "epoch": 0.2987673573936336,
-      "grad_norm": 583.646728515625,
-      "learning_rate": 4.43846384877774e-05,
-      "loss": 49.2991,
-      "step": 73950
-    },
-    {
-      "epoch": 0.2988077586590012,
-      "grad_norm": 1229.034423828125,
-      "learning_rate": 4.4382433989397895e-05,
-      "loss": 56.1411,
-      "step": 73960
-    },
-    {
-      "epoch": 0.2988481599243688,
-      "grad_norm": 556.5376586914062,
-      "learning_rate": 4.4380229113147866e-05,
-      "loss": 69.2976,
-      "step": 73970
-    },
-    {
-      "epoch": 0.29888856118973645,
-      "grad_norm": 448.5850524902344,
-      "learning_rate": 4.437802385907029e-05,
-      "loss": 65.7849,
-      "step": 73980
-    },
-    {
-      "epoch": 0.2989289624551041,
-      "grad_norm": 412.0335388183594,
-      "learning_rate": 4.4375818227208164e-05,
-      "loss": 47.8312,
-      "step": 73990
-    },
-    {
-      "epoch": 0.2989693637204717,
-      "grad_norm": 549.283935546875,
-      "learning_rate": 4.4373612217604496e-05,
-      "loss": 60.2066,
-      "step": 74000
-    },
-    {
-      "epoch": 0.29900976498583937,
-      "grad_norm": 641.086181640625,
-      "learning_rate": 4.437140583030227e-05,
-      "loss": 81.0014,
-      "step": 74010
-    },
-    {
-      "epoch": 0.299050166251207,
-      "grad_norm": 470.0632629394531,
-      "learning_rate": 4.4369199065344525e-05,
-      "loss": 48.1158,
-      "step": 74020
-    },
-    {
-      "epoch": 0.29909056751657465,
-      "grad_norm": 479.7843933105469,
-      "learning_rate": 4.436699192277426e-05,
-      "loss": 62.3095,
-      "step": 74030
-    },
-    {
-      "epoch": 0.29913096878194223,
-      "grad_norm": 953.147705078125,
-      "learning_rate": 4.436478440263453e-05,
-      "loss": 69.0217,
-      "step": 74040
-    },
-    {
-      "epoch": 0.29917137004730987,
-      "grad_norm": 609.3707885742188,
-      "learning_rate": 4.436257650496834e-05,
-      "loss": 71.6211,
-      "step": 74050
-    },
-    {
-      "epoch": 0.2992117713126775,
-      "grad_norm": 1487.8802490234375,
-      "learning_rate": 4.436036822981877e-05,
-      "loss": 84.6675,
-      "step": 74060
-    },
-    {
-      "epoch": 0.29925217257804515,
-      "grad_norm": 894.4046630859375,
-      "learning_rate": 4.435815957722885e-05,
-      "loss": 51.4052,
-      "step": 74070
-    },
-    {
-      "epoch": 0.2992925738434128,
-      "grad_norm": 1185.888671875,
-      "learning_rate": 4.4355950547241645e-05,
-      "loss": 79.9474,
-      "step": 74080
-    },
-    {
-      "epoch": 0.29933297510878043,
-      "grad_norm": 2298.0419921875,
-      "learning_rate": 4.435374113990021e-05,
-      "loss": 62.4108,
-      "step": 74090
-    },
-    {
-      "epoch": 0.299373376374148,
-      "grad_norm": 1640.901611328125,
-      "learning_rate": 4.435153135524763e-05,
-      "loss": 108.3077,
-      "step": 74100
-    },
-    {
-      "epoch": 0.29941377763951565,
-      "grad_norm": 509.5463562011719,
-      "learning_rate": 4.434932119332699e-05,
-      "loss": 74.0899,
-      "step": 74110
-    },
-    {
-      "epoch": 0.2994541789048833,
-      "grad_norm": 1251.0087890625,
-      "learning_rate": 4.434711065418137e-05,
-      "loss": 117.1789,
-      "step": 74120
-    },
-    {
-      "epoch": 0.29949458017025093,
-      "grad_norm": 838.3406982421875,
-      "learning_rate": 4.434489973785386e-05,
-      "loss": 99.4823,
-      "step": 74130
-    },
-    {
-      "epoch": 0.2995349814356186,
-      "grad_norm": 924.2326049804688,
-      "learning_rate": 4.434268844438758e-05,
-      "loss": 60.6183,
-      "step": 74140
-    },
-    {
-      "epoch": 0.2995753827009862,
-      "grad_norm": 602.224365234375,
-      "learning_rate": 4.4340476773825625e-05,
-      "loss": 65.4386,
-      "step": 74150
-    },
-    {
-      "epoch": 0.2996157839663538,
-      "grad_norm": 1039.2386474609375,
-      "learning_rate": 4.433826472621112e-05,
-      "loss": 72.3676,
-      "step": 74160
-    },
-    {
-      "epoch": 0.29965618523172144,
-      "grad_norm": 1209.8310546875,
-      "learning_rate": 4.4336052301587185e-05,
-      "loss": 86.5389,
-      "step": 74170
-    },
-    {
-      "epoch": 0.2996965864970891,
-      "grad_norm": 664.8455200195312,
-      "learning_rate": 4.4333839499996954e-05,
-      "loss": 77.3693,
-      "step": 74180
-    },
-    {
-      "epoch": 0.2997369877624567,
-      "grad_norm": 373.7446594238281,
-      "learning_rate": 4.4331626321483575e-05,
-      "loss": 37.4625,
-      "step": 74190
-    },
-    {
-      "epoch": 0.29977738902782436,
-      "grad_norm": 587.56396484375,
-      "learning_rate": 4.432941276609018e-05,
-      "loss": 53.5302,
-      "step": 74200
-    },
-    {
-      "epoch": 0.299817790293192,
-      "grad_norm": 399.6770935058594,
-      "learning_rate": 4.432719883385994e-05,
-      "loss": 43.0968,
-      "step": 74210
-    },
-    {
-      "epoch": 0.29985819155855964,
-      "grad_norm": 455.2350158691406,
-      "learning_rate": 4.4324984524836e-05,
-      "loss": 65.6205,
-      "step": 74220
-    },
-    {
-      "epoch": 0.2998985928239272,
-      "grad_norm": 1229.9195556640625,
-      "learning_rate": 4.432276983906155e-05,
-      "loss": 59.1207,
-      "step": 74230
-    },
-    {
-      "epoch": 0.29993899408929486,
-      "grad_norm": 1179.552734375,
-      "learning_rate": 4.4320554776579747e-05,
-      "loss": 68.4554,
-      "step": 74240
-    },
-    {
-      "epoch": 0.2999793953546625,
-      "grad_norm": 563.2706298828125,
-      "learning_rate": 4.431833933743378e-05,
-      "loss": 51.8571,
-      "step": 74250
-    },
-    {
-      "epoch": 0.30001979662003014,
-      "grad_norm": 779.345947265625,
-      "learning_rate": 4.431612352166684e-05,
-      "loss": 81.9162,
-      "step": 74260
-    },
-    {
-      "epoch": 0.3000601978853978,
-      "grad_norm": 363.64324951171875,
-      "learning_rate": 4.431390732932213e-05,
-      "loss": 55.0468,
-      "step": 74270
-    },
-    {
-      "epoch": 0.3001005991507654,
-      "grad_norm": 1637.340087890625,
-      "learning_rate": 4.431169076044286e-05,
-      "loss": 81.3846,
-      "step": 74280
-    },
-    {
-      "epoch": 0.300141000416133,
-      "grad_norm": 563.9585571289062,
-      "learning_rate": 4.4309473815072225e-05,
-      "loss": 60.6945,
-      "step": 74290
-    },
-    {
-      "epoch": 0.30018140168150065,
-      "grad_norm": 749.6682739257812,
-      "learning_rate": 4.4307256493253457e-05,
-      "loss": 68.1788,
-      "step": 74300
-    },
-    {
-      "epoch": 0.3002218029468683,
-      "grad_norm": 5840.30810546875,
-      "learning_rate": 4.4305038795029794e-05,
-      "loss": 86.4264,
-      "step": 74310
-    },
-    {
-      "epoch": 0.30026220421223593,
-      "grad_norm": 910.1226806640625,
-      "learning_rate": 4.4302820720444456e-05,
-      "loss": 77.6066,
-      "step": 74320
-    },
-    {
-      "epoch": 0.30030260547760357,
-      "grad_norm": 709.5213012695312,
-      "learning_rate": 4.430060226954069e-05,
-      "loss": 120.0873,
-      "step": 74330
-    },
-    {
-      "epoch": 0.3003430067429712,
-      "grad_norm": 907.42431640625,
-      "learning_rate": 4.429838344236174e-05,
-      "loss": 82.4217,
-      "step": 74340
-    },
-    {
-      "epoch": 0.30038340800833885,
-      "grad_norm": 595.138671875,
-      "learning_rate": 4.4296164238950874e-05,
-      "loss": 77.8745,
-      "step": 74350
-    },
-    {
-      "epoch": 0.30042380927370643,
-      "grad_norm": 1740.2891845703125,
-      "learning_rate": 4.429394465935136e-05,
-      "loss": 70.9948,
-      "step": 74360
-    },
-    {
-      "epoch": 0.3004642105390741,
-      "grad_norm": 624.7880249023438,
-      "learning_rate": 4.429172470360645e-05,
-      "loss": 84.4048,
-      "step": 74370
-    },
-    {
-      "epoch": 0.3005046118044417,
-      "grad_norm": 632.6150512695312,
-      "learning_rate": 4.428950437175944e-05,
-      "loss": 65.9942,
-      "step": 74380
-    },
-    {
-      "epoch": 0.30054501306980935,
-      "grad_norm": 217.38211059570312,
-      "learning_rate": 4.428728366385361e-05,
-      "loss": 73.9153,
-      "step": 74390
-    },
-    {
-      "epoch": 0.300585414335177,
-      "grad_norm": 1739.6029052734375,
-      "learning_rate": 4.428506257993226e-05,
-      "loss": 78.2645,
-      "step": 74400
-    },
-    {
-      "epoch": 0.30062581560054463,
-      "grad_norm": 1436.2427978515625,
-      "learning_rate": 4.428284112003868e-05,
-      "loss": 71.5589,
-      "step": 74410
-    },
-    {
-      "epoch": 0.3006662168659122,
-      "grad_norm": 316.6112060546875,
-      "learning_rate": 4.428061928421618e-05,
-      "loss": 77.6914,
-      "step": 74420
-    },
-    {
-      "epoch": 0.30070661813127986,
-      "grad_norm": 711.2620849609375,
-      "learning_rate": 4.427839707250809e-05,
-      "loss": 77.8412,
-      "step": 74430
-    },
-    {
-      "epoch": 0.3007470193966475,
-      "grad_norm": 1620.897705078125,
-      "learning_rate": 4.427617448495772e-05,
-      "loss": 78.4815,
-      "step": 74440
-    },
-    {
-      "epoch": 0.30078742066201514,
-      "grad_norm": 519.2609252929688,
-      "learning_rate": 4.427395152160841e-05,
-      "loss": 41.6989,
-      "step": 74450
-    },
-    {
-      "epoch": 0.3008278219273828,
-      "grad_norm": 666.103515625,
-      "learning_rate": 4.427172818250349e-05,
-      "loss": 61.7915,
-      "step": 74460
-    },
-    {
-      "epoch": 0.3008682231927504,
-      "grad_norm": 611.0293579101562,
-      "learning_rate": 4.42695044676863e-05,
-      "loss": 63.6666,
-      "step": 74470
-    },
-    {
-      "epoch": 0.300908624458118,
-      "grad_norm": 789.2578125,
-      "learning_rate": 4.4267280377200205e-05,
-      "loss": 82.4072,
-      "step": 74480
-    },
-    {
-      "epoch": 0.30094902572348564,
-      "grad_norm": 1049.16650390625,
-      "learning_rate": 4.426505591108856e-05,
-      "loss": 67.3771,
-      "step": 74490
-    },
-    {
-      "epoch": 0.3009894269888533,
-      "grad_norm": 565.5245361328125,
-      "learning_rate": 4.426283106939474e-05,
-      "loss": 52.4517,
-      "step": 74500
-    },
-    {
-      "epoch": 0.3010298282542209,
-      "grad_norm": 1094.810546875,
-      "learning_rate": 4.42606058521621e-05,
-      "loss": 89.7887,
-      "step": 74510
-    },
-    {
-      "epoch": 0.30107022951958856,
-      "grad_norm": 938.249267578125,
-      "learning_rate": 4.425838025943403e-05,
-      "loss": 67.3641,
-      "step": 74520
-    },
-    {
-      "epoch": 0.3011106307849562,
-      "grad_norm": 857.2614135742188,
-      "learning_rate": 4.4256154291253925e-05,
-      "loss": 102.7043,
-      "step": 74530
-    },
-    {
-      "epoch": 0.30115103205032384,
-      "grad_norm": 520.708251953125,
-      "learning_rate": 4.4253927947665185e-05,
-      "loss": 66.7792,
-      "step": 74540
-    },
-    {
-      "epoch": 0.3011914333156914,
-      "grad_norm": 764.5960693359375,
-      "learning_rate": 4.42517012287112e-05,
-      "loss": 69.1683,
-      "step": 74550
-    },
-    {
-      "epoch": 0.30123183458105907,
-      "grad_norm": 1161.567138671875,
-      "learning_rate": 4.424947413443539e-05,
-      "loss": 75.7145,
-      "step": 74560
-    },
-    {
-      "epoch": 0.3012722358464267,
-      "grad_norm": 810.30126953125,
-      "learning_rate": 4.424724666488117e-05,
-      "loss": 60.572,
-      "step": 74570
-    },
-    {
-      "epoch": 0.30131263711179435,
-      "grad_norm": 1023.4842529296875,
-      "learning_rate": 4.424501882009198e-05,
-      "loss": 60.1247,
-      "step": 74580
-    },
-    {
-      "epoch": 0.301353038377162,
-      "grad_norm": 572.3932495117188,
-      "learning_rate": 4.424279060011123e-05,
-      "loss": 58.7828,
-      "step": 74590
-    },
-    {
-      "epoch": 0.3013934396425296,
-      "grad_norm": 1139.3538818359375,
-      "learning_rate": 4.4240562004982364e-05,
-      "loss": 89.5676,
-      "step": 74600
-    },
-    {
-      "epoch": 0.3014338409078972,
-      "grad_norm": 685.8850708007812,
-      "learning_rate": 4.423833303474884e-05,
-      "loss": 74.0729,
-      "step": 74610
-    },
-    {
-      "epoch": 0.30147424217326485,
-      "grad_norm": 358.7518005371094,
-      "learning_rate": 4.423610368945411e-05,
-      "loss": 64.4605,
-      "step": 74620
-    },
-    {
-      "epoch": 0.3015146434386325,
-      "grad_norm": 2760.231201171875,
-      "learning_rate": 4.423387396914164e-05,
-      "loss": 63.1559,
-      "step": 74630
-    },
-    {
-      "epoch": 0.30155504470400013,
-      "grad_norm": 1156.1490478515625,
-      "learning_rate": 4.423164387385489e-05,
-      "loss": 68.2249,
-      "step": 74640
-    },
-    {
-      "epoch": 0.30159544596936777,
-      "grad_norm": 397.64569091796875,
-      "learning_rate": 4.4229413403637345e-05,
-      "loss": 70.5398,
-      "step": 74650
-    },
-    {
-      "epoch": 0.3016358472347354,
-      "grad_norm": 488.1241760253906,
-      "learning_rate": 4.422718255853248e-05,
-      "loss": 63.8304,
-      "step": 74660
-    },
-    {
-      "epoch": 0.30167624850010305,
-      "grad_norm": 623.3941650390625,
-      "learning_rate": 4.42249513385838e-05,
-      "loss": 93.5955,
-      "step": 74670
-    },
-    {
-      "epoch": 0.30171664976547063,
-      "grad_norm": 584.9109497070312,
-      "learning_rate": 4.422271974383479e-05,
-      "loss": 63.7987,
-      "step": 74680
-    },
-    {
-      "epoch": 0.3017570510308383,
-      "grad_norm": 1291.9105224609375,
-      "learning_rate": 4.4220487774328964e-05,
-      "loss": 80.7659,
-      "step": 74690
-    },
-    {
-      "epoch": 0.3017974522962059,
-      "grad_norm": 921.5263671875,
-      "learning_rate": 4.421825543010983e-05,
-      "loss": 57.1126,
-      "step": 74700
-    },
-    {
-      "epoch": 0.30183785356157355,
-      "grad_norm": 840.3859252929688,
-      "learning_rate": 4.4216022711220916e-05,
-      "loss": 47.0774,
-      "step": 74710
-    },
-    {
-      "epoch": 0.3018782548269412,
-      "grad_norm": 1399.228759765625,
-      "learning_rate": 4.4213789617705746e-05,
-      "loss": 60.6638,
-      "step": 74720
-    },
-    {
-      "epoch": 0.30191865609230883,
-      "grad_norm": 574.983642578125,
-      "learning_rate": 4.421155614960785e-05,
-      "loss": 64.1311,
-      "step": 74730
-    },
-    {
-      "epoch": 0.3019590573576764,
-      "grad_norm": 1133.6746826171875,
-      "learning_rate": 4.420932230697079e-05,
-      "loss": 66.0676,
-      "step": 74740
-    },
-    {
-      "epoch": 0.30199945862304406,
-      "grad_norm": 2249.599853515625,
-      "learning_rate": 4.420708808983809e-05,
-      "loss": 72.025,
-      "step": 74750
-    },
-    {
-      "epoch": 0.3020398598884117,
-      "grad_norm": 837.6932983398438,
-      "learning_rate": 4.420485349825332e-05,
-      "loss": 68.1134,
-      "step": 74760
-    },
-    {
-      "epoch": 0.30208026115377934,
-      "grad_norm": 984.0842895507812,
-      "learning_rate": 4.4202618532260046e-05,
-      "loss": 91.3187,
-      "step": 74770
-    },
-    {
-      "epoch": 0.302120662419147,
-      "grad_norm": 727.6422729492188,
-      "learning_rate": 4.420038319190184e-05,
-      "loss": 63.3439,
-      "step": 74780
-    },
-    {
-      "epoch": 0.3021610636845146,
-      "grad_norm": 412.24908447265625,
-      "learning_rate": 4.4198147477222274e-05,
-      "loss": 60.6722,
-      "step": 74790
-    },
-    {
-      "epoch": 0.3022014649498822,
-      "grad_norm": 1043.6190185546875,
-      "learning_rate": 4.4195911388264946e-05,
-      "loss": 50.6668,
-      "step": 74800
-    },
-    {
-      "epoch": 0.30224186621524984,
-      "grad_norm": 432.61956787109375,
-      "learning_rate": 4.419367492507343e-05,
-      "loss": 53.3677,
-      "step": 74810
-    },
-    {
-      "epoch": 0.3022822674806175,
-      "grad_norm": 554.3751831054688,
-      "learning_rate": 4.419143808769135e-05,
-      "loss": 45.9557,
-      "step": 74820
-    },
-    {
-      "epoch": 0.3023226687459851,
-      "grad_norm": 880.563720703125,
-      "learning_rate": 4.4189200876162295e-05,
-      "loss": 79.8755,
-      "step": 74830
-    },
-    {
-      "epoch": 0.30236307001135276,
-      "grad_norm": 766.0863647460938,
-      "learning_rate": 4.41869632905299e-05,
-      "loss": 81.7201,
-      "step": 74840
-    },
-    {
-      "epoch": 0.3024034712767204,
-      "grad_norm": 512.21240234375,
-      "learning_rate": 4.418472533083777e-05,
-      "loss": 64.7566,
-      "step": 74850
-    },
-    {
-      "epoch": 0.30244387254208804,
-      "grad_norm": 1144.3148193359375,
-      "learning_rate": 4.418248699712955e-05,
-      "loss": 119.2652,
-      "step": 74860
-    },
-    {
-      "epoch": 0.30248427380745563,
-      "grad_norm": 535.4666137695312,
-      "learning_rate": 4.418024828944886e-05,
-      "loss": 60.0249,
-      "step": 74870
-    },
-    {
-      "epoch": 0.30252467507282327,
-      "grad_norm": 3031.205322265625,
-      "learning_rate": 4.417800920783937e-05,
-      "loss": 56.3927,
-      "step": 74880
-    },
-    {
-      "epoch": 0.3025650763381909,
-      "grad_norm": 1405.0643310546875,
-      "learning_rate": 4.4175769752344706e-05,
-      "loss": 78.4172,
-      "step": 74890
-    },
-    {
-      "epoch": 0.30260547760355855,
-      "grad_norm": 779.4258422851562,
-      "learning_rate": 4.417352992300854e-05,
-      "loss": 59.0052,
-      "step": 74900
-    },
-    {
-      "epoch": 0.3026458788689262,
-      "grad_norm": 1063.1361083984375,
-      "learning_rate": 4.4171289719874543e-05,
-      "loss": 66.4332,
-      "step": 74910
-    },
-    {
-      "epoch": 0.30268628013429383,
-      "grad_norm": 479.256103515625,
-      "learning_rate": 4.4169049142986376e-05,
-      "loss": 51.7122,
-      "step": 74920
-    },
-    {
-      "epoch": 0.3027266813996614,
-      "grad_norm": 482.245849609375,
-      "learning_rate": 4.416680819238773e-05,
-      "loss": 71.5557,
-      "step": 74930
-    },
-    {
-      "epoch": 0.30276708266502905,
-      "grad_norm": 1661.60791015625,
-      "learning_rate": 4.4164566868122286e-05,
-      "loss": 110.3273,
-      "step": 74940
-    },
-    {
-      "epoch": 0.3028074839303967,
-      "grad_norm": 763.4310913085938,
-      "learning_rate": 4.4162325170233745e-05,
-      "loss": 63.9054,
-      "step": 74950
-    },
-    {
-      "epoch": 0.30284788519576433,
-      "grad_norm": 581.7908325195312,
-      "learning_rate": 4.4160083098765815e-05,
-      "loss": 73.6799,
-      "step": 74960
-    },
-    {
-      "epoch": 0.302888286461132,
-      "grad_norm": 522.9912109375,
-      "learning_rate": 4.4157840653762196e-05,
-      "loss": 63.8081,
-      "step": 74970
-    },
-    {
-      "epoch": 0.3029286877264996,
-      "grad_norm": 671.7311401367188,
-      "learning_rate": 4.4155597835266616e-05,
-      "loss": 43.7649,
-      "step": 74980
-    },
-    {
-      "epoch": 0.30296908899186725,
-      "grad_norm": 655.3804931640625,
-      "learning_rate": 4.415335464332279e-05,
-      "loss": 68.5104,
-      "step": 74990
-    },
-    {
-      "epoch": 0.30300949025723484,
-      "grad_norm": 705.6524047851562,
-      "learning_rate": 4.415111107797445e-05,
-      "loss": 55.7934,
-      "step": 75000
-    },
-    {
-      "epoch": 0.3030498915226025,
-      "grad_norm": 2905.262939453125,
-      "learning_rate": 4.4148867139265345e-05,
-      "loss": 111.0617,
-      "step": 75010
-    },
-    {
-      "epoch": 0.3030902927879701,
-      "grad_norm": 972.8043823242188,
-      "learning_rate": 4.414662282723922e-05,
-      "loss": 63.5983,
-      "step": 75020
-    },
-    {
-      "epoch": 0.30313069405333776,
-      "grad_norm": 1290.5452880859375,
-      "learning_rate": 4.414437814193982e-05,
-      "loss": 58.9312,
-      "step": 75030
-    },
-    {
-      "epoch": 0.3031710953187054,
-      "grad_norm": 1295.46044921875,
-      "learning_rate": 4.414213308341092e-05,
-      "loss": 60.9342,
-      "step": 75040
-    },
-    {
-      "epoch": 0.30321149658407304,
-      "grad_norm": 778.5823364257812,
-      "learning_rate": 4.4139887651696265e-05,
-      "loss": 65.9136,
-      "step": 75050
-    },
-    {
-      "epoch": 0.3032518978494406,
-      "grad_norm": 515.2728271484375,
-      "learning_rate": 4.413764184683966e-05,
-      "loss": 73.2117,
-      "step": 75060
-    },
-    {
-      "epoch": 0.30329229911480826,
-      "grad_norm": 1737.145751953125,
-      "learning_rate": 4.413539566888487e-05,
-      "loss": 122.1578,
-      "step": 75070
-    },
-    {
-      "epoch": 0.3033327003801759,
-      "grad_norm": 606.1090698242188,
-      "learning_rate": 4.413314911787569e-05,
-      "loss": 62.4606,
-      "step": 75080
-    },
-    {
-      "epoch": 0.30337310164554354,
-      "grad_norm": 760.4569702148438,
-      "learning_rate": 4.413090219385592e-05,
-      "loss": 42.6394,
-      "step": 75090
-    },
-    {
-      "epoch": 0.3034135029109112,
-      "grad_norm": 809.2393188476562,
-      "learning_rate": 4.412865489686936e-05,
-      "loss": 61.619,
-      "step": 75100
-    },
-    {
-      "epoch": 0.3034539041762788,
-      "grad_norm": 555.9096069335938,
-      "learning_rate": 4.412640722695982e-05,
-      "loss": 79.0981,
-      "step": 75110
-    },
-    {
-      "epoch": 0.3034943054416464,
-      "grad_norm": 745.3788452148438,
-      "learning_rate": 4.4124159184171134e-05,
-      "loss": 80.1419,
-      "step": 75120
-    },
-    {
-      "epoch": 0.30353470670701405,
-      "grad_norm": 915.6812744140625,
-      "learning_rate": 4.412191076854711e-05,
-      "loss": 80.7349,
-      "step": 75130
-    },
-    {
-      "epoch": 0.3035751079723817,
-      "grad_norm": 857.411865234375,
-      "learning_rate": 4.41196619801316e-05,
-      "loss": 43.2552,
-      "step": 75140
-    },
-    {
-      "epoch": 0.3036155092377493,
-      "grad_norm": 1478.0841064453125,
-      "learning_rate": 4.4117412818968426e-05,
-      "loss": 71.5397,
-      "step": 75150
-    },
-    {
-      "epoch": 0.30365591050311697,
-      "grad_norm": 818.4799194335938,
-      "learning_rate": 4.411516328510145e-05,
-      "loss": 42.5693,
-      "step": 75160
-    },
-    {
-      "epoch": 0.3036963117684846,
-      "grad_norm": 1053.192626953125,
-      "learning_rate": 4.411291337857453e-05,
-      "loss": 67.0812,
-      "step": 75170
-    },
-    {
-      "epoch": 0.30373671303385225,
-      "grad_norm": 399.8765563964844,
-      "learning_rate": 4.4110663099431514e-05,
-      "loss": 65.8789,
-      "step": 75180
-    },
-    {
-      "epoch": 0.30377711429921983,
-      "grad_norm": 869.1029052734375,
-      "learning_rate": 4.41084124477163e-05,
-      "loss": 86.4271,
-      "step": 75190
-    },
-    {
-      "epoch": 0.30381751556458747,
-      "grad_norm": 561.1309814453125,
-      "learning_rate": 4.410616142347273e-05,
-      "loss": 65.8079,
-      "step": 75200
-    },
-    {
-      "epoch": 0.3038579168299551,
-      "grad_norm": 1289.060302734375,
-      "learning_rate": 4.410391002674471e-05,
-      "loss": 74.7418,
-      "step": 75210
-    },
-    {
-      "epoch": 0.30389831809532275,
-      "grad_norm": 566.0087280273438,
-      "learning_rate": 4.410165825757613e-05,
-      "loss": 74.0141,
-      "step": 75220
-    },
-    {
-      "epoch": 0.3039387193606904,
-      "grad_norm": 0.0,
-      "learning_rate": 4.409940611601089e-05,
-      "loss": 52.371,
-      "step": 75230
-    },
-    {
-      "epoch": 0.30397912062605803,
-      "grad_norm": 628.7798461914062,
-      "learning_rate": 4.409715360209289e-05,
-      "loss": 70.8489,
-      "step": 75240
-    },
-    {
-      "epoch": 0.3040195218914256,
-      "grad_norm": 644.141845703125,
-      "learning_rate": 4.4094900715866064e-05,
-      "loss": 41.9273,
-      "step": 75250
-    },
-    {
-      "epoch": 0.30405992315679325,
-      "grad_norm": 727.8847045898438,
-      "learning_rate": 4.40926474573743e-05,
-      "loss": 67.9712,
-      "step": 75260
-    },
-    {
-      "epoch": 0.3041003244221609,
-      "grad_norm": 0.0,
-      "learning_rate": 4.409039382666155e-05,
-      "loss": 52.6979,
-      "step": 75270
-    },
-    {
-      "epoch": 0.30414072568752853,
-      "grad_norm": 521.0494384765625,
-      "learning_rate": 4.4088139823771744e-05,
-      "loss": 48.4199,
-      "step": 75280
-    },
-    {
-      "epoch": 0.3041811269528962,
-      "grad_norm": 494.1226501464844,
-      "learning_rate": 4.408588544874882e-05,
-      "loss": 65.0988,
-      "step": 75290
-    },
-    {
-      "epoch": 0.3042215282182638,
-      "grad_norm": 513.5343627929688,
-      "learning_rate": 4.408363070163675e-05,
-      "loss": 42.8742,
-      "step": 75300
-    },
-    {
-      "epoch": 0.30426192948363145,
-      "grad_norm": 654.8699951171875,
-      "learning_rate": 4.408137558247946e-05,
-      "loss": 64.6675,
-      "step": 75310
-    },
-    {
-      "epoch": 0.30430233074899904,
-      "grad_norm": 1542.4345703125,
-      "learning_rate": 4.407912009132093e-05,
-      "loss": 58.0917,
-      "step": 75320
-    },
-    {
-      "epoch": 0.3043427320143667,
-      "grad_norm": 1139.7677001953125,
-      "learning_rate": 4.4076864228205136e-05,
-      "loss": 62.5426,
-      "step": 75330
-    },
-    {
-      "epoch": 0.3043831332797343,
-      "grad_norm": 480.1134033203125,
-      "learning_rate": 4.407460799317604e-05,
-      "loss": 64.8383,
-      "step": 75340
-    },
-    {
-      "epoch": 0.30442353454510196,
-      "grad_norm": 1814.2696533203125,
-      "learning_rate": 4.4072351386277654e-05,
-      "loss": 78.7744,
-      "step": 75350
-    },
-    {
-      "epoch": 0.3044639358104696,
-      "grad_norm": 1329.923095703125,
-      "learning_rate": 4.407009440755396e-05,
-      "loss": 72.4044,
-      "step": 75360
-    },
-    {
-      "epoch": 0.30450433707583724,
-      "grad_norm": 3605.839599609375,
-      "learning_rate": 4.4067837057048956e-05,
-      "loss": 79.2061,
-      "step": 75370
-    },
-    {
-      "epoch": 0.3045447383412048,
-      "grad_norm": 1073.45166015625,
-      "learning_rate": 4.406557933480664e-05,
-      "loss": 81.1469,
-      "step": 75380
-    },
-    {
-      "epoch": 0.30458513960657246,
-      "grad_norm": 751.8543090820312,
-      "learning_rate": 4.406332124087105e-05,
-      "loss": 77.547,
-      "step": 75390
-    },
-    {
-      "epoch": 0.3046255408719401,
-      "grad_norm": 531.0945434570312,
-      "learning_rate": 4.40610627752862e-05,
-      "loss": 66.1161,
-      "step": 75400
-    },
-    {
-      "epoch": 0.30466594213730774,
-      "grad_norm": 738.7662963867188,
-      "learning_rate": 4.405880393809612e-05,
-      "loss": 64.749,
-      "step": 75410
-    },
-    {
-      "epoch": 0.3047063434026754,
-      "grad_norm": 877.9468383789062,
-      "learning_rate": 4.405654472934483e-05,
-      "loss": 51.235,
-      "step": 75420
-    },
-    {
-      "epoch": 0.304746744668043,
-      "grad_norm": 653.334716796875,
-      "learning_rate": 4.4054285149076404e-05,
-      "loss": 55.1535,
-      "step": 75430
-    },
-    {
-      "epoch": 0.3047871459334106,
-      "grad_norm": 1065.5577392578125,
-      "learning_rate": 4.4052025197334864e-05,
-      "loss": 75.8509,
-      "step": 75440
-    },
-    {
-      "epoch": 0.30482754719877825,
-      "grad_norm": 1329.488525390625,
-      "learning_rate": 4.40497648741643e-05,
-      "loss": 78.9796,
-      "step": 75450
-    },
-    {
-      "epoch": 0.3048679484641459,
-      "grad_norm": 1260.999755859375,
-      "learning_rate": 4.4047504179608755e-05,
-      "loss": 91.8325,
-      "step": 75460
-    },
-    {
-      "epoch": 0.30490834972951353,
-      "grad_norm": 656.1160888671875,
-      "learning_rate": 4.404524311371231e-05,
-      "loss": 35.623,
-      "step": 75470
-    },
-    {
-      "epoch": 0.30494875099488117,
-      "grad_norm": 882.13623046875,
-      "learning_rate": 4.404298167651905e-05,
-      "loss": 58.8389,
-      "step": 75480
-    },
-    {
-      "epoch": 0.3049891522602488,
-      "grad_norm": 1063.432861328125,
-      "learning_rate": 4.4040719868073055e-05,
-      "loss": 55.5148,
-      "step": 75490
-    },
-    {
-      "epoch": 0.30502955352561645,
-      "grad_norm": 1446.3270263671875,
-      "learning_rate": 4.403845768841842e-05,
-      "loss": 93.3359,
-      "step": 75500
-    },
-    {
-      "epoch": 0.30506995479098403,
-      "grad_norm": 652.8603515625,
-      "learning_rate": 4.403619513759926e-05,
-      "loss": 88.8308,
-      "step": 75510
-    },
-    {
-      "epoch": 0.3051103560563517,
-      "grad_norm": 848.1201782226562,
-      "learning_rate": 4.403393221565966e-05,
-      "loss": 51.726,
-      "step": 75520
-    },
-    {
-      "epoch": 0.3051507573217193,
-      "grad_norm": 823.25390625,
-      "learning_rate": 4.403166892264376e-05,
-      "loss": 60.7063,
-      "step": 75530
-    },
-    {
-      "epoch": 0.30519115858708695,
-      "grad_norm": 837.02978515625,
-      "learning_rate": 4.402940525859568e-05,
-      "loss": 62.3497,
-      "step": 75540
-    },
-    {
-      "epoch": 0.3052315598524546,
-      "grad_norm": 1130.3653564453125,
-      "learning_rate": 4.402714122355955e-05,
-      "loss": 86.2569,
-      "step": 75550
-    },
-    {
-      "epoch": 0.30527196111782223,
-      "grad_norm": 313.84063720703125,
-      "learning_rate": 4.40248768175795e-05,
-      "loss": 85.372,
-      "step": 75560
-    },
-    {
-      "epoch": 0.3053123623831898,
-      "grad_norm": 1272.7034912109375,
-      "learning_rate": 4.4022612040699676e-05,
-      "loss": 59.0401,
-      "step": 75570
-    },
-    {
-      "epoch": 0.30535276364855746,
-      "grad_norm": 259.1897888183594,
-      "learning_rate": 4.4020346892964246e-05,
-      "loss": 42.5053,
-      "step": 75580
-    },
-    {
-      "epoch": 0.3053931649139251,
-      "grad_norm": 541.4120483398438,
-      "learning_rate": 4.401808137441736e-05,
-      "loss": 59.4345,
-      "step": 75590
-    },
-    {
-      "epoch": 0.30543356617929274,
-      "grad_norm": 923.3859252929688,
-      "learning_rate": 4.401581548510318e-05,
-      "loss": 47.9736,
-      "step": 75600
-    },
-    {
-      "epoch": 0.3054739674446604,
-      "grad_norm": 1377.25,
-      "learning_rate": 4.40135492250659e-05,
-      "loss": 78.1078,
-      "step": 75610
-    },
-    {
-      "epoch": 0.305514368710028,
-      "grad_norm": 1074.262939453125,
-      "learning_rate": 4.401128259434968e-05,
-      "loss": 57.5512,
-      "step": 75620
-    },
-    {
-      "epoch": 0.30555476997539566,
-      "grad_norm": 616.6300048828125,
-      "learning_rate": 4.400901559299871e-05,
-      "loss": 38.326,
-      "step": 75630
-    },
-    {
-      "epoch": 0.30559517124076324,
-      "grad_norm": 1104.368408203125,
-      "learning_rate": 4.4006748221057206e-05,
-      "loss": 94.1468,
-      "step": 75640
-    },
-    {
-      "epoch": 0.3056355725061309,
-      "grad_norm": 581.0299072265625,
-      "learning_rate": 4.4004480478569353e-05,
-      "loss": 87.8687,
-      "step": 75650
-    },
-    {
-      "epoch": 0.3056759737714985,
-      "grad_norm": 1138.6673583984375,
-      "learning_rate": 4.400221236557938e-05,
-      "loss": 66.9354,
-      "step": 75660
-    },
-    {
-      "epoch": 0.30571637503686616,
-      "grad_norm": 568.2470703125,
-      "learning_rate": 4.399994388213149e-05,
-      "loss": 88.6248,
-      "step": 75670
-    },
-    {
-      "epoch": 0.3057567763022338,
-      "grad_norm": 598.5287475585938,
-      "learning_rate": 4.3997675028269906e-05,
-      "loss": 55.7092,
-      "step": 75680
-    },
-    {
-      "epoch": 0.30579717756760144,
-      "grad_norm": 971.5711669921875,
-      "learning_rate": 4.399540580403887e-05,
-      "loss": 67.1653,
-      "step": 75690
-    },
-    {
-      "epoch": 0.305837578832969,
-      "grad_norm": 1012.9927368164062,
-      "learning_rate": 4.399313620948262e-05,
-      "loss": 51.6907,
-      "step": 75700
-    },
-    {
-      "epoch": 0.30587798009833667,
-      "grad_norm": 1266.41748046875,
-      "learning_rate": 4.39908662446454e-05,
-      "loss": 59.8522,
-      "step": 75710
-    },
-    {
-      "epoch": 0.3059183813637043,
-      "grad_norm": 580.0430297851562,
-      "learning_rate": 4.3988595909571464e-05,
-      "loss": 51.2898,
-      "step": 75720
-    },
-    {
-      "epoch": 0.30595878262907195,
-      "grad_norm": 503.0343322753906,
-      "learning_rate": 4.3986325204305076e-05,
-      "loss": 69.8676,
-      "step": 75730
-    },
-    {
-      "epoch": 0.3059991838944396,
-      "grad_norm": 246.7170867919922,
-      "learning_rate": 4.398405412889051e-05,
-      "loss": 60.3378,
-      "step": 75740
-    },
-    {
-      "epoch": 0.3060395851598072,
-      "grad_norm": 440.064208984375,
-      "learning_rate": 4.3981782683372016e-05,
-      "loss": 52.1341,
-      "step": 75750
-    },
-    {
-      "epoch": 0.3060799864251748,
-      "grad_norm": 684.9658203125,
-      "learning_rate": 4.3979510867793917e-05,
-      "loss": 83.72,
-      "step": 75760
-    },
-    {
-      "epoch": 0.30612038769054245,
-      "grad_norm": 0.0,
-      "learning_rate": 4.397723868220047e-05,
-      "loss": 64.8489,
-      "step": 75770
-    },
-    {
-      "epoch": 0.3061607889559101,
-      "grad_norm": 971.6740112304688,
-      "learning_rate": 4.397496612663599e-05,
-      "loss": 59.74,
-      "step": 75780
-    },
-    {
-      "epoch": 0.30620119022127773,
-      "grad_norm": 1435.0618896484375,
-      "learning_rate": 4.397269320114478e-05,
-      "loss": 92.3261,
-      "step": 75790
-    },
-    {
-      "epoch": 0.30624159148664537,
-      "grad_norm": 1401.4566650390625,
-      "learning_rate": 4.3970419905771145e-05,
-      "loss": 86.528,
-      "step": 75800
-    },
-    {
-      "epoch": 0.306281992752013,
-      "grad_norm": 423.3729553222656,
-      "learning_rate": 4.39681462405594e-05,
-      "loss": 106.0067,
-      "step": 75810
-    },
-    {
-      "epoch": 0.30632239401738065,
-      "grad_norm": 0.0,
-      "learning_rate": 4.3965872205553885e-05,
-      "loss": 70.0686,
-      "step": 75820
-    },
-    {
-      "epoch": 0.30636279528274823,
-      "grad_norm": 1514.8172607421875,
-      "learning_rate": 4.3963597800798927e-05,
-      "loss": 85.976,
-      "step": 75830
-    },
-    {
-      "epoch": 0.3064031965481159,
-      "grad_norm": 286.7713928222656,
-      "learning_rate": 4.396132302633886e-05,
-      "loss": 47.0414,
-      "step": 75840
-    },
-    {
-      "epoch": 0.3064435978134835,
-      "grad_norm": 1182.9954833984375,
-      "learning_rate": 4.395904788221805e-05,
-      "loss": 59.4661,
-      "step": 75850
-    },
-    {
-      "epoch": 0.30648399907885115,
-      "grad_norm": 1122.1630859375,
-      "learning_rate": 4.3956772368480836e-05,
-      "loss": 94.7752,
-      "step": 75860
-    },
-    {
-      "epoch": 0.3065244003442188,
-      "grad_norm": 555.4927368164062,
-      "learning_rate": 4.395449648517158e-05,
-      "loss": 52.1087,
-      "step": 75870
-    },
-    {
-      "epoch": 0.30656480160958643,
-      "grad_norm": 698.0592041015625,
-      "learning_rate": 4.395222023233466e-05,
-      "loss": 52.4741,
-      "step": 75880
-    },
-    {
-      "epoch": 0.306605202874954,
-      "grad_norm": 628.822509765625,
-      "learning_rate": 4.3949943610014455e-05,
-      "loss": 86.0263,
-      "step": 75890
-    },
-    {
-      "epoch": 0.30664560414032166,
-      "grad_norm": 1240.1202392578125,
-      "learning_rate": 4.394766661825533e-05,
-      "loss": 46.8542,
-      "step": 75900
-    },
-    {
-      "epoch": 0.3066860054056893,
-      "grad_norm": 816.1203002929688,
-      "learning_rate": 4.3945389257101704e-05,
-      "loss": 38.3735,
-      "step": 75910
-    },
-    {
-      "epoch": 0.30672640667105694,
-      "grad_norm": 889.33349609375,
-      "learning_rate": 4.394311152659796e-05,
-      "loss": 73.3283,
-      "step": 75920
-    },
-    {
-      "epoch": 0.3067668079364246,
-      "grad_norm": 1453.927978515625,
-      "learning_rate": 4.3940833426788496e-05,
-      "loss": 62.5195,
-      "step": 75930
-    },
-    {
-      "epoch": 0.3068072092017922,
-      "grad_norm": 987.9056396484375,
-      "learning_rate": 4.393855495771774e-05,
-      "loss": 77.4361,
-      "step": 75940
-    },
-    {
-      "epoch": 0.3068476104671598,
-      "grad_norm": 1086.7833251953125,
-      "learning_rate": 4.3936276119430096e-05,
-      "loss": 87.7577,
-      "step": 75950
-    },
-    {
-      "epoch": 0.30688801173252744,
-      "grad_norm": 491.1762390136719,
-      "learning_rate": 4.393399691197e-05,
-      "loss": 71.9018,
-      "step": 75960
-    },
-    {
-      "epoch": 0.3069284129978951,
-      "grad_norm": 834.448974609375,
-      "learning_rate": 4.3931717335381894e-05,
-      "loss": 56.691,
-      "step": 75970
-    },
-    {
-      "epoch": 0.3069688142632627,
-      "grad_norm": 1463.418212890625,
-      "learning_rate": 4.392943738971021e-05,
-      "loss": 66.625,
-      "step": 75980
-    },
-    {
-      "epoch": 0.30700921552863036,
-      "grad_norm": 709.5184936523438,
-      "learning_rate": 4.39271570749994e-05,
-      "loss": 72.5582,
-      "step": 75990
-    },
-    {
-      "epoch": 0.307049616793998,
-      "grad_norm": 802.2734985351562,
-      "learning_rate": 4.3924876391293915e-05,
-      "loss": 80.8964,
-      "step": 76000
-    },
-    {
-      "epoch": 0.30709001805936564,
-      "grad_norm": 1081.5029296875,
-      "learning_rate": 4.3922595338638214e-05,
-      "loss": 64.753,
-      "step": 76010
-    },
-    {
-      "epoch": 0.30713041932473323,
-      "grad_norm": 747.075439453125,
-      "learning_rate": 4.3920313917076794e-05,
-      "loss": 48.6337,
-      "step": 76020
-    },
-    {
-      "epoch": 0.30717082059010087,
-      "grad_norm": 371.2689208984375,
-      "learning_rate": 4.3918032126654095e-05,
-      "loss": 56.0596,
-      "step": 76030
-    },
-    {
-      "epoch": 0.3072112218554685,
-      "grad_norm": 852.907958984375,
-      "learning_rate": 4.391574996741463e-05,
-      "loss": 66.6618,
-      "step": 76040
-    },
-    {
-      "epoch": 0.30725162312083615,
-      "grad_norm": 1084.6485595703125,
-      "learning_rate": 4.391346743940288e-05,
-      "loss": 66.4222,
-      "step": 76050
-    },
-    {
-      "epoch": 0.3072920243862038,
-      "grad_norm": 1998.0355224609375,
-      "learning_rate": 4.3911184542663344e-05,
-      "loss": 77.1086,
-      "step": 76060
-    },
-    {
-      "epoch": 0.30733242565157143,
-      "grad_norm": 789.2811889648438,
-      "learning_rate": 4.390890127724053e-05,
-      "loss": 68.8511,
-      "step": 76070
-    },
-    {
-      "epoch": 0.307372826916939,
-      "grad_norm": 666.281494140625,
-      "learning_rate": 4.390661764317895e-05,
-      "loss": 68.8774,
-      "step": 76080
-    },
-    {
-      "epoch": 0.30741322818230665,
-      "grad_norm": 593.8773803710938,
-      "learning_rate": 4.390433364052312e-05,
-      "loss": 45.1297,
-      "step": 76090
-    },
-    {
-      "epoch": 0.3074536294476743,
-      "grad_norm": 1460.0093994140625,
-      "learning_rate": 4.390204926931758e-05,
-      "loss": 79.1225,
-      "step": 76100
-    },
-    {
-      "epoch": 0.30749403071304193,
-      "grad_norm": 653.6718139648438,
-      "learning_rate": 4.389976452960686e-05,
-      "loss": 72.7947,
-      "step": 76110
-    },
-    {
-      "epoch": 0.3075344319784096,
-      "grad_norm": 906.63134765625,
-      "learning_rate": 4.38974794214355e-05,
-      "loss": 72.8858,
-      "step": 76120
-    },
-    {
-      "epoch": 0.3075748332437772,
-      "grad_norm": 1931.9991455078125,
-      "learning_rate": 4.3895193944848034e-05,
-      "loss": 64.3772,
-      "step": 76130
-    },
-    {
-      "epoch": 0.30761523450914485,
-      "grad_norm": 798.5737915039062,
-      "learning_rate": 4.3892908099889054e-05,
-      "loss": 55.5147,
-      "step": 76140
-    },
-    {
-      "epoch": 0.30765563577451244,
-      "grad_norm": 696.617919921875,
-      "learning_rate": 4.389062188660309e-05,
-      "loss": 62.2774,
-      "step": 76150
-    },
-    {
-      "epoch": 0.3076960370398801,
-      "grad_norm": 840.3499145507812,
-      "learning_rate": 4.388833530503473e-05,
-      "loss": 98.8143,
-      "step": 76160
-    },
-    {
-      "epoch": 0.3077364383052477,
-      "grad_norm": 764.2418212890625,
-      "learning_rate": 4.388604835522855e-05,
-      "loss": 69.4767,
-      "step": 76170
-    },
-    {
-      "epoch": 0.30777683957061536,
-      "grad_norm": 663.6424560546875,
-      "learning_rate": 4.3883761037229146e-05,
-      "loss": 69.4592,
-      "step": 76180
-    },
-    {
-      "epoch": 0.307817240835983,
-      "grad_norm": 624.4500732421875,
-      "learning_rate": 4.388147335108108e-05,
-      "loss": 53.6257,
-      "step": 76190
-    },
-    {
-      "epoch": 0.30785764210135064,
-      "grad_norm": 1174.481201171875,
-      "learning_rate": 4.387918529682898e-05,
-      "loss": 70.0419,
-      "step": 76200
-    },
-    {
-      "epoch": 0.3078980433667182,
-      "grad_norm": 466.0067443847656,
-      "learning_rate": 4.3876896874517434e-05,
-      "loss": 112.7296,
-      "step": 76210
-    },
-    {
-      "epoch": 0.30793844463208586,
-      "grad_norm": 576.7889404296875,
-      "learning_rate": 4.387460808419108e-05,
-      "loss": 56.6612,
-      "step": 76220
-    },
-    {
-      "epoch": 0.3079788458974535,
-      "grad_norm": 809.4537353515625,
-      "learning_rate": 4.387231892589452e-05,
-      "loss": 84.8902,
-      "step": 76230
-    },
-    {
-      "epoch": 0.30801924716282114,
-      "grad_norm": 1000.173095703125,
-      "learning_rate": 4.387002939967237e-05,
-      "loss": 69.7403,
-      "step": 76240
-    },
-    {
-      "epoch": 0.3080596484281888,
-      "grad_norm": 756.7673950195312,
-      "learning_rate": 4.386773950556931e-05,
-      "loss": 51.9468,
-      "step": 76250
-    },
-    {
-      "epoch": 0.3081000496935564,
-      "grad_norm": 922.7078247070312,
-      "learning_rate": 4.386544924362993e-05,
-      "loss": 80.4618,
-      "step": 76260
-    },
-    {
-      "epoch": 0.308140450958924,
-      "grad_norm": 928.8509521484375,
-      "learning_rate": 4.3863158613898915e-05,
-      "loss": 90.6427,
-      "step": 76270
-    },
-    {
-      "epoch": 0.30818085222429165,
-      "grad_norm": 990.8674926757812,
-      "learning_rate": 4.386086761642091e-05,
-      "loss": 54.2147,
-      "step": 76280
-    },
-    {
-      "epoch": 0.3082212534896593,
-      "grad_norm": 973.0326538085938,
-      "learning_rate": 4.385857625124058e-05,
-      "loss": 86.8008,
-      "step": 76290
-    },
-    {
-      "epoch": 0.3082616547550269,
-      "grad_norm": 831.694580078125,
-      "learning_rate": 4.3856284518402594e-05,
-      "loss": 77.8919,
-      "step": 76300
-    },
-    {
-      "epoch": 0.30830205602039457,
-      "grad_norm": 1294.062255859375,
-      "learning_rate": 4.385399241795164e-05,
-      "loss": 89.092,
-      "step": 76310
-    },
-    {
-      "epoch": 0.3083424572857622,
-      "grad_norm": 899.0340576171875,
-      "learning_rate": 4.3851699949932396e-05,
-      "loss": 66.3644,
-      "step": 76320
-    },
-    {
-      "epoch": 0.30838285855112985,
-      "grad_norm": 2219.54296875,
-      "learning_rate": 4.384940711438955e-05,
-      "loss": 43.2072,
-      "step": 76330
-    },
-    {
-      "epoch": 0.30842325981649743,
-      "grad_norm": 1805.721923828125,
-      "learning_rate": 4.384711391136781e-05,
-      "loss": 68.598,
-      "step": 76340
-    },
-    {
-      "epoch": 0.30846366108186507,
-      "grad_norm": 701.5916137695312,
-      "learning_rate": 4.384482034091189e-05,
-      "loss": 49.2661,
-      "step": 76350
-    },
-    {
-      "epoch": 0.3085040623472327,
-      "grad_norm": 435.8236389160156,
-      "learning_rate": 4.3842526403066486e-05,
-      "loss": 52.0933,
-      "step": 76360
-    },
-    {
-      "epoch": 0.30854446361260035,
-      "grad_norm": 2196.045166015625,
-      "learning_rate": 4.384023209787633e-05,
-      "loss": 93.9429,
-      "step": 76370
-    },
-    {
-      "epoch": 0.308584864877968,
-      "grad_norm": 780.8671264648438,
-      "learning_rate": 4.383793742538616e-05,
-      "loss": 60.4108,
-      "step": 76380
-    },
-    {
-      "epoch": 0.30862526614333563,
-      "grad_norm": 1023.3828125,
-      "learning_rate": 4.383564238564068e-05,
-      "loss": 65.5329,
-      "step": 76390
-    },
-    {
-      "epoch": 0.3086656674087032,
-      "grad_norm": 677.0743408203125,
-      "learning_rate": 4.3833346978684675e-05,
-      "loss": 89.5438,
-      "step": 76400
-    },
-    {
-      "epoch": 0.30870606867407085,
-      "grad_norm": 650.9645385742188,
-      "learning_rate": 4.383105120456287e-05,
-      "loss": 50.4317,
-      "step": 76410
-    },
-    {
-      "epoch": 0.3087464699394385,
-      "grad_norm": 543.072021484375,
-      "learning_rate": 4.3828755063320016e-05,
-      "loss": 84.4733,
-      "step": 76420
-    },
-    {
-      "epoch": 0.30878687120480613,
-      "grad_norm": 773.248046875,
-      "learning_rate": 4.38264585550009e-05,
-      "loss": 64.6917,
-      "step": 76430
-    },
-    {
-      "epoch": 0.3088272724701738,
-      "grad_norm": 365.4747009277344,
-      "learning_rate": 4.382416167965028e-05,
-      "loss": 62.2631,
-      "step": 76440
-    },
-    {
-      "epoch": 0.3088676737355414,
-      "grad_norm": 674.434326171875,
-      "learning_rate": 4.382186443731293e-05,
-      "loss": 54.153,
-      "step": 76450
-    },
-    {
-      "epoch": 0.30890807500090905,
-      "grad_norm": 1216.2164306640625,
-      "learning_rate": 4.381956682803365e-05,
-      "loss": 77.9781,
-      "step": 76460
-    },
-    {
-      "epoch": 0.30894847626627664,
-      "grad_norm": 615.4111938476562,
-      "learning_rate": 4.381726885185722e-05,
-      "loss": 66.789,
-      "step": 76470
-    },
-    {
-      "epoch": 0.3089888775316443,
-      "grad_norm": 1128.7919921875,
-      "learning_rate": 4.381497050882845e-05,
-      "loss": 61.4229,
-      "step": 76480
-    },
-    {
-      "epoch": 0.3090292787970119,
-      "grad_norm": 645.344482421875,
-      "learning_rate": 4.381267179899214e-05,
-      "loss": 47.0934,
-      "step": 76490
-    },
-    {
-      "epoch": 0.30906968006237956,
-      "grad_norm": 397.85980224609375,
-      "learning_rate": 4.381037272239311e-05,
-      "loss": 63.7962,
-      "step": 76500
-    },
-    {
-      "epoch": 0.3091100813277472,
-      "grad_norm": 463.2906188964844,
-      "learning_rate": 4.380807327907618e-05,
-      "loss": 60.8565,
-      "step": 76510
-    },
-    {
-      "epoch": 0.30915048259311484,
-      "grad_norm": 470.76104736328125,
-      "learning_rate": 4.380577346908618e-05,
-      "loss": 71.5916,
-      "step": 76520
-    },
-    {
-      "epoch": 0.3091908838584824,
-      "grad_norm": 1406.014404296875,
-      "learning_rate": 4.380347329246794e-05,
-      "loss": 76.5488,
-      "step": 76530
-    },
-    {
-      "epoch": 0.30923128512385006,
-      "grad_norm": 898.1220703125,
-      "learning_rate": 4.380117274926631e-05,
-      "loss": 75.5578,
-      "step": 76540
-    },
-    {
-      "epoch": 0.3092716863892177,
-      "grad_norm": 1027.6002197265625,
-      "learning_rate": 4.379887183952614e-05,
-      "loss": 73.7126,
-      "step": 76550
-    },
-    {
-      "epoch": 0.30931208765458534,
-      "grad_norm": 2206.342529296875,
-      "learning_rate": 4.379657056329228e-05,
-      "loss": 121.1146,
-      "step": 76560
-    },
-    {
-      "epoch": 0.309352488919953,
-      "grad_norm": 986.2785034179688,
-      "learning_rate": 4.3794268920609605e-05,
-      "loss": 59.564,
-      "step": 76570
-    },
-    {
-      "epoch": 0.3093928901853206,
-      "grad_norm": 847.2274169921875,
-      "learning_rate": 4.379196691152298e-05,
-      "loss": 69.7664,
-      "step": 76580
-    },
-    {
-      "epoch": 0.3094332914506882,
-      "grad_norm": 938.61767578125,
-      "learning_rate": 4.3789664536077286e-05,
-      "loss": 54.9465,
-      "step": 76590
-    },
-    {
-      "epoch": 0.30947369271605585,
-      "grad_norm": 879.3148193359375,
-      "learning_rate": 4.3787361794317405e-05,
-      "loss": 82.9992,
-      "step": 76600
-    },
-    {
-      "epoch": 0.3095140939814235,
-      "grad_norm": 394.82000732421875,
-      "learning_rate": 4.378505868628823e-05,
-      "loss": 79.9965,
-      "step": 76610
-    },
-    {
-      "epoch": 0.30955449524679113,
-      "grad_norm": 458.2685241699219,
-      "learning_rate": 4.3782755212034675e-05,
-      "loss": 90.1063,
-      "step": 76620
-    },
-    {
-      "epoch": 0.30959489651215877,
-      "grad_norm": 1523.7799072265625,
-      "learning_rate": 4.3780451371601626e-05,
-      "loss": 63.1249,
-      "step": 76630
-    },
-    {
-      "epoch": 0.3096352977775264,
-      "grad_norm": 656.118896484375,
-      "learning_rate": 4.3778147165034025e-05,
-      "loss": 75.1157,
-      "step": 76640
-    },
-    {
-      "epoch": 0.30967569904289405,
-      "grad_norm": 1332.3353271484375,
-      "learning_rate": 4.377584259237676e-05,
-      "loss": 98.9753,
-      "step": 76650
-    },
-    {
-      "epoch": 0.30971610030826163,
-      "grad_norm": 1163.6890869140625,
-      "learning_rate": 4.377353765367479e-05,
-      "loss": 82.146,
-      "step": 76660
-    },
-    {
-      "epoch": 0.3097565015736293,
-      "grad_norm": 994.2089233398438,
-      "learning_rate": 4.377123234897303e-05,
-      "loss": 47.6127,
-      "step": 76670
-    },
-    {
-      "epoch": 0.3097969028389969,
-      "grad_norm": 427.42059326171875,
-      "learning_rate": 4.376892667831644e-05,
-      "loss": 61.8331,
-      "step": 76680
-    },
-    {
-      "epoch": 0.30983730410436455,
-      "grad_norm": 1147.4427490234375,
-      "learning_rate": 4.376662064174994e-05,
-      "loss": 65.2715,
-      "step": 76690
-    },
-    {
-      "epoch": 0.3098777053697322,
-      "grad_norm": 1077.2244873046875,
-      "learning_rate": 4.376431423931853e-05,
-      "loss": 111.3696,
-      "step": 76700
-    },
-    {
-      "epoch": 0.30991810663509983,
-      "grad_norm": 384.9374084472656,
-      "learning_rate": 4.3762007471067146e-05,
-      "loss": 72.2117,
-      "step": 76710
-    },
-    {
-      "epoch": 0.3099585079004674,
-      "grad_norm": 0.0,
-      "learning_rate": 4.375970033704077e-05,
-      "loss": 53.84,
-      "step": 76720
-    },
-    {
-      "epoch": 0.30999890916583506,
-      "grad_norm": 514.8012084960938,
-      "learning_rate": 4.375739283728437e-05,
-      "loss": 43.7522,
-      "step": 76730
-    },
-    {
-      "epoch": 0.3100393104312027,
-      "grad_norm": 1148.413818359375,
-      "learning_rate": 4.3755084971842954e-05,
-      "loss": 73.6758,
-      "step": 76740
-    },
-    {
-      "epoch": 0.31007971169657034,
-      "grad_norm": 1197.4542236328125,
-      "learning_rate": 4.375277674076149e-05,
-      "loss": 51.1232,
-      "step": 76750
-    },
-    {
-      "epoch": 0.310120112961938,
-      "grad_norm": 1706.7191162109375,
-      "learning_rate": 4.375046814408499e-05,
-      "loss": 102.9833,
-      "step": 76760
-    },
-    {
-      "epoch": 0.3101605142273056,
-      "grad_norm": 696.5614624023438,
-      "learning_rate": 4.374815918185846e-05,
-      "loss": 58.2234,
-      "step": 76770
-    },
-    {
-      "epoch": 0.31020091549267326,
-      "grad_norm": 1848.7935791015625,
-      "learning_rate": 4.374584985412692e-05,
-      "loss": 67.9964,
-      "step": 76780
-    },
-    {
-      "epoch": 0.31024131675804084,
-      "grad_norm": 469.37255859375,
-      "learning_rate": 4.374354016093538e-05,
-      "loss": 72.5402,
-      "step": 76790
-    },
-    {
-      "epoch": 0.3102817180234085,
-      "grad_norm": 1170.890380859375,
-      "learning_rate": 4.374123010232888e-05,
-      "loss": 67.4718,
-      "step": 76800
-    },
-    {
-      "epoch": 0.3103221192887761,
-      "grad_norm": 468.8921813964844,
-      "learning_rate": 4.373891967835245e-05,
-      "loss": 50.772,
-      "step": 76810
-    },
-    {
-      "epoch": 0.31036252055414376,
-      "grad_norm": 1059.63623046875,
-      "learning_rate": 4.373660888905113e-05,
-      "loss": 74.093,
-      "step": 76820
-    },
-    {
-      "epoch": 0.3104029218195114,
-      "grad_norm": 1103.974609375,
-      "learning_rate": 4.373429773446998e-05,
-      "loss": 65.5093,
-      "step": 76830
-    },
-    {
-      "epoch": 0.31044332308487904,
-      "grad_norm": 885.3768920898438,
-      "learning_rate": 4.373198621465404e-05,
-      "loss": 77.8017,
-      "step": 76840
-    },
-    {
-      "epoch": 0.3104837243502466,
-      "grad_norm": 968.3536376953125,
-      "learning_rate": 4.372967432964838e-05,
-      "loss": 53.5939,
-      "step": 76850
-    },
-    {
-      "epoch": 0.31052412561561427,
-      "grad_norm": 365.57904052734375,
-      "learning_rate": 4.372736207949809e-05,
-      "loss": 94.1086,
-      "step": 76860
-    },
-    {
-      "epoch": 0.3105645268809819,
-      "grad_norm": 974.5148315429688,
-      "learning_rate": 4.3725049464248235e-05,
-      "loss": 62.9441,
-      "step": 76870
-    },
-    {
-      "epoch": 0.31060492814634955,
-      "grad_norm": 191.68748474121094,
-      "learning_rate": 4.372273648394389e-05,
-      "loss": 61.9872,
-      "step": 76880
-    },
-    {
-      "epoch": 0.3106453294117172,
-      "grad_norm": 661.6585693359375,
-      "learning_rate": 4.372042313863017e-05,
-      "loss": 50.0335,
-      "step": 76890
-    },
-    {
-      "epoch": 0.3106857306770848,
-      "grad_norm": 774.6177978515625,
-      "learning_rate": 4.371810942835215e-05,
-      "loss": 48.9739,
-      "step": 76900
-    },
-    {
-      "epoch": 0.3107261319424524,
-      "grad_norm": 813.4871826171875,
-      "learning_rate": 4.371579535315496e-05,
-      "loss": 69.586,
-      "step": 76910
-    },
-    {
-      "epoch": 0.31076653320782005,
-      "grad_norm": 1050.9583740234375,
-      "learning_rate": 4.37134809130837e-05,
-      "loss": 67.1586,
-      "step": 76920
-    },
-    {
-      "epoch": 0.3108069344731877,
-      "grad_norm": 747.37939453125,
-      "learning_rate": 4.37111661081835e-05,
-      "loss": 56.411,
-      "step": 76930
-    },
-    {
-      "epoch": 0.31084733573855533,
-      "grad_norm": 499.1912536621094,
-      "learning_rate": 4.370885093849948e-05,
-      "loss": 36.8895,
-      "step": 76940
-    },
-    {
-      "epoch": 0.31088773700392297,
-      "grad_norm": 1127.0404052734375,
-      "learning_rate": 4.3706535404076784e-05,
-      "loss": 100.2089,
-      "step": 76950
-    },
-    {
-      "epoch": 0.3109281382692906,
-      "grad_norm": 696.9407958984375,
-      "learning_rate": 4.370421950496054e-05,
-      "loss": 55.8497,
-      "step": 76960
-    },
-    {
-      "epoch": 0.31096853953465825,
-      "grad_norm": 867.03173828125,
-      "learning_rate": 4.3701903241195916e-05,
-      "loss": 67.1302,
-      "step": 76970
-    },
-    {
-      "epoch": 0.31100894080002583,
-      "grad_norm": 972.6212768554688,
-      "learning_rate": 4.369958661282805e-05,
-      "loss": 54.7901,
-      "step": 76980
-    },
-    {
-      "epoch": 0.3110493420653935,
-      "grad_norm": 979.4427490234375,
-      "learning_rate": 4.369726961990213e-05,
-      "loss": 63.9195,
-      "step": 76990
-    },
-    {
-      "epoch": 0.3110897433307611,
-      "grad_norm": 1254.9720458984375,
-      "learning_rate": 4.36949522624633e-05,
-      "loss": 110.9768,
-      "step": 77000
-    },
-    {
-      "epoch": 0.31113014459612875,
-      "grad_norm": 1459.2310791015625,
-      "learning_rate": 4.369263454055675e-05,
-      "loss": 81.1262,
-      "step": 77010
-    },
-    {
-      "epoch": 0.3111705458614964,
-      "grad_norm": 968.3046875,
-      "learning_rate": 4.3690316454227674e-05,
-      "loss": 64.1656,
-      "step": 77020
-    },
-    {
-      "epoch": 0.31121094712686403,
-      "grad_norm": 818.0725708007812,
-      "learning_rate": 4.368799800352126e-05,
-      "loss": 93.2603,
-      "step": 77030
-    },
-    {
-      "epoch": 0.3112513483922316,
-      "grad_norm": 521.28125,
-      "learning_rate": 4.368567918848269e-05,
-      "loss": 45.2758,
-      "step": 77040
-    },
-    {
-      "epoch": 0.31129174965759926,
-      "grad_norm": 273.4820556640625,
-      "learning_rate": 4.368336000915719e-05,
-      "loss": 69.712,
-      "step": 77050
-    },
-    {
-      "epoch": 0.3113321509229669,
-      "grad_norm": 905.5136108398438,
-      "learning_rate": 4.3681040465589976e-05,
-      "loss": 88.944,
-      "step": 77060
-    },
-    {
-      "epoch": 0.31137255218833454,
-      "grad_norm": 809.4381713867188,
-      "learning_rate": 4.3678720557826247e-05,
-      "loss": 72.0091,
-      "step": 77070
-    },
-    {
-      "epoch": 0.3114129534537022,
-      "grad_norm": 439.21533203125,
-      "learning_rate": 4.3676400285911256e-05,
-      "loss": 68.7604,
-      "step": 77080
-    },
-    {
-      "epoch": 0.3114533547190698,
-      "grad_norm": 4371.494140625,
-      "learning_rate": 4.367407964989022e-05,
-      "loss": 73.8835,
-      "step": 77090
-    },
-    {
-      "epoch": 0.31149375598443746,
-      "grad_norm": 1301.020751953125,
-      "learning_rate": 4.367175864980839e-05,
-      "loss": 71.925,
-      "step": 77100
-    },
-    {
-      "epoch": 0.31153415724980504,
-      "grad_norm": 608.1272583007812,
-      "learning_rate": 4.366943728571101e-05,
-      "loss": 61.2593,
-      "step": 77110
-    },
-    {
-      "epoch": 0.3115745585151727,
-      "grad_norm": 529.8942260742188,
-      "learning_rate": 4.3667115557643336e-05,
-      "loss": 68.4521,
-      "step": 77120
-    },
-    {
-      "epoch": 0.3116149597805403,
-      "grad_norm": 499.2430725097656,
-      "learning_rate": 4.366479346565064e-05,
-      "loss": 47.7916,
-      "step": 77130
-    },
-    {
-      "epoch": 0.31165536104590796,
-      "grad_norm": 1138.603759765625,
-      "learning_rate": 4.366247100977818e-05,
-      "loss": 58.4289,
-      "step": 77140
-    },
-    {
-      "epoch": 0.3116957623112756,
-      "grad_norm": 558.7338256835938,
-      "learning_rate": 4.366014819007124e-05,
-      "loss": 40.2173,
-      "step": 77150
-    },
-    {
-      "epoch": 0.31173616357664324,
-      "grad_norm": 0.0,
-      "learning_rate": 4.3657825006575106e-05,
-      "loss": 76.302,
-      "step": 77160
-    },
-    {
-      "epoch": 0.31177656484201083,
-      "grad_norm": 1142.35986328125,
-      "learning_rate": 4.365550145933507e-05,
-      "loss": 59.8843,
-      "step": 77170
-    },
-    {
-      "epoch": 0.31181696610737847,
-      "grad_norm": 437.07196044921875,
-      "learning_rate": 4.3653177548396426e-05,
-      "loss": 84.331,
-      "step": 77180
-    },
-    {
-      "epoch": 0.3118573673727461,
-      "grad_norm": 1512.7645263671875,
-      "learning_rate": 4.365085327380448e-05,
-      "loss": 68.1001,
-      "step": 77190
-    },
-    {
-      "epoch": 0.31189776863811375,
-      "grad_norm": 595.816650390625,
-      "learning_rate": 4.3648528635604556e-05,
-      "loss": 61.2616,
-      "step": 77200
-    },
-    {
-      "epoch": 0.3119381699034814,
-      "grad_norm": 307.5999450683594,
-      "learning_rate": 4.364620363384196e-05,
-      "loss": 66.6272,
-      "step": 77210
-    },
-    {
-      "epoch": 0.31197857116884903,
-      "grad_norm": 1286.98095703125,
-      "learning_rate": 4.364387826856202e-05,
-      "loss": 78.2909,
-      "step": 77220
-    },
-    {
-      "epoch": 0.3120189724342166,
-      "grad_norm": 1373.5860595703125,
-      "learning_rate": 4.364155253981008e-05,
-      "loss": 72.0954,
-      "step": 77230
-    },
-    {
-      "epoch": 0.31205937369958425,
-      "grad_norm": 392.8745422363281,
-      "learning_rate": 4.363922644763147e-05,
-      "loss": 71.9362,
-      "step": 77240
-    },
-    {
-      "epoch": 0.3120997749649519,
-      "grad_norm": 891.7858276367188,
-      "learning_rate": 4.363689999207156e-05,
-      "loss": 62.0088,
-      "step": 77250
-    },
-    {
-      "epoch": 0.31214017623031953,
-      "grad_norm": 861.945556640625,
-      "learning_rate": 4.363457317317567e-05,
-      "loss": 105.5122,
-      "step": 77260
-    },
-    {
-      "epoch": 0.3121805774956872,
-      "grad_norm": 843.31689453125,
-      "learning_rate": 4.3632245990989194e-05,
-      "loss": 52.6803,
-      "step": 77270
-    },
-    {
-      "epoch": 0.3122209787610548,
-      "grad_norm": 649.4727172851562,
-      "learning_rate": 4.362991844555749e-05,
-      "loss": 44.0785,
-      "step": 77280
-    },
-    {
-      "epoch": 0.31226138002642245,
-      "grad_norm": 961.2431640625,
-      "learning_rate": 4.362759053692593e-05,
-      "loss": 61.1215,
-      "step": 77290
-    },
-    {
-      "epoch": 0.31230178129179004,
-      "grad_norm": 832.6133422851562,
-      "learning_rate": 4.3625262265139906e-05,
-      "loss": 71.7721,
-      "step": 77300
-    },
-    {
-      "epoch": 0.3123421825571577,
-      "grad_norm": 1031.405517578125,
-      "learning_rate": 4.36229336302448e-05,
-      "loss": 61.8297,
-      "step": 77310
-    },
-    {
-      "epoch": 0.3123825838225253,
-      "grad_norm": 821.444580078125,
-      "learning_rate": 4.3620604632286024e-05,
-      "loss": 92.3522,
-      "step": 77320
-    },
-    {
-      "epoch": 0.31242298508789296,
-      "grad_norm": 809.0938720703125,
-      "learning_rate": 4.361827527130896e-05,
-      "loss": 75.1257,
-      "step": 77330
-    },
-    {
-      "epoch": 0.3124633863532606,
-      "grad_norm": 725.1190795898438,
-      "learning_rate": 4.361594554735905e-05,
-      "loss": 77.7145,
-      "step": 77340
-    },
-    {
-      "epoch": 0.31250378761862824,
-      "grad_norm": 888.7479858398438,
-      "learning_rate": 4.361361546048169e-05,
-      "loss": 95.2185,
-      "step": 77350
-    },
-    {
-      "epoch": 0.3125441888839958,
-      "grad_norm": 2857.910400390625,
-      "learning_rate": 4.361128501072231e-05,
-      "loss": 68.9192,
-      "step": 77360
-    },
-    {
-      "epoch": 0.31258459014936346,
-      "grad_norm": 506.87982177734375,
-      "learning_rate": 4.360895419812635e-05,
-      "loss": 58.4488,
-      "step": 77370
-    },
-    {
-      "epoch": 0.3126249914147311,
-      "grad_norm": 749.842041015625,
-      "learning_rate": 4.360662302273925e-05,
-      "loss": 71.4121,
-      "step": 77380
-    },
-    {
-      "epoch": 0.31266539268009874,
-      "grad_norm": 1351.6341552734375,
-      "learning_rate": 4.360429148460645e-05,
-      "loss": 75.3694,
-      "step": 77390
-    },
-    {
-      "epoch": 0.3127057939454664,
-      "grad_norm": 460.1432189941406,
-      "learning_rate": 4.3601959583773415e-05,
-      "loss": 89.7496,
-      "step": 77400
-    },
-    {
-      "epoch": 0.312746195210834,
-      "grad_norm": 1023.0693969726562,
-      "learning_rate": 4.3599627320285596e-05,
-      "loss": 110.523,
-      "step": 77410
-    },
-    {
-      "epoch": 0.31278659647620166,
-      "grad_norm": 385.4729919433594,
-      "learning_rate": 4.3597294694188475e-05,
-      "loss": 37.7279,
-      "step": 77420
-    },
-    {
-      "epoch": 0.31282699774156925,
-      "grad_norm": 1083.031005859375,
-      "learning_rate": 4.359496170552751e-05,
-      "loss": 65.8572,
-      "step": 77430
-    },
-    {
-      "epoch": 0.3128673990069369,
-      "grad_norm": 571.319580078125,
-      "learning_rate": 4.35926283543482e-05,
-      "loss": 54.1978,
-      "step": 77440
-    },
-    {
-      "epoch": 0.3129078002723045,
-      "grad_norm": 535.0450439453125,
-      "learning_rate": 4.3590294640696025e-05,
-      "loss": 88.7871,
-      "step": 77450
-    },
-    {
-      "epoch": 0.31294820153767217,
-      "grad_norm": 625.9801025390625,
-      "learning_rate": 4.358796056461648e-05,
-      "loss": 73.7608,
-      "step": 77460
-    },
-    {
-      "epoch": 0.3129886028030398,
-      "grad_norm": 569.495361328125,
-      "learning_rate": 4.3585626126155084e-05,
-      "loss": 78.8239,
-      "step": 77470
-    },
-    {
-      "epoch": 0.31302900406840745,
-      "grad_norm": 632.259521484375,
-      "learning_rate": 4.358329132535733e-05,
-      "loss": 119.5843,
-      "step": 77480
-    },
-    {
-      "epoch": 0.31306940533377503,
-      "grad_norm": 752.5601806640625,
-      "learning_rate": 4.3580956162268746e-05,
-      "loss": 41.869,
-      "step": 77490
-    },
-    {
-      "epoch": 0.31310980659914267,
-      "grad_norm": 1639.3475341796875,
-      "learning_rate": 4.357862063693486e-05,
-      "loss": 104.7077,
-      "step": 77500
-    },
-    {
-      "epoch": 0.3131502078645103,
-      "grad_norm": 1603.3714599609375,
-      "learning_rate": 4.35762847494012e-05,
-      "loss": 133.0148,
-      "step": 77510
-    },
-    {
-      "epoch": 0.31319060912987795,
-      "grad_norm": 921.978515625,
-      "learning_rate": 4.35739484997133e-05,
-      "loss": 54.5854,
-      "step": 77520
-    },
-    {
-      "epoch": 0.3132310103952456,
-      "grad_norm": 567.6099853515625,
-      "learning_rate": 4.3571611887916705e-05,
-      "loss": 45.054,
-      "step": 77530
-    },
-    {
-      "epoch": 0.31327141166061323,
-      "grad_norm": 1245.0203857421875,
-      "learning_rate": 4.356927491405699e-05,
-      "loss": 54.2814,
-      "step": 77540
-    },
-    {
-      "epoch": 0.3133118129259808,
-      "grad_norm": 798.68359375,
-      "learning_rate": 4.356693757817969e-05,
-      "loss": 68.1666,
-      "step": 77550
-    },
-    {
-      "epoch": 0.31335221419134845,
-      "grad_norm": 838.5280151367188,
-      "learning_rate": 4.356459988033039e-05,
-      "loss": 55.8854,
-      "step": 77560
-    },
-    {
-      "epoch": 0.3133926154567161,
-      "grad_norm": 497.4013366699219,
-      "learning_rate": 4.356226182055465e-05,
-      "loss": 43.9287,
-      "step": 77570
-    },
-    {
-      "epoch": 0.31343301672208373,
-      "grad_norm": 2377.817626953125,
-      "learning_rate": 4.355992339889806e-05,
-      "loss": 89.1516,
-      "step": 77580
-    },
-    {
-      "epoch": 0.3134734179874514,
-      "grad_norm": 1571.2337646484375,
-      "learning_rate": 4.355758461540622e-05,
-      "loss": 64.8556,
-      "step": 77590
-    },
-    {
-      "epoch": 0.313513819252819,
-      "grad_norm": 2129.327880859375,
-      "learning_rate": 4.355524547012471e-05,
-      "loss": 88.4741,
-      "step": 77600
-    },
-    {
-      "epoch": 0.31355422051818665,
-      "grad_norm": 445.593505859375,
-      "learning_rate": 4.355290596309912e-05,
-      "loss": 59.0199,
-      "step": 77610
-    },
-    {
-      "epoch": 0.31359462178355424,
-      "grad_norm": 660.9097290039062,
-      "learning_rate": 4.3550566094375086e-05,
-      "loss": 65.1326,
-      "step": 77620
-    },
-    {
-      "epoch": 0.3136350230489219,
-      "grad_norm": 670.1815185546875,
-      "learning_rate": 4.3548225863998224e-05,
-      "loss": 76.792,
-      "step": 77630
-    },
-    {
-      "epoch": 0.3136754243142895,
-      "grad_norm": 882.048828125,
-      "learning_rate": 4.354588527201414e-05,
-      "loss": 70.3437,
-      "step": 77640
-    },
-    {
-      "epoch": 0.31371582557965716,
-      "grad_norm": 831.5132446289062,
-      "learning_rate": 4.3543544318468485e-05,
-      "loss": 60.2682,
-      "step": 77650
-    },
-    {
-      "epoch": 0.3137562268450248,
-      "grad_norm": 1422.9718017578125,
-      "learning_rate": 4.354120300340688e-05,
-      "loss": 69.6125,
-      "step": 77660
-    },
-    {
-      "epoch": 0.31379662811039244,
-      "grad_norm": 703.7569580078125,
-      "learning_rate": 4.353886132687497e-05,
-      "loss": 46.6462,
-      "step": 77670
-    },
-    {
-      "epoch": 0.31383702937576,
-      "grad_norm": 620.9691772460938,
-      "learning_rate": 4.353651928891842e-05,
-      "loss": 86.715,
-      "step": 77680
-    },
-    {
-      "epoch": 0.31387743064112766,
-      "grad_norm": 586.15625,
-      "learning_rate": 4.353417688958289e-05,
-      "loss": 58.7201,
-      "step": 77690
-    },
-    {
-      "epoch": 0.3139178319064953,
-      "grad_norm": 1025.6397705078125,
-      "learning_rate": 4.3531834128914025e-05,
-      "loss": 46.0605,
-      "step": 77700
-    },
-    {
-      "epoch": 0.31395823317186294,
-      "grad_norm": 1030.073974609375,
-      "learning_rate": 4.352949100695752e-05,
-      "loss": 79.4799,
-      "step": 77710
-    },
-    {
-      "epoch": 0.3139986344372306,
-      "grad_norm": 862.0059204101562,
-      "learning_rate": 4.352714752375906e-05,
-      "loss": 94.2031,
-      "step": 77720
-    },
-    {
-      "epoch": 0.3140390357025982,
-      "grad_norm": 968.3876342773438,
-      "learning_rate": 4.352480367936431e-05,
-      "loss": 63.7685,
-      "step": 77730
-    },
-    {
-      "epoch": 0.31407943696796586,
-      "grad_norm": 819.8076782226562,
-      "learning_rate": 4.352245947381897e-05,
-      "loss": 77.5755,
-      "step": 77740
-    },
-    {
-      "epoch": 0.31411983823333345,
-      "grad_norm": 1108.0177001953125,
-      "learning_rate": 4.352011490716875e-05,
-      "loss": 70.4065,
-      "step": 77750
-    },
-    {
-      "epoch": 0.3141602394987011,
-      "grad_norm": 698.9972534179688,
-      "learning_rate": 4.351776997945936e-05,
-      "loss": 72.4767,
-      "step": 77760
-    },
-    {
-      "epoch": 0.31420064076406873,
-      "grad_norm": 1234.1883544921875,
-      "learning_rate": 4.351542469073651e-05,
-      "loss": 68.6958,
-      "step": 77770
-    },
-    {
-      "epoch": 0.31424104202943637,
-      "grad_norm": 493.3515625,
-      "learning_rate": 4.351307904104592e-05,
-      "loss": 51.1366,
-      "step": 77780
-    },
-    {
-      "epoch": 0.314281443294804,
-      "grad_norm": 1404.8785400390625,
-      "learning_rate": 4.351073303043332e-05,
-      "loss": 85.1898,
-      "step": 77790
-    },
-    {
-      "epoch": 0.31432184456017165,
-      "grad_norm": 2445.363037109375,
-      "learning_rate": 4.350838665894446e-05,
-      "loss": 67.7399,
-      "step": 77800
-    },
-    {
-      "epoch": 0.31436224582553923,
-      "grad_norm": 1352.8099365234375,
-      "learning_rate": 4.350603992662506e-05,
-      "loss": 52.9771,
-      "step": 77810
-    },
-    {
-      "epoch": 0.3144026470909069,
-      "grad_norm": 969.1926879882812,
-      "learning_rate": 4.3503692833520894e-05,
-      "loss": 62.1888,
-      "step": 77820
-    },
-    {
-      "epoch": 0.3144430483562745,
-      "grad_norm": 486.71185302734375,
-      "learning_rate": 4.350134537967771e-05,
-      "loss": 54.4507,
-      "step": 77830
-    },
-    {
-      "epoch": 0.31448344962164215,
-      "grad_norm": 817.6029663085938,
-      "learning_rate": 4.3498997565141267e-05,
-      "loss": 42.0353,
-      "step": 77840
-    },
-    {
-      "epoch": 0.3145238508870098,
-      "grad_norm": 1096.86669921875,
-      "learning_rate": 4.349664938995734e-05,
-      "loss": 79.652,
-      "step": 77850
-    },
-    {
-      "epoch": 0.31456425215237743,
-      "grad_norm": 1449.87744140625,
-      "learning_rate": 4.3494300854171715e-05,
-      "loss": 93.2444,
-      "step": 77860
-    },
-    {
-      "epoch": 0.314604653417745,
-      "grad_norm": 821.5718994140625,
-      "learning_rate": 4.349195195783017e-05,
-      "loss": 53.875,
-      "step": 77870
-    },
-    {
-      "epoch": 0.31464505468311266,
-      "grad_norm": 1154.4615478515625,
-      "learning_rate": 4.348960270097851e-05,
-      "loss": 62.7061,
-      "step": 77880
-    },
-    {
-      "epoch": 0.3146854559484803,
-      "grad_norm": 792.2052001953125,
-      "learning_rate": 4.348725308366252e-05,
-      "loss": 83.3605,
-      "step": 77890
-    },
-    {
-      "epoch": 0.31472585721384794,
-      "grad_norm": 3039.34228515625,
-      "learning_rate": 4.348490310592801e-05,
-      "loss": 87.3251,
-      "step": 77900
-    },
-    {
-      "epoch": 0.3147662584792156,
-      "grad_norm": 886.7009887695312,
-      "learning_rate": 4.34825527678208e-05,
-      "loss": 57.0282,
-      "step": 77910
-    },
-    {
-      "epoch": 0.3148066597445832,
-      "grad_norm": 786.5516967773438,
-      "learning_rate": 4.348020206938672e-05,
-      "loss": 87.8153,
-      "step": 77920
-    },
-    {
-      "epoch": 0.31484706100995086,
-      "grad_norm": 739.66015625,
-      "learning_rate": 4.347785101067157e-05,
-      "loss": 58.632,
-      "step": 77930
-    },
-    {
-      "epoch": 0.31488746227531844,
-      "grad_norm": 578.0338134765625,
-      "learning_rate": 4.347549959172121e-05,
-      "loss": 80.8651,
-      "step": 77940
-    },
-    {
-      "epoch": 0.3149278635406861,
-      "grad_norm": 1119.5306396484375,
-      "learning_rate": 4.347314781258147e-05,
-      "loss": 68.4437,
-      "step": 77950
-    },
-    {
-      "epoch": 0.3149682648060537,
-      "grad_norm": 0.0,
-      "learning_rate": 4.3470795673298206e-05,
-      "loss": 57.3587,
-      "step": 77960
-    },
-    {
-      "epoch": 0.31500866607142136,
-      "grad_norm": 681.5150146484375,
-      "learning_rate": 4.3468443173917267e-05,
-      "loss": 52.9524,
-      "step": 77970
-    },
-    {
-      "epoch": 0.315049067336789,
-      "grad_norm": 412.8719787597656,
-      "learning_rate": 4.346609031448452e-05,
-      "loss": 36.4723,
-      "step": 77980
-    },
-    {
-      "epoch": 0.31508946860215664,
-      "grad_norm": 945.4547119140625,
-      "learning_rate": 4.346373709504584e-05,
-      "loss": 67.0522,
-      "step": 77990
-    },
-    {
-      "epoch": 0.3151298698675242,
-      "grad_norm": 992.1616821289062,
-      "learning_rate": 4.3461383515647106e-05,
-      "loss": 78.1435,
-      "step": 78000
-    },
-    {
-      "epoch": 0.31517027113289187,
-      "grad_norm": 503.8204650878906,
-      "learning_rate": 4.345902957633418e-05,
-      "loss": 48.7498,
-      "step": 78010
-    },
-    {
-      "epoch": 0.3152106723982595,
-      "grad_norm": 780.5662841796875,
-      "learning_rate": 4.3456675277152973e-05,
-      "loss": 56.2306,
-      "step": 78020
-    },
-    {
-      "epoch": 0.31525107366362715,
-      "grad_norm": 454.04132080078125,
-      "learning_rate": 4.345432061814938e-05,
-      "loss": 85.0091,
-      "step": 78030
-    },
-    {
-      "epoch": 0.3152914749289948,
-      "grad_norm": 1036.3134765625,
-      "learning_rate": 4.345196559936932e-05,
-      "loss": 53.9506,
-      "step": 78040
-    },
-    {
-      "epoch": 0.3153318761943624,
-      "grad_norm": 929.426513671875,
-      "learning_rate": 4.344961022085867e-05,
-      "loss": 70.4333,
-      "step": 78050
-    },
-    {
-      "epoch": 0.31537227745973007,
-      "grad_norm": 597.806640625,
-      "learning_rate": 4.344725448266338e-05,
-      "loss": 41.9415,
-      "step": 78060
-    },
-    {
-      "epoch": 0.31541267872509765,
-      "grad_norm": 391.077392578125,
-      "learning_rate": 4.3444898384829364e-05,
-      "loss": 54.7683,
-      "step": 78070
-    },
-    {
-      "epoch": 0.3154530799904653,
-      "grad_norm": 873.1908569335938,
-      "learning_rate": 4.3442541927402566e-05,
-      "loss": 60.2897,
-      "step": 78080
-    },
-    {
-      "epoch": 0.31549348125583293,
-      "grad_norm": 464.908447265625,
-      "learning_rate": 4.344018511042891e-05,
-      "loss": 57.7049,
-      "step": 78090
-    },
-    {
-      "epoch": 0.31553388252120057,
-      "grad_norm": 481.4813537597656,
-      "learning_rate": 4.343782793395435e-05,
-      "loss": 67.3648,
-      "step": 78100
-    },
-    {
-      "epoch": 0.3155742837865682,
-      "grad_norm": 1816.134521484375,
-      "learning_rate": 4.343547039802485e-05,
-      "loss": 79.2945,
-      "step": 78110
-    },
-    {
-      "epoch": 0.31561468505193585,
-      "grad_norm": 561.038330078125,
-      "learning_rate": 4.3433112502686355e-05,
-      "loss": 50.8454,
-      "step": 78120
-    },
-    {
-      "epoch": 0.31565508631730343,
-      "grad_norm": 685.545654296875,
-      "learning_rate": 4.3430754247984845e-05,
-      "loss": 54.5997,
-      "step": 78130
-    },
-    {
-      "epoch": 0.3156954875826711,
-      "grad_norm": 618.0608520507812,
-      "learning_rate": 4.342839563396629e-05,
-      "loss": 76.7737,
-      "step": 78140
-    },
-    {
-      "epoch": 0.3157358888480387,
-      "grad_norm": 1280.976806640625,
-      "learning_rate": 4.3426036660676686e-05,
-      "loss": 54.933,
-      "step": 78150
-    },
-    {
-      "epoch": 0.31577629011340635,
-      "grad_norm": 650.5822143554688,
-      "learning_rate": 4.3423677328161996e-05,
-      "loss": 71.7501,
-      "step": 78160
-    },
-    {
-      "epoch": 0.315816691378774,
-      "grad_norm": 597.2694091796875,
-      "learning_rate": 4.342131763646824e-05,
-      "loss": 68.88,
-      "step": 78170
-    },
-    {
-      "epoch": 0.31585709264414163,
-      "grad_norm": 1208.90576171875,
-      "learning_rate": 4.341895758564141e-05,
-      "loss": 67.8843,
-      "step": 78180
-    },
-    {
-      "epoch": 0.3158974939095092,
-      "grad_norm": 830.9584350585938,
-      "learning_rate": 4.3416597175727514e-05,
-      "loss": 89.2304,
-      "step": 78190
-    },
-    {
-      "epoch": 0.31593789517487686,
-      "grad_norm": 1917.3353271484375,
-      "learning_rate": 4.3414236406772584e-05,
-      "loss": 111.667,
-      "step": 78200
-    },
-    {
-      "epoch": 0.3159782964402445,
-      "grad_norm": 924.7677612304688,
-      "learning_rate": 4.3411875278822635e-05,
-      "loss": 49.8145,
-      "step": 78210
-    },
-    {
-      "epoch": 0.31601869770561214,
-      "grad_norm": 469.1351318359375,
-      "learning_rate": 4.340951379192369e-05,
-      "loss": 116.5447,
-      "step": 78220
-    },
-    {
-      "epoch": 0.3160590989709798,
-      "grad_norm": 495.249755859375,
-      "learning_rate": 4.34071519461218e-05,
-      "loss": 63.0928,
-      "step": 78230
-    },
-    {
-      "epoch": 0.3160995002363474,
-      "grad_norm": 1259.7349853515625,
-      "learning_rate": 4.3404789741463e-05,
-      "loss": 54.0104,
-      "step": 78240
-    },
-    {
-      "epoch": 0.31613990150171506,
-      "grad_norm": 512.0302124023438,
-      "learning_rate": 4.3402427177993366e-05,
-      "loss": 48.6828,
-      "step": 78250
-    },
-    {
-      "epoch": 0.31618030276708264,
-      "grad_norm": 1092.3133544921875,
-      "learning_rate": 4.340006425575892e-05,
-      "loss": 51.555,
-      "step": 78260
-    },
-    {
-      "epoch": 0.3162207040324503,
-      "grad_norm": 668.2493896484375,
-      "learning_rate": 4.339770097480576e-05,
-      "loss": 67.2536,
-      "step": 78270
-    },
-    {
-      "epoch": 0.3162611052978179,
-      "grad_norm": 832.0966796875,
-      "learning_rate": 4.3395337335179945e-05,
-      "loss": 78.7102,
-      "step": 78280
-    },
-    {
-      "epoch": 0.31630150656318556,
-      "grad_norm": 931.7671508789062,
-      "learning_rate": 4.339297333692756e-05,
-      "loss": 57.0442,
-      "step": 78290
-    },
-    {
-      "epoch": 0.3163419078285532,
-      "grad_norm": 1212.2379150390625,
-      "learning_rate": 4.339060898009469e-05,
-      "loss": 83.3376,
-      "step": 78300
-    },
-    {
-      "epoch": 0.31638230909392084,
-      "grad_norm": 393.2463073730469,
-      "learning_rate": 4.338824426472743e-05,
-      "loss": 48.1462,
-      "step": 78310
-    },
-    {
-      "epoch": 0.31642271035928843,
-      "grad_norm": 422.79998779296875,
-      "learning_rate": 4.338587919087187e-05,
-      "loss": 57.0146,
-      "step": 78320
-    },
-    {
-      "epoch": 0.31646311162465607,
-      "grad_norm": 1542.5203857421875,
-      "learning_rate": 4.3383513758574143e-05,
-      "loss": 69.819,
-      "step": 78330
-    },
-    {
-      "epoch": 0.3165035128900237,
-      "grad_norm": 858.1940307617188,
-      "learning_rate": 4.338114796788035e-05,
-      "loss": 95.8173,
-      "step": 78340
-    },
-    {
-      "epoch": 0.31654391415539135,
-      "grad_norm": 393.74505615234375,
-      "learning_rate": 4.337878181883661e-05,
-      "loss": 46.5559,
-      "step": 78350
-    },
-    {
-      "epoch": 0.316584315420759,
-      "grad_norm": 786.9148559570312,
-      "learning_rate": 4.3376415311489056e-05,
-      "loss": 68.8837,
-      "step": 78360
-    },
-    {
-      "epoch": 0.31662471668612663,
-      "grad_norm": 1402.4737548828125,
-      "learning_rate": 4.337404844588382e-05,
-      "loss": 89.8514,
-      "step": 78370
-    },
-    {
-      "epoch": 0.31666511795149427,
-      "grad_norm": 773.1746826171875,
-      "learning_rate": 4.337168122206706e-05,
-      "loss": 74.1317,
-      "step": 78380
-    },
-    {
-      "epoch": 0.31670551921686185,
-      "grad_norm": 653.9239501953125,
-      "learning_rate": 4.3369313640084916e-05,
-      "loss": 51.6219,
-      "step": 78390
-    },
-    {
-      "epoch": 0.3167459204822295,
-      "grad_norm": 664.2848510742188,
-      "learning_rate": 4.336694569998354e-05,
-      "loss": 74.1221,
-      "step": 78400
-    },
-    {
-      "epoch": 0.31678632174759713,
-      "grad_norm": 609.3822021484375,
-      "learning_rate": 4.3364577401809105e-05,
-      "loss": 71.547,
-      "step": 78410
-    },
-    {
-      "epoch": 0.3168267230129648,
-      "grad_norm": 1231.6363525390625,
-      "learning_rate": 4.336220874560778e-05,
-      "loss": 77.0097,
-      "step": 78420
-    },
-    {
-      "epoch": 0.3168671242783324,
-      "grad_norm": 205.0146484375,
-      "learning_rate": 4.3359839731425735e-05,
-      "loss": 50.0911,
-      "step": 78430
-    },
-    {
-      "epoch": 0.31690752554370005,
-      "grad_norm": 389.177978515625,
-      "learning_rate": 4.335747035930916e-05,
-      "loss": 59.2829,
-      "step": 78440
-    },
-    {
-      "epoch": 0.31694792680906764,
-      "grad_norm": 600.1491088867188,
-      "learning_rate": 4.3355100629304254e-05,
-      "loss": 79.876,
-      "step": 78450
-    },
-    {
-      "epoch": 0.3169883280744353,
-      "grad_norm": 544.903564453125,
-      "learning_rate": 4.335273054145722e-05,
-      "loss": 82.2558,
-      "step": 78460
-    },
-    {
-      "epoch": 0.3170287293398029,
-      "grad_norm": 909.1273803710938,
-      "learning_rate": 4.335036009581425e-05,
-      "loss": 70.1611,
-      "step": 78470
-    },
-    {
-      "epoch": 0.31706913060517056,
-      "grad_norm": 639.6483764648438,
-      "learning_rate": 4.334798929242155e-05,
-      "loss": 69.5685,
-      "step": 78480
-    },
-    {
-      "epoch": 0.3171095318705382,
-      "grad_norm": 615.6561889648438,
-      "learning_rate": 4.3345618131325374e-05,
-      "loss": 76.2855,
-      "step": 78490
-    },
-    {
-      "epoch": 0.31714993313590584,
-      "grad_norm": 3859.427490234375,
-      "learning_rate": 4.334324661257191e-05,
-      "loss": 76.5629,
-      "step": 78500
-    },
-    {
-      "epoch": 0.3171903344012734,
-      "grad_norm": 1610.2098388671875,
-      "learning_rate": 4.334087473620742e-05,
-      "loss": 103.2507,
-      "step": 78510
-    },
-    {
-      "epoch": 0.31723073566664106,
-      "grad_norm": 588.2677612304688,
-      "learning_rate": 4.3338502502278134e-05,
-      "loss": 51.7838,
-      "step": 78520
-    },
-    {
-      "epoch": 0.3172711369320087,
-      "grad_norm": 192.1394500732422,
-      "learning_rate": 4.333612991083029e-05,
-      "loss": 68.0169,
-      "step": 78530
-    },
-    {
-      "epoch": 0.31731153819737634,
-      "grad_norm": 960.3283081054688,
-      "learning_rate": 4.3333756961910166e-05,
-      "loss": 47.2101,
-      "step": 78540
-    },
-    {
-      "epoch": 0.317351939462744,
-      "grad_norm": 1771.0201416015625,
-      "learning_rate": 4.3331383655564006e-05,
-      "loss": 60.5402,
-      "step": 78550
-    },
-    {
-      "epoch": 0.3173923407281116,
-      "grad_norm": 1430.1416015625,
-      "learning_rate": 4.3329009991838084e-05,
-      "loss": 53.3842,
-      "step": 78560
-    },
-    {
-      "epoch": 0.31743274199347926,
-      "grad_norm": 429.119873046875,
-      "learning_rate": 4.3326635970778676e-05,
-      "loss": 48.8891,
-      "step": 78570
-    },
-    {
-      "epoch": 0.31747314325884685,
-      "grad_norm": 798.0321655273438,
-      "learning_rate": 4.3324261592432056e-05,
-      "loss": 68.9384,
-      "step": 78580
-    },
-    {
-      "epoch": 0.3175135445242145,
-      "grad_norm": 999.2288208007812,
-      "learning_rate": 4.3321886856844534e-05,
-      "loss": 57.1692,
-      "step": 78590
-    },
-    {
-      "epoch": 0.3175539457895821,
-      "grad_norm": 923.92724609375,
-      "learning_rate": 4.331951176406239e-05,
-      "loss": 43.5955,
-      "step": 78600
-    },
-    {
-      "epoch": 0.31759434705494977,
-      "grad_norm": 967.3873901367188,
-      "learning_rate": 4.331713631413194e-05,
-      "loss": 47.511,
-      "step": 78610
-    },
-    {
-      "epoch": 0.3176347483203174,
-      "grad_norm": 884.6947021484375,
-      "learning_rate": 4.331476050709948e-05,
-      "loss": 57.999,
-      "step": 78620
-    },
-    {
-      "epoch": 0.31767514958568505,
-      "grad_norm": 611.1873168945312,
-      "learning_rate": 4.331238434301134e-05,
-      "loss": 77.5249,
-      "step": 78630
-    },
-    {
-      "epoch": 0.31771555085105263,
-      "grad_norm": 913.7862548828125,
-      "learning_rate": 4.3310007821913836e-05,
-      "loss": 54.6169,
-      "step": 78640
-    },
-    {
-      "epoch": 0.31775595211642027,
-      "grad_norm": 1171.2972412109375,
-      "learning_rate": 4.330763094385329e-05,
-      "loss": 80.6045,
-      "step": 78650
-    },
-    {
-      "epoch": 0.3177963533817879,
-      "grad_norm": 507.2369384765625,
-      "learning_rate": 4.330525370887607e-05,
-      "loss": 46.0367,
-      "step": 78660
-    },
-    {
-      "epoch": 0.31783675464715555,
-      "grad_norm": 564.16259765625,
-      "learning_rate": 4.33028761170285e-05,
-      "loss": 73.1167,
-      "step": 78670
-    },
-    {
-      "epoch": 0.3178771559125232,
-      "grad_norm": 897.42919921875,
-      "learning_rate": 4.330049816835694e-05,
-      "loss": 92.9207,
-      "step": 78680
-    },
-    {
-      "epoch": 0.31791755717789083,
-      "grad_norm": 3515.646240234375,
-      "learning_rate": 4.3298119862907744e-05,
-      "loss": 119.0089,
-      "step": 78690
-    },
-    {
-      "epoch": 0.31795795844325847,
-      "grad_norm": 1024.9483642578125,
-      "learning_rate": 4.329574120072728e-05,
-      "loss": 58.4423,
-      "step": 78700
-    },
-    {
-      "epoch": 0.31799835970862605,
-      "grad_norm": 969.68115234375,
-      "learning_rate": 4.329336218186192e-05,
-      "loss": 55.7534,
-      "step": 78710
-    },
-    {
-      "epoch": 0.3180387609739937,
-      "grad_norm": 602.466552734375,
-      "learning_rate": 4.3290982806358046e-05,
-      "loss": 75.1919,
-      "step": 78720
-    },
-    {
-      "epoch": 0.31807916223936133,
-      "grad_norm": 599.2578735351562,
-      "learning_rate": 4.3288603074262054e-05,
-      "loss": 105.2766,
-      "step": 78730
-    },
-    {
-      "epoch": 0.318119563504729,
-      "grad_norm": 790.10693359375,
-      "learning_rate": 4.328622298562033e-05,
-      "loss": 53.7875,
-      "step": 78740
-    },
-    {
-      "epoch": 0.3181599647700966,
-      "grad_norm": 551.5433349609375,
-      "learning_rate": 4.3283842540479264e-05,
-      "loss": 63.0458,
-      "step": 78750
-    },
-    {
-      "epoch": 0.31820036603546425,
-      "grad_norm": 0.0,
-      "learning_rate": 4.3281461738885274e-05,
-      "loss": 61.5794,
-      "step": 78760
-    },
-    {
-      "epoch": 0.31824076730083184,
-      "grad_norm": 681.5363159179688,
-      "learning_rate": 4.327908058088479e-05,
-      "loss": 67.191,
-      "step": 78770
-    },
-    {
-      "epoch": 0.3182811685661995,
-      "grad_norm": 1063.3834228515625,
-      "learning_rate": 4.327669906652421e-05,
-      "loss": 82.7017,
-      "step": 78780
-    },
-    {
-      "epoch": 0.3183215698315671,
-      "grad_norm": 500.39361572265625,
-      "learning_rate": 4.327431719584997e-05,
-      "loss": 85.2169,
-      "step": 78790
-    },
-    {
-      "epoch": 0.31836197109693476,
-      "grad_norm": 1306.898681640625,
-      "learning_rate": 4.3271934968908514e-05,
-      "loss": 72.2316,
-      "step": 78800
-    },
-    {
-      "epoch": 0.3184023723623024,
-      "grad_norm": 815.1703491210938,
-      "learning_rate": 4.326955238574627e-05,
-      "loss": 50.3086,
-      "step": 78810
-    },
-    {
-      "epoch": 0.31844277362767004,
-      "grad_norm": 460.31005859375,
-      "learning_rate": 4.32671694464097e-05,
-      "loss": 64.6161,
-      "step": 78820
-    },
-    {
-      "epoch": 0.3184831748930376,
-      "grad_norm": 2179.723388671875,
-      "learning_rate": 4.326478615094526e-05,
-      "loss": 49.6086,
-      "step": 78830
-    },
-    {
-      "epoch": 0.31852357615840526,
-      "grad_norm": 1665.5799560546875,
-      "learning_rate": 4.3262402499399404e-05,
-      "loss": 97.2382,
-      "step": 78840
-    },
-    {
-      "epoch": 0.3185639774237729,
-      "grad_norm": 1162.87841796875,
-      "learning_rate": 4.326001849181862e-05,
-      "loss": 57.7263,
-      "step": 78850
-    },
-    {
-      "epoch": 0.31860437868914054,
-      "grad_norm": 333.8135681152344,
-      "learning_rate": 4.325763412824937e-05,
-      "loss": 49.9047,
-      "step": 78860
-    },
-    {
-      "epoch": 0.3186447799545082,
-      "grad_norm": 510.8147888183594,
-      "learning_rate": 4.325524940873814e-05,
-      "loss": 82.3427,
-      "step": 78870
-    },
-    {
-      "epoch": 0.3186851812198758,
-      "grad_norm": 434.4716491699219,
-      "learning_rate": 4.325286433333142e-05,
-      "loss": 71.3787,
-      "step": 78880
-    },
-    {
-      "epoch": 0.31872558248524346,
-      "grad_norm": 725.61279296875,
-      "learning_rate": 4.325047890207572e-05,
-      "loss": 63.1463,
-      "step": 78890
-    },
-    {
-      "epoch": 0.31876598375061105,
-      "grad_norm": 366.7152404785156,
-      "learning_rate": 4.324809311501754e-05,
-      "loss": 66.2424,
-      "step": 78900
-    },
-    {
-      "epoch": 0.3188063850159787,
-      "grad_norm": 1318.6060791015625,
-      "learning_rate": 4.3245706972203385e-05,
-      "loss": 67.3569,
-      "step": 78910
-    },
-    {
-      "epoch": 0.31884678628134633,
-      "grad_norm": 727.7313842773438,
-      "learning_rate": 4.3243320473679785e-05,
-      "loss": 63.7311,
-      "step": 78920
-    },
-    {
-      "epoch": 0.31888718754671397,
-      "grad_norm": 1180.8690185546875,
-      "learning_rate": 4.324093361949325e-05,
-      "loss": 54.1359,
-      "step": 78930
-    },
-    {
-      "epoch": 0.3189275888120816,
-      "grad_norm": 652.950439453125,
-      "learning_rate": 4.323854640969033e-05,
-      "loss": 66.1783,
-      "step": 78940
-    },
-    {
-      "epoch": 0.31896799007744925,
-      "grad_norm": 554.9948120117188,
-      "learning_rate": 4.323615884431756e-05,
-      "loss": 77.2896,
-      "step": 78950
-    },
-    {
-      "epoch": 0.31900839134281683,
-      "grad_norm": 657.7294921875,
-      "learning_rate": 4.323377092342148e-05,
-      "loss": 56.7781,
-      "step": 78960
-    },
-    {
-      "epoch": 0.3190487926081845,
-      "grad_norm": 1056.711181640625,
-      "learning_rate": 4.323138264704864e-05,
-      "loss": 51.1476,
-      "step": 78970
-    },
-    {
-      "epoch": 0.3190891938735521,
-      "grad_norm": 1820.9912109375,
-      "learning_rate": 4.322899401524563e-05,
-      "loss": 90.834,
-      "step": 78980
-    },
-    {
-      "epoch": 0.31912959513891975,
-      "grad_norm": 1768.5189208984375,
-      "learning_rate": 4.322660502805899e-05,
-      "loss": 79.8078,
-      "step": 78990
-    },
-    {
-      "epoch": 0.3191699964042874,
-      "grad_norm": 601.8983154296875,
-      "learning_rate": 4.3224215685535294e-05,
-      "loss": 58.5694,
-      "step": 79000
-    },
-    {
-      "epoch": 0.31921039766965503,
-      "grad_norm": 1303.4908447265625,
-      "learning_rate": 4.322182598772113e-05,
-      "loss": 45.729,
-      "step": 79010
-    },
-    {
-      "epoch": 0.3192507989350226,
-      "grad_norm": 480.8334655761719,
-      "learning_rate": 4.321943593466309e-05,
-      "loss": 84.9618,
-      "step": 79020
-    },
-    {
-      "epoch": 0.31929120020039026,
-      "grad_norm": 932.1970825195312,
-      "learning_rate": 4.321704552640777e-05,
-      "loss": 67.6896,
-      "step": 79030
-    },
-    {
-      "epoch": 0.3193316014657579,
-      "grad_norm": 1057.0179443359375,
-      "learning_rate": 4.321465476300177e-05,
-      "loss": 100.6307,
-      "step": 79040
-    },
-    {
-      "epoch": 0.31937200273112554,
-      "grad_norm": 996.9810791015625,
-      "learning_rate": 4.321226364449169e-05,
-      "loss": 71.2474,
-      "step": 79050
-    },
-    {
-      "epoch": 0.3194124039964932,
-      "grad_norm": 568.8705444335938,
-      "learning_rate": 4.320987217092416e-05,
-      "loss": 43.2181,
-      "step": 79060
-    },
-    {
-      "epoch": 0.3194528052618608,
-      "grad_norm": 420.3672180175781,
-      "learning_rate": 4.320748034234579e-05,
-      "loss": 57.1254,
-      "step": 79070
-    },
-    {
-      "epoch": 0.31949320652722846,
-      "grad_norm": 1757.289306640625,
-      "learning_rate": 4.3205088158803226e-05,
-      "loss": 68.5235,
-      "step": 79080
-    },
-    {
-      "epoch": 0.31953360779259604,
-      "grad_norm": 660.7200927734375,
-      "learning_rate": 4.3202695620343083e-05,
-      "loss": 50.2219,
-      "step": 79090
-    },
-    {
-      "epoch": 0.3195740090579637,
-      "grad_norm": 1262.357421875,
-      "learning_rate": 4.320030272701203e-05,
-      "loss": 58.4075,
-      "step": 79100
-    },
-    {
-      "epoch": 0.3196144103233313,
-      "grad_norm": 898.780029296875,
-      "learning_rate": 4.31979094788567e-05,
-      "loss": 88.9766,
-      "step": 79110
-    },
-    {
-      "epoch": 0.31965481158869896,
-      "grad_norm": 535.6188354492188,
-      "learning_rate": 4.319551587592376e-05,
-      "loss": 30.2677,
-      "step": 79120
-    },
-    {
-      "epoch": 0.3196952128540666,
-      "grad_norm": 559.2113647460938,
-      "learning_rate": 4.319312191825987e-05,
-      "loss": 78.3262,
-      "step": 79130
-    },
-    {
-      "epoch": 0.31973561411943424,
-      "grad_norm": 952.9019775390625,
-      "learning_rate": 4.31907276059117e-05,
-      "loss": 67.9382,
-      "step": 79140
-    },
-    {
-      "epoch": 0.3197760153848018,
-      "grad_norm": 959.8912353515625,
-      "learning_rate": 4.318833293892592e-05,
-      "loss": 68.8999,
-      "step": 79150
-    },
-    {
-      "epoch": 0.31981641665016947,
-      "grad_norm": 985.2177124023438,
-      "learning_rate": 4.318593791734924e-05,
-      "loss": 73.2616,
-      "step": 79160
-    },
-    {
-      "epoch": 0.3198568179155371,
-      "grad_norm": 1765.3321533203125,
-      "learning_rate": 4.318354254122833e-05,
-      "loss": 101.728,
-      "step": 79170
-    },
-    {
-      "epoch": 0.31989721918090475,
-      "grad_norm": 2424.275390625,
-      "learning_rate": 4.31811468106099e-05,
-      "loss": 40.6484,
-      "step": 79180
-    },
-    {
-      "epoch": 0.3199376204462724,
-      "grad_norm": 529.9296264648438,
-      "learning_rate": 4.317875072554065e-05,
-      "loss": 56.8446,
-      "step": 79190
-    },
-    {
-      "epoch": 0.31997802171164,
-      "grad_norm": 988.1986083984375,
-      "learning_rate": 4.31763542860673e-05,
-      "loss": 89.1671,
-      "step": 79200
-    },
-    {
-      "epoch": 0.32001842297700767,
-      "grad_norm": 954.2215576171875,
-      "learning_rate": 4.317395749223656e-05,
-      "loss": 62.5503,
-      "step": 79210
-    },
-    {
-      "epoch": 0.32005882424237525,
-      "grad_norm": 1343.7386474609375,
-      "learning_rate": 4.3171560344095164e-05,
-      "loss": 85.5978,
-      "step": 79220
-    },
-    {
-      "epoch": 0.3200992255077429,
-      "grad_norm": 992.0386962890625,
-      "learning_rate": 4.3169162841689846e-05,
-      "loss": 77.3141,
-      "step": 79230
-    },
-    {
-      "epoch": 0.32013962677311053,
-      "grad_norm": 684.0596313476562,
-      "learning_rate": 4.3166764985067343e-05,
-      "loss": 68.0206,
-      "step": 79240
-    },
-    {
-      "epoch": 0.32018002803847817,
-      "grad_norm": 544.3250732421875,
-      "learning_rate": 4.31643667742744e-05,
-      "loss": 64.4232,
-      "step": 79250
-    },
-    {
-      "epoch": 0.3202204293038458,
-      "grad_norm": 485.05413818359375,
-      "learning_rate": 4.3161968209357776e-05,
-      "loss": 72.3071,
-      "step": 79260
-    },
-    {
-      "epoch": 0.32026083056921345,
-      "grad_norm": 1109.7979736328125,
-      "learning_rate": 4.315956929036423e-05,
-      "loss": 87.8763,
-      "step": 79270
-    },
-    {
-      "epoch": 0.32030123183458103,
-      "grad_norm": 735.7939453125,
-      "learning_rate": 4.3157170017340545e-05,
-      "loss": 60.9775,
-      "step": 79280
-    },
-    {
-      "epoch": 0.3203416330999487,
-      "grad_norm": 515.8373413085938,
-      "learning_rate": 4.3154770390333463e-05,
-      "loss": 57.4723,
-      "step": 79290
-    },
-    {
-      "epoch": 0.3203820343653163,
-      "grad_norm": 11893.99609375,
-      "learning_rate": 4.3152370409389795e-05,
-      "loss": 97.6898,
-      "step": 79300
-    },
-    {
-      "epoch": 0.32042243563068395,
-      "grad_norm": 583.2001953125,
-      "learning_rate": 4.3149970074556324e-05,
-      "loss": 104.1178,
-      "step": 79310
-    },
-    {
-      "epoch": 0.3204628368960516,
-      "grad_norm": 1154.815185546875,
-      "learning_rate": 4.314756938587984e-05,
-      "loss": 47.7879,
-      "step": 79320
-    },
-    {
-      "epoch": 0.32050323816141923,
-      "grad_norm": 1132.3023681640625,
-      "learning_rate": 4.314516834340715e-05,
-      "loss": 59.4125,
-      "step": 79330
-    },
-    {
-      "epoch": 0.3205436394267868,
-      "grad_norm": 530.7276611328125,
-      "learning_rate": 4.3142766947185056e-05,
-      "loss": 76.4587,
-      "step": 79340
-    },
-    {
-      "epoch": 0.32058404069215446,
-      "grad_norm": 943.7814331054688,
-      "learning_rate": 4.314036519726038e-05,
-      "loss": 83.0291,
-      "step": 79350
-    },
-    {
-      "epoch": 0.3206244419575221,
-      "grad_norm": 1113.9444580078125,
-      "learning_rate": 4.3137963093679945e-05,
-      "loss": 62.54,
-      "step": 79360
-    },
-    {
-      "epoch": 0.32066484322288974,
-      "grad_norm": 839.0965576171875,
-      "learning_rate": 4.313556063649059e-05,
-      "loss": 65.0847,
-      "step": 79370
-    },
-    {
-      "epoch": 0.3207052444882574,
-      "grad_norm": 645.9111938476562,
-      "learning_rate": 4.313315782573913e-05,
-      "loss": 66.1717,
-      "step": 79380
-    },
-    {
-      "epoch": 0.320745645753625,
-      "grad_norm": 1160.1456298828125,
-      "learning_rate": 4.3130754661472435e-05,
-      "loss": 72.6717,
-      "step": 79390
-    },
-    {
-      "epoch": 0.32078604701899266,
-      "grad_norm": 861.6610107421875,
-      "learning_rate": 4.3128351143737335e-05,
-      "loss": 94.8734,
-      "step": 79400
-    },
-    {
-      "epoch": 0.32082644828436024,
-      "grad_norm": 1239.6209716796875,
-      "learning_rate": 4.31259472725807e-05,
-      "loss": 65.5956,
-      "step": 79410
-    },
-    {
-      "epoch": 0.3208668495497279,
-      "grad_norm": 1918.2100830078125,
-      "learning_rate": 4.312354304804939e-05,
-      "loss": 98.4281,
-      "step": 79420
-    },
-    {
-      "epoch": 0.3209072508150955,
-      "grad_norm": 496.2858581542969,
-      "learning_rate": 4.312113847019028e-05,
-      "loss": 64.9097,
-      "step": 79430
-    },
-    {
-      "epoch": 0.32094765208046316,
-      "grad_norm": 568.856689453125,
-      "learning_rate": 4.3118733539050244e-05,
-      "loss": 80.5664,
-      "step": 79440
-    },
-    {
-      "epoch": 0.3209880533458308,
-      "grad_norm": 647.578369140625,
-      "learning_rate": 4.311632825467617e-05,
-      "loss": 81.0642,
-      "step": 79450
-    },
-    {
-      "epoch": 0.32102845461119844,
-      "grad_norm": 606.3375854492188,
-      "learning_rate": 4.311392261711495e-05,
-      "loss": 87.3034,
-      "step": 79460
-    },
-    {
-      "epoch": 0.32106885587656603,
-      "grad_norm": 486.86273193359375,
-      "learning_rate": 4.3111516626413485e-05,
-      "loss": 36.4118,
-      "step": 79470
-    },
-    {
-      "epoch": 0.32110925714193367,
-      "grad_norm": 527.7968139648438,
-      "learning_rate": 4.310911028261867e-05,
-      "loss": 75.8701,
-      "step": 79480
-    },
-    {
-      "epoch": 0.3211496584073013,
-      "grad_norm": 899.1015014648438,
-      "learning_rate": 4.310670358577744e-05,
-      "loss": 64.9072,
-      "step": 79490
-    },
-    {
-      "epoch": 0.32119005967266895,
-      "grad_norm": 491.0596923828125,
-      "learning_rate": 4.3104296535936695e-05,
-      "loss": 62.814,
-      "step": 79500
-    },
-    {
-      "epoch": 0.3212304609380366,
-      "grad_norm": 643.3750610351562,
-      "learning_rate": 4.3101889133143365e-05,
-      "loss": 62.7745,
-      "step": 79510
-    },
-    {
-      "epoch": 0.32127086220340423,
-      "grad_norm": 584.9881591796875,
-      "learning_rate": 4.3099481377444384e-05,
-      "loss": 70.7787,
-      "step": 79520
-    },
-    {
-      "epoch": 0.32131126346877187,
-      "grad_norm": 0.0,
-      "learning_rate": 4.30970732688867e-05,
-      "loss": 72.2548,
-      "step": 79530
-    },
-    {
-      "epoch": 0.32135166473413945,
-      "grad_norm": 551.8460693359375,
-      "learning_rate": 4.309466480751726e-05,
-      "loss": 57.843,
-      "step": 79540
-    },
-    {
-      "epoch": 0.3213920659995071,
-      "grad_norm": 1137.9078369140625,
-      "learning_rate": 4.309225599338301e-05,
-      "loss": 67.0156,
-      "step": 79550
-    },
-    {
-      "epoch": 0.32143246726487473,
-      "grad_norm": 1040.0809326171875,
-      "learning_rate": 4.308984682653092e-05,
-      "loss": 51.4768,
-      "step": 79560
-    },
-    {
-      "epoch": 0.3214728685302424,
-      "grad_norm": 496.1695556640625,
-      "learning_rate": 4.308743730700795e-05,
-      "loss": 49.8229,
-      "step": 79570
-    },
-    {
-      "epoch": 0.32151326979561,
-      "grad_norm": 928.9419555664062,
-      "learning_rate": 4.308502743486107e-05,
-      "loss": 68.5493,
-      "step": 79580
-    },
-    {
-      "epoch": 0.32155367106097765,
-      "grad_norm": 554.7418212890625,
-      "learning_rate": 4.308261721013728e-05,
-      "loss": 58.4424,
-      "step": 79590
-    },
-    {
-      "epoch": 0.32159407232634524,
-      "grad_norm": 530.4016723632812,
-      "learning_rate": 4.3080206632883554e-05,
-      "loss": 71.472,
-      "step": 79600
-    },
-    {
-      "epoch": 0.3216344735917129,
-      "grad_norm": 823.0756225585938,
-      "learning_rate": 4.307779570314689e-05,
-      "loss": 54.4288,
-      "step": 79610
-    },
-    {
-      "epoch": 0.3216748748570805,
-      "grad_norm": 523.12744140625,
-      "learning_rate": 4.307538442097429e-05,
-      "loss": 52.3814,
-      "step": 79620
-    },
-    {
-      "epoch": 0.32171527612244816,
-      "grad_norm": 1036.486083984375,
-      "learning_rate": 4.307297278641277e-05,
-      "loss": 56.4284,
-      "step": 79630
-    },
-    {
-      "epoch": 0.3217556773878158,
-      "grad_norm": 886.7937622070312,
-      "learning_rate": 4.307056079950934e-05,
-      "loss": 61.4247,
-      "step": 79640
-    },
-    {
-      "epoch": 0.32179607865318344,
-      "grad_norm": 1947.759033203125,
-      "learning_rate": 4.306814846031102e-05,
-      "loss": 69.1207,
-      "step": 79650
-    },
-    {
-      "epoch": 0.321836479918551,
-      "grad_norm": 1469.478759765625,
-      "learning_rate": 4.306573576886484e-05,
-      "loss": 76.3007,
-      "step": 79660
-    },
-    {
-      "epoch": 0.32187688118391866,
-      "grad_norm": 883.7667846679688,
-      "learning_rate": 4.306332272521785e-05,
-      "loss": 45.0061,
-      "step": 79670
-    },
-    {
-      "epoch": 0.3219172824492863,
-      "grad_norm": 418.3061828613281,
-      "learning_rate": 4.306090932941708e-05,
-      "loss": 45.161,
-      "step": 79680
-    },
-    {
-      "epoch": 0.32195768371465394,
-      "grad_norm": 504.8469543457031,
-      "learning_rate": 4.3058495581509586e-05,
-      "loss": 89.2983,
-      "step": 79690
-    },
-    {
-      "epoch": 0.3219980849800216,
-      "grad_norm": 510.06805419921875,
-      "learning_rate": 4.305608148154242e-05,
-      "loss": 57.7559,
-      "step": 79700
-    },
-    {
-      "epoch": 0.3220384862453892,
-      "grad_norm": 1332.9613037109375,
-      "learning_rate": 4.305366702956265e-05,
-      "loss": 54.3312,
-      "step": 79710
-    },
-    {
-      "epoch": 0.32207888751075686,
-      "grad_norm": 608.1598510742188,
-      "learning_rate": 4.305125222561736e-05,
-      "loss": 63.2103,
-      "step": 79720
-    },
-    {
-      "epoch": 0.32211928877612445,
-      "grad_norm": 504.0339050292969,
-      "learning_rate": 4.304883706975359e-05,
-      "loss": 45.3614,
-      "step": 79730
-    },
-    {
-      "epoch": 0.3221596900414921,
-      "grad_norm": 865.779052734375,
-      "learning_rate": 4.304642156201847e-05,
-      "loss": 77.652,
-      "step": 79740
-    },
-    {
-      "epoch": 0.3222000913068597,
-      "grad_norm": 1158.892333984375,
-      "learning_rate": 4.304400570245906e-05,
-      "loss": 46.2974,
-      "step": 79750
-    },
-    {
-      "epoch": 0.32224049257222737,
-      "grad_norm": 1002.193603515625,
-      "learning_rate": 4.304158949112247e-05,
-      "loss": 78.2484,
-      "step": 79760
-    },
-    {
-      "epoch": 0.322280893837595,
-      "grad_norm": 1321.477294921875,
-      "learning_rate": 4.3039172928055805e-05,
-      "loss": 72.8844,
-      "step": 79770
-    },
-    {
-      "epoch": 0.32232129510296265,
-      "grad_norm": 188.91424560546875,
-      "learning_rate": 4.303675601330618e-05,
-      "loss": 47.1185,
-      "step": 79780
-    },
-    {
-      "epoch": 0.32236169636833023,
-      "grad_norm": 535.0440673828125,
-      "learning_rate": 4.3034338746920707e-05,
-      "loss": 63.9909,
-      "step": 79790
-    },
-    {
-      "epoch": 0.32240209763369787,
-      "grad_norm": 474.7372741699219,
-      "learning_rate": 4.303192112894652e-05,
-      "loss": 73.8586,
-      "step": 79800
-    },
-    {
-      "epoch": 0.3224424988990655,
-      "grad_norm": 813.8701782226562,
-      "learning_rate": 4.302950315943074e-05,
-      "loss": 90.6143,
-      "step": 79810
-    },
-    {
-      "epoch": 0.32248290016443315,
-      "grad_norm": 344.40325927734375,
-      "learning_rate": 4.3027084838420516e-05,
-      "loss": 36.8543,
-      "step": 79820
-    },
-    {
-      "epoch": 0.3225233014298008,
-      "grad_norm": 300.6861267089844,
-      "learning_rate": 4.302466616596299e-05,
-      "loss": 48.8195,
-      "step": 79830
-    },
-    {
-      "epoch": 0.32256370269516843,
-      "grad_norm": 782.69970703125,
-      "learning_rate": 4.302224714210532e-05,
-      "loss": 59.1578,
-      "step": 79840
-    },
-    {
-      "epoch": 0.32260410396053607,
-      "grad_norm": 801.529541015625,
-      "learning_rate": 4.301982776689467e-05,
-      "loss": 44.7871,
-      "step": 79850
-    },
-    {
-      "epoch": 0.32264450522590365,
-      "grad_norm": 926.8488159179688,
-      "learning_rate": 4.301740804037819e-05,
-      "loss": 65.3786,
-      "step": 79860
-    },
-    {
-      "epoch": 0.3226849064912713,
-      "grad_norm": 917.0177001953125,
-      "learning_rate": 4.301498796260307e-05,
-      "loss": 55.9066,
-      "step": 79870
-    },
-    {
-      "epoch": 0.32272530775663893,
-      "grad_norm": 0.0,
-      "learning_rate": 4.301256753361649e-05,
-      "loss": 50.5934,
-      "step": 79880
-    },
-    {
-      "epoch": 0.3227657090220066,
-      "grad_norm": 771.5294799804688,
-      "learning_rate": 4.301014675346562e-05,
-      "loss": 109.5298,
-      "step": 79890
-    },
-    {
-      "epoch": 0.3228061102873742,
-      "grad_norm": 562.3530883789062,
-      "learning_rate": 4.3007725622197674e-05,
-      "loss": 43.964,
-      "step": 79900
-    },
-    {
-      "epoch": 0.32284651155274185,
-      "grad_norm": 468.8666076660156,
-      "learning_rate": 4.300530413985985e-05,
-      "loss": 49.0986,
-      "step": 79910
-    },
-    {
-      "epoch": 0.32288691281810944,
-      "grad_norm": 678.0904541015625,
-      "learning_rate": 4.3002882306499345e-05,
-      "loss": 91.667,
-      "step": 79920
-    },
-    {
-      "epoch": 0.3229273140834771,
-      "grad_norm": 601.5751342773438,
-      "learning_rate": 4.300046012216338e-05,
-      "loss": 52.2733,
-      "step": 79930
-    },
-    {
-      "epoch": 0.3229677153488447,
-      "grad_norm": 1330.5103759765625,
-      "learning_rate": 4.299803758689919e-05,
-      "loss": 82.389,
-      "step": 79940
-    },
-    {
-      "epoch": 0.32300811661421236,
-      "grad_norm": 509.53277587890625,
-      "learning_rate": 4.299561470075397e-05,
-      "loss": 78.0836,
-      "step": 79950
-    },
-    {
-      "epoch": 0.32304851787958,
-      "grad_norm": 589.202392578125,
-      "learning_rate": 4.2993191463774997e-05,
-      "loss": 60.7983,
-      "step": 79960
-    },
-    {
-      "epoch": 0.32308891914494764,
-      "grad_norm": 738.194580078125,
-      "learning_rate": 4.299076787600948e-05,
-      "loss": 77.6573,
-      "step": 79970
-    },
-    {
-      "epoch": 0.3231293204103152,
-      "grad_norm": 1307.095947265625,
-      "learning_rate": 4.2988343937504686e-05,
-      "loss": 61.8344,
-      "step": 79980
-    },
-    {
-      "epoch": 0.32316972167568286,
-      "grad_norm": 656.2578125,
-      "learning_rate": 4.298591964830787e-05,
-      "loss": 95.2794,
-      "step": 79990
-    },
-    {
-      "epoch": 0.3232101229410505,
-      "grad_norm": 1574.1217041015625,
-      "learning_rate": 4.2983495008466276e-05,
-      "loss": 46.9904,
-      "step": 80000
-    },
-    {
-      "epoch": 0.32325052420641814,
-      "grad_norm": 678.7640380859375,
-      "learning_rate": 4.2981070018027204e-05,
-      "loss": 83.523,
-      "step": 80010
-    },
-    {
-      "epoch": 0.3232909254717858,
-      "grad_norm": 802.5941772460938,
-      "learning_rate": 4.29786446770379e-05,
-      "loss": 47.3945,
-      "step": 80020
-    },
-    {
-      "epoch": 0.3233313267371534,
-      "grad_norm": 880.6734008789062,
-      "learning_rate": 4.297621898554568e-05,
-      "loss": 69.5324,
-      "step": 80030
-    },
-    {
-      "epoch": 0.32337172800252106,
-      "grad_norm": 1087.5975341796875,
-      "learning_rate": 4.297379294359781e-05,
-      "loss": 68.7424,
-      "step": 80040
-    },
-    {
-      "epoch": 0.32341212926788865,
-      "grad_norm": 1480.21826171875,
-      "learning_rate": 4.297136655124159e-05,
-      "loss": 68.1467,
-      "step": 80050
-    },
-    {
-      "epoch": 0.3234525305332563,
-      "grad_norm": 375.3006286621094,
-      "learning_rate": 4.2968939808524323e-05,
-      "loss": 58.7787,
-      "step": 80060
-    },
-    {
-      "epoch": 0.32349293179862393,
-      "grad_norm": 811.9970092773438,
-      "learning_rate": 4.296651271549333e-05,
-      "loss": 72.5431,
-      "step": 80070
-    },
-    {
-      "epoch": 0.32353333306399157,
-      "grad_norm": 533.86474609375,
-      "learning_rate": 4.296408527219592e-05,
-      "loss": 87.0758,
-      "step": 80080
-    },
-    {
-      "epoch": 0.3235737343293592,
-      "grad_norm": 850.876220703125,
-      "learning_rate": 4.296165747867942e-05,
-      "loss": 41.6743,
-      "step": 80090
-    },
-    {
-      "epoch": 0.32361413559472685,
-      "grad_norm": 924.040283203125,
-      "learning_rate": 4.2959229334991156e-05,
-      "loss": 69.6225,
-      "step": 80100
-    },
-    {
-      "epoch": 0.32365453686009443,
-      "grad_norm": 351.706298828125,
-      "learning_rate": 4.295680084117847e-05,
-      "loss": 60.5032,
-      "step": 80110
-    },
-    {
-      "epoch": 0.3236949381254621,
-      "grad_norm": 893.626708984375,
-      "learning_rate": 4.295437199728871e-05,
-      "loss": 77.7679,
-      "step": 80120
-    },
-    {
-      "epoch": 0.3237353393908297,
-      "grad_norm": 652.7449951171875,
-      "learning_rate": 4.2951942803369225e-05,
-      "loss": 54.3826,
-      "step": 80130
-    },
-    {
-      "epoch": 0.32377574065619735,
-      "grad_norm": 1125.824462890625,
-      "learning_rate": 4.294951325946737e-05,
-      "loss": 93.5955,
-      "step": 80140
-    },
-    {
-      "epoch": 0.323816141921565,
-      "grad_norm": 1012.917236328125,
-      "learning_rate": 4.2947083365630514e-05,
-      "loss": 61.9179,
-      "step": 80150
-    },
-    {
-      "epoch": 0.32385654318693263,
-      "grad_norm": 807.2591552734375,
-      "learning_rate": 4.294465312190603e-05,
-      "loss": 46.5419,
-      "step": 80160
-    },
-    {
-      "epoch": 0.3238969444523003,
-      "grad_norm": 317.6124572753906,
-      "learning_rate": 4.294222252834129e-05,
-      "loss": 68.4578,
-      "step": 80170
-    },
-    {
-      "epoch": 0.32393734571766786,
-      "grad_norm": 2099.4423828125,
-      "learning_rate": 4.293979158498369e-05,
-      "loss": 87.771,
-      "step": 80180
-    },
-    {
-      "epoch": 0.3239777469830355,
-      "grad_norm": 482.209228515625,
-      "learning_rate": 4.293736029188061e-05,
-      "loss": 73.7765,
-      "step": 80190
-    },
-    {
-      "epoch": 0.32401814824840314,
-      "grad_norm": 1821.595947265625,
-      "learning_rate": 4.293492864907947e-05,
-      "loss": 84.6861,
-      "step": 80200
-    },
-    {
-      "epoch": 0.3240585495137708,
-      "grad_norm": 640.3442993164062,
-      "learning_rate": 4.293249665662765e-05,
-      "loss": 55.7724,
-      "step": 80210
-    },
-    {
-      "epoch": 0.3240989507791384,
-      "grad_norm": 974.443115234375,
-      "learning_rate": 4.293006431457258e-05,
-      "loss": 89.1751,
-      "step": 80220
-    },
-    {
-      "epoch": 0.32413935204450606,
-      "grad_norm": 625.10888671875,
-      "learning_rate": 4.2927631622961674e-05,
-      "loss": 77.496,
-      "step": 80230
-    },
-    {
-      "epoch": 0.32417975330987364,
-      "grad_norm": 267.2085266113281,
-      "learning_rate": 4.292519858184236e-05,
-      "loss": 54.031,
-      "step": 80240
-    },
-    {
-      "epoch": 0.3242201545752413,
-      "grad_norm": 543.6681518554688,
-      "learning_rate": 4.292276519126207e-05,
-      "loss": 61.7213,
-      "step": 80250
-    },
-    {
-      "epoch": 0.3242605558406089,
-      "grad_norm": 1130.407958984375,
-      "learning_rate": 4.292033145126825e-05,
-      "loss": 61.5954,
-      "step": 80260
-    },
-    {
-      "epoch": 0.32430095710597656,
-      "grad_norm": 590.1688842773438,
-      "learning_rate": 4.2917897361908335e-05,
-      "loss": 54.5805,
-      "step": 80270
-    },
-    {
-      "epoch": 0.3243413583713442,
-      "grad_norm": 1286.2869873046875,
-      "learning_rate": 4.291546292322979e-05,
-      "loss": 94.4061,
-      "step": 80280
-    },
-    {
-      "epoch": 0.32438175963671184,
-      "grad_norm": 520.0972290039062,
-      "learning_rate": 4.2913028135280076e-05,
-      "loss": 49.1347,
-      "step": 80290
-    },
-    {
-      "epoch": 0.3244221609020794,
-      "grad_norm": 775.9698486328125,
-      "learning_rate": 4.291059299810665e-05,
-      "loss": 58.7011,
-      "step": 80300
-    },
-    {
-      "epoch": 0.32446256216744707,
-      "grad_norm": 276.6414794921875,
-      "learning_rate": 4.2908157511757e-05,
-      "loss": 70.6107,
-      "step": 80310
-    },
-    {
-      "epoch": 0.3245029634328147,
-      "grad_norm": 446.62158203125,
-      "learning_rate": 4.290572167627859e-05,
-      "loss": 59.438,
-      "step": 80320
-    },
-    {
-      "epoch": 0.32454336469818235,
-      "grad_norm": 692.0556030273438,
-      "learning_rate": 4.290328549171893e-05,
-      "loss": 62.021,
-      "step": 80330
-    },
-    {
-      "epoch": 0.32458376596355,
-      "grad_norm": 446.0535583496094,
-      "learning_rate": 4.2900848958125485e-05,
-      "loss": 56.7572,
-      "step": 80340
-    },
-    {
-      "epoch": 0.3246241672289176,
-      "grad_norm": 477.32568359375,
-      "learning_rate": 4.289841207554578e-05,
-      "loss": 110.483,
-      "step": 80350
-    },
-    {
-      "epoch": 0.32466456849428527,
-      "grad_norm": 700.8925170898438,
-      "learning_rate": 4.289597484402732e-05,
-      "loss": 65.4878,
-      "step": 80360
-    },
-    {
-      "epoch": 0.32470496975965285,
-      "grad_norm": 1176.870849609375,
-      "learning_rate": 4.289353726361762e-05,
-      "loss": 61.7405,
-      "step": 80370
-    },
-    {
-      "epoch": 0.3247453710250205,
-      "grad_norm": 574.3339233398438,
-      "learning_rate": 4.289109933436419e-05,
-      "loss": 58.5538,
-      "step": 80380
-    },
-    {
-      "epoch": 0.32478577229038813,
-      "grad_norm": 928.0294799804688,
-      "learning_rate": 4.2888661056314574e-05,
-      "loss": 62.1923,
-      "step": 80390
-    },
-    {
-      "epoch": 0.32482617355575577,
-      "grad_norm": 1341.2864990234375,
-      "learning_rate": 4.2886222429516296e-05,
-      "loss": 67.9961,
-      "step": 80400
-    },
-    {
-      "epoch": 0.3248665748211234,
-      "grad_norm": 575.213134765625,
-      "learning_rate": 4.2883783454016915e-05,
-      "loss": 43.7556,
-      "step": 80410
-    },
-    {
-      "epoch": 0.32490697608649105,
-      "grad_norm": 637.3678588867188,
-      "learning_rate": 4.288134412986395e-05,
-      "loss": 72.743,
-      "step": 80420
-    },
-    {
-      "epoch": 0.32494737735185864,
-      "grad_norm": 936.275146484375,
-      "learning_rate": 4.287890445710499e-05,
-      "loss": 89.083,
-      "step": 80430
-    },
-    {
-      "epoch": 0.3249877786172263,
-      "grad_norm": 587.6963500976562,
-      "learning_rate": 4.287646443578758e-05,
-      "loss": 40.0728,
-      "step": 80440
-    },
-    {
-      "epoch": 0.3250281798825939,
-      "grad_norm": 465.3719177246094,
-      "learning_rate": 4.287402406595929e-05,
-      "loss": 53.9417,
-      "step": 80450
-    },
-    {
-      "epoch": 0.32506858114796156,
-      "grad_norm": 862.9938354492188,
-      "learning_rate": 4.28715833476677e-05,
-      "loss": 59.9648,
-      "step": 80460
-    },
-    {
-      "epoch": 0.3251089824133292,
-      "grad_norm": 902.8772583007812,
-      "learning_rate": 4.2869142280960396e-05,
-      "loss": 53.2347,
-      "step": 80470
-    },
-    {
-      "epoch": 0.32514938367869683,
-      "grad_norm": 827.6386108398438,
-      "learning_rate": 4.2866700865884954e-05,
-      "loss": 64.4303,
-      "step": 80480
-    },
-    {
-      "epoch": 0.3251897849440645,
-      "grad_norm": 1176.3564453125,
-      "learning_rate": 4.2864259102488984e-05,
-      "loss": 70.9066,
-      "step": 80490
-    },
-    {
-      "epoch": 0.32523018620943206,
-      "grad_norm": 2174.75244140625,
-      "learning_rate": 4.2861816990820084e-05,
-      "loss": 82.248,
-      "step": 80500
-    },
-    {
-      "epoch": 0.3252705874747997,
-      "grad_norm": 1319.863037109375,
-      "learning_rate": 4.285937453092587e-05,
-      "loss": 73.5724,
-      "step": 80510
-    },
-    {
-      "epoch": 0.32531098874016734,
-      "grad_norm": 802.91845703125,
-      "learning_rate": 4.285693172285396e-05,
-      "loss": 62.4872,
-      "step": 80520
-    },
-    {
-      "epoch": 0.325351390005535,
-      "grad_norm": 735.5004272460938,
-      "learning_rate": 4.2854488566651965e-05,
-      "loss": 80.5372,
-      "step": 80530
-    },
-    {
-      "epoch": 0.3253917912709026,
-      "grad_norm": 813.9871215820312,
-      "learning_rate": 4.2852045062367516e-05,
-      "loss": 59.1823,
-      "step": 80540
-    },
-    {
-      "epoch": 0.32543219253627026,
-      "grad_norm": 911.4254150390625,
-      "learning_rate": 4.2849601210048274e-05,
-      "loss": 73.9285,
-      "step": 80550
-    },
-    {
-      "epoch": 0.32547259380163784,
-      "grad_norm": 1268.9906005859375,
-      "learning_rate": 4.2847157009741856e-05,
-      "loss": 63.6592,
-      "step": 80560
-    },
-    {
-      "epoch": 0.3255129950670055,
-      "grad_norm": 514.0250244140625,
-      "learning_rate": 4.2844712461495926e-05,
-      "loss": 66.8111,
-      "step": 80570
-    },
-    {
-      "epoch": 0.3255533963323731,
-      "grad_norm": 1467.4144287109375,
-      "learning_rate": 4.284226756535814e-05,
-      "loss": 66.5214,
-      "step": 80580
-    },
-    {
-      "epoch": 0.32559379759774076,
-      "grad_norm": 416.23828125,
-      "learning_rate": 4.283982232137617e-05,
-      "loss": 47.4183,
-      "step": 80590
-    },
-    {
-      "epoch": 0.3256341988631084,
-      "grad_norm": 1595.3319091796875,
-      "learning_rate": 4.283737672959766e-05,
-      "loss": 72.357,
-      "step": 80600
-    },
-    {
-      "epoch": 0.32567460012847604,
-      "grad_norm": 1119.335693359375,
-      "learning_rate": 4.283493079007032e-05,
-      "loss": 51.9857,
-      "step": 80610
-    },
-    {
-      "epoch": 0.32571500139384363,
-      "grad_norm": 838.667236328125,
-      "learning_rate": 4.283248450284182e-05,
-      "loss": 55.8164,
-      "step": 80620
-    },
-    {
-      "epoch": 0.32575540265921127,
-      "grad_norm": 394.1136779785156,
-      "learning_rate": 4.283003786795986e-05,
-      "loss": 49.545,
-      "step": 80630
-    },
-    {
-      "epoch": 0.3257958039245789,
-      "grad_norm": 874.3743286132812,
-      "learning_rate": 4.2827590885472125e-05,
-      "loss": 64.6963,
-      "step": 80640
-    },
-    {
-      "epoch": 0.32583620518994655,
-      "grad_norm": 1160.2918701171875,
-      "learning_rate": 4.2825143555426326e-05,
-      "loss": 61.9271,
-      "step": 80650
-    },
-    {
-      "epoch": 0.3258766064553142,
-      "grad_norm": 1602.0166015625,
-      "learning_rate": 4.2822695877870177e-05,
-      "loss": 51.2635,
-      "step": 80660
-    },
-    {
-      "epoch": 0.32591700772068183,
-      "grad_norm": 382.5502014160156,
-      "learning_rate": 4.28202478528514e-05,
-      "loss": 54.0187,
-      "step": 80670
-    },
-    {
-      "epoch": 0.32595740898604947,
-      "grad_norm": 1229.9798583984375,
-      "learning_rate": 4.281779948041772e-05,
-      "loss": 60.0021,
-      "step": 80680
-    },
-    {
-      "epoch": 0.32599781025141705,
-      "grad_norm": 565.9646606445312,
-      "learning_rate": 4.2815350760616864e-05,
-      "loss": 61.7397,
-      "step": 80690
-    },
-    {
-      "epoch": 0.3260382115167847,
-      "grad_norm": 769.6617431640625,
-      "learning_rate": 4.2812901693496564e-05,
-      "loss": 60.3849,
-      "step": 80700
-    },
-    {
-      "epoch": 0.32607861278215233,
-      "grad_norm": 648.3060913085938,
-      "learning_rate": 4.281045227910459e-05,
-      "loss": 75.0846,
-      "step": 80710
-    },
-    {
-      "epoch": 0.32611901404752,
-      "grad_norm": 3067.703857421875,
-      "learning_rate": 4.2808002517488667e-05,
-      "loss": 90.7922,
-      "step": 80720
-    },
-    {
-      "epoch": 0.3261594153128876,
-      "grad_norm": 109.4122085571289,
-      "learning_rate": 4.280555240869657e-05,
-      "loss": 69.8678,
-      "step": 80730
-    },
-    {
-      "epoch": 0.32619981657825525,
-      "grad_norm": 733.5319213867188,
-      "learning_rate": 4.280310195277606e-05,
-      "loss": 81.5929,
-      "step": 80740
-    },
-    {
-      "epoch": 0.32624021784362284,
-      "grad_norm": 610.775634765625,
-      "learning_rate": 4.280065114977492e-05,
-      "loss": 57.4617,
-      "step": 80750
-    },
-    {
-      "epoch": 0.3262806191089905,
-      "grad_norm": 551.3561401367188,
-      "learning_rate": 4.279819999974091e-05,
-      "loss": 61.8333,
-      "step": 80760
-    },
-    {
-      "epoch": 0.3263210203743581,
-      "grad_norm": 685.814697265625,
-      "learning_rate": 4.279574850272183e-05,
-      "loss": 67.4173,
-      "step": 80770
-    },
-    {
-      "epoch": 0.32636142163972576,
-      "grad_norm": 1033.493896484375,
-      "learning_rate": 4.279329665876548e-05,
-      "loss": 53.4654,
-      "step": 80780
-    },
-    {
-      "epoch": 0.3264018229050934,
-      "grad_norm": 485.32745361328125,
-      "learning_rate": 4.2790844467919646e-05,
-      "loss": 54.8732,
-      "step": 80790
-    },
-    {
-      "epoch": 0.32644222417046104,
-      "grad_norm": 0.0,
-      "learning_rate": 4.278839193023214e-05,
-      "loss": 75.1927,
-      "step": 80800
-    },
-    {
-      "epoch": 0.3264826254358287,
-      "grad_norm": 743.0775756835938,
-      "learning_rate": 4.278593904575077e-05,
-      "loss": 44.6109,
-      "step": 80810
-    },
-    {
-      "epoch": 0.32652302670119626,
-      "grad_norm": 1801.4681396484375,
-      "learning_rate": 4.278348581452337e-05,
-      "loss": 76.2488,
-      "step": 80820
-    },
-    {
-      "epoch": 0.3265634279665639,
-      "grad_norm": 1143.9945068359375,
-      "learning_rate": 4.278103223659775e-05,
-      "loss": 68.8127,
-      "step": 80830
-    },
-    {
-      "epoch": 0.32660382923193154,
-      "grad_norm": 568.1923217773438,
-      "learning_rate": 4.2778578312021754e-05,
-      "loss": 46.9349,
-      "step": 80840
-    },
-    {
-      "epoch": 0.3266442304972992,
-      "grad_norm": 758.5953979492188,
-      "learning_rate": 4.277612404084322e-05,
-      "loss": 67.1757,
-      "step": 80850
-    },
-    {
-      "epoch": 0.3266846317626668,
-      "grad_norm": 696.4761352539062,
-      "learning_rate": 4.277366942311001e-05,
-      "loss": 62.2708,
-      "step": 80860
-    },
-    {
-      "epoch": 0.32672503302803446,
-      "grad_norm": 664.263671875,
-      "learning_rate": 4.277121445886995e-05,
-      "loss": 70.6147,
-      "step": 80870
-    },
-    {
-      "epoch": 0.32676543429340205,
-      "grad_norm": 184.5323944091797,
-      "learning_rate": 4.2768759148170915e-05,
-      "loss": 42.6514,
-      "step": 80880
-    },
-    {
-      "epoch": 0.3268058355587697,
-      "grad_norm": 586.1389770507812,
-      "learning_rate": 4.276630349106078e-05,
-      "loss": 38.8002,
-      "step": 80890
-    },
-    {
-      "epoch": 0.3268462368241373,
-      "grad_norm": 1424.904541015625,
-      "learning_rate": 4.276384748758741e-05,
-      "loss": 77.3322,
-      "step": 80900
-    },
-    {
-      "epoch": 0.32688663808950497,
-      "grad_norm": 859.6981811523438,
-      "learning_rate": 4.2761391137798676e-05,
-      "loss": 86.453,
-      "step": 80910
-    },
-    {
-      "epoch": 0.3269270393548726,
-      "grad_norm": 1004.2837524414062,
-      "learning_rate": 4.2758934441742496e-05,
-      "loss": 56.4649,
-      "step": 80920
-    },
-    {
-      "epoch": 0.32696744062024025,
-      "grad_norm": 1162.806396484375,
-      "learning_rate": 4.2756477399466735e-05,
-      "loss": 97.9132,
-      "step": 80930
-    },
-    {
-      "epoch": 0.32700784188560783,
-      "grad_norm": 425.3879699707031,
-      "learning_rate": 4.275402001101931e-05,
-      "loss": 68.8998,
-      "step": 80940
-    },
-    {
-      "epoch": 0.32704824315097547,
-      "grad_norm": 645.8928833007812,
-      "learning_rate": 4.2751562276448124e-05,
-      "loss": 51.0803,
-      "step": 80950
-    },
-    {
-      "epoch": 0.3270886444163431,
-      "grad_norm": 2057.58837890625,
-      "learning_rate": 4.274910419580108e-05,
-      "loss": 60.8407,
-      "step": 80960
-    },
-    {
-      "epoch": 0.32712904568171075,
-      "grad_norm": 1685.50048828125,
-      "learning_rate": 4.274664576912613e-05,
-      "loss": 66.7414,
-      "step": 80970
-    },
-    {
-      "epoch": 0.3271694469470784,
-      "grad_norm": 1081.0294189453125,
-      "learning_rate": 4.2744186996471174e-05,
-      "loss": 87.3173,
-      "step": 80980
-    },
-    {
-      "epoch": 0.32720984821244603,
-      "grad_norm": 1577.8040771484375,
-      "learning_rate": 4.2741727877884155e-05,
-      "loss": 85.7177,
-      "step": 80990
-    },
-    {
-      "epoch": 0.32725024947781367,
-      "grad_norm": 547.41259765625,
-      "learning_rate": 4.273926841341302e-05,
-      "loss": 47.8796,
-      "step": 81000
-    },
-    {
-      "epoch": 0.32729065074318126,
-      "grad_norm": 990.9943237304688,
-      "learning_rate": 4.273680860310572e-05,
-      "loss": 71.2299,
-      "step": 81010
-    },
-    {
-      "epoch": 0.3273310520085489,
-      "grad_norm": 379.1767883300781,
-      "learning_rate": 4.2734348447010206e-05,
-      "loss": 53.3839,
-      "step": 81020
-    },
-    {
-      "epoch": 0.32737145327391654,
-      "grad_norm": 280.44287109375,
-      "learning_rate": 4.2731887945174434e-05,
-      "loss": 66.233,
-      "step": 81030
-    },
-    {
-      "epoch": 0.3274118545392842,
-      "grad_norm": 659.9508056640625,
-      "learning_rate": 4.272942709764638e-05,
-      "loss": 80.5297,
-      "step": 81040
-    },
-    {
-      "epoch": 0.3274522558046518,
-      "grad_norm": 688.0112915039062,
-      "learning_rate": 4.2726965904474e-05,
-      "loss": 58.7139,
-      "step": 81050
-    },
-    {
-      "epoch": 0.32749265707001946,
-      "grad_norm": 742.99169921875,
-      "learning_rate": 4.2724504365705314e-05,
-      "loss": 46.189,
-      "step": 81060
-    },
-    {
-      "epoch": 0.32753305833538704,
-      "grad_norm": 1215.4967041015625,
-      "learning_rate": 4.272204248138828e-05,
-      "loss": 48.7161,
-      "step": 81070
-    },
-    {
-      "epoch": 0.3275734596007547,
-      "grad_norm": 575.4824829101562,
-      "learning_rate": 4.2719580251570915e-05,
-      "loss": 78.7691,
-      "step": 81080
-    },
-    {
-      "epoch": 0.3276138608661223,
-      "grad_norm": 547.03759765625,
-      "learning_rate": 4.2717117676301196e-05,
-      "loss": 42.8608,
-      "step": 81090
-    },
-    {
-      "epoch": 0.32765426213148996,
-      "grad_norm": 605.3917846679688,
-      "learning_rate": 4.271465475562716e-05,
-      "loss": 52.3641,
-      "step": 81100
-    },
-    {
-      "epoch": 0.3276946633968576,
-      "grad_norm": 402.33050537109375,
-      "learning_rate": 4.2712191489596796e-05,
-      "loss": 81.6275,
-      "step": 81110
-    },
-    {
-      "epoch": 0.32773506466222524,
-      "grad_norm": 944.0283813476562,
-      "learning_rate": 4.270972787825815e-05,
-      "loss": 44.0819,
-      "step": 81120
-    },
-    {
-      "epoch": 0.3277754659275929,
-      "grad_norm": 2794.85986328125,
-      "learning_rate": 4.2707263921659236e-05,
-      "loss": 77.4065,
-      "step": 81130
-    },
-    {
-      "epoch": 0.32781586719296046,
-      "grad_norm": 706.5042724609375,
-      "learning_rate": 4.27047996198481e-05,
-      "loss": 84.9347,
-      "step": 81140
-    },
-    {
-      "epoch": 0.3278562684583281,
-      "grad_norm": 1184.057861328125,
-      "learning_rate": 4.2702334972872776e-05,
-      "loss": 70.5365,
-      "step": 81150
-    },
-    {
-      "epoch": 0.32789666972369574,
-      "grad_norm": 548.59716796875,
-      "learning_rate": 4.269986998078132e-05,
-      "loss": 57.2449,
-      "step": 81160
-    },
-    {
-      "epoch": 0.3279370709890634,
-      "grad_norm": 712.474609375,
-      "learning_rate": 4.2697404643621786e-05,
-      "loss": 68.5629,
-      "step": 81170
-    },
-    {
-      "epoch": 0.327977472254431,
-      "grad_norm": 727.8631591796875,
-      "learning_rate": 4.269493896144224e-05,
-      "loss": 63.5557,
-      "step": 81180
-    },
-    {
-      "epoch": 0.32801787351979866,
-      "grad_norm": 1237.4130859375,
-      "learning_rate": 4.2692472934290746e-05,
-      "loss": 70.6992,
-      "step": 81190
-    },
-    {
-      "epoch": 0.32805827478516625,
-      "grad_norm": 625.5675048828125,
-      "learning_rate": 4.2690006562215384e-05,
-      "loss": 80.8208,
-      "step": 81200
-    },
-    {
-      "epoch": 0.3280986760505339,
-      "grad_norm": 561.9657592773438,
-      "learning_rate": 4.2687539845264235e-05,
-      "loss": 62.1039,
-      "step": 81210
-    },
-    {
-      "epoch": 0.32813907731590153,
-      "grad_norm": 1042.281494140625,
-      "learning_rate": 4.268507278348539e-05,
-      "loss": 39.331,
-      "step": 81220
-    },
-    {
-      "epoch": 0.32817947858126917,
-      "grad_norm": 401.18792724609375,
-      "learning_rate": 4.2682605376926955e-05,
-      "loss": 61.7103,
-      "step": 81230
-    },
-    {
-      "epoch": 0.3282198798466368,
-      "grad_norm": 1227.1231689453125,
-      "learning_rate": 4.268013762563702e-05,
-      "loss": 72.3686,
-      "step": 81240
-    },
-    {
-      "epoch": 0.32826028111200445,
-      "grad_norm": 2529.718994140625,
-      "learning_rate": 4.267766952966369e-05,
-      "loss": 111.4047,
-      "step": 81250
-    },
-    {
-      "epoch": 0.32830068237737203,
-      "grad_norm": 762.4780883789062,
-      "learning_rate": 4.2675201089055096e-05,
-      "loss": 44.7507,
-      "step": 81260
-    },
-    {
-      "epoch": 0.3283410836427397,
-      "grad_norm": 917.88720703125,
-      "learning_rate": 4.2672732303859365e-05,
-      "loss": 74.2715,
-      "step": 81270
-    },
-    {
-      "epoch": 0.3283814849081073,
-      "grad_norm": 520.3221435546875,
-      "learning_rate": 4.267026317412461e-05,
-      "loss": 54.3115,
-      "step": 81280
-    },
-    {
-      "epoch": 0.32842188617347495,
-      "grad_norm": 185.31436157226562,
-      "learning_rate": 4.266779369989899e-05,
-      "loss": 50.5796,
-      "step": 81290
-    },
-    {
-      "epoch": 0.3284622874388426,
-      "grad_norm": 1354.3519287109375,
-      "learning_rate": 4.2665323881230624e-05,
-      "loss": 101.209,
-      "step": 81300
-    },
-    {
-      "epoch": 0.32850268870421023,
-      "grad_norm": 387.4610900878906,
-      "learning_rate": 4.266285371816767e-05,
-      "loss": 80.6324,
-      "step": 81310
-    },
-    {
-      "epoch": 0.3285430899695779,
-      "grad_norm": 1534.5985107421875,
-      "learning_rate": 4.266038321075831e-05,
-      "loss": 55.0623,
-      "step": 81320
-    },
-    {
-      "epoch": 0.32858349123494546,
-      "grad_norm": 750.3333129882812,
-      "learning_rate": 4.265791235905067e-05,
-      "loss": 65.2265,
-      "step": 81330
-    },
-    {
-      "epoch": 0.3286238925003131,
-      "grad_norm": 438.5273742675781,
-      "learning_rate": 4.265544116309294e-05,
-      "loss": 82.4092,
-      "step": 81340
-    },
-    {
-      "epoch": 0.32866429376568074,
-      "grad_norm": 1133.52099609375,
-      "learning_rate": 4.265296962293329e-05,
-      "loss": 58.1583,
-      "step": 81350
-    },
-    {
-      "epoch": 0.3287046950310484,
-      "grad_norm": 632.3345336914062,
-      "learning_rate": 4.265049773861991e-05,
-      "loss": 79.6419,
-      "step": 81360
-    },
-    {
-      "epoch": 0.328745096296416,
-      "grad_norm": 778.0379028320312,
-      "learning_rate": 4.2648025510201e-05,
-      "loss": 49.8477,
-      "step": 81370
-    },
-    {
-      "epoch": 0.32878549756178366,
-      "grad_norm": 662.5733642578125,
-      "learning_rate": 4.2645552937724744e-05,
-      "loss": 58.1828,
-      "step": 81380
-    },
-    {
-      "epoch": 0.32882589882715124,
-      "grad_norm": 1248.02001953125,
-      "learning_rate": 4.264308002123935e-05,
-      "loss": 57.9085,
-      "step": 81390
-    },
-    {
-      "epoch": 0.3288663000925189,
-      "grad_norm": 645.2049560546875,
-      "learning_rate": 4.264060676079302e-05,
-      "loss": 56.5214,
-      "step": 81400
-    },
-    {
-      "epoch": 0.3289067013578865,
-      "grad_norm": 2653.940673828125,
-      "learning_rate": 4.2638133156433986e-05,
-      "loss": 68.7137,
-      "step": 81410
-    },
-    {
-      "epoch": 0.32894710262325416,
-      "grad_norm": 756.0985717773438,
-      "learning_rate": 4.263565920821046e-05,
-      "loss": 53.9748,
-      "step": 81420
-    },
-    {
-      "epoch": 0.3289875038886218,
-      "grad_norm": 449.9473571777344,
-      "learning_rate": 4.2633184916170677e-05,
-      "loss": 69.4224,
-      "step": 81430
-    },
-    {
-      "epoch": 0.32902790515398944,
-      "grad_norm": 1535.21923828125,
-      "learning_rate": 4.263071028036288e-05,
-      "loss": 66.2684,
-      "step": 81440
-    },
-    {
-      "epoch": 0.3290683064193571,
-      "grad_norm": 899.3699951171875,
-      "learning_rate": 4.2628235300835314e-05,
-      "loss": 65.152,
-      "step": 81450
-    },
-    {
-      "epoch": 0.32910870768472467,
-      "grad_norm": 956.332275390625,
-      "learning_rate": 4.2625759977636214e-05,
-      "loss": 68.6298,
-      "step": 81460
-    },
-    {
-      "epoch": 0.3291491089500923,
-      "grad_norm": 815.9563598632812,
-      "learning_rate": 4.262328431081386e-05,
-      "loss": 80.0543,
-      "step": 81470
-    },
-    {
-      "epoch": 0.32918951021545995,
-      "grad_norm": 536.765625,
-      "learning_rate": 4.26208083004165e-05,
-      "loss": 47.5851,
-      "step": 81480
-    },
-    {
-      "epoch": 0.3292299114808276,
-      "grad_norm": 564.234375,
-      "learning_rate": 4.261833194649241e-05,
-      "loss": 73.5911,
-      "step": 81490
-    },
-    {
-      "epoch": 0.3292703127461952,
-      "grad_norm": 999.4091186523438,
-      "learning_rate": 4.261585524908987e-05,
-      "loss": 68.8592,
-      "step": 81500
-    },
-    {
-      "epoch": 0.32931071401156287,
-      "grad_norm": 1350.5775146484375,
-      "learning_rate": 4.261337820825716e-05,
-      "loss": 70.5121,
-      "step": 81510
-    },
-    {
-      "epoch": 0.32935111527693045,
-      "grad_norm": 735.5888061523438,
-      "learning_rate": 4.261090082404258e-05,
-      "loss": 75.4246,
-      "step": 81520
-    },
-    {
-      "epoch": 0.3293915165422981,
-      "grad_norm": 365.2147521972656,
-      "learning_rate": 4.2608423096494406e-05,
-      "loss": 61.5511,
-      "step": 81530
-    },
-    {
-      "epoch": 0.32943191780766573,
-      "grad_norm": 1004.5228881835938,
-      "learning_rate": 4.260594502566097e-05,
-      "loss": 66.0472,
-      "step": 81540
-    },
-    {
-      "epoch": 0.32947231907303337,
-      "grad_norm": 725.521484375,
-      "learning_rate": 4.260346661159058e-05,
-      "loss": 61.7147,
-      "step": 81550
-    },
-    {
-      "epoch": 0.329512720338401,
-      "grad_norm": 1222.5792236328125,
-      "learning_rate": 4.260098785433154e-05,
-      "loss": 60.5714,
-      "step": 81560
-    },
-    {
-      "epoch": 0.32955312160376865,
-      "grad_norm": 1317.46728515625,
-      "learning_rate": 4.259850875393217e-05,
-      "loss": 75.1795,
-      "step": 81570
-    },
-    {
-      "epoch": 0.32959352286913624,
-      "grad_norm": 1155.3607177734375,
-      "learning_rate": 4.2596029310440824e-05,
-      "loss": 65.329,
-      "step": 81580
-    },
-    {
-      "epoch": 0.3296339241345039,
-      "grad_norm": 549.3975219726562,
-      "learning_rate": 4.259354952390582e-05,
-      "loss": 65.329,
-      "step": 81590
-    },
-    {
-      "epoch": 0.3296743253998715,
-      "grad_norm": 1814.1624755859375,
-      "learning_rate": 4.259106939437551e-05,
-      "loss": 66.1722,
-      "step": 81600
-    },
-    {
-      "epoch": 0.32971472666523916,
-      "grad_norm": 355.4609680175781,
-      "learning_rate": 4.258858892189825e-05,
-      "loss": 80.2931,
-      "step": 81610
-    },
-    {
-      "epoch": 0.3297551279306068,
-      "grad_norm": 570.4844360351562,
-      "learning_rate": 4.258610810652239e-05,
-      "loss": 44.5537,
-      "step": 81620
-    },
-    {
-      "epoch": 0.32979552919597444,
-      "grad_norm": 719.2767333984375,
-      "learning_rate": 4.258362694829629e-05,
-      "loss": 63.4508,
-      "step": 81630
-    },
-    {
-      "epoch": 0.3298359304613421,
-      "grad_norm": 494.90911865234375,
-      "learning_rate": 4.258114544726835e-05,
-      "loss": 67.4852,
-      "step": 81640
-    },
-    {
-      "epoch": 0.32987633172670966,
-      "grad_norm": 730.2485961914062,
-      "learning_rate": 4.257866360348692e-05,
-      "loss": 58.3942,
-      "step": 81650
-    },
-    {
-      "epoch": 0.3299167329920773,
-      "grad_norm": 528.159423828125,
-      "learning_rate": 4.257618141700039e-05,
-      "loss": 55.3444,
-      "step": 81660
-    },
-    {
-      "epoch": 0.32995713425744494,
-      "grad_norm": 911.97998046875,
-      "learning_rate": 4.257369888785715e-05,
-      "loss": 69.4855,
-      "step": 81670
-    },
-    {
-      "epoch": 0.3299975355228126,
-      "grad_norm": 1036.72021484375,
-      "learning_rate": 4.2571216016105614e-05,
-      "loss": 90.7197,
-      "step": 81680
-    },
-    {
-      "epoch": 0.3300379367881802,
-      "grad_norm": 1021.642333984375,
-      "learning_rate": 4.256873280179416e-05,
-      "loss": 96.8164,
-      "step": 81690
-    },
-    {
-      "epoch": 0.33007833805354786,
-      "grad_norm": 1377.108642578125,
-      "learning_rate": 4.256624924497123e-05,
-      "loss": 73.541,
-      "step": 81700
-    },
-    {
-      "epoch": 0.33011873931891544,
-      "grad_norm": 702.5912475585938,
-      "learning_rate": 4.256376534568522e-05,
-      "loss": 98.5258,
-      "step": 81710
-    },
-    {
-      "epoch": 0.3301591405842831,
-      "grad_norm": 3800.063720703125,
-      "learning_rate": 4.256128110398457e-05,
-      "loss": 64.3711,
-      "step": 81720
-    },
-    {
-      "epoch": 0.3301995418496507,
-      "grad_norm": 806.7174072265625,
-      "learning_rate": 4.25587965199177e-05,
-      "loss": 77.5143,
-      "step": 81730
-    },
-    {
-      "epoch": 0.33023994311501836,
-      "grad_norm": 753.8306274414062,
-      "learning_rate": 4.255631159353305e-05,
-      "loss": 85.3019,
-      "step": 81740
-    },
-    {
-      "epoch": 0.330280344380386,
-      "grad_norm": 1084.2166748046875,
-      "learning_rate": 4.2553826324879064e-05,
-      "loss": 70.9181,
-      "step": 81750
-    },
-    {
-      "epoch": 0.33032074564575364,
-      "grad_norm": 1052.1856689453125,
-      "learning_rate": 4.2551340714004203e-05,
-      "loss": 71.8444,
-      "step": 81760
-    },
-    {
-      "epoch": 0.3303611469111213,
-      "grad_norm": 1021.4315185546875,
-      "learning_rate": 4.254885476095691e-05,
-      "loss": 72.4072,
-      "step": 81770
-    },
-    {
-      "epoch": 0.33040154817648887,
-      "grad_norm": 433.3638000488281,
-      "learning_rate": 4.254636846578566e-05,
-      "loss": 95.435,
-      "step": 81780
-    },
-    {
-      "epoch": 0.3304419494418565,
-      "grad_norm": 778.073974609375,
-      "learning_rate": 4.254388182853894e-05,
-      "loss": 72.969,
-      "step": 81790
-    },
-    {
-      "epoch": 0.33048235070722415,
-      "grad_norm": 1745.2950439453125,
-      "learning_rate": 4.254139484926519e-05,
-      "loss": 81.5817,
-      "step": 81800
-    },
-    {
-      "epoch": 0.3305227519725918,
-      "grad_norm": 273.6972961425781,
-      "learning_rate": 4.253890752801293e-05,
-      "loss": 73.0842,
-      "step": 81810
-    },
-    {
-      "epoch": 0.33056315323795943,
-      "grad_norm": 711.11328125,
-      "learning_rate": 4.253641986483062e-05,
-      "loss": 66.8901,
-      "step": 81820
-    },
-    {
-      "epoch": 0.33060355450332707,
-      "grad_norm": 810.6575927734375,
-      "learning_rate": 4.2533931859766794e-05,
-      "loss": 80.1247,
-      "step": 81830
-    },
-    {
-      "epoch": 0.33064395576869465,
-      "grad_norm": 0.0,
-      "learning_rate": 4.253144351286994e-05,
-      "loss": 72.4615,
-      "step": 81840
-    },
-    {
-      "epoch": 0.3306843570340623,
-      "grad_norm": 2913.29443359375,
-      "learning_rate": 4.252895482418856e-05,
-      "loss": 76.5862,
-      "step": 81850
-    },
-    {
-      "epoch": 0.33072475829942993,
-      "grad_norm": 1155.0936279296875,
-      "learning_rate": 4.252646579377119e-05,
-      "loss": 84.624,
-      "step": 81860
-    },
-    {
-      "epoch": 0.3307651595647976,
-      "grad_norm": 587.8722534179688,
-      "learning_rate": 4.252397642166633e-05,
-      "loss": 72.8901,
-      "step": 81870
-    },
-    {
-      "epoch": 0.3308055608301652,
-      "grad_norm": 840.6516723632812,
-      "learning_rate": 4.252148670792254e-05,
-      "loss": 92.0446,
-      "step": 81880
-    },
-    {
-      "epoch": 0.33084596209553285,
-      "grad_norm": 701.7250366210938,
-      "learning_rate": 4.251899665258835e-05,
-      "loss": 30.5136,
-      "step": 81890
-    },
-    {
-      "epoch": 0.33088636336090044,
-      "grad_norm": 0.0,
-      "learning_rate": 4.2516506255712296e-05,
-      "loss": 71.1064,
-      "step": 81900
-    },
-    {
-      "epoch": 0.3309267646262681,
-      "grad_norm": 485.8135986328125,
-      "learning_rate": 4.251401551734293e-05,
-      "loss": 77.7925,
-      "step": 81910
-    },
-    {
-      "epoch": 0.3309671658916357,
-      "grad_norm": 918.968994140625,
-      "learning_rate": 4.2511524437528825e-05,
-      "loss": 79.0183,
-      "step": 81920
-    },
-    {
-      "epoch": 0.33100756715700336,
-      "grad_norm": 808.7003173828125,
-      "learning_rate": 4.250903301631853e-05,
-      "loss": 45.6422,
-      "step": 81930
-    },
-    {
-      "epoch": 0.331047968422371,
-      "grad_norm": 1006.4502563476562,
-      "learning_rate": 4.250654125376062e-05,
-      "loss": 81.9783,
-      "step": 81940
-    },
-    {
-      "epoch": 0.33108836968773864,
-      "grad_norm": 926.7799682617188,
-      "learning_rate": 4.250404914990367e-05,
-      "loss": 51.5458,
-      "step": 81950
-    },
-    {
-      "epoch": 0.3311287709531063,
-      "grad_norm": 422.4704284667969,
-      "learning_rate": 4.250155670479628e-05,
-      "loss": 68.1897,
-      "step": 81960
-    },
-    {
-      "epoch": 0.33116917221847386,
-      "grad_norm": 661.1273193359375,
-      "learning_rate": 4.2499063918487034e-05,
-      "loss": 57.1841,
-      "step": 81970
-    },
-    {
-      "epoch": 0.3312095734838415,
-      "grad_norm": 1525.314208984375,
-      "learning_rate": 4.2496570791024513e-05,
-      "loss": 85.9705,
-      "step": 81980
-    },
-    {
-      "epoch": 0.33124997474920914,
-      "grad_norm": 677.7116088867188,
-      "learning_rate": 4.2494077322457346e-05,
-      "loss": 49.9874,
-      "step": 81990
-    },
-    {
-      "epoch": 0.3312903760145768,
-      "grad_norm": 854.898193359375,
-      "learning_rate": 4.249158351283414e-05,
-      "loss": 67.6385,
-      "step": 82000
-    },
-    {
-      "epoch": 0.3313307772799444,
-      "grad_norm": 836.3336791992188,
-      "learning_rate": 4.24890893622035e-05,
-      "loss": 79.6089,
-      "step": 82010
-    },
-    {
-      "epoch": 0.33137117854531206,
-      "grad_norm": 509.12847900390625,
-      "learning_rate": 4.248659487061406e-05,
-      "loss": 30.2002,
-      "step": 82020
-    },
-    {
-      "epoch": 0.33141157981067965,
-      "grad_norm": 623.1883544921875,
-      "learning_rate": 4.248410003811445e-05,
-      "loss": 54.6345,
-      "step": 82030
-    },
-    {
-      "epoch": 0.3314519810760473,
-      "grad_norm": 1047.8133544921875,
-      "learning_rate": 4.248160486475331e-05,
-      "loss": 69.4941,
-      "step": 82040
-    },
-    {
-      "epoch": 0.3314923823414149,
-      "grad_norm": 2248.522216796875,
-      "learning_rate": 4.247910935057929e-05,
-      "loss": 75.4848,
-      "step": 82050
-    },
-    {
-      "epoch": 0.33153278360678257,
-      "grad_norm": 834.0875854492188,
-      "learning_rate": 4.2476613495641026e-05,
-      "loss": 56.126,
-      "step": 82060
-    },
-    {
-      "epoch": 0.3315731848721502,
-      "grad_norm": 629.9591064453125,
-      "learning_rate": 4.247411729998718e-05,
-      "loss": 64.6742,
-      "step": 82070
-    },
-    {
-      "epoch": 0.33161358613751785,
-      "grad_norm": 738.5386352539062,
-      "learning_rate": 4.247162076366643e-05,
-      "loss": 56.9134,
-      "step": 82080
-    },
-    {
-      "epoch": 0.33165398740288543,
-      "grad_norm": 734.9909057617188,
-      "learning_rate": 4.246912388672744e-05,
-      "loss": 58.0093,
-      "step": 82090
-    },
-    {
-      "epoch": 0.33169438866825307,
-      "grad_norm": 315.8343505859375,
-      "learning_rate": 4.246662666921888e-05,
-      "loss": 68.2145,
-      "step": 82100
-    },
-    {
-      "epoch": 0.3317347899336207,
-      "grad_norm": 471.6864318847656,
-      "learning_rate": 4.2464129111189444e-05,
-      "loss": 98.6437,
-      "step": 82110
-    },
-    {
-      "epoch": 0.33177519119898835,
-      "grad_norm": 913.5321044921875,
-      "learning_rate": 4.2461631212687816e-05,
-      "loss": 50.3258,
-      "step": 82120
-    },
-    {
-      "epoch": 0.331815592464356,
-      "grad_norm": 837.1478271484375,
-      "learning_rate": 4.24591329737627e-05,
-      "loss": 79.771,
-      "step": 82130
-    },
-    {
-      "epoch": 0.33185599372972363,
-      "grad_norm": 681.1831665039062,
-      "learning_rate": 4.24566343944628e-05,
-      "loss": 54.0234,
-      "step": 82140
-    },
-    {
-      "epoch": 0.33189639499509127,
-      "grad_norm": 497.15582275390625,
-      "learning_rate": 4.245413547483682e-05,
-      "loss": 79.7911,
-      "step": 82150
-    },
-    {
-      "epoch": 0.33193679626045886,
-      "grad_norm": 463.25885009765625,
-      "learning_rate": 4.245163621493349e-05,
-      "loss": 60.4234,
-      "step": 82160
-    },
-    {
-      "epoch": 0.3319771975258265,
-      "grad_norm": 566.7998046875,
-      "learning_rate": 4.244913661480152e-05,
-      "loss": 47.7919,
-      "step": 82170
-    },
-    {
-      "epoch": 0.33201759879119414,
-      "grad_norm": 1045.619384765625,
-      "learning_rate": 4.2446636674489645e-05,
-      "loss": 67.3056,
-      "step": 82180
-    },
-    {
-      "epoch": 0.3320580000565618,
-      "grad_norm": 605.4349975585938,
-      "learning_rate": 4.244413639404662e-05,
-      "loss": 68.3609,
-      "step": 82190
-    },
-    {
-      "epoch": 0.3320984013219294,
-      "grad_norm": 562.8687133789062,
-      "learning_rate": 4.244163577352116e-05,
-      "loss": 64.8594,
-      "step": 82200
-    },
-    {
-      "epoch": 0.33213880258729706,
-      "grad_norm": 830.25927734375,
-      "learning_rate": 4.243913481296205e-05,
-      "loss": 42.151,
-      "step": 82210
-    },
-    {
-      "epoch": 0.33217920385266464,
-      "grad_norm": 570.14453125,
-      "learning_rate": 4.243663351241801e-05,
-      "loss": 91.821,
-      "step": 82220
-    },
-    {
-      "epoch": 0.3322196051180323,
-      "grad_norm": 499.4716796875,
-      "learning_rate": 4.243413187193783e-05,
-      "loss": 56.719,
-      "step": 82230
-    },
-    {
-      "epoch": 0.3322600063833999,
-      "grad_norm": 536.1835327148438,
-      "learning_rate": 4.2431629891570266e-05,
-      "loss": 71.9568,
-      "step": 82240
-    },
-    {
-      "epoch": 0.33230040764876756,
-      "grad_norm": 907.7794799804688,
-      "learning_rate": 4.242912757136412e-05,
-      "loss": 80.2039,
-      "step": 82250
-    },
-    {
-      "epoch": 0.3323408089141352,
-      "grad_norm": 896.99267578125,
-      "learning_rate": 4.2426624911368146e-05,
-      "loss": 61.2477,
-      "step": 82260
-    },
-    {
-      "epoch": 0.33238121017950284,
-      "grad_norm": 3104.49169921875,
-      "learning_rate": 4.242412191163115e-05,
-      "loss": 64.1988,
-      "step": 82270
-    },
-    {
-      "epoch": 0.3324216114448705,
-      "grad_norm": 739.8251342773438,
-      "learning_rate": 4.242161857220193e-05,
-      "loss": 99.7195,
-      "step": 82280
-    },
-    {
-      "epoch": 0.33246201271023806,
-      "grad_norm": 717.78662109375,
-      "learning_rate": 4.241911489312927e-05,
-      "loss": 69.5272,
-      "step": 82290
-    },
-    {
-      "epoch": 0.3325024139756057,
-      "grad_norm": 494.52197265625,
-      "learning_rate": 4.241661087446202e-05,
-      "loss": 46.656,
-      "step": 82300
-    },
-    {
-      "epoch": 0.33254281524097334,
-      "grad_norm": 355.43536376953125,
-      "learning_rate": 4.2414106516248964e-05,
-      "loss": 56.7551,
-      "step": 82310
-    },
-    {
-      "epoch": 0.332583216506341,
-      "grad_norm": 1103.3970947265625,
-      "learning_rate": 4.241160181853894e-05,
-      "loss": 68.9124,
-      "step": 82320
-    },
-    {
-      "epoch": 0.3326236177717086,
-      "grad_norm": 742.4725952148438,
-      "learning_rate": 4.240909678138077e-05,
-      "loss": 83.4965,
-      "step": 82330
-    },
-    {
-      "epoch": 0.33266401903707626,
-      "grad_norm": 1117.4141845703125,
-      "learning_rate": 4.24065914048233e-05,
-      "loss": 71.4723,
-      "step": 82340
-    },
-    {
-      "epoch": 0.33270442030244385,
-      "grad_norm": 539.0693969726562,
-      "learning_rate": 4.2404085688915364e-05,
-      "loss": 53.2493,
-      "step": 82350
-    },
-    {
-      "epoch": 0.3327448215678115,
-      "grad_norm": 4054.252197265625,
-      "learning_rate": 4.240157963370582e-05,
-      "loss": 99.1258,
-      "step": 82360
-    },
-    {
-      "epoch": 0.33278522283317913,
-      "grad_norm": 826.72509765625,
-      "learning_rate": 4.2399073239243526e-05,
-      "loss": 68.9222,
-      "step": 82370
-    },
-    {
-      "epoch": 0.33282562409854677,
-      "grad_norm": 831.2509155273438,
-      "learning_rate": 4.239656650557734e-05,
-      "loss": 73.0231,
-      "step": 82380
-    },
-    {
-      "epoch": 0.3328660253639144,
-      "grad_norm": 409.9931335449219,
-      "learning_rate": 4.239405943275613e-05,
-      "loss": 45.4566,
-      "step": 82390
-    },
-    {
-      "epoch": 0.33290642662928205,
-      "grad_norm": 926.7503051757812,
-      "learning_rate": 4.2391552020828775e-05,
-      "loss": 77.0996,
-      "step": 82400
-    },
-    {
-      "epoch": 0.33294682789464963,
-      "grad_norm": 1059.4488525390625,
-      "learning_rate": 4.2389044269844155e-05,
-      "loss": 95.9397,
-      "step": 82410
-    },
-    {
-      "epoch": 0.3329872291600173,
-      "grad_norm": 0.0,
-      "learning_rate": 4.238653617985118e-05,
-      "loss": 49.413,
-      "step": 82420
-    },
-    {
-      "epoch": 0.3330276304253849,
-      "grad_norm": 1252.39453125,
-      "learning_rate": 4.238402775089871e-05,
-      "loss": 66.4441,
-      "step": 82430
-    },
-    {
-      "epoch": 0.33306803169075255,
-      "grad_norm": 1263.342529296875,
-      "learning_rate": 4.238151898303569e-05,
-      "loss": 76.9728,
-      "step": 82440
-    },
-    {
-      "epoch": 0.3331084329561202,
-      "grad_norm": 473.049072265625,
-      "learning_rate": 4.2379009876311e-05,
-      "loss": 63.0306,
-      "step": 82450
-    },
-    {
-      "epoch": 0.33314883422148783,
-      "grad_norm": 1100.3658447265625,
-      "learning_rate": 4.237650043077357e-05,
-      "loss": 107.0959,
-      "step": 82460
-    },
-    {
-      "epoch": 0.3331892354868555,
-      "grad_norm": 731.7568359375,
-      "learning_rate": 4.237399064647231e-05,
-      "loss": 105.6519,
-      "step": 82470
-    },
-    {
-      "epoch": 0.33322963675222306,
-      "grad_norm": 569.7168579101562,
-      "learning_rate": 4.237148052345616e-05,
-      "loss": 89.5767,
-      "step": 82480
-    },
-    {
-      "epoch": 0.3332700380175907,
-      "grad_norm": 650.82958984375,
-      "learning_rate": 4.236897006177405e-05,
-      "loss": 84.1082,
-      "step": 82490
-    },
-    {
-      "epoch": 0.33331043928295834,
-      "grad_norm": 2010.1734619140625,
-      "learning_rate": 4.2366459261474933e-05,
-      "loss": 60.609,
-      "step": 82500
-    },
-    {
-      "epoch": 0.333350840548326,
-      "grad_norm": 1313.4893798828125,
-      "learning_rate": 4.2363948122607756e-05,
-      "loss": 72.3615,
-      "step": 82510
-    },
-    {
-      "epoch": 0.3333912418136936,
-      "grad_norm": 538.3892211914062,
-      "learning_rate": 4.236143664522146e-05,
-      "loss": 78.7254,
-      "step": 82520
-    },
-    {
-      "epoch": 0.33343164307906126,
-      "grad_norm": 678.4175415039062,
-      "learning_rate": 4.235892482936502e-05,
-      "loss": 67.0122,
-      "step": 82530
-    },
-    {
-      "epoch": 0.33347204434442884,
-      "grad_norm": 1527.4241943359375,
-      "learning_rate": 4.2356412675087406e-05,
-      "loss": 46.1357,
-      "step": 82540
-    },
-    {
-      "epoch": 0.3335124456097965,
-      "grad_norm": 889.2577514648438,
-      "learning_rate": 4.23539001824376e-05,
-      "loss": 85.7127,
-      "step": 82550
-    },
-    {
-      "epoch": 0.3335528468751641,
-      "grad_norm": 621.6854858398438,
-      "learning_rate": 4.2351387351464565e-05,
-      "loss": 62.1339,
-      "step": 82560
-    },
-    {
-      "epoch": 0.33359324814053176,
-      "grad_norm": 606.4068603515625,
-      "learning_rate": 4.2348874182217305e-05,
-      "loss": 56.7306,
-      "step": 82570
-    },
-    {
-      "epoch": 0.3336336494058994,
-      "grad_norm": 785.7623901367188,
-      "learning_rate": 4.2346360674744815e-05,
-      "loss": 49.6323,
-      "step": 82580
-    },
-    {
-      "epoch": 0.33367405067126704,
-      "grad_norm": 1377.6890869140625,
-      "learning_rate": 4.234384682909608e-05,
-      "loss": 84.6193,
-      "step": 82590
-    },
-    {
-      "epoch": 0.3337144519366347,
-      "grad_norm": 2395.7802734375,
-      "learning_rate": 4.234133264532012e-05,
-      "loss": 70.3222,
-      "step": 82600
-    },
-    {
-      "epoch": 0.33375485320200227,
-      "grad_norm": 746.2875366210938,
-      "learning_rate": 4.2338818123465966e-05,
-      "loss": 41.8322,
-      "step": 82610
-    },
-    {
-      "epoch": 0.3337952544673699,
-      "grad_norm": 461.2134094238281,
-      "learning_rate": 4.2336303263582624e-05,
-      "loss": 40.6642,
-      "step": 82620
-    },
-    {
-      "epoch": 0.33383565573273755,
-      "grad_norm": 469.310546875,
-      "learning_rate": 4.233378806571912e-05,
-      "loss": 67.8555,
-      "step": 82630
-    },
-    {
-      "epoch": 0.3338760569981052,
-      "grad_norm": 827.2962036132812,
-      "learning_rate": 4.2331272529924495e-05,
-      "loss": 74.3388,
-      "step": 82640
-    },
-    {
-      "epoch": 0.3339164582634728,
-      "grad_norm": 544.0149536132812,
-      "learning_rate": 4.2328756656247795e-05,
-      "loss": 79.4444,
-      "step": 82650
-    },
-    {
-      "epoch": 0.33395685952884047,
-      "grad_norm": 383.1783142089844,
-      "learning_rate": 4.2326240444738055e-05,
-      "loss": 51.7502,
-      "step": 82660
-    },
-    {
-      "epoch": 0.33399726079420805,
-      "grad_norm": 1237.60009765625,
-      "learning_rate": 4.232372389544434e-05,
-      "loss": 73.6413,
-      "step": 82670
-    },
-    {
-      "epoch": 0.3340376620595757,
-      "grad_norm": 309.4773864746094,
-      "learning_rate": 4.232120700841571e-05,
-      "loss": 59.2111,
-      "step": 82680
-    },
-    {
-      "epoch": 0.33407806332494333,
-      "grad_norm": 1590.2628173828125,
-      "learning_rate": 4.2318689783701224e-05,
-      "loss": 59.3465,
-      "step": 82690
-    },
-    {
-      "epoch": 0.33411846459031097,
-      "grad_norm": 464.6656799316406,
-      "learning_rate": 4.2316172221349973e-05,
-      "loss": 38.0717,
-      "step": 82700
-    },
-    {
-      "epoch": 0.3341588658556786,
-      "grad_norm": 2365.112060546875,
-      "learning_rate": 4.231365432141103e-05,
-      "loss": 52.5256,
-      "step": 82710
-    },
-    {
-      "epoch": 0.33419926712104625,
-      "grad_norm": 905.1858520507812,
-      "learning_rate": 4.231113608393348e-05,
-      "loss": 78.6967,
-      "step": 82720
-    },
-    {
-      "epoch": 0.33423966838641384,
-      "grad_norm": 399.5769348144531,
-      "learning_rate": 4.2308617508966414e-05,
-      "loss": 79.4581,
-      "step": 82730
-    },
-    {
-      "epoch": 0.3342800696517815,
-      "grad_norm": 2698.52685546875,
-      "learning_rate": 4.230609859655895e-05,
-      "loss": 78.356,
-      "step": 82740
-    },
-    {
-      "epoch": 0.3343204709171491,
-      "grad_norm": 1030.6182861328125,
-      "learning_rate": 4.230357934676017e-05,
-      "loss": 57.3243,
-      "step": 82750
-    },
-    {
-      "epoch": 0.33436087218251676,
-      "grad_norm": 1275.8216552734375,
-      "learning_rate": 4.230105975961921e-05,
-      "loss": 77.0945,
-      "step": 82760
-    },
-    {
-      "epoch": 0.3344012734478844,
-      "grad_norm": 846.1837768554688,
-      "learning_rate": 4.229853983518518e-05,
-      "loss": 58.413,
-      "step": 82770
-    },
-    {
-      "epoch": 0.33444167471325204,
-      "grad_norm": 502.4288330078125,
-      "learning_rate": 4.229601957350722e-05,
-      "loss": 55.0714,
-      "step": 82780
-    },
-    {
-      "epoch": 0.3344820759786197,
-      "grad_norm": 1429.5887451171875,
-      "learning_rate": 4.229349897463445e-05,
-      "loss": 74.8495,
-      "step": 82790
-    },
-    {
-      "epoch": 0.33452247724398726,
-      "grad_norm": 534.1192016601562,
-      "learning_rate": 4.2290978038616e-05,
-      "loss": 69.4769,
-      "step": 82800
-    },
-    {
-      "epoch": 0.3345628785093549,
-      "grad_norm": 625.2440185546875,
-      "learning_rate": 4.228845676550105e-05,
-      "loss": 68.7517,
-      "step": 82810
-    },
-    {
-      "epoch": 0.33460327977472254,
-      "grad_norm": 529.8661499023438,
-      "learning_rate": 4.2285935155338724e-05,
-      "loss": 77.1349,
-      "step": 82820
-    },
-    {
-      "epoch": 0.3346436810400902,
-      "grad_norm": 1316.5428466796875,
-      "learning_rate": 4.22834132081782e-05,
-      "loss": 61.7906,
-      "step": 82830
-    },
-    {
-      "epoch": 0.3346840823054578,
-      "grad_norm": 483.9060974121094,
-      "learning_rate": 4.2280890924068625e-05,
-      "loss": 62.0083,
-      "step": 82840
-    },
-    {
-      "epoch": 0.33472448357082546,
-      "grad_norm": 882.261962890625,
-      "learning_rate": 4.22783683030592e-05,
-      "loss": 62.3824,
-      "step": 82850
-    },
-    {
-      "epoch": 0.33476488483619304,
-      "grad_norm": 733.6650390625,
-      "learning_rate": 4.227584534519907e-05,
-      "loss": 81.4946,
-      "step": 82860
-    },
-    {
-      "epoch": 0.3348052861015607,
-      "grad_norm": 604.4740600585938,
-      "learning_rate": 4.227332205053746e-05,
-      "loss": 72.0598,
-      "step": 82870
-    },
-    {
-      "epoch": 0.3348456873669283,
-      "grad_norm": 4985.6357421875,
-      "learning_rate": 4.2270798419123534e-05,
-      "loss": 66.5251,
-      "step": 82880
-    },
-    {
-      "epoch": 0.33488608863229596,
-      "grad_norm": 477.78289794921875,
-      "learning_rate": 4.2268274451006506e-05,
-      "loss": 116.5259,
-      "step": 82890
-    },
-    {
-      "epoch": 0.3349264898976636,
-      "grad_norm": 1549.6126708984375,
-      "learning_rate": 4.226575014623557e-05,
-      "loss": 49.6759,
-      "step": 82900
-    },
-    {
-      "epoch": 0.33496689116303124,
-      "grad_norm": 416.32879638671875,
-      "learning_rate": 4.2263225504859955e-05,
-      "loss": 69.3102,
-      "step": 82910
-    },
-    {
-      "epoch": 0.3350072924283989,
-      "grad_norm": 575.8262939453125,
-      "learning_rate": 4.226070052692886e-05,
-      "loss": 54.391,
-      "step": 82920
-    },
-    {
-      "epoch": 0.33504769369376647,
-      "grad_norm": 318.236572265625,
-      "learning_rate": 4.2258175212491537e-05,
-      "loss": 58.843,
-      "step": 82930
-    },
-    {
-      "epoch": 0.3350880949591341,
-      "grad_norm": 439.44488525390625,
-      "learning_rate": 4.2255649561597186e-05,
-      "loss": 62.2686,
-      "step": 82940
-    },
-    {
-      "epoch": 0.33512849622450175,
-      "grad_norm": 857.3777465820312,
-      "learning_rate": 4.225312357429508e-05,
-      "loss": 68.3515,
-      "step": 82950
-    },
-    {
-      "epoch": 0.3351688974898694,
-      "grad_norm": 1923.6248779296875,
-      "learning_rate": 4.225059725063444e-05,
-      "loss": 75.4591,
-      "step": 82960
-    },
-    {
-      "epoch": 0.33520929875523703,
-      "grad_norm": 1054.3583984375,
-      "learning_rate": 4.2248070590664525e-05,
-      "loss": 57.7459,
-      "step": 82970
-    },
-    {
-      "epoch": 0.33524970002060467,
-      "grad_norm": 974.0528564453125,
-      "learning_rate": 4.224554359443459e-05,
-      "loss": 52.0075,
-      "step": 82980
-    },
-    {
-      "epoch": 0.33529010128597225,
-      "grad_norm": 1275.845947265625,
-      "learning_rate": 4.22430162619939e-05,
-      "loss": 94.6089,
-      "step": 82990
-    },
-    {
-      "epoch": 0.3353305025513399,
-      "grad_norm": 1231.898681640625,
-      "learning_rate": 4.224048859339175e-05,
-      "loss": 65.2636,
-      "step": 83000
-    },
-    {
-      "epoch": 0.33537090381670753,
-      "grad_norm": 592.545166015625,
-      "learning_rate": 4.223796058867738e-05,
-      "loss": 74.3102,
-      "step": 83010
-    },
-    {
-      "epoch": 0.3354113050820752,
-      "grad_norm": 570.7125244140625,
-      "learning_rate": 4.22354322479001e-05,
-      "loss": 77.6862,
-      "step": 83020
-    },
-    {
-      "epoch": 0.3354517063474428,
-      "grad_norm": 814.4917602539062,
-      "learning_rate": 4.22329035711092e-05,
-      "loss": 78.4505,
-      "step": 83030
-    },
-    {
-      "epoch": 0.33549210761281045,
-      "grad_norm": 1354.454345703125,
-      "learning_rate": 4.223037455835397e-05,
-      "loss": 56.5575,
-      "step": 83040
-    },
-    {
-      "epoch": 0.33553250887817804,
-      "grad_norm": 1030.1226806640625,
-      "learning_rate": 4.2227845209683716e-05,
-      "loss": 53.6785,
-      "step": 83050
-    },
-    {
-      "epoch": 0.3355729101435457,
-      "grad_norm": 1466.65966796875,
-      "learning_rate": 4.222531552514775e-05,
-      "loss": 68.2507,
-      "step": 83060
-    },
-    {
-      "epoch": 0.3356133114089133,
-      "grad_norm": 705.6240844726562,
-      "learning_rate": 4.2222785504795394e-05,
-      "loss": 92.5086,
-      "step": 83070
-    },
-    {
-      "epoch": 0.33565371267428096,
-      "grad_norm": 703.1721801757812,
-      "learning_rate": 4.2220255148675956e-05,
-      "loss": 59.856,
-      "step": 83080
-    },
-    {
-      "epoch": 0.3356941139396486,
-      "grad_norm": 523.6635131835938,
-      "learning_rate": 4.221772445683878e-05,
-      "loss": 93.9165,
-      "step": 83090
-    },
-    {
-      "epoch": 0.33573451520501624,
-      "grad_norm": 915.0527954101562,
-      "learning_rate": 4.221519342933321e-05,
-      "loss": 62.3113,
-      "step": 83100
-    },
-    {
-      "epoch": 0.3357749164703839,
-      "grad_norm": 5522.94091796875,
-      "learning_rate": 4.221266206620859e-05,
-      "loss": 62.9414,
-      "step": 83110
-    },
-    {
-      "epoch": 0.33581531773575146,
-      "grad_norm": 619.2235717773438,
-      "learning_rate": 4.221013036751424e-05,
-      "loss": 51.0544,
-      "step": 83120
-    },
-    {
-      "epoch": 0.3358557190011191,
-      "grad_norm": 860.1320190429688,
-      "learning_rate": 4.220759833329955e-05,
-      "loss": 57.2342,
-      "step": 83130
-    },
-    {
-      "epoch": 0.33589612026648674,
-      "grad_norm": 1275.2235107421875,
-      "learning_rate": 4.2205065963613864e-05,
-      "loss": 72.2249,
-      "step": 83140
-    },
-    {
-      "epoch": 0.3359365215318544,
-      "grad_norm": 1409.1829833984375,
-      "learning_rate": 4.220253325850657e-05,
-      "loss": 75.4809,
-      "step": 83150
-    },
-    {
-      "epoch": 0.335976922797222,
-      "grad_norm": 155.7539825439453,
-      "learning_rate": 4.220000021802702e-05,
-      "loss": 66.316,
-      "step": 83160
-    },
-    {
-      "epoch": 0.33601732406258966,
-      "grad_norm": 519.6751098632812,
-      "learning_rate": 4.219746684222462e-05,
-      "loss": 46.7393,
-      "step": 83170
-    },
-    {
-      "epoch": 0.33605772532795725,
-      "grad_norm": 3975.830078125,
-      "learning_rate": 4.219493313114875e-05,
-      "loss": 113.0992,
-      "step": 83180
-    },
-    {
-      "epoch": 0.3360981265933249,
-      "grad_norm": 741.3836059570312,
-      "learning_rate": 4.219239908484881e-05,
-      "loss": 62.3929,
-      "step": 83190
-    },
-    {
-      "epoch": 0.3361385278586925,
-      "grad_norm": 1000.0194091796875,
-      "learning_rate": 4.218986470337419e-05,
-      "loss": 63.8103,
-      "step": 83200
-    },
-    {
-      "epoch": 0.33617892912406017,
-      "grad_norm": 915.8265380859375,
-      "learning_rate": 4.21873299867743e-05,
-      "loss": 64.819,
-      "step": 83210
-    },
-    {
-      "epoch": 0.3362193303894278,
-      "grad_norm": 1064.971923828125,
-      "learning_rate": 4.218479493509858e-05,
-      "loss": 75.7704,
-      "step": 83220
-    },
-    {
-      "epoch": 0.33625973165479545,
-      "grad_norm": 381.1853942871094,
-      "learning_rate": 4.218225954839643e-05,
-      "loss": 44.5471,
-      "step": 83230
-    },
-    {
-      "epoch": 0.3363001329201631,
-      "grad_norm": 2388.5380859375,
-      "learning_rate": 4.217972382671729e-05,
-      "loss": 56.8206,
-      "step": 83240
-    },
-    {
-      "epoch": 0.33634053418553067,
-      "grad_norm": 2224.6982421875,
-      "learning_rate": 4.2177187770110576e-05,
-      "loss": 75.5993,
-      "step": 83250
-    },
-    {
-      "epoch": 0.3363809354508983,
-      "grad_norm": 584.9492797851562,
-      "learning_rate": 4.2174651378625754e-05,
-      "loss": 76.4168,
-      "step": 83260
-    },
-    {
-      "epoch": 0.33642133671626595,
-      "grad_norm": 576.220947265625,
-      "learning_rate": 4.217211465231226e-05,
-      "loss": 69.955,
-      "step": 83270
-    },
-    {
-      "epoch": 0.3364617379816336,
-      "grad_norm": 419.0504455566406,
-      "learning_rate": 4.2169577591219545e-05,
-      "loss": 61.2648,
-      "step": 83280
-    },
-    {
-      "epoch": 0.33650213924700123,
-      "grad_norm": 649.897705078125,
-      "learning_rate": 4.216704019539707e-05,
-      "loss": 50.4646,
-      "step": 83290
-    },
-    {
-      "epoch": 0.33654254051236887,
-      "grad_norm": 1460.0035400390625,
-      "learning_rate": 4.2164502464894316e-05,
-      "loss": 66.0741,
-      "step": 83300
-    },
-    {
-      "epoch": 0.33658294177773646,
-      "grad_norm": 618.8856201171875,
-      "learning_rate": 4.216196439976076e-05,
-      "loss": 57.7886,
-      "step": 83310
-    },
-    {
-      "epoch": 0.3366233430431041,
-      "grad_norm": 540.6898193359375,
-      "learning_rate": 4.2159426000045854e-05,
-      "loss": 62.811,
-      "step": 83320
-    },
-    {
-      "epoch": 0.33666374430847174,
-      "grad_norm": 629.9652099609375,
-      "learning_rate": 4.215688726579911e-05,
-      "loss": 61.0531,
-      "step": 83330
-    },
-    {
-      "epoch": 0.3367041455738394,
-      "grad_norm": 3191.65185546875,
-      "learning_rate": 4.2154348197070017e-05,
-      "loss": 70.6365,
-      "step": 83340
-    },
-    {
-      "epoch": 0.336744546839207,
-      "grad_norm": 483.31396484375,
-      "learning_rate": 4.215180879390808e-05,
-      "loss": 71.8596,
-      "step": 83350
-    },
-    {
-      "epoch": 0.33678494810457466,
-      "grad_norm": 633.4132080078125,
-      "learning_rate": 4.2149269056362794e-05,
-      "loss": 65.0201,
-      "step": 83360
-    },
-    {
-      "epoch": 0.33682534936994224,
-      "grad_norm": 1588.8582763671875,
-      "learning_rate": 4.214672898448367e-05,
-      "loss": 81.9684,
-      "step": 83370
-    },
-    {
-      "epoch": 0.3368657506353099,
-      "grad_norm": 1885.4788818359375,
-      "learning_rate": 4.214418857832025e-05,
-      "loss": 81.2159,
-      "step": 83380
-    },
-    {
-      "epoch": 0.3369061519006775,
-      "grad_norm": 2098.720947265625,
-      "learning_rate": 4.214164783792205e-05,
-      "loss": 78.4835,
-      "step": 83390
-    },
-    {
-      "epoch": 0.33694655316604516,
-      "grad_norm": 934.3864135742188,
-      "learning_rate": 4.213910676333859e-05,
-      "loss": 51.1806,
-      "step": 83400
-    },
-    {
-      "epoch": 0.3369869544314128,
-      "grad_norm": 561.3240356445312,
-      "learning_rate": 4.213656535461942e-05,
-      "loss": 83.0985,
-      "step": 83410
-    },
-    {
-      "epoch": 0.33702735569678044,
-      "grad_norm": 517.56982421875,
-      "learning_rate": 4.213402361181409e-05,
-      "loss": 52.2328,
-      "step": 83420
-    },
-    {
-      "epoch": 0.3370677569621481,
-      "grad_norm": 501.65325927734375,
-      "learning_rate": 4.213148153497215e-05,
-      "loss": 45.5168,
-      "step": 83430
-    },
-    {
-      "epoch": 0.33710815822751566,
-      "grad_norm": 619.2743530273438,
-      "learning_rate": 4.212893912414316e-05,
-      "loss": 54.3897,
-      "step": 83440
-    },
-    {
-      "epoch": 0.3371485594928833,
-      "grad_norm": 281.7502746582031,
-      "learning_rate": 4.212639637937668e-05,
-      "loss": 48.6456,
-      "step": 83450
-    },
-    {
-      "epoch": 0.33718896075825094,
-      "grad_norm": 1214.2874755859375,
-      "learning_rate": 4.212385330072228e-05,
-      "loss": 80.0758,
-      "step": 83460
-    },
-    {
-      "epoch": 0.3372293620236186,
-      "grad_norm": 639.830322265625,
-      "learning_rate": 4.2121309888229544e-05,
-      "loss": 64.5304,
-      "step": 83470
-    },
-    {
-      "epoch": 0.3372697632889862,
-      "grad_norm": 823.2178955078125,
-      "learning_rate": 4.2118766141948066e-05,
-      "loss": 52.8296,
-      "step": 83480
-    },
-    {
-      "epoch": 0.33731016455435386,
-      "grad_norm": 429.5492858886719,
-      "learning_rate": 4.211622206192742e-05,
-      "loss": 63.3038,
-      "step": 83490
-    },
-    {
-      "epoch": 0.33735056581972145,
-      "grad_norm": 0.0,
-      "learning_rate": 4.211367764821722e-05,
-      "loss": 82.3349,
-      "step": 83500
-    },
-    {
-      "epoch": 0.3373909670850891,
-      "grad_norm": 974.5023803710938,
-      "learning_rate": 4.211113290086706e-05,
-      "loss": 43.6207,
-      "step": 83510
-    },
-    {
-      "epoch": 0.33743136835045673,
-      "grad_norm": 804.0608520507812,
-      "learning_rate": 4.2108587819926554e-05,
-      "loss": 66.1226,
-      "step": 83520
-    },
-    {
-      "epoch": 0.33747176961582437,
-      "grad_norm": 269.8977355957031,
-      "learning_rate": 4.2106042405445325e-05,
-      "loss": 82.0022,
-      "step": 83530
-    },
-    {
-      "epoch": 0.337512170881192,
-      "grad_norm": 1011.4495849609375,
-      "learning_rate": 4.210349665747299e-05,
-      "loss": 62.9397,
-      "step": 83540
-    },
-    {
-      "epoch": 0.33755257214655965,
-      "grad_norm": 568.0720825195312,
-      "learning_rate": 4.210095057605917e-05,
-      "loss": 80.0488,
-      "step": 83550
-    },
-    {
-      "epoch": 0.3375929734119273,
-      "grad_norm": 3479.609375,
-      "learning_rate": 4.209840416125353e-05,
-      "loss": 88.0878,
-      "step": 83560
-    },
-    {
-      "epoch": 0.3376333746772949,
-      "grad_norm": 0.0,
-      "learning_rate": 4.20958574131057e-05,
-      "loss": 100.4558,
-      "step": 83570
-    },
-    {
-      "epoch": 0.3376737759426625,
-      "grad_norm": 2387.772216796875,
-      "learning_rate": 4.209331033166531e-05,
-      "loss": 77.3918,
-      "step": 83580
-    },
-    {
-      "epoch": 0.33771417720803015,
-      "grad_norm": 710.8726806640625,
-      "learning_rate": 4.209076291698205e-05,
-      "loss": 53.3156,
-      "step": 83590
-    },
-    {
-      "epoch": 0.3377545784733978,
-      "grad_norm": 847.5494384765625,
-      "learning_rate": 4.208821516910557e-05,
-      "loss": 64.4423,
-      "step": 83600
-    },
-    {
-      "epoch": 0.33779497973876543,
-      "grad_norm": 540.9591674804688,
-      "learning_rate": 4.208566708808554e-05,
-      "loss": 54.3956,
-      "step": 83610
-    },
-    {
-      "epoch": 0.3378353810041331,
-      "grad_norm": 426.9322814941406,
-      "learning_rate": 4.2083118673971613e-05,
-      "loss": 80.7343,
-      "step": 83620
-    },
-    {
-      "epoch": 0.33787578226950066,
-      "grad_norm": 530.9072875976562,
-      "learning_rate": 4.2080569926813503e-05,
-      "loss": 62.1767,
-      "step": 83630
-    },
-    {
-      "epoch": 0.3379161835348683,
-      "grad_norm": 3503.007080078125,
-      "learning_rate": 4.20780208466609e-05,
-      "loss": 95.5601,
-      "step": 83640
-    },
-    {
-      "epoch": 0.33795658480023594,
-      "grad_norm": 764.01708984375,
-      "learning_rate": 4.207547143356347e-05,
-      "loss": 44.9169,
-      "step": 83650
-    },
-    {
-      "epoch": 0.3379969860656036,
-      "grad_norm": 898.1088256835938,
-      "learning_rate": 4.207292168757095e-05,
-      "loss": 42.9797,
-      "step": 83660
-    },
-    {
-      "epoch": 0.3380373873309712,
-      "grad_norm": 616.1493530273438,
-      "learning_rate": 4.2070371608733025e-05,
-      "loss": 55.619,
-      "step": 83670
-    },
-    {
-      "epoch": 0.33807778859633886,
-      "grad_norm": 1211.4007568359375,
-      "learning_rate": 4.206782119709942e-05,
-      "loss": 92.8507,
-      "step": 83680
-    },
-    {
-      "epoch": 0.33811818986170644,
-      "grad_norm": 1338.9266357421875,
-      "learning_rate": 4.206527045271985e-05,
-      "loss": 60.5956,
-      "step": 83690
-    },
-    {
-      "epoch": 0.3381585911270741,
-      "grad_norm": 5582.28515625,
-      "learning_rate": 4.206271937564405e-05,
-      "loss": 60.9006,
-      "step": 83700
-    },
-    {
-      "epoch": 0.3381989923924417,
-      "grad_norm": 1023.3121337890625,
-      "learning_rate": 4.206016796592174e-05,
-      "loss": 69.0761,
-      "step": 83710
-    },
-    {
-      "epoch": 0.33823939365780936,
-      "grad_norm": 665.3640747070312,
-      "learning_rate": 4.2057616223602684e-05,
-      "loss": 63.9492,
-      "step": 83720
-    },
-    {
-      "epoch": 0.338279794923177,
-      "grad_norm": 360.83563232421875,
-      "learning_rate": 4.205506414873661e-05,
-      "loss": 40.3897,
-      "step": 83730
-    },
-    {
-      "epoch": 0.33832019618854464,
-      "grad_norm": 1596.3583984375,
-      "learning_rate": 4.205251174137329e-05,
-      "loss": 72.7595,
-      "step": 83740
-    },
-    {
-      "epoch": 0.3383605974539123,
-      "grad_norm": 361.8247985839844,
-      "learning_rate": 4.2049959001562464e-05,
-      "loss": 98.5569,
-      "step": 83750
-    },
-    {
-      "epoch": 0.33840099871927987,
-      "grad_norm": 818.7006225585938,
-      "learning_rate": 4.204740592935392e-05,
-      "loss": 88.2147,
-      "step": 83760
-    },
-    {
-      "epoch": 0.3384413999846475,
-      "grad_norm": 434.5075378417969,
-      "learning_rate": 4.2044852524797406e-05,
-      "loss": 104.2029,
-      "step": 83770
-    },
-    {
-      "epoch": 0.33848180125001515,
-      "grad_norm": 722.4409790039062,
-      "learning_rate": 4.204229878794273e-05,
-      "loss": 48.2412,
-      "step": 83780
-    },
-    {
-      "epoch": 0.3385222025153828,
-      "grad_norm": 328.6600036621094,
-      "learning_rate": 4.203974471883966e-05,
-      "loss": 50.057,
-      "step": 83790
-    },
-    {
-      "epoch": 0.3385626037807504,
-      "grad_norm": 381.5543518066406,
-      "learning_rate": 4.2037190317538e-05,
-      "loss": 62.9453,
-      "step": 83800
-    },
-    {
-      "epoch": 0.33860300504611807,
-      "grad_norm": 1003.5863037109375,
-      "learning_rate": 4.2034635584087535e-05,
-      "loss": 97.3556,
-      "step": 83810
-    },
-    {
-      "epoch": 0.33864340631148565,
-      "grad_norm": 257.1349182128906,
-      "learning_rate": 4.203208051853808e-05,
-      "loss": 65.2049,
-      "step": 83820
-    },
-    {
-      "epoch": 0.3386838075768533,
-      "grad_norm": 542.9520263671875,
-      "learning_rate": 4.202952512093945e-05,
-      "loss": 60.1434,
-      "step": 83830
-    },
-    {
-      "epoch": 0.33872420884222093,
-      "grad_norm": 742.0493774414062,
-      "learning_rate": 4.202696939134146e-05,
-      "loss": 75.3315,
-      "step": 83840
-    },
-    {
-      "epoch": 0.33876461010758857,
-      "grad_norm": 839.355712890625,
-      "learning_rate": 4.202441332979394e-05,
-      "loss": 53.2438,
-      "step": 83850
-    },
-    {
-      "epoch": 0.3388050113729562,
-      "grad_norm": 616.1432495117188,
-      "learning_rate": 4.20218569363467e-05,
-      "loss": 64.6932,
-      "step": 83860
-    },
-    {
-      "epoch": 0.33884541263832385,
-      "grad_norm": 956.6621704101562,
-      "learning_rate": 4.2019300211049615e-05,
-      "loss": 66.2041,
-      "step": 83870
-    },
-    {
-      "epoch": 0.3388858139036915,
-      "grad_norm": 1587.0950927734375,
-      "learning_rate": 4.2016743153952505e-05,
-      "loss": 72.81,
-      "step": 83880
-    },
-    {
-      "epoch": 0.3389262151690591,
-      "grad_norm": 1762.2025146484375,
-      "learning_rate": 4.201418576510523e-05,
-      "loss": 55.4421,
-      "step": 83890
-    },
-    {
-      "epoch": 0.3389666164344267,
-      "grad_norm": 874.0090942382812,
-      "learning_rate": 4.201162804455763e-05,
-      "loss": 62.957,
-      "step": 83900
-    },
-    {
-      "epoch": 0.33900701769979436,
-      "grad_norm": 457.5177307128906,
-      "learning_rate": 4.2009069992359595e-05,
-      "loss": 100.793,
-      "step": 83910
-    },
-    {
-      "epoch": 0.339047418965162,
-      "grad_norm": 469.1245422363281,
-      "learning_rate": 4.200651160856098e-05,
-      "loss": 66.5053,
-      "step": 83920
-    },
-    {
-      "epoch": 0.33908782023052964,
-      "grad_norm": 394.78167724609375,
-      "learning_rate": 4.200395289321167e-05,
-      "loss": 35.0095,
-      "step": 83930
-    },
-    {
-      "epoch": 0.3391282214958973,
-      "grad_norm": 1044.5858154296875,
-      "learning_rate": 4.2001393846361536e-05,
-      "loss": 99.2867,
-      "step": 83940
-    },
-    {
-      "epoch": 0.33916862276126486,
-      "grad_norm": 793.2132568359375,
-      "learning_rate": 4.199883446806048e-05,
-      "loss": 31.189,
-      "step": 83950
-    },
-    {
-      "epoch": 0.3392090240266325,
-      "grad_norm": 441.0299072265625,
-      "learning_rate": 4.19962747583584e-05,
-      "loss": 78.0899,
-      "step": 83960
-    },
-    {
-      "epoch": 0.33924942529200014,
-      "grad_norm": 1687.0687255859375,
-      "learning_rate": 4.1993714717305185e-05,
-      "loss": 110.2709,
-      "step": 83970
-    },
-    {
-      "epoch": 0.3392898265573678,
-      "grad_norm": 645.5243530273438,
-      "learning_rate": 4.199115434495076e-05,
-      "loss": 70.5179,
-      "step": 83980
-    },
-    {
-      "epoch": 0.3393302278227354,
-      "grad_norm": 889.2882690429688,
-      "learning_rate": 4.1988593641345024e-05,
-      "loss": 59.327,
-      "step": 83990
-    },
-    {
-      "epoch": 0.33937062908810306,
-      "grad_norm": 545.3060913085938,
-      "learning_rate": 4.198603260653792e-05,
-      "loss": 62.0888,
-      "step": 84000
-    },
-    {
-      "epoch": 0.33941103035347064,
-      "grad_norm": 437.35064697265625,
-      "learning_rate": 4.1983471240579356e-05,
-      "loss": 48.4227,
-      "step": 84010
-    },
-    {
-      "epoch": 0.3394514316188383,
-      "grad_norm": 662.9463500976562,
-      "learning_rate": 4.198090954351928e-05,
-      "loss": 80.6067,
-      "step": 84020
-    },
-    {
-      "epoch": 0.3394918328842059,
-      "grad_norm": 952.4712524414062,
-      "learning_rate": 4.197834751540762e-05,
-      "loss": 84.5905,
-      "step": 84030
-    },
-    {
-      "epoch": 0.33953223414957356,
-      "grad_norm": 781.6859130859375,
-      "learning_rate": 4.197578515629435e-05,
-      "loss": 60.8523,
-      "step": 84040
-    },
-    {
-      "epoch": 0.3395726354149412,
-      "grad_norm": 975.0667724609375,
-      "learning_rate": 4.1973222466229404e-05,
-      "loss": 54.1651,
-      "step": 84050
-    },
-    {
-      "epoch": 0.33961303668030884,
-      "grad_norm": 2085.77392578125,
-      "learning_rate": 4.197065944526275e-05,
-      "loss": 60.0403,
-      "step": 84060
-    },
-    {
-      "epoch": 0.3396534379456765,
-      "grad_norm": 599.7440185546875,
-      "learning_rate": 4.196809609344434e-05,
-      "loss": 56.4642,
-      "step": 84070
-    },
-    {
-      "epoch": 0.33969383921104407,
-      "grad_norm": 1469.0364990234375,
-      "learning_rate": 4.196553241082418e-05,
-      "loss": 57.0351,
-      "step": 84080
-    },
-    {
-      "epoch": 0.3397342404764117,
-      "grad_norm": 579.5526123046875,
-      "learning_rate": 4.1962968397452216e-05,
-      "loss": 58.5323,
-      "step": 84090
-    },
-    {
-      "epoch": 0.33977464174177935,
-      "grad_norm": 639.9765625,
-      "learning_rate": 4.1960404053378454e-05,
-      "loss": 65.8241,
-      "step": 84100
-    },
-    {
-      "epoch": 0.339815043007147,
-      "grad_norm": 1099.93017578125,
-      "learning_rate": 4.1957839378652886e-05,
-      "loss": 81.618,
-      "step": 84110
-    },
-    {
-      "epoch": 0.33985544427251463,
-      "grad_norm": 490.1031494140625,
-      "learning_rate": 4.1955274373325506e-05,
-      "loss": 71.3405,
-      "step": 84120
-    },
-    {
-      "epoch": 0.33989584553788227,
-      "grad_norm": 2551.330322265625,
-      "learning_rate": 4.1952709037446324e-05,
-      "loss": 63.7435,
-      "step": 84130
-    },
-    {
-      "epoch": 0.33993624680324985,
-      "grad_norm": 930.4701538085938,
-      "learning_rate": 4.1950143371065355e-05,
-      "loss": 58.0226,
-      "step": 84140
-    },
-    {
-      "epoch": 0.3399766480686175,
-      "grad_norm": 856.0897216796875,
-      "learning_rate": 4.194757737423261e-05,
-      "loss": 58.936,
-      "step": 84150
-    },
-    {
-      "epoch": 0.34001704933398513,
-      "grad_norm": 592.7817993164062,
-      "learning_rate": 4.194501104699812e-05,
-      "loss": 55.8844,
-      "step": 84160
-    },
-    {
-      "epoch": 0.3400574505993528,
-      "grad_norm": 1155.2713623046875,
-      "learning_rate": 4.194244438941192e-05,
-      "loss": 62.0951,
-      "step": 84170
-    },
-    {
-      "epoch": 0.3400978518647204,
-      "grad_norm": 353.23583984375,
-      "learning_rate": 4.193987740152404e-05,
-      "loss": 59.3539,
-      "step": 84180
-    },
-    {
-      "epoch": 0.34013825313008805,
-      "grad_norm": 1153.6064453125,
-      "learning_rate": 4.193731008338453e-05,
-      "loss": 73.7305,
-      "step": 84190
-    },
-    {
-      "epoch": 0.3401786543954557,
-      "grad_norm": 524.4332885742188,
-      "learning_rate": 4.193474243504343e-05,
-      "loss": 77.0145,
-      "step": 84200
-    },
-    {
-      "epoch": 0.3402190556608233,
-      "grad_norm": 718.3153686523438,
-      "learning_rate": 4.193217445655082e-05,
-      "loss": 61.5601,
-      "step": 84210
-    },
-    {
-      "epoch": 0.3402594569261909,
-      "grad_norm": 1005.5006713867188,
-      "learning_rate": 4.192960614795675e-05,
-      "loss": 53.4194,
-      "step": 84220
-    },
-    {
-      "epoch": 0.34029985819155856,
-      "grad_norm": 1067.0313720703125,
-      "learning_rate": 4.192703750931129e-05,
-      "loss": 62.9559,
-      "step": 84230
-    },
-    {
-      "epoch": 0.3403402594569262,
-      "grad_norm": 794.4739990234375,
-      "learning_rate": 4.192446854066452e-05,
-      "loss": 46.9849,
-      "step": 84240
-    },
-    {
-      "epoch": 0.34038066072229384,
-      "grad_norm": 821.4888305664062,
-      "learning_rate": 4.192189924206652e-05,
-      "loss": 54.5941,
-      "step": 84250
-    },
-    {
-      "epoch": 0.3404210619876615,
-      "grad_norm": 876.5982666015625,
-      "learning_rate": 4.191932961356739e-05,
-      "loss": 91.2899,
-      "step": 84260
-    },
-    {
-      "epoch": 0.34046146325302906,
-      "grad_norm": 780.8380737304688,
-      "learning_rate": 4.1916759655217206e-05,
-      "loss": 53.8854,
-      "step": 84270
-    },
-    {
-      "epoch": 0.3405018645183967,
-      "grad_norm": 2006.7110595703125,
-      "learning_rate": 4.1914189367066094e-05,
-      "loss": 64.6178,
-      "step": 84280
-    },
-    {
-      "epoch": 0.34054226578376434,
-      "grad_norm": 663.4932861328125,
-      "learning_rate": 4.191161874916415e-05,
-      "loss": 71.5971,
-      "step": 84290
-    },
-    {
-      "epoch": 0.340582667049132,
-      "grad_norm": 1891.99658203125,
-      "learning_rate": 4.1909047801561484e-05,
-      "loss": 87.5498,
-      "step": 84300
-    },
-    {
-      "epoch": 0.3406230683144996,
-      "grad_norm": 483.8343811035156,
-      "learning_rate": 4.1906476524308235e-05,
-      "loss": 87.025,
-      "step": 84310
-    },
-    {
-      "epoch": 0.34066346957986726,
-      "grad_norm": 1536.576171875,
-      "learning_rate": 4.1903904917454516e-05,
-      "loss": 88.3288,
-      "step": 84320
-    },
-    {
-      "epoch": 0.34070387084523485,
-      "grad_norm": 708.4118041992188,
-      "learning_rate": 4.190133298105047e-05,
-      "loss": 59.6797,
-      "step": 84330
-    },
-    {
-      "epoch": 0.3407442721106025,
-      "grad_norm": 837.2510986328125,
-      "learning_rate": 4.189876071514624e-05,
-      "loss": 60.1374,
-      "step": 84340
-    },
-    {
-      "epoch": 0.3407846733759701,
-      "grad_norm": 2082.153564453125,
-      "learning_rate": 4.189618811979197e-05,
-      "loss": 65.0576,
-      "step": 84350
-    },
-    {
-      "epoch": 0.34082507464133777,
-      "grad_norm": 1336.408447265625,
-      "learning_rate": 4.18936151950378e-05,
-      "loss": 76.66,
-      "step": 84360
-    },
-    {
-      "epoch": 0.3408654759067054,
-      "grad_norm": 2176.240234375,
-      "learning_rate": 4.189104194093392e-05,
-      "loss": 84.4399,
-      "step": 84370
-    },
-    {
-      "epoch": 0.34090587717207305,
-      "grad_norm": 602.0006103515625,
-      "learning_rate": 4.1888468357530476e-05,
-      "loss": 68.7765,
-      "step": 84380
-    },
-    {
-      "epoch": 0.3409462784374407,
-      "grad_norm": 614.801513671875,
-      "learning_rate": 4.188589444487765e-05,
-      "loss": 63.4444,
-      "step": 84390
-    },
-    {
-      "epoch": 0.34098667970280827,
-      "grad_norm": 1069.0560302734375,
-      "learning_rate": 4.188332020302561e-05,
-      "loss": 65.1551,
-      "step": 84400
-    },
-    {
-      "epoch": 0.3410270809681759,
-      "grad_norm": 667.7435913085938,
-      "learning_rate": 4.1880745632024554e-05,
-      "loss": 62.9196,
-      "step": 84410
-    },
-    {
-      "epoch": 0.34106748223354355,
-      "grad_norm": 668.0552368164062,
-      "learning_rate": 4.187817073192468e-05,
-      "loss": 40.3692,
-      "step": 84420
-    },
-    {
-      "epoch": 0.3411078834989112,
-      "grad_norm": 943.7819213867188,
-      "learning_rate": 4.187559550277617e-05,
-      "loss": 57.6345,
-      "step": 84430
-    },
-    {
-      "epoch": 0.34114828476427883,
-      "grad_norm": 314.5076599121094,
-      "learning_rate": 4.187301994462924e-05,
-      "loss": 59.5329,
-      "step": 84440
-    },
-    {
-      "epoch": 0.34118868602964647,
-      "grad_norm": 759.6746826171875,
-      "learning_rate": 4.1870444057534095e-05,
-      "loss": 58.5424,
-      "step": 84450
-    },
-    {
-      "epoch": 0.34122908729501406,
-      "grad_norm": 802.6051635742188,
-      "learning_rate": 4.1867867841540964e-05,
-      "loss": 58.658,
-      "step": 84460
-    },
-    {
-      "epoch": 0.3412694885603817,
-      "grad_norm": 1856.4595947265625,
-      "learning_rate": 4.186529129670006e-05,
-      "loss": 88.8393,
-      "step": 84470
-    },
-    {
-      "epoch": 0.34130988982574934,
-      "grad_norm": 921.8723754882812,
-      "learning_rate": 4.1862714423061624e-05,
-      "loss": 62.6543,
-      "step": 84480
-    },
-    {
-      "epoch": 0.341350291091117,
-      "grad_norm": 756.6406860351562,
-      "learning_rate": 4.186013722067588e-05,
-      "loss": 70.8543,
-      "step": 84490
-    },
-    {
-      "epoch": 0.3413906923564846,
-      "grad_norm": 2013.621826171875,
-      "learning_rate": 4.185755968959308e-05,
-      "loss": 62.2787,
-      "step": 84500
-    },
-    {
-      "epoch": 0.34143109362185226,
-      "grad_norm": 831.4722900390625,
-      "learning_rate": 4.185498182986349e-05,
-      "loss": 42.0386,
-      "step": 84510
-    },
-    {
-      "epoch": 0.3414714948872199,
-      "grad_norm": 723.7086181640625,
-      "learning_rate": 4.185240364153734e-05,
-      "loss": 67.1348,
-      "step": 84520
-    },
-    {
-      "epoch": 0.3415118961525875,
-      "grad_norm": 754.1615600585938,
-      "learning_rate": 4.184982512466491e-05,
-      "loss": 80.2152,
-      "step": 84530
-    },
-    {
-      "epoch": 0.3415522974179551,
-      "grad_norm": 326.5327453613281,
-      "learning_rate": 4.1847246279296464e-05,
-      "loss": 81.1519,
-      "step": 84540
-    },
-    {
-      "epoch": 0.34159269868332276,
-      "grad_norm": 528.33935546875,
-      "learning_rate": 4.184466710548227e-05,
-      "loss": 64.1323,
-      "step": 84550
-    },
-    {
-      "epoch": 0.3416330999486904,
-      "grad_norm": 682.2901611328125,
-      "learning_rate": 4.184208760327263e-05,
-      "loss": 64.297,
-      "step": 84560
-    },
-    {
-      "epoch": 0.34167350121405804,
-      "grad_norm": 516.5701904296875,
-      "learning_rate": 4.183950777271781e-05,
-      "loss": 54.8298,
-      "step": 84570
-    },
-    {
-      "epoch": 0.3417139024794257,
-      "grad_norm": 674.0258178710938,
-      "learning_rate": 4.183692761386813e-05,
-      "loss": 66.7959,
-      "step": 84580
-    },
-    {
-      "epoch": 0.34175430374479326,
-      "grad_norm": 317.5458984375,
-      "learning_rate": 4.183434712677387e-05,
-      "loss": 43.3866,
-      "step": 84590
-    },
-    {
-      "epoch": 0.3417947050101609,
-      "grad_norm": 1114.966064453125,
-      "learning_rate": 4.183176631148534e-05,
-      "loss": 63.7891,
-      "step": 84600
-    },
-    {
-      "epoch": 0.34183510627552854,
-      "grad_norm": 887.722412109375,
-      "learning_rate": 4.1829185168052877e-05,
-      "loss": 87.129,
-      "step": 84610
-    },
-    {
-      "epoch": 0.3418755075408962,
-      "grad_norm": 735.17822265625,
-      "learning_rate": 4.182660369652677e-05,
-      "loss": 52.123,
-      "step": 84620
-    },
-    {
-      "epoch": 0.3419159088062638,
-      "grad_norm": 470.5625305175781,
-      "learning_rate": 4.182402189695736e-05,
-      "loss": 82.246,
-      "step": 84630
-    },
-    {
-      "epoch": 0.34195631007163146,
-      "grad_norm": 1080.039794921875,
-      "learning_rate": 4.1821439769395e-05,
-      "loss": 85.6906,
-      "step": 84640
-    },
-    {
-      "epoch": 0.34199671133699905,
-      "grad_norm": 529.3787841796875,
-      "learning_rate": 4.181885731388999e-05,
-      "loss": 66.6203,
-      "step": 84650
-    },
-    {
-      "epoch": 0.3420371126023667,
-      "grad_norm": 710.7547607421875,
-      "learning_rate": 4.1816274530492713e-05,
-      "loss": 64.3684,
-      "step": 84660
-    },
-    {
-      "epoch": 0.34207751386773433,
-      "grad_norm": 1438.0509033203125,
-      "learning_rate": 4.18136914192535e-05,
-      "loss": 71.3715,
-      "step": 84670
-    },
-    {
-      "epoch": 0.34211791513310197,
-      "grad_norm": 779.0742797851562,
-      "learning_rate": 4.181110798022271e-05,
-      "loss": 64.4421,
-      "step": 84680
-    },
-    {
-      "epoch": 0.3421583163984696,
-      "grad_norm": 415.0228576660156,
-      "learning_rate": 4.180852421345072e-05,
-      "loss": 58.738,
-      "step": 84690
-    },
-    {
-      "epoch": 0.34219871766383725,
-      "grad_norm": 530.8126220703125,
-      "learning_rate": 4.180594011898791e-05,
-      "loss": 64.9684,
-      "step": 84700
-    },
-    {
-      "epoch": 0.3422391189292049,
-      "grad_norm": 553.169677734375,
-      "learning_rate": 4.1803355696884625e-05,
-      "loss": 58.3446,
-      "step": 84710
-    },
-    {
-      "epoch": 0.3422795201945725,
-      "grad_norm": 766.584228515625,
-      "learning_rate": 4.180077094719128e-05,
-      "loss": 66.1009,
-      "step": 84720
-    },
-    {
-      "epoch": 0.3423199214599401,
-      "grad_norm": 801.9248657226562,
-      "learning_rate": 4.179818586995825e-05,
-      "loss": 49.8386,
-      "step": 84730
-    },
-    {
-      "epoch": 0.34236032272530775,
-      "grad_norm": 1037.901611328125,
-      "learning_rate": 4.1795600465235947e-05,
-      "loss": 79.0902,
-      "step": 84740
-    },
-    {
-      "epoch": 0.3424007239906754,
-      "grad_norm": 871.1192626953125,
-      "learning_rate": 4.179301473307476e-05,
-      "loss": 89.6048,
-      "step": 84750
-    },
-    {
-      "epoch": 0.34244112525604303,
-      "grad_norm": 257.207763671875,
-      "learning_rate": 4.179042867352511e-05,
-      "loss": 53.3181,
-      "step": 84760
-    },
-    {
-      "epoch": 0.3424815265214107,
-      "grad_norm": 1137.479736328125,
-      "learning_rate": 4.17878422866374e-05,
-      "loss": 64.5483,
-      "step": 84770
-    },
-    {
-      "epoch": 0.34252192778677826,
-      "grad_norm": 1104.8631591796875,
-      "learning_rate": 4.1785255572462066e-05,
-      "loss": 50.5376,
-      "step": 84780
-    },
-    {
-      "epoch": 0.3425623290521459,
-      "grad_norm": 674.0195922851562,
-      "learning_rate": 4.178266853104954e-05,
-      "loss": 61.1695,
-      "step": 84790
-    },
-    {
-      "epoch": 0.34260273031751354,
-      "grad_norm": 647.3651733398438,
-      "learning_rate": 4.178008116245024e-05,
-      "loss": 73.4516,
-      "step": 84800
-    },
-    {
-      "epoch": 0.3426431315828812,
-      "grad_norm": 1168.34228515625,
-      "learning_rate": 4.1777493466714624e-05,
-      "loss": 62.1275,
-      "step": 84810
-    },
-    {
-      "epoch": 0.3426835328482488,
-      "grad_norm": 2343.4892578125,
-      "learning_rate": 4.177490544389313e-05,
-      "loss": 64.6482,
-      "step": 84820
-    },
-    {
-      "epoch": 0.34272393411361646,
-      "grad_norm": 817.9552001953125,
-      "learning_rate": 4.177231709403622e-05,
-      "loss": 64.0418,
-      "step": 84830
-    },
-    {
-      "epoch": 0.3427643353789841,
-      "grad_norm": 753.8648681640625,
-      "learning_rate": 4.176972841719435e-05,
-      "loss": 57.1917,
-      "step": 84840
-    },
-    {
-      "epoch": 0.3428047366443517,
-      "grad_norm": 510.9037780761719,
-      "learning_rate": 4.1767139413418e-05,
-      "loss": 54.1073,
-      "step": 84850
-    },
-    {
-      "epoch": 0.3428451379097193,
-      "grad_norm": 641.147705078125,
-      "learning_rate": 4.176455008275764e-05,
-      "loss": 81.8066,
-      "step": 84860
-    },
-    {
-      "epoch": 0.34288553917508696,
-      "grad_norm": 1563.3798828125,
-      "learning_rate": 4.1761960425263735e-05,
-      "loss": 71.83,
-      "step": 84870
-    },
-    {
-      "epoch": 0.3429259404404546,
-      "grad_norm": 1007.089111328125,
-      "learning_rate": 4.1759370440986775e-05,
-      "loss": 65.9258,
-      "step": 84880
-    },
-    {
-      "epoch": 0.34296634170582224,
-      "grad_norm": 694.405517578125,
-      "learning_rate": 4.175678012997727e-05,
-      "loss": 53.0368,
-      "step": 84890
-    },
-    {
-      "epoch": 0.3430067429711899,
-      "grad_norm": 787.7047119140625,
-      "learning_rate": 4.1754189492285714e-05,
-      "loss": 78.8049,
-      "step": 84900
-    },
-    {
-      "epoch": 0.34304714423655747,
-      "grad_norm": 1134.1265869140625,
-      "learning_rate": 4.17515985279626e-05,
-      "loss": 52.2765,
-      "step": 84910
-    },
-    {
-      "epoch": 0.3430875455019251,
-      "grad_norm": 984.3347778320312,
-      "learning_rate": 4.174900723705845e-05,
-      "loss": 79.6157,
-      "step": 84920
-    },
-    {
-      "epoch": 0.34312794676729275,
-      "grad_norm": 671.3541259765625,
-      "learning_rate": 4.174641561962378e-05,
-      "loss": 68.477,
-      "step": 84930
-    },
-    {
-      "epoch": 0.3431683480326604,
-      "grad_norm": 870.8594970703125,
-      "learning_rate": 4.174382367570912e-05,
-      "loss": 60.4037,
-      "step": 84940
-    },
-    {
-      "epoch": 0.343208749298028,
-      "grad_norm": 11791.9111328125,
-      "learning_rate": 4.174123140536499e-05,
-      "loss": 129.9077,
-      "step": 84950
-    },
-    {
-      "epoch": 0.34324915056339567,
-      "grad_norm": 656.0787963867188,
-      "learning_rate": 4.1738638808641936e-05,
-      "loss": 68.1256,
-      "step": 84960
-    },
-    {
-      "epoch": 0.34328955182876325,
-      "grad_norm": 549.08935546875,
-      "learning_rate": 4.17360458855905e-05,
-      "loss": 46.7033,
-      "step": 84970
-    },
-    {
-      "epoch": 0.3433299530941309,
-      "grad_norm": 748.6014404296875,
-      "learning_rate": 4.1733452636261244e-05,
-      "loss": 37.716,
-      "step": 84980
-    },
-    {
-      "epoch": 0.34337035435949853,
-      "grad_norm": 870.7546997070312,
-      "learning_rate": 4.173085906070471e-05,
-      "loss": 46.5251,
-      "step": 84990
-    },
-    {
-      "epoch": 0.34341075562486617,
-      "grad_norm": 525.0897216796875,
-      "learning_rate": 4.172826515897146e-05,
-      "loss": 88.4015,
-      "step": 85000
-    },
-    {
-      "epoch": 0.3434511568902338,
-      "grad_norm": 1008.6449584960938,
-      "learning_rate": 4.172567093111207e-05,
-      "loss": 69.9759,
-      "step": 85010
-    },
-    {
-      "epoch": 0.34349155815560145,
-      "grad_norm": 694.2141723632812,
-      "learning_rate": 4.172307637717711e-05,
-      "loss": 64.9771,
-      "step": 85020
-    },
-    {
-      "epoch": 0.3435319594209691,
-      "grad_norm": 601.18505859375,
-      "learning_rate": 4.172048149721717e-05,
-      "loss": 72.7544,
-      "step": 85030
-    },
-    {
-      "epoch": 0.3435723606863367,
-      "grad_norm": 1000.3665771484375,
-      "learning_rate": 4.171788629128284e-05,
-      "loss": 79.2617,
-      "step": 85040
-    },
-    {
-      "epoch": 0.3436127619517043,
-      "grad_norm": 380.0011901855469,
-      "learning_rate": 4.1715290759424705e-05,
-      "loss": 77.5077,
-      "step": 85050
-    },
-    {
-      "epoch": 0.34365316321707196,
-      "grad_norm": 674.123779296875,
-      "learning_rate": 4.1712694901693374e-05,
-      "loss": 104.6444,
-      "step": 85060
-    },
-    {
-      "epoch": 0.3436935644824396,
-      "grad_norm": 2173.708251953125,
-      "learning_rate": 4.171009871813944e-05,
-      "loss": 66.0904,
-      "step": 85070
-    },
-    {
-      "epoch": 0.34373396574780724,
-      "grad_norm": 1144.228759765625,
-      "learning_rate": 4.170750220881354e-05,
-      "loss": 66.5176,
-      "step": 85080
-    },
-    {
-      "epoch": 0.3437743670131749,
-      "grad_norm": 648.3857421875,
-      "learning_rate": 4.1704905373766286e-05,
-      "loss": 58.1772,
-      "step": 85090
-    },
-    {
-      "epoch": 0.34381476827854246,
-      "grad_norm": 1063.3370361328125,
-      "learning_rate": 4.170230821304829e-05,
-      "loss": 56.7559,
-      "step": 85100
-    },
-    {
-      "epoch": 0.3438551695439101,
-      "grad_norm": 704.5231323242188,
-      "learning_rate": 4.169971072671021e-05,
-      "loss": 68.8396,
-      "step": 85110
-    },
-    {
-      "epoch": 0.34389557080927774,
-      "grad_norm": 857.602294921875,
-      "learning_rate": 4.169711291480266e-05,
-      "loss": 49.2389,
-      "step": 85120
-    },
-    {
-      "epoch": 0.3439359720746454,
-      "grad_norm": 1297.68701171875,
-      "learning_rate": 4.16945147773763e-05,
-      "loss": 61.9332,
-      "step": 85130
-    },
-    {
-      "epoch": 0.343976373340013,
-      "grad_norm": 603.5067138671875,
-      "learning_rate": 4.169191631448178e-05,
-      "loss": 83.4856,
-      "step": 85140
-    },
-    {
-      "epoch": 0.34401677460538066,
-      "grad_norm": 541.5191040039062,
-      "learning_rate": 4.1689317526169766e-05,
-      "loss": 44.5568,
-      "step": 85150
-    },
-    {
-      "epoch": 0.34405717587074824,
-      "grad_norm": 589.892578125,
-      "learning_rate": 4.168671841249091e-05,
-      "loss": 56.9382,
-      "step": 85160
-    },
-    {
-      "epoch": 0.3440975771361159,
-      "grad_norm": 555.4111328125,
-      "learning_rate": 4.168411897349588e-05,
-      "loss": 83.5068,
-      "step": 85170
-    },
-    {
-      "epoch": 0.3441379784014835,
-      "grad_norm": 547.9937133789062,
-      "learning_rate": 4.168151920923536e-05,
-      "loss": 34.5336,
-      "step": 85180
-    },
-    {
-      "epoch": 0.34417837966685116,
-      "grad_norm": 1771.55078125,
-      "learning_rate": 4.1678919119760054e-05,
-      "loss": 90.6765,
-      "step": 85190
-    },
-    {
-      "epoch": 0.3442187809322188,
-      "grad_norm": 696.42236328125,
-      "learning_rate": 4.1676318705120616e-05,
-      "loss": 59.9467,
-      "step": 85200
-    },
-    {
-      "epoch": 0.34425918219758644,
-      "grad_norm": 877.6763916015625,
-      "learning_rate": 4.167371796536777e-05,
-      "loss": 95.2035,
-      "step": 85210
-    },
-    {
-      "epoch": 0.3442995834629541,
-      "grad_norm": 1338.104248046875,
-      "learning_rate": 4.1671116900552194e-05,
-      "loss": 66.9077,
-      "step": 85220
-    },
-    {
-      "epoch": 0.34433998472832167,
-      "grad_norm": 992.7899169921875,
-      "learning_rate": 4.166851551072462e-05,
-      "loss": 60.782,
-      "step": 85230
-    },
-    {
-      "epoch": 0.3443803859936893,
-      "grad_norm": 1082.4932861328125,
-      "learning_rate": 4.166591379593575e-05,
-      "loss": 61.3283,
-      "step": 85240
-    },
-    {
-      "epoch": 0.34442078725905695,
-      "grad_norm": 1248.29052734375,
-      "learning_rate": 4.166331175623631e-05,
-      "loss": 65.6705,
-      "step": 85250
-    },
-    {
-      "epoch": 0.3444611885244246,
-      "grad_norm": 651.8319091796875,
-      "learning_rate": 4.166070939167703e-05,
-      "loss": 107.7256,
-      "step": 85260
-    },
-    {
-      "epoch": 0.34450158978979223,
-      "grad_norm": 950.817138671875,
-      "learning_rate": 4.165810670230865e-05,
-      "loss": 53.1864,
-      "step": 85270
-    },
-    {
-      "epoch": 0.34454199105515987,
-      "grad_norm": 871.7650756835938,
-      "learning_rate": 4.16555036881819e-05,
-      "loss": 97.4988,
-      "step": 85280
-    },
-    {
-      "epoch": 0.34458239232052745,
-      "grad_norm": 581.1294555664062,
-      "learning_rate": 4.1652900349347533e-05,
-      "loss": 52.556,
-      "step": 85290
-    },
-    {
-      "epoch": 0.3446227935858951,
-      "grad_norm": 919.3056640625,
-      "learning_rate": 4.165029668585629e-05,
-      "loss": 55.3191,
-      "step": 85300
-    },
-    {
-      "epoch": 0.34466319485126273,
-      "grad_norm": 622.2424926757812,
-      "learning_rate": 4.164769269775896e-05,
-      "loss": 107.1292,
-      "step": 85310
-    },
-    {
-      "epoch": 0.3447035961166304,
-      "grad_norm": 0.0,
-      "learning_rate": 4.1645088385106266e-05,
-      "loss": 51.1484,
-      "step": 85320
-    },
-    {
-      "epoch": 0.344743997381998,
-      "grad_norm": 2131.662353515625,
-      "learning_rate": 4.164248374794902e-05,
-      "loss": 83.7406,
-      "step": 85330
-    },
-    {
-      "epoch": 0.34478439864736565,
-      "grad_norm": 527.97509765625,
-      "learning_rate": 4.163987878633798e-05,
-      "loss": 57.3865,
-      "step": 85340
-    },
-    {
-      "epoch": 0.3448247999127333,
-      "grad_norm": 460.5814208984375,
-      "learning_rate": 4.163727350032394e-05,
-      "loss": 35.4961,
-      "step": 85350
-    },
-    {
-      "epoch": 0.3448652011781009,
-      "grad_norm": 1813.7447509765625,
-      "learning_rate": 4.1634667889957676e-05,
-      "loss": 93.811,
-      "step": 85360
-    },
-    {
-      "epoch": 0.3449056024434685,
-      "grad_norm": 516.6041259765625,
-      "learning_rate": 4.1632061955290017e-05,
-      "loss": 66.2047,
-      "step": 85370
-    },
-    {
-      "epoch": 0.34494600370883616,
-      "grad_norm": 1379.9407958984375,
-      "learning_rate": 4.1629455696371734e-05,
-      "loss": 62.7511,
-      "step": 85380
-    },
-    {
-      "epoch": 0.3449864049742038,
-      "grad_norm": 573.064453125,
-      "learning_rate": 4.162684911325365e-05,
-      "loss": 68.7898,
-      "step": 85390
-    },
-    {
-      "epoch": 0.34502680623957144,
-      "grad_norm": 620.5324096679688,
-      "learning_rate": 4.162424220598658e-05,
-      "loss": 62.7597,
-      "step": 85400
-    },
-    {
-      "epoch": 0.3450672075049391,
-      "grad_norm": 522.501708984375,
-      "learning_rate": 4.162163497462136e-05,
-      "loss": 63.0242,
-      "step": 85410
-    },
-    {
-      "epoch": 0.34510760877030666,
-      "grad_norm": 687.868896484375,
-      "learning_rate": 4.161902741920881e-05,
-      "loss": 68.5159,
-      "step": 85420
-    },
-    {
-      "epoch": 0.3451480100356743,
-      "grad_norm": 842.2858276367188,
-      "learning_rate": 4.1616419539799754e-05,
-      "loss": 46.8133,
-      "step": 85430
-    },
-    {
-      "epoch": 0.34518841130104194,
-      "grad_norm": 808.839111328125,
-      "learning_rate": 4.161381133644505e-05,
-      "loss": 58.4971,
-      "step": 85440
-    },
-    {
-      "epoch": 0.3452288125664096,
-      "grad_norm": 2333.049072265625,
-      "learning_rate": 4.161120280919554e-05,
-      "loss": 99.8404,
-      "step": 85450
-    },
-    {
-      "epoch": 0.3452692138317772,
-      "grad_norm": 667.53125,
-      "learning_rate": 4.160859395810208e-05,
-      "loss": 74.1404,
-      "step": 85460
-    },
-    {
-      "epoch": 0.34530961509714486,
-      "grad_norm": 886.0628051757812,
-      "learning_rate": 4.160598478321553e-05,
-      "loss": 62.568,
-      "step": 85470
-    },
-    {
-      "epoch": 0.34535001636251245,
-      "grad_norm": 421.406005859375,
-      "learning_rate": 4.160337528458676e-05,
-      "loss": 75.3051,
-      "step": 85480
-    },
-    {
-      "epoch": 0.3453904176278801,
-      "grad_norm": 852.6929931640625,
-      "learning_rate": 4.160076546226663e-05,
-      "loss": 67.8079,
-      "step": 85490
-    },
-    {
-      "epoch": 0.3454308188932477,
-      "grad_norm": 1610.6650390625,
-      "learning_rate": 4.1598155316306044e-05,
-      "loss": 67.7026,
-      "step": 85500
-    },
-    {
-      "epoch": 0.34547122015861537,
-      "grad_norm": 740.6041259765625,
-      "learning_rate": 4.1595544846755865e-05,
-      "loss": 69.3296,
-      "step": 85510
-    },
-    {
-      "epoch": 0.345511621423983,
-      "grad_norm": 713.525390625,
-      "learning_rate": 4.1592934053667004e-05,
-      "loss": 66.8102,
-      "step": 85520
-    },
-    {
-      "epoch": 0.34555202268935065,
-      "grad_norm": 1433.9876708984375,
-      "learning_rate": 4.1590322937090345e-05,
-      "loss": 49.2965,
-      "step": 85530
-    },
-    {
-      "epoch": 0.3455924239547183,
-      "grad_norm": 974.6792602539062,
-      "learning_rate": 4.15877114970768e-05,
-      "loss": 71.5766,
-      "step": 85540
-    },
-    {
-      "epoch": 0.34563282522008587,
-      "grad_norm": 996.3238525390625,
-      "learning_rate": 4.158509973367728e-05,
-      "loss": 82.5995,
-      "step": 85550
-    },
-    {
-      "epoch": 0.3456732264854535,
-      "grad_norm": 3350.908935546875,
-      "learning_rate": 4.1582487646942706e-05,
-      "loss": 86.8919,
-      "step": 85560
-    },
-    {
-      "epoch": 0.34571362775082115,
-      "grad_norm": 1187.7080078125,
-      "learning_rate": 4.157987523692399e-05,
-      "loss": 59.8579,
-      "step": 85570
-    },
-    {
-      "epoch": 0.3457540290161888,
-      "grad_norm": 713.9551391601562,
-      "learning_rate": 4.157726250367207e-05,
-      "loss": 61.2336,
-      "step": 85580
-    },
-    {
-      "epoch": 0.34579443028155643,
-      "grad_norm": 932.2216186523438,
-      "learning_rate": 4.157464944723789e-05,
-      "loss": 72.413,
-      "step": 85590
-    },
-    {
-      "epoch": 0.34583483154692407,
-      "grad_norm": 813.9993286132812,
-      "learning_rate": 4.157203606767238e-05,
-      "loss": 65.8375,
-      "step": 85600
-    },
-    {
-      "epoch": 0.34587523281229166,
-      "grad_norm": 648.2496948242188,
-      "learning_rate": 4.15694223650265e-05,
-      "loss": 56.1922,
-      "step": 85610
-    },
-    {
-      "epoch": 0.3459156340776593,
-      "grad_norm": 1841.92236328125,
-      "learning_rate": 4.156680833935119e-05,
-      "loss": 73.8934,
-      "step": 85620
-    },
-    {
-      "epoch": 0.34595603534302694,
-      "grad_norm": 1668.5313720703125,
-      "learning_rate": 4.156419399069744e-05,
-      "loss": 98.3084,
-      "step": 85630
-    },
-    {
-      "epoch": 0.3459964366083946,
-      "grad_norm": 451.40338134765625,
-      "learning_rate": 4.156157931911619e-05,
-      "loss": 41.3341,
-      "step": 85640
-    },
-    {
-      "epoch": 0.3460368378737622,
-      "grad_norm": 740.665771484375,
-      "learning_rate": 4.155896432465842e-05,
-      "loss": 93.8652,
-      "step": 85650
-    },
-    {
-      "epoch": 0.34607723913912986,
-      "grad_norm": 716.4408569335938,
-      "learning_rate": 4.155634900737513e-05,
-      "loss": 57.4022,
-      "step": 85660
-    },
-    {
-      "epoch": 0.3461176404044975,
-      "grad_norm": 543.2172241210938,
-      "learning_rate": 4.155373336731728e-05,
-      "loss": 49.6871,
-      "step": 85670
-    },
-    {
-      "epoch": 0.3461580416698651,
-      "grad_norm": 1185.981689453125,
-      "learning_rate": 4.155111740453588e-05,
-      "loss": 69.9553,
-      "step": 85680
-    },
-    {
-      "epoch": 0.3461984429352327,
-      "grad_norm": 817.763671875,
-      "learning_rate": 4.154850111908192e-05,
-      "loss": 54.951,
-      "step": 85690
-    },
-    {
-      "epoch": 0.34623884420060036,
-      "grad_norm": 793.4094848632812,
-      "learning_rate": 4.154588451100642e-05,
-      "loss": 98.9682,
-      "step": 85700
-    },
-    {
-      "epoch": 0.346279245465968,
-      "grad_norm": 643.4264526367188,
-      "learning_rate": 4.1543267580360374e-05,
-      "loss": 52.4739,
-      "step": 85710
-    },
-    {
-      "epoch": 0.34631964673133564,
-      "grad_norm": 1268.8812255859375,
-      "learning_rate": 4.154065032719481e-05,
-      "loss": 83.5729,
-      "step": 85720
-    },
-    {
-      "epoch": 0.3463600479967033,
-      "grad_norm": 532.7398681640625,
-      "learning_rate": 4.153803275156076e-05,
-      "loss": 52.5248,
-      "step": 85730
-    },
-    {
-      "epoch": 0.34640044926207086,
-      "grad_norm": 2376.361083984375,
-      "learning_rate": 4.153541485350924e-05,
-      "loss": 97.7747,
-      "step": 85740
-    },
-    {
-      "epoch": 0.3464408505274385,
-      "grad_norm": 2388.05908203125,
-      "learning_rate": 4.1532796633091296e-05,
-      "loss": 77.3507,
-      "step": 85750
-    },
-    {
-      "epoch": 0.34648125179280614,
-      "grad_norm": 625.3255004882812,
-      "learning_rate": 4.1530178090357976e-05,
-      "loss": 56.9691,
-      "step": 85760
-    },
-    {
-      "epoch": 0.3465216530581738,
-      "grad_norm": 792.126953125,
-      "learning_rate": 4.152755922536032e-05,
-      "loss": 51.1804,
-      "step": 85770
-    },
-    {
-      "epoch": 0.3465620543235414,
-      "grad_norm": 379.2236328125,
-      "learning_rate": 4.1524940038149384e-05,
-      "loss": 53.4376,
-      "step": 85780
-    },
-    {
-      "epoch": 0.34660245558890906,
-      "grad_norm": 1024.7281494140625,
-      "learning_rate": 4.152232052877624e-05,
-      "loss": 101.2286,
-      "step": 85790
-    },
-    {
-      "epoch": 0.34664285685427665,
-      "grad_norm": 1685.01611328125,
-      "learning_rate": 4.1519700697291944e-05,
-      "loss": 58.5797,
-      "step": 85800
-    },
-    {
-      "epoch": 0.3466832581196443,
-      "grad_norm": 642.328857421875,
-      "learning_rate": 4.1517080543747584e-05,
-      "loss": 80.0744,
-      "step": 85810
-    },
-    {
-      "epoch": 0.34672365938501193,
-      "grad_norm": 760.1344604492188,
-      "learning_rate": 4.151446006819423e-05,
-      "loss": 65.5687,
-      "step": 85820
-    },
-    {
-      "epoch": 0.34676406065037957,
-      "grad_norm": 499.570556640625,
-      "learning_rate": 4.151183927068298e-05,
-      "loss": 57.8297,
-      "step": 85830
-    },
-    {
-      "epoch": 0.3468044619157472,
-      "grad_norm": 1389.1029052734375,
-      "learning_rate": 4.150921815126493e-05,
-      "loss": 74.6479,
-      "step": 85840
-    },
-    {
-      "epoch": 0.34684486318111485,
-      "grad_norm": 1510.54150390625,
-      "learning_rate": 4.150659670999116e-05,
-      "loss": 82.4179,
-      "step": 85850
-    },
-    {
-      "epoch": 0.3468852644464825,
-      "grad_norm": 728.5758056640625,
-      "learning_rate": 4.150397494691279e-05,
-      "loss": 70.3151,
-      "step": 85860
-    },
-    {
-      "epoch": 0.3469256657118501,
-      "grad_norm": 1455.7420654296875,
-      "learning_rate": 4.150135286208093e-05,
-      "loss": 73.2886,
-      "step": 85870
-    },
-    {
-      "epoch": 0.3469660669772177,
-      "grad_norm": 328.35015869140625,
-      "learning_rate": 4.149873045554671e-05,
-      "loss": 42.4879,
-      "step": 85880
-    },
-    {
-      "epoch": 0.34700646824258535,
-      "grad_norm": 434.59527587890625,
-      "learning_rate": 4.1496107727361235e-05,
-      "loss": 45.5354,
-      "step": 85890
-    },
-    {
-      "epoch": 0.347046869507953,
-      "grad_norm": 3333.067138671875,
-      "learning_rate": 4.149348467757566e-05,
-      "loss": 98.8698,
-      "step": 85900
-    },
-    {
-      "epoch": 0.34708727077332063,
-      "grad_norm": 605.552978515625,
-      "learning_rate": 4.1490861306241096e-05,
-      "loss": 74.4398,
-      "step": 85910
-    },
-    {
-      "epoch": 0.3471276720386883,
-      "grad_norm": 1759.8612060546875,
-      "learning_rate": 4.148823761340871e-05,
-      "loss": 76.9145,
-      "step": 85920
-    },
-    {
-      "epoch": 0.34716807330405586,
-      "grad_norm": 567.1038818359375,
-      "learning_rate": 4.1485613599129636e-05,
-      "loss": 40.8438,
-      "step": 85930
-    },
-    {
-      "epoch": 0.3472084745694235,
-      "grad_norm": 3881.6103515625,
-      "learning_rate": 4.148298926345504e-05,
-      "loss": 66.3889,
-      "step": 85940
-    },
-    {
-      "epoch": 0.34724887583479114,
-      "grad_norm": 621.2008666992188,
-      "learning_rate": 4.148036460643608e-05,
-      "loss": 59.3511,
-      "step": 85950
-    },
-    {
-      "epoch": 0.3472892771001588,
-      "grad_norm": 468.0349426269531,
-      "learning_rate": 4.1477739628123934e-05,
-      "loss": 101.2036,
-      "step": 85960
-    },
-    {
-      "epoch": 0.3473296783655264,
-      "grad_norm": 1429.98095703125,
-      "learning_rate": 4.1475114328569776e-05,
-      "loss": 56.234,
-      "step": 85970
-    },
-    {
-      "epoch": 0.34737007963089406,
-      "grad_norm": 695.315673828125,
-      "learning_rate": 4.147248870782477e-05,
-      "loss": 54.0909,
-      "step": 85980
-    },
-    {
-      "epoch": 0.3474104808962617,
-      "grad_norm": 1032.339111328125,
-      "learning_rate": 4.146986276594012e-05,
-      "loss": 73.7455,
-      "step": 85990
-    },
-    {
-      "epoch": 0.3474508821616293,
-      "grad_norm": 353.8186950683594,
-      "learning_rate": 4.146723650296701e-05,
-      "loss": 55.2589,
-      "step": 86000
-    },
-    {
-      "epoch": 0.3474912834269969,
-      "grad_norm": 1116.111572265625,
-      "learning_rate": 4.146460991895666e-05,
-      "loss": 58.2723,
-      "step": 86010
-    },
-    {
-      "epoch": 0.34753168469236456,
-      "grad_norm": 416.8829040527344,
-      "learning_rate": 4.1461983013960245e-05,
-      "loss": 47.4193,
-      "step": 86020
-    },
-    {
-      "epoch": 0.3475720859577322,
-      "grad_norm": 794.2473754882812,
-      "learning_rate": 4.1459355788029013e-05,
-      "loss": 67.6769,
-      "step": 86030
-    },
-    {
-      "epoch": 0.34761248722309984,
-      "grad_norm": 727.6878051757812,
-      "learning_rate": 4.145672824121416e-05,
-      "loss": 84.4826,
-      "step": 86040
-    },
-    {
-      "epoch": 0.3476528884884675,
-      "grad_norm": 2156.322265625,
-      "learning_rate": 4.145410037356692e-05,
-      "loss": 82.6664,
-      "step": 86050
-    },
-    {
-      "epoch": 0.34769328975383507,
-      "grad_norm": 511.3225402832031,
-      "learning_rate": 4.145147218513852e-05,
-      "loss": 43.9569,
-      "step": 86060
-    },
-    {
-      "epoch": 0.3477336910192027,
-      "grad_norm": 298.62005615234375,
-      "learning_rate": 4.14488436759802e-05,
-      "loss": 45.4608,
-      "step": 86070
-    },
-    {
-      "epoch": 0.34777409228457035,
-      "grad_norm": 1355.5355224609375,
-      "learning_rate": 4.144621484614319e-05,
-      "loss": 124.826,
-      "step": 86080
-    },
-    {
-      "epoch": 0.347814493549938,
-      "grad_norm": 1087.8583984375,
-      "learning_rate": 4.1443585695678774e-05,
-      "loss": 54.7751,
-      "step": 86090
-    },
-    {
-      "epoch": 0.3478548948153056,
-      "grad_norm": 431.3752746582031,
-      "learning_rate": 4.1440956224638184e-05,
-      "loss": 63.882,
-      "step": 86100
-    },
-    {
-      "epoch": 0.34789529608067327,
-      "grad_norm": 1415.1689453125,
-      "learning_rate": 4.143832643307269e-05,
-      "loss": 55.6841,
-      "step": 86110
-    },
-    {
-      "epoch": 0.34793569734604085,
-      "grad_norm": 500.28759765625,
-      "learning_rate": 4.1435696321033554e-05,
-      "loss": 75.8191,
-      "step": 86120
-    },
-    {
-      "epoch": 0.3479760986114085,
-      "grad_norm": 725.256591796875,
-      "learning_rate": 4.143306588857206e-05,
-      "loss": 68.6684,
-      "step": 86130
-    },
-    {
-      "epoch": 0.34801649987677613,
-      "grad_norm": 1193.91357421875,
-      "learning_rate": 4.143043513573949e-05,
-      "loss": 68.4111,
-      "step": 86140
-    },
-    {
-      "epoch": 0.34805690114214377,
-      "grad_norm": 311.3026428222656,
-      "learning_rate": 4.1427804062587116e-05,
-      "loss": 60.3588,
-      "step": 86150
-    },
-    {
-      "epoch": 0.3480973024075114,
-      "grad_norm": 2180.667236328125,
-      "learning_rate": 4.142517266916625e-05,
-      "loss": 71.0806,
-      "step": 86160
-    },
-    {
-      "epoch": 0.34813770367287905,
-      "grad_norm": 561.2362060546875,
-      "learning_rate": 4.1422540955528186e-05,
-      "loss": 65.3,
-      "step": 86170
-    },
-    {
-      "epoch": 0.3481781049382467,
-      "grad_norm": 674.5610961914062,
-      "learning_rate": 4.141990892172424e-05,
-      "loss": 77.8949,
-      "step": 86180
-    },
-    {
-      "epoch": 0.3482185062036143,
-      "grad_norm": 1787.229248046875,
-      "learning_rate": 4.14172765678057e-05,
-      "loss": 88.2457,
-      "step": 86190
-    },
-    {
-      "epoch": 0.3482589074689819,
-      "grad_norm": 1689.93505859375,
-      "learning_rate": 4.1414643893823914e-05,
-      "loss": 63.5292,
-      "step": 86200
-    },
-    {
-      "epoch": 0.34829930873434956,
-      "grad_norm": 1198.643310546875,
-      "learning_rate": 4.141201089983019e-05,
-      "loss": 57.5542,
-      "step": 86210
-    },
-    {
-      "epoch": 0.3483397099997172,
-      "grad_norm": 301.3227233886719,
-      "learning_rate": 4.1409377585875865e-05,
-      "loss": 59.0172,
-      "step": 86220
-    },
-    {
-      "epoch": 0.34838011126508484,
-      "grad_norm": 4092.246826171875,
-      "learning_rate": 4.1406743952012275e-05,
-      "loss": 72.4229,
-      "step": 86230
-    },
-    {
-      "epoch": 0.3484205125304525,
-      "grad_norm": 2412.6689453125,
-      "learning_rate": 4.140410999829076e-05,
-      "loss": 118.2052,
-      "step": 86240
-    },
-    {
-      "epoch": 0.34846091379582006,
-      "grad_norm": 649.2863159179688,
-      "learning_rate": 4.140147572476268e-05,
-      "loss": 78.7629,
-      "step": 86250
-    },
-    {
-      "epoch": 0.3485013150611877,
-      "grad_norm": 634.6868286132812,
-      "learning_rate": 4.1398841131479395e-05,
-      "loss": 59.1645,
-      "step": 86260
-    },
-    {
-      "epoch": 0.34854171632655534,
-      "grad_norm": 908.8325805664062,
-      "learning_rate": 4.139620621849225e-05,
-      "loss": 60.6888,
-      "step": 86270
-    },
-    {
-      "epoch": 0.348582117591923,
-      "grad_norm": 1681.2099609375,
-      "learning_rate": 4.139357098585262e-05,
-      "loss": 82.6059,
-      "step": 86280
-    },
-    {
-      "epoch": 0.3486225188572906,
-      "grad_norm": 743.29443359375,
-      "learning_rate": 4.1390935433611886e-05,
-      "loss": 86.1345,
-      "step": 86290
-    },
-    {
-      "epoch": 0.34866292012265826,
-      "grad_norm": 906.7951049804688,
-      "learning_rate": 4.138829956182144e-05,
-      "loss": 63.9069,
-      "step": 86300
-    },
-    {
-      "epoch": 0.3487033213880259,
-      "grad_norm": 821.8445434570312,
-      "learning_rate": 4.138566337053264e-05,
-      "loss": 64.0225,
-      "step": 86310
-    },
-    {
-      "epoch": 0.3487437226533935,
-      "grad_norm": 1009.7755737304688,
-      "learning_rate": 4.1383026859796905e-05,
-      "loss": 71.5133,
-      "step": 86320
-    },
-    {
-      "epoch": 0.3487841239187611,
-      "grad_norm": 260.2582702636719,
-      "learning_rate": 4.138039002966563e-05,
-      "loss": 52.0272,
-      "step": 86330
-    },
-    {
-      "epoch": 0.34882452518412876,
-      "grad_norm": 847.3204956054688,
-      "learning_rate": 4.137775288019021e-05,
-      "loss": 59.5128,
-      "step": 86340
-    },
-    {
-      "epoch": 0.3488649264494964,
-      "grad_norm": 414.3377380371094,
-      "learning_rate": 4.137511541142207e-05,
-      "loss": 55.5041,
-      "step": 86350
-    },
-    {
-      "epoch": 0.34890532771486404,
-      "grad_norm": 2023.2427978515625,
-      "learning_rate": 4.137247762341262e-05,
-      "loss": 67.8408,
-      "step": 86360
-    },
-    {
-      "epoch": 0.3489457289802317,
-      "grad_norm": 1327.11767578125,
-      "learning_rate": 4.136983951621329e-05,
-      "loss": 50.12,
-      "step": 86370
-    },
-    {
-      "epoch": 0.34898613024559927,
-      "grad_norm": 608.749267578125,
-      "learning_rate": 4.136720108987552e-05,
-      "loss": 61.7877,
-      "step": 86380
-    },
-    {
-      "epoch": 0.3490265315109669,
-      "grad_norm": 864.80322265625,
-      "learning_rate": 4.136456234445073e-05,
-      "loss": 72.582,
-      "step": 86390
-    },
-    {
-      "epoch": 0.34906693277633455,
-      "grad_norm": 749.7403564453125,
-      "learning_rate": 4.136192327999037e-05,
-      "loss": 68.246,
-      "step": 86400
-    },
-    {
-      "epoch": 0.3491073340417022,
-      "grad_norm": 1620.177734375,
-      "learning_rate": 4.1359283896545895e-05,
-      "loss": 94.8168,
-      "step": 86410
-    },
-    {
-      "epoch": 0.34914773530706983,
-      "grad_norm": 2137.778564453125,
-      "learning_rate": 4.135664419416877e-05,
-      "loss": 54.6499,
-      "step": 86420
-    },
-    {
-      "epoch": 0.34918813657243747,
-      "grad_norm": 355.5881042480469,
-      "learning_rate": 4.1354004172910434e-05,
-      "loss": 71.6114,
-      "step": 86430
-    },
-    {
-      "epoch": 0.34922853783780505,
-      "grad_norm": 579.06005859375,
-      "learning_rate": 4.135136383282237e-05,
-      "loss": 83.3694,
-      "step": 86440
-    },
-    {
-      "epoch": 0.3492689391031727,
-      "grad_norm": 363.8780822753906,
-      "learning_rate": 4.134872317395604e-05,
-      "loss": 71.6161,
-      "step": 86450
-    },
-    {
-      "epoch": 0.34930934036854033,
-      "grad_norm": 734.9833984375,
-      "learning_rate": 4.134608219636294e-05,
-      "loss": 63.2569,
-      "step": 86460
-    },
-    {
-      "epoch": 0.349349741633908,
-      "grad_norm": 659.5712890625,
-      "learning_rate": 4.134344090009455e-05,
-      "loss": 59.2688,
-      "step": 86470
-    },
-    {
-      "epoch": 0.3493901428992756,
-      "grad_norm": 658.69677734375,
-      "learning_rate": 4.1340799285202376e-05,
-      "loss": 49.0295,
-      "step": 86480
-    },
-    {
-      "epoch": 0.34943054416464325,
-      "grad_norm": 687.8778076171875,
-      "learning_rate": 4.13381573517379e-05,
-      "loss": 84.1034,
-      "step": 86490
-    },
-    {
-      "epoch": 0.3494709454300109,
-      "grad_norm": 576.1513061523438,
-      "learning_rate": 4.133551509975264e-05,
-      "loss": 46.6568,
-      "step": 86500
-    },
-    {
-      "epoch": 0.3495113466953785,
-      "grad_norm": 435.50640869140625,
-      "learning_rate": 4.13328725292981e-05,
-      "loss": 51.9577,
-      "step": 86510
-    },
-    {
-      "epoch": 0.3495517479607461,
-      "grad_norm": 538.1187744140625,
-      "learning_rate": 4.13302296404258e-05,
-      "loss": 159.9702,
-      "step": 86520
-    },
-    {
-      "epoch": 0.34959214922611376,
-      "grad_norm": 989.9334716796875,
-      "learning_rate": 4.132758643318726e-05,
-      "loss": 90.6167,
-      "step": 86530
-    },
-    {
-      "epoch": 0.3496325504914814,
-      "grad_norm": 538.1937255859375,
-      "learning_rate": 4.132494290763403e-05,
-      "loss": 52.8154,
-      "step": 86540
-    },
-    {
-      "epoch": 0.34967295175684904,
-      "grad_norm": 2396.126708984375,
-      "learning_rate": 4.1322299063817624e-05,
-      "loss": 60.857,
-      "step": 86550
-    },
-    {
-      "epoch": 0.3497133530222167,
-      "grad_norm": 909.927978515625,
-      "learning_rate": 4.131965490178959e-05,
-      "loss": 127.2578,
-      "step": 86560
-    },
-    {
-      "epoch": 0.34975375428758426,
-      "grad_norm": 535.4166259765625,
-      "learning_rate": 4.131701042160149e-05,
-      "loss": 77.5207,
-      "step": 86570
-    },
-    {
-      "epoch": 0.3497941555529519,
-      "grad_norm": 947.1220703125,
-      "learning_rate": 4.131436562330487e-05,
-      "loss": 85.3897,
-      "step": 86580
-    },
-    {
-      "epoch": 0.34983455681831954,
-      "grad_norm": 1153.6351318359375,
-      "learning_rate": 4.13117205069513e-05,
-      "loss": 85.3414,
-      "step": 86590
-    },
-    {
-      "epoch": 0.3498749580836872,
-      "grad_norm": 1040.203857421875,
-      "learning_rate": 4.130907507259233e-05,
-      "loss": 79.24,
-      "step": 86600
-    },
-    {
-      "epoch": 0.3499153593490548,
-      "grad_norm": 737.9971313476562,
-      "learning_rate": 4.130642932027955e-05,
-      "loss": 47.7125,
-      "step": 86610
-    },
-    {
-      "epoch": 0.34995576061442246,
-      "grad_norm": 1519.4359130859375,
-      "learning_rate": 4.130378325006453e-05,
-      "loss": 59.9955,
-      "step": 86620
-    },
-    {
-      "epoch": 0.3499961618797901,
-      "grad_norm": 734.98828125,
-      "learning_rate": 4.130113686199887e-05,
-      "loss": 44.1955,
-      "step": 86630
-    },
-    {
-      "epoch": 0.3500365631451577,
-      "grad_norm": 1075.6939697265625,
-      "learning_rate": 4.129849015613415e-05,
-      "loss": 73.9399,
-      "step": 86640
-    },
-    {
-      "epoch": 0.3500769644105253,
-      "grad_norm": 3161.4248046875,
-      "learning_rate": 4.1295843132521973e-05,
-      "loss": 81.6022,
-      "step": 86650
-    },
-    {
-      "epoch": 0.35011736567589297,
-      "grad_norm": 1227.13134765625,
-      "learning_rate": 4.129319579121394e-05,
-      "loss": 67.1966,
-      "step": 86660
-    },
-    {
-      "epoch": 0.3501577669412606,
-      "grad_norm": 779.573974609375,
-      "learning_rate": 4.129054813226167e-05,
-      "loss": 72.8276,
-      "step": 86670
-    },
-    {
-      "epoch": 0.35019816820662825,
-      "grad_norm": 378.6111755371094,
-      "learning_rate": 4.1287900155716784e-05,
-      "loss": 43.1125,
-      "step": 86680
-    },
-    {
-      "epoch": 0.3502385694719959,
-      "grad_norm": 865.598876953125,
-      "learning_rate": 4.128525186163089e-05,
-      "loss": 75.8998,
-      "step": 86690
-    },
-    {
-      "epoch": 0.35027897073736347,
-      "grad_norm": 1067.35693359375,
-      "learning_rate": 4.128260325005564e-05,
-      "loss": 66.426,
-      "step": 86700
-    },
-    {
-      "epoch": 0.3503193720027311,
-      "grad_norm": 1469.167236328125,
-      "learning_rate": 4.127995432104264e-05,
-      "loss": 95.7886,
-      "step": 86710
-    },
-    {
-      "epoch": 0.35035977326809875,
-      "grad_norm": 554.118408203125,
-      "learning_rate": 4.127730507464356e-05,
-      "loss": 83.7477,
-      "step": 86720
-    },
-    {
-      "epoch": 0.3504001745334664,
-      "grad_norm": 1134.07470703125,
-      "learning_rate": 4.127465551091003e-05,
-      "loss": 123.3934,
-      "step": 86730
-    },
-    {
-      "epoch": 0.35044057579883403,
-      "grad_norm": 1663.503173828125,
-      "learning_rate": 4.1272005629893714e-05,
-      "loss": 51.4834,
-      "step": 86740
-    },
-    {
-      "epoch": 0.35048097706420167,
-      "grad_norm": 464.7097473144531,
-      "learning_rate": 4.1269355431646274e-05,
-      "loss": 47.998,
-      "step": 86750
-    },
-    {
-      "epoch": 0.35052137832956926,
-      "grad_norm": 533.934814453125,
-      "learning_rate": 4.126670491621938e-05,
-      "loss": 89.6148,
-      "step": 86760
-    },
-    {
-      "epoch": 0.3505617795949369,
-      "grad_norm": 1182.53466796875,
-      "learning_rate": 4.126405408366468e-05,
-      "loss": 58.9152,
-      "step": 86770
-    },
-    {
-      "epoch": 0.35060218086030454,
-      "grad_norm": 1748.3175048828125,
-      "learning_rate": 4.1261402934033886e-05,
-      "loss": 76.4899,
-      "step": 86780
-    },
-    {
-      "epoch": 0.3506425821256722,
-      "grad_norm": 503.2514343261719,
-      "learning_rate": 4.125875146737868e-05,
-      "loss": 94.3811,
-      "step": 86790
-    },
-    {
-      "epoch": 0.3506829833910398,
-      "grad_norm": 602.4920043945312,
-      "learning_rate": 4.125609968375072e-05,
-      "loss": 41.265,
-      "step": 86800
-    },
-    {
-      "epoch": 0.35072338465640746,
-      "grad_norm": 1253.5115966796875,
-      "learning_rate": 4.125344758320174e-05,
-      "loss": 56.0734,
-      "step": 86810
-    },
-    {
-      "epoch": 0.3507637859217751,
-      "grad_norm": 1205.50634765625,
-      "learning_rate": 4.125079516578344e-05,
-      "loss": 90.1057,
-      "step": 86820
-    },
-    {
-      "epoch": 0.3508041871871427,
-      "grad_norm": 422.2942810058594,
-      "learning_rate": 4.12481424315475e-05,
-      "loss": 60.9675,
-      "step": 86830
-    },
-    {
-      "epoch": 0.3508445884525103,
-      "grad_norm": 702.0094604492188,
-      "learning_rate": 4.124548938054568e-05,
-      "loss": 53.5536,
-      "step": 86840
-    },
-    {
-      "epoch": 0.35088498971787796,
-      "grad_norm": 604.8543090820312,
-      "learning_rate": 4.1242836012829665e-05,
-      "loss": 64.9087,
-      "step": 86850
-    },
-    {
-      "epoch": 0.3509253909832456,
-      "grad_norm": 307.5919494628906,
-      "learning_rate": 4.1240182328451204e-05,
-      "loss": 72.9601,
-      "step": 86860
-    },
-    {
-      "epoch": 0.35096579224861324,
-      "grad_norm": 1943.6920166015625,
-      "learning_rate": 4.123752832746203e-05,
-      "loss": 87.0177,
-      "step": 86870
-    },
-    {
-      "epoch": 0.3510061935139809,
-      "grad_norm": 1072.779296875,
-      "learning_rate": 4.123487400991388e-05,
-      "loss": 73.2874,
-      "step": 86880
-    },
-    {
-      "epoch": 0.35104659477934846,
-      "grad_norm": 644.7705078125,
-      "learning_rate": 4.1232219375858504e-05,
-      "loss": 82.4604,
-      "step": 86890
-    },
-    {
-      "epoch": 0.3510869960447161,
-      "grad_norm": 214.5071563720703,
-      "learning_rate": 4.1229564425347654e-05,
-      "loss": 50.0135,
-      "step": 86900
-    },
-    {
-      "epoch": 0.35112739731008374,
-      "grad_norm": 662.0516357421875,
-      "learning_rate": 4.122690915843309e-05,
-      "loss": 45.6385,
-      "step": 86910
-    },
-    {
-      "epoch": 0.3511677985754514,
-      "grad_norm": 1001.0264282226562,
-      "learning_rate": 4.122425357516658e-05,
-      "loss": 63.8673,
-      "step": 86920
-    },
-    {
-      "epoch": 0.351208199840819,
-      "grad_norm": 1112.534912109375,
-      "learning_rate": 4.1221597675599886e-05,
-      "loss": 54.2994,
-      "step": 86930
-    },
-    {
-      "epoch": 0.35124860110618666,
-      "grad_norm": 565.2496948242188,
-      "learning_rate": 4.1218941459784796e-05,
-      "loss": 50.4045,
-      "step": 86940
-    },
-    {
-      "epoch": 0.3512890023715543,
-      "grad_norm": 823.8052978515625,
-      "learning_rate": 4.121628492777311e-05,
-      "loss": 66.9651,
-      "step": 86950
-    },
-    {
-      "epoch": 0.3513294036369219,
-      "grad_norm": 875.3696899414062,
-      "learning_rate": 4.121362807961658e-05,
-      "loss": 84.8828,
-      "step": 86960
-    },
-    {
-      "epoch": 0.35136980490228953,
-      "grad_norm": 412.0173034667969,
-      "learning_rate": 4.1210970915367026e-05,
-      "loss": 78.3683,
-      "step": 86970
-    },
-    {
-      "epoch": 0.35141020616765717,
-      "grad_norm": 1158.7080078125,
-      "learning_rate": 4.120831343507625e-05,
-      "loss": 77.7848,
-      "step": 86980
-    },
-    {
-      "epoch": 0.3514506074330248,
-      "grad_norm": 261.9429016113281,
-      "learning_rate": 4.1205655638796065e-05,
-      "loss": 43.2981,
-      "step": 86990
-    },
-    {
-      "epoch": 0.35149100869839245,
-      "grad_norm": 779.2062377929688,
-      "learning_rate": 4.1202997526578276e-05,
-      "loss": 57.5505,
-      "step": 87000
-    },
-    {
-      "epoch": 0.3515314099637601,
-      "grad_norm": 742.50048828125,
-      "learning_rate": 4.120033909847471e-05,
-      "loss": 68.5733,
-      "step": 87010
-    },
-    {
-      "epoch": 0.3515718112291277,
-      "grad_norm": 610.2131958007812,
-      "learning_rate": 4.1197680354537186e-05,
-      "loss": 64.115,
-      "step": 87020
-    },
-    {
-      "epoch": 0.3516122124944953,
-      "grad_norm": 994.4957275390625,
-      "learning_rate": 4.119502129481755e-05,
-      "loss": 42.9862,
-      "step": 87030
-    },
-    {
-      "epoch": 0.35165261375986295,
-      "grad_norm": 1622.78076171875,
-      "learning_rate": 4.119236191936764e-05,
-      "loss": 55.3836,
-      "step": 87040
-    },
-    {
-      "epoch": 0.3516930150252306,
-      "grad_norm": 1582.70458984375,
-      "learning_rate": 4.118970222823929e-05,
-      "loss": 71.9621,
-      "step": 87050
-    },
-    {
-      "epoch": 0.35173341629059823,
-      "grad_norm": 1007.8423461914062,
-      "learning_rate": 4.118704222148436e-05,
-      "loss": 66.6885,
-      "step": 87060
-    },
-    {
-      "epoch": 0.3517738175559659,
-      "grad_norm": 431.42828369140625,
-      "learning_rate": 4.118438189915471e-05,
-      "loss": 83.4902,
-      "step": 87070
-    },
-    {
-      "epoch": 0.35181421882133346,
-      "grad_norm": 1616.8265380859375,
-      "learning_rate": 4.118172126130221e-05,
-      "loss": 49.96,
-      "step": 87080
-    },
-    {
-      "epoch": 0.3518546200867011,
-      "grad_norm": 358.4612121582031,
-      "learning_rate": 4.117906030797871e-05,
-      "loss": 85.6933,
-      "step": 87090
-    },
-    {
-      "epoch": 0.35189502135206874,
-      "grad_norm": 595.7027587890625,
-      "learning_rate": 4.1176399039236116e-05,
-      "loss": 67.9241,
-      "step": 87100
-    },
-    {
-      "epoch": 0.3519354226174364,
-      "grad_norm": 826.05517578125,
-      "learning_rate": 4.117373745512628e-05,
-      "loss": 92.1798,
-      "step": 87110
-    },
-    {
-      "epoch": 0.351975823882804,
-      "grad_norm": 2575.160888671875,
-      "learning_rate": 4.117107555570111e-05,
-      "loss": 80.2855,
-      "step": 87120
-    },
-    {
-      "epoch": 0.35201622514817166,
-      "grad_norm": 635.9917602539062,
-      "learning_rate": 4.1168413341012496e-05,
-      "loss": 51.3892,
-      "step": 87130
-    },
-    {
-      "epoch": 0.3520566264135393,
-      "grad_norm": 586.3646240234375,
-      "learning_rate": 4.116575081111235e-05,
-      "loss": 74.7427,
-      "step": 87140
-    },
-    {
-      "epoch": 0.3520970276789069,
-      "grad_norm": 379.6501770019531,
-      "learning_rate": 4.116308796605255e-05,
-      "loss": 42.7718,
-      "step": 87150
-    },
-    {
-      "epoch": 0.3521374289442745,
-      "grad_norm": 757.53076171875,
-      "learning_rate": 4.116042480588505e-05,
-      "loss": 79.6147,
-      "step": 87160
-    },
-    {
-      "epoch": 0.35217783020964216,
-      "grad_norm": 521.4224853515625,
-      "learning_rate": 4.1157761330661734e-05,
-      "loss": 114.8519,
-      "step": 87170
-    },
-    {
-      "epoch": 0.3522182314750098,
-      "grad_norm": 543.6570434570312,
-      "learning_rate": 4.115509754043454e-05,
-      "loss": 67.7495,
-      "step": 87180
-    },
-    {
-      "epoch": 0.35225863274037744,
-      "grad_norm": 635.4970703125,
-      "learning_rate": 4.115243343525541e-05,
-      "loss": 78.6889,
-      "step": 87190
-    },
-    {
-      "epoch": 0.3522990340057451,
-      "grad_norm": 1206.6654052734375,
-      "learning_rate": 4.1149769015176275e-05,
-      "loss": 56.5697,
-      "step": 87200
-    },
-    {
-      "epoch": 0.35233943527111267,
-      "grad_norm": 811.91455078125,
-      "learning_rate": 4.114710428024907e-05,
-      "loss": 78.0935,
-      "step": 87210
-    },
-    {
-      "epoch": 0.3523798365364803,
-      "grad_norm": 1116.96240234375,
-      "learning_rate": 4.114443923052577e-05,
-      "loss": 59.2988,
-      "step": 87220
-    },
-    {
-      "epoch": 0.35242023780184795,
-      "grad_norm": 623.7386474609375,
-      "learning_rate": 4.11417738660583e-05,
-      "loss": 56.9475,
-      "step": 87230
-    },
-    {
-      "epoch": 0.3524606390672156,
-      "grad_norm": 1740.817138671875,
-      "learning_rate": 4.113910818689864e-05,
-      "loss": 65.6938,
-      "step": 87240
-    },
-    {
-      "epoch": 0.3525010403325832,
-      "grad_norm": 752.4931640625,
-      "learning_rate": 4.113644219309877e-05,
-      "loss": 83.3021,
-      "step": 87250
-    },
-    {
-      "epoch": 0.35254144159795087,
-      "grad_norm": 1242.556640625,
-      "learning_rate": 4.1133775884710634e-05,
-      "loss": 59.4292,
-      "step": 87260
-    },
-    {
-      "epoch": 0.3525818428633185,
-      "grad_norm": 1105.26318359375,
-      "learning_rate": 4.113110926178625e-05,
-      "loss": 52.1607,
-      "step": 87270
-    },
-    {
-      "epoch": 0.3526222441286861,
-      "grad_norm": 941.704345703125,
-      "learning_rate": 4.112844232437757e-05,
-      "loss": 44.7782,
-      "step": 87280
-    },
-    {
-      "epoch": 0.35266264539405373,
-      "grad_norm": 975.8741455078125,
-      "learning_rate": 4.112577507253661e-05,
-      "loss": 63.2205,
-      "step": 87290
-    },
-    {
-      "epoch": 0.35270304665942137,
-      "grad_norm": 267.49029541015625,
-      "learning_rate": 4.112310750631536e-05,
-      "loss": 55.718,
-      "step": 87300
-    },
-    {
-      "epoch": 0.352743447924789,
-      "grad_norm": 492.7359619140625,
-      "learning_rate": 4.112043962576583e-05,
-      "loss": 87.6296,
-      "step": 87310
-    },
-    {
-      "epoch": 0.35278384919015665,
-      "grad_norm": 723.97412109375,
-      "learning_rate": 4.1117771430940035e-05,
-      "loss": 72.1795,
-      "step": 87320
-    },
-    {
-      "epoch": 0.3528242504555243,
-      "grad_norm": 1166.4947509765625,
-      "learning_rate": 4.111510292188998e-05,
-      "loss": 61.6153,
-      "step": 87330
-    },
-    {
-      "epoch": 0.3528646517208919,
-      "grad_norm": 943.3729248046875,
-      "learning_rate": 4.111243409866769e-05,
-      "loss": 78.3224,
-      "step": 87340
-    },
-    {
-      "epoch": 0.3529050529862595,
-      "grad_norm": 1874.787353515625,
-      "learning_rate": 4.110976496132522e-05,
-      "loss": 43.5677,
-      "step": 87350
-    },
-    {
-      "epoch": 0.35294545425162716,
-      "grad_norm": 567.2123413085938,
-      "learning_rate": 4.1107095509914584e-05,
-      "loss": 114.9888,
-      "step": 87360
-    },
-    {
-      "epoch": 0.3529858555169948,
-      "grad_norm": 1504.210205078125,
-      "learning_rate": 4.1104425744487826e-05,
-      "loss": 63.7791,
-      "step": 87370
-    },
-    {
-      "epoch": 0.35302625678236244,
-      "grad_norm": 252.47015380859375,
-      "learning_rate": 4.1101755665096996e-05,
-      "loss": 46.304,
-      "step": 87380
-    },
-    {
-      "epoch": 0.3530666580477301,
-      "grad_norm": 952.9064331054688,
-      "learning_rate": 4.109908527179415e-05,
-      "loss": 68.7049,
-      "step": 87390
-    },
-    {
-      "epoch": 0.35310705931309766,
-      "grad_norm": 3070.967041015625,
-      "learning_rate": 4.109641456463135e-05,
-      "loss": 70.3882,
-      "step": 87400
-    },
-    {
-      "epoch": 0.3531474605784653,
-      "grad_norm": 1895.307861328125,
-      "learning_rate": 4.109374354366066e-05,
-      "loss": 44.8805,
-      "step": 87410
-    },
-    {
-      "epoch": 0.35318786184383294,
-      "grad_norm": 1331.394287109375,
-      "learning_rate": 4.109107220893415e-05,
-      "loss": 59.9866,
-      "step": 87420
-    },
-    {
-      "epoch": 0.3532282631092006,
-      "grad_norm": 249.96646118164062,
-      "learning_rate": 4.1088400560503905e-05,
-      "loss": 66.2172,
-      "step": 87430
-    },
-    {
-      "epoch": 0.3532686643745682,
-      "grad_norm": 947.6691284179688,
-      "learning_rate": 4.108572859842201e-05,
-      "loss": 90.3438,
-      "step": 87440
-    },
-    {
-      "epoch": 0.35330906563993586,
-      "grad_norm": 573.4298706054688,
-      "learning_rate": 4.108305632274055e-05,
-      "loss": 34.3865,
-      "step": 87450
-    },
-    {
-      "epoch": 0.3533494669053035,
-      "grad_norm": 602.94384765625,
-      "learning_rate": 4.108038373351163e-05,
-      "loss": 51.0906,
-      "step": 87460
-    },
-    {
-      "epoch": 0.3533898681706711,
-      "grad_norm": 3697.798095703125,
-      "learning_rate": 4.107771083078735e-05,
-      "loss": 70.8239,
-      "step": 87470
-    },
-    {
-      "epoch": 0.3534302694360387,
-      "grad_norm": 549.9475708007812,
-      "learning_rate": 4.107503761461983e-05,
-      "loss": 48.9701,
-      "step": 87480
-    },
-    {
-      "epoch": 0.35347067070140636,
-      "grad_norm": 651.7734985351562,
-      "learning_rate": 4.107236408506116e-05,
-      "loss": 87.1631,
-      "step": 87490
-    },
-    {
-      "epoch": 0.353511071966774,
-      "grad_norm": 854.0162963867188,
-      "learning_rate": 4.1069690242163484e-05,
-      "loss": 39.3587,
-      "step": 87500
-    },
-    {
-      "epoch": 0.35355147323214164,
-      "grad_norm": 845.365234375,
-      "learning_rate": 4.106701608597893e-05,
-      "loss": 80.0646,
-      "step": 87510
-    },
-    {
-      "epoch": 0.3535918744975093,
-      "grad_norm": 571.5369262695312,
-      "learning_rate": 4.106434161655962e-05,
-      "loss": 54.8925,
-      "step": 87520
-    },
-    {
-      "epoch": 0.35363227576287687,
-      "grad_norm": 514.8745727539062,
-      "learning_rate": 4.106166683395769e-05,
-      "loss": 80.812,
-      "step": 87530
-    },
-    {
-      "epoch": 0.3536726770282445,
-      "grad_norm": 895.7685546875,
-      "learning_rate": 4.105899173822531e-05,
-      "loss": 52.7344,
-      "step": 87540
-    },
-    {
-      "epoch": 0.35371307829361215,
-      "grad_norm": 408.03485107421875,
-      "learning_rate": 4.1056316329414616e-05,
-      "loss": 63.1606,
-      "step": 87550
-    },
-    {
-      "epoch": 0.3537534795589798,
-      "grad_norm": 794.3991088867188,
-      "learning_rate": 4.105364060757776e-05,
-      "loss": 81.9505,
-      "step": 87560
-    },
-    {
-      "epoch": 0.35379388082434743,
-      "grad_norm": 308.87799072265625,
-      "learning_rate": 4.1050964572766923e-05,
-      "loss": 71.2676,
-      "step": 87570
-    },
-    {
-      "epoch": 0.35383428208971507,
-      "grad_norm": 743.8997192382812,
-      "learning_rate": 4.104828822503427e-05,
-      "loss": 54.9144,
-      "step": 87580
-    },
-    {
-      "epoch": 0.3538746833550827,
-      "grad_norm": 800.4918212890625,
-      "learning_rate": 4.104561156443197e-05,
-      "loss": 48.0048,
-      "step": 87590
-    },
-    {
-      "epoch": 0.3539150846204503,
-      "grad_norm": 217.0595245361328,
-      "learning_rate": 4.104293459101222e-05,
-      "loss": 71.4233,
-      "step": 87600
-    },
-    {
-      "epoch": 0.35395548588581793,
-      "grad_norm": 778.3648071289062,
-      "learning_rate": 4.104025730482719e-05,
-      "loss": 66.6853,
-      "step": 87610
-    },
-    {
-      "epoch": 0.3539958871511856,
-      "grad_norm": 712.687255859375,
-      "learning_rate": 4.103757970592909e-05,
-      "loss": 59.8492,
-      "step": 87620
-    },
-    {
-      "epoch": 0.3540362884165532,
-      "grad_norm": 366.0815734863281,
-      "learning_rate": 4.1034901794370116e-05,
-      "loss": 49.8096,
-      "step": 87630
-    },
-    {
-      "epoch": 0.35407668968192085,
-      "grad_norm": 1185.479736328125,
-      "learning_rate": 4.1032223570202474e-05,
-      "loss": 83.2932,
-      "step": 87640
-    },
-    {
-      "epoch": 0.3541170909472885,
-      "grad_norm": 411.51373291015625,
-      "learning_rate": 4.102954503347839e-05,
-      "loss": 45.1826,
-      "step": 87650
-    },
-    {
-      "epoch": 0.3541574922126561,
-      "grad_norm": 749.7902221679688,
-      "learning_rate": 4.102686618425006e-05,
-      "loss": 56.4452,
-      "step": 87660
-    },
-    {
-      "epoch": 0.3541978934780237,
-      "grad_norm": 573.7445068359375,
-      "learning_rate": 4.102418702256973e-05,
-      "loss": 59.8702,
-      "step": 87670
-    },
-    {
-      "epoch": 0.35423829474339136,
-      "grad_norm": 1168.4649658203125,
-      "learning_rate": 4.1021507548489625e-05,
-      "loss": 66.0522,
-      "step": 87680
-    },
-    {
-      "epoch": 0.354278696008759,
-      "grad_norm": 949.6201171875,
-      "learning_rate": 4.1018827762061985e-05,
-      "loss": 82.0665,
-      "step": 87690
-    },
-    {
-      "epoch": 0.35431909727412664,
-      "grad_norm": 1565.3529052734375,
-      "learning_rate": 4.101614766333904e-05,
-      "loss": 102.3075,
-      "step": 87700
-    },
-    {
-      "epoch": 0.3543594985394943,
-      "grad_norm": 1542.8006591796875,
-      "learning_rate": 4.101346725237305e-05,
-      "loss": 66.2509,
-      "step": 87710
-    },
-    {
-      "epoch": 0.35439989980486186,
-      "grad_norm": 527.8439331054688,
-      "learning_rate": 4.1010786529216284e-05,
-      "loss": 71.4895,
-      "step": 87720
-    },
-    {
-      "epoch": 0.3544403010702295,
-      "grad_norm": 1168.476318359375,
-      "learning_rate": 4.100810549392099e-05,
-      "loss": 85.4038,
-      "step": 87730
-    },
-    {
-      "epoch": 0.35448070233559714,
-      "grad_norm": 714.1607666015625,
-      "learning_rate": 4.100542414653943e-05,
-      "loss": 51.9524,
-      "step": 87740
-    },
-    {
-      "epoch": 0.3545211036009648,
-      "grad_norm": 2820.892578125,
-      "learning_rate": 4.100274248712389e-05,
-      "loss": 84.763,
-      "step": 87750
-    },
-    {
-      "epoch": 0.3545615048663324,
-      "grad_norm": 1056.6544189453125,
-      "learning_rate": 4.1000060515726647e-05,
-      "loss": 46.4493,
-      "step": 87760
-    },
-    {
-      "epoch": 0.35460190613170006,
-      "grad_norm": 637.3196411132812,
-      "learning_rate": 4.0997378232399984e-05,
-      "loss": 64.2221,
-      "step": 87770
-    },
-    {
-      "epoch": 0.3546423073970677,
-      "grad_norm": 1249.30078125,
-      "learning_rate": 4.09946956371962e-05,
-      "loss": 62.3628,
-      "step": 87780
-    },
-    {
-      "epoch": 0.3546827086624353,
-      "grad_norm": 562.5096435546875,
-      "learning_rate": 4.0992012730167584e-05,
-      "loss": 56.4057,
-      "step": 87790
-    },
-    {
-      "epoch": 0.3547231099278029,
-      "grad_norm": 515.3623046875,
-      "learning_rate": 4.098932951136645e-05,
-      "loss": 75.1756,
-      "step": 87800
-    },
-    {
-      "epoch": 0.35476351119317057,
-      "grad_norm": 1004.6653442382812,
-      "learning_rate": 4.098664598084511e-05,
-      "loss": 87.7658,
-      "step": 87810
-    },
-    {
-      "epoch": 0.3548039124585382,
-      "grad_norm": 514.6812133789062,
-      "learning_rate": 4.0983962138655873e-05,
-      "loss": 63.0205,
-      "step": 87820
-    },
-    {
-      "epoch": 0.35484431372390585,
-      "grad_norm": 327.2781677246094,
-      "learning_rate": 4.0981277984851066e-05,
-      "loss": 64.0254,
-      "step": 87830
-    },
-    {
-      "epoch": 0.3548847149892735,
-      "grad_norm": 945.9813842773438,
-      "learning_rate": 4.097859351948301e-05,
-      "loss": 71.4724,
-      "step": 87840
-    },
-    {
-      "epoch": 0.35492511625464107,
-      "grad_norm": 1146.121337890625,
-      "learning_rate": 4.0975908742604055e-05,
-      "loss": 74.1327,
-      "step": 87850
-    },
-    {
-      "epoch": 0.3549655175200087,
-      "grad_norm": 2756.161376953125,
-      "learning_rate": 4.097322365426653e-05,
-      "loss": 60.8623,
-      "step": 87860
-    },
-    {
-      "epoch": 0.35500591878537635,
-      "grad_norm": 616.4971923828125,
-      "learning_rate": 4.097053825452278e-05,
-      "loss": 74.3813,
-      "step": 87870
-    },
-    {
-      "epoch": 0.355046320050744,
-      "grad_norm": 796.4453735351562,
-      "learning_rate": 4.0967852543425175e-05,
-      "loss": 65.7487,
-      "step": 87880
-    },
-    {
-      "epoch": 0.35508672131611163,
-      "grad_norm": 377.42138671875,
-      "learning_rate": 4.0965166521026065e-05,
-      "loss": 46.341,
-      "step": 87890
-    },
-    {
-      "epoch": 0.35512712258147927,
-      "grad_norm": 605.7762451171875,
-      "learning_rate": 4.096248018737781e-05,
-      "loss": 81.4626,
-      "step": 87900
-    },
-    {
-      "epoch": 0.3551675238468469,
-      "grad_norm": 408.6437072753906,
-      "learning_rate": 4.095979354253279e-05,
-      "loss": 85.3279,
-      "step": 87910
-    },
-    {
-      "epoch": 0.3552079251122145,
-      "grad_norm": 673.0602416992188,
-      "learning_rate": 4.095710658654337e-05,
-      "loss": 35.6571,
-      "step": 87920
-    },
-    {
-      "epoch": 0.35524832637758214,
-      "grad_norm": 562.7644653320312,
-      "learning_rate": 4.0954419319461946e-05,
-      "loss": 68.498,
-      "step": 87930
-    },
-    {
-      "epoch": 0.3552887276429498,
-      "grad_norm": 1345.530029296875,
-      "learning_rate": 4.09517317413409e-05,
-      "loss": 54.9481,
-      "step": 87940
-    },
-    {
-      "epoch": 0.3553291289083174,
-      "grad_norm": 670.7591552734375,
-      "learning_rate": 4.094904385223264e-05,
-      "loss": 95.4115,
-      "step": 87950
-    },
-    {
-      "epoch": 0.35536953017368506,
-      "grad_norm": 6252.8359375,
-      "learning_rate": 4.094635565218955e-05,
-      "loss": 85.2645,
-      "step": 87960
-    },
-    {
-      "epoch": 0.3554099314390527,
-      "grad_norm": 699.56982421875,
-      "learning_rate": 4.094366714126405e-05,
-      "loss": 61.0998,
-      "step": 87970
-    },
-    {
-      "epoch": 0.3554503327044203,
-      "grad_norm": 572.2498779296875,
-      "learning_rate": 4.094097831950855e-05,
-      "loss": 55.9745,
-      "step": 87980
-    },
-    {
-      "epoch": 0.3554907339697879,
-      "grad_norm": 477.888671875,
-      "learning_rate": 4.093828918697547e-05,
-      "loss": 64.4699,
-      "step": 87990
-    },
-    {
-      "epoch": 0.35553113523515556,
-      "grad_norm": 1221.31982421875,
-      "learning_rate": 4.093559974371725e-05,
-      "loss": 60.6648,
-      "step": 88000
-    },
-    {
-      "epoch": 0.3555715365005232,
-      "grad_norm": 390.103759765625,
-      "learning_rate": 4.09329099897863e-05,
-      "loss": 51.6583,
-      "step": 88010
-    },
-    {
-      "epoch": 0.35561193776589084,
-      "grad_norm": 1909.0634765625,
-      "learning_rate": 4.0930219925235056e-05,
-      "loss": 62.6641,
-      "step": 88020
-    },
-    {
-      "epoch": 0.3556523390312585,
-      "grad_norm": 0.0,
-      "learning_rate": 4.0927529550115986e-05,
-      "loss": 54.3005,
-      "step": 88030
-    },
-    {
-      "epoch": 0.35569274029662606,
-      "grad_norm": 1166.9232177734375,
-      "learning_rate": 4.0924838864481516e-05,
-      "loss": 77.7679,
-      "step": 88040
-    },
-    {
-      "epoch": 0.3557331415619937,
-      "grad_norm": 736.1539916992188,
-      "learning_rate": 4.092214786838413e-05,
-      "loss": 55.0295,
-      "step": 88050
-    },
-    {
-      "epoch": 0.35577354282736134,
-      "grad_norm": 957.0504150390625,
-      "learning_rate": 4.0919456561876256e-05,
-      "loss": 72.6961,
-      "step": 88060
-    },
-    {
-      "epoch": 0.355813944092729,
-      "grad_norm": 687.1393432617188,
-      "learning_rate": 4.091676494501039e-05,
-      "loss": 60.3476,
-      "step": 88070
-    },
-    {
-      "epoch": 0.3558543453580966,
-      "grad_norm": 1779.8883056640625,
-      "learning_rate": 4.0914073017838996e-05,
-      "loss": 85.0388,
-      "step": 88080
-    },
-    {
-      "epoch": 0.35589474662346426,
-      "grad_norm": 1815.7283935546875,
-      "learning_rate": 4.091138078041455e-05,
-      "loss": 67.7276,
-      "step": 88090
-    },
-    {
-      "epoch": 0.3559351478888319,
-      "grad_norm": 510.8993835449219,
-      "learning_rate": 4.090868823278956e-05,
-      "loss": 74.9138,
-      "step": 88100
-    },
-    {
-      "epoch": 0.3559755491541995,
-      "grad_norm": 1010.1596069335938,
-      "learning_rate": 4.090599537501649e-05,
-      "loss": 70.1757,
-      "step": 88110
-    },
-    {
-      "epoch": 0.35601595041956713,
-      "grad_norm": 806.6387329101562,
-      "learning_rate": 4.090330220714785e-05,
-      "loss": 64.2958,
-      "step": 88120
-    },
-    {
-      "epoch": 0.35605635168493477,
-      "grad_norm": 1274.768310546875,
-      "learning_rate": 4.090060872923615e-05,
-      "loss": 98.0021,
-      "step": 88130
-    },
-    {
-      "epoch": 0.3560967529503024,
-      "grad_norm": 819.7002563476562,
-      "learning_rate": 4.089791494133389e-05,
-      "loss": 52.4494,
-      "step": 88140
-    },
-    {
-      "epoch": 0.35613715421567005,
-      "grad_norm": 1136.931396484375,
-      "learning_rate": 4.0895220843493606e-05,
-      "loss": 105.6924,
-      "step": 88150
-    },
-    {
-      "epoch": 0.3561775554810377,
-      "grad_norm": 2234.421630859375,
-      "learning_rate": 4.0892526435767795e-05,
-      "loss": 103.7004,
-      "step": 88160
-    },
-    {
-      "epoch": 0.3562179567464053,
-      "grad_norm": 841.349853515625,
-      "learning_rate": 4.088983171820901e-05,
-      "loss": 52.9514,
-      "step": 88170
-    },
-    {
-      "epoch": 0.3562583580117729,
-      "grad_norm": 725.3030395507812,
-      "learning_rate": 4.088713669086977e-05,
-      "loss": 49.6578,
-      "step": 88180
-    },
-    {
-      "epoch": 0.35629875927714055,
-      "grad_norm": 1026.1910400390625,
-      "learning_rate": 4.088444135380262e-05,
-      "loss": 63.2285,
-      "step": 88190
-    },
-    {
-      "epoch": 0.3563391605425082,
-      "grad_norm": 1608.8538818359375,
-      "learning_rate": 4.088174570706011e-05,
-      "loss": 65.9221,
-      "step": 88200
-    },
-    {
-      "epoch": 0.35637956180787583,
-      "grad_norm": 608.7267456054688,
-      "learning_rate": 4.0879049750694795e-05,
-      "loss": 59.0359,
-      "step": 88210
-    },
-    {
-      "epoch": 0.3564199630732435,
-      "grad_norm": 3048.426513671875,
-      "learning_rate": 4.0876353484759224e-05,
-      "loss": 100.1788,
-      "step": 88220
-    },
-    {
-      "epoch": 0.35646036433861106,
-      "grad_norm": 540.722412109375,
-      "learning_rate": 4.087365690930597e-05,
-      "loss": 66.9113,
-      "step": 88230
-    },
-    {
-      "epoch": 0.3565007656039787,
-      "grad_norm": 781.4616088867188,
-      "learning_rate": 4.0870960024387596e-05,
-      "loss": 103.4365,
-      "step": 88240
-    },
-    {
-      "epoch": 0.35654116686934634,
-      "grad_norm": 1404.780029296875,
-      "learning_rate": 4.086826283005669e-05,
-      "loss": 58.6043,
-      "step": 88250
-    },
-    {
-      "epoch": 0.356581568134714,
-      "grad_norm": 440.78472900390625,
-      "learning_rate": 4.0865565326365835e-05,
-      "loss": 43.4884,
-      "step": 88260
-    },
-    {
-      "epoch": 0.3566219694000816,
-      "grad_norm": 425.8787841796875,
-      "learning_rate": 4.086286751336761e-05,
-      "loss": 44.2717,
-      "step": 88270
-    },
-    {
-      "epoch": 0.35666237066544926,
-      "grad_norm": 386.0218200683594,
-      "learning_rate": 4.0860169391114625e-05,
-      "loss": 52.1067,
-      "step": 88280
-    },
-    {
-      "epoch": 0.3567027719308169,
-      "grad_norm": 421.0408935546875,
-      "learning_rate": 4.085747095965946e-05,
-      "loss": 63.7093,
-      "step": 88290
-    },
-    {
-      "epoch": 0.3567431731961845,
-      "grad_norm": 688.8589477539062,
-      "learning_rate": 4.085477221905474e-05,
-      "loss": 63.6186,
-      "step": 88300
-    },
-    {
-      "epoch": 0.3567835744615521,
-      "grad_norm": 2102.650634765625,
-      "learning_rate": 4.085207316935308e-05,
-      "loss": 120.8342,
-      "step": 88310
-    },
-    {
-      "epoch": 0.35682397572691976,
-      "grad_norm": 644.7640380859375,
-      "learning_rate": 4.084937381060708e-05,
-      "loss": 71.6069,
-      "step": 88320
-    },
-    {
-      "epoch": 0.3568643769922874,
-      "grad_norm": 718.741455078125,
-      "learning_rate": 4.084667414286939e-05,
-      "loss": 49.9685,
-      "step": 88330
-    },
-    {
-      "epoch": 0.35690477825765504,
-      "grad_norm": 1353.2684326171875,
-      "learning_rate": 4.0843974166192614e-05,
-      "loss": 89.7978,
-      "step": 88340
-    },
-    {
-      "epoch": 0.3569451795230227,
-      "grad_norm": 518.8037109375,
-      "learning_rate": 4.0841273880629416e-05,
-      "loss": 68.2225,
-      "step": 88350
-    },
-    {
-      "epoch": 0.35698558078839027,
-      "grad_norm": 491.72265625,
-      "learning_rate": 4.083857328623243e-05,
-      "loss": 62.1392,
-      "step": 88360
-    },
-    {
-      "epoch": 0.3570259820537579,
-      "grad_norm": 713.7077026367188,
-      "learning_rate": 4.0835872383054296e-05,
-      "loss": 59.1912,
-      "step": 88370
-    },
-    {
-      "epoch": 0.35706638331912555,
-      "grad_norm": 546.5416259765625,
-      "learning_rate": 4.083317117114768e-05,
-      "loss": 65.1588,
-      "step": 88380
-    },
-    {
-      "epoch": 0.3571067845844932,
-      "grad_norm": 749.3577270507812,
-      "learning_rate": 4.083046965056524e-05,
-      "loss": 65.049,
-      "step": 88390
-    },
-    {
-      "epoch": 0.3571471858498608,
-      "grad_norm": 808.5428466796875,
-      "learning_rate": 4.082776782135964e-05,
-      "loss": 59.8705,
-      "step": 88400
-    },
-    {
-      "epoch": 0.35718758711522847,
-      "grad_norm": 372.74261474609375,
-      "learning_rate": 4.082506568358357e-05,
-      "loss": 62.2298,
-      "step": 88410
-    },
-    {
-      "epoch": 0.3572279883805961,
-      "grad_norm": 574.08447265625,
-      "learning_rate": 4.082236323728968e-05,
-      "loss": 56.4968,
-      "step": 88420
-    },
-    {
-      "epoch": 0.3572683896459637,
-      "grad_norm": 901.8555297851562,
-      "learning_rate": 4.0819660482530684e-05,
-      "loss": 67.4959,
-      "step": 88430
-    },
-    {
-      "epoch": 0.35730879091133133,
-      "grad_norm": 2012.2177734375,
-      "learning_rate": 4.0816957419359264e-05,
-      "loss": 60.3428,
-      "step": 88440
-    },
-    {
-      "epoch": 0.35734919217669897,
-      "grad_norm": 434.2742614746094,
-      "learning_rate": 4.0814254047828116e-05,
-      "loss": 55.9326,
-      "step": 88450
-    },
-    {
-      "epoch": 0.3573895934420666,
-      "grad_norm": 1581.9569091796875,
-      "learning_rate": 4.081155036798994e-05,
-      "loss": 80.8084,
-      "step": 88460
-    },
-    {
-      "epoch": 0.35742999470743425,
-      "grad_norm": 980.014404296875,
-      "learning_rate": 4.080884637989745e-05,
-      "loss": 53.9677,
-      "step": 88470
-    },
-    {
-      "epoch": 0.3574703959728019,
-      "grad_norm": 438.5036926269531,
-      "learning_rate": 4.080614208360336e-05,
-      "loss": 46.5922,
-      "step": 88480
-    },
-    {
-      "epoch": 0.3575107972381695,
-      "grad_norm": 1167.0660400390625,
-      "learning_rate": 4.080343747916039e-05,
-      "loss": 60.1899,
-      "step": 88490
-    },
-    {
-      "epoch": 0.3575511985035371,
-      "grad_norm": 1493.02734375,
-      "learning_rate": 4.080073256662127e-05,
-      "loss": 55.8727,
-      "step": 88500
-    },
-    {
-      "epoch": 0.35759159976890476,
-      "grad_norm": 643.5103759765625,
-      "learning_rate": 4.079802734603874e-05,
-      "loss": 75.9446,
-      "step": 88510
-    },
-    {
-      "epoch": 0.3576320010342724,
-      "grad_norm": 619.2804565429688,
-      "learning_rate": 4.079532181746553e-05,
-      "loss": 59.9599,
-      "step": 88520
-    },
-    {
-      "epoch": 0.35767240229964004,
-      "grad_norm": 699.5809936523438,
-      "learning_rate": 4.079261598095439e-05,
-      "loss": 75.2107,
-      "step": 88530
-    },
-    {
-      "epoch": 0.3577128035650077,
-      "grad_norm": 2605.017578125,
-      "learning_rate": 4.078990983655807e-05,
-      "loss": 78.6188,
-      "step": 88540
-    },
-    {
-      "epoch": 0.35775320483037526,
-      "grad_norm": 483.28961181640625,
-      "learning_rate": 4.078720338432933e-05,
-      "loss": 52.8732,
-      "step": 88550
-    },
-    {
-      "epoch": 0.3577936060957429,
-      "grad_norm": 371.9551086425781,
-      "learning_rate": 4.078449662432093e-05,
-      "loss": 67.858,
-      "step": 88560
-    },
-    {
-      "epoch": 0.35783400736111054,
-      "grad_norm": 539.9357299804688,
-      "learning_rate": 4.078178955658565e-05,
-      "loss": 62.4049,
-      "step": 88570
-    },
-    {
-      "epoch": 0.3578744086264782,
-      "grad_norm": 571.8364868164062,
-      "learning_rate": 4.077908218117625e-05,
-      "loss": 80.4749,
-      "step": 88580
-    },
-    {
-      "epoch": 0.3579148098918458,
-      "grad_norm": 1833.5181884765625,
-      "learning_rate": 4.077637449814552e-05,
-      "loss": 59.1497,
-      "step": 88590
-    },
-    {
-      "epoch": 0.35795521115721346,
-      "grad_norm": 1636.6201171875,
-      "learning_rate": 4.077366650754624e-05,
-      "loss": 69.4785,
-      "step": 88600
-    },
-    {
-      "epoch": 0.3579956124225811,
-      "grad_norm": 2520.92822265625,
-      "learning_rate": 4.077095820943122e-05,
-      "loss": 92.7431,
-      "step": 88610
-    },
-    {
-      "epoch": 0.3580360136879487,
-      "grad_norm": 781.56298828125,
-      "learning_rate": 4.0768249603853245e-05,
-      "loss": 66.4768,
-      "step": 88620
-    },
-    {
-      "epoch": 0.3580764149533163,
-      "grad_norm": 462.7001037597656,
-      "learning_rate": 4.0765540690865134e-05,
-      "loss": 39.243,
-      "step": 88630
-    },
-    {
-      "epoch": 0.35811681621868396,
-      "grad_norm": 1026.7464599609375,
-      "learning_rate": 4.076283147051968e-05,
-      "loss": 68.4941,
-      "step": 88640
-    },
-    {
-      "epoch": 0.3581572174840516,
-      "grad_norm": 558.1005859375,
-      "learning_rate": 4.0760121942869725e-05,
-      "loss": 75.5133,
-      "step": 88650
-    },
-    {
-      "epoch": 0.35819761874941924,
-      "grad_norm": 1109.0501708984375,
-      "learning_rate": 4.075741210796806e-05,
-      "loss": 89.089,
-      "step": 88660
-    },
-    {
-      "epoch": 0.3582380200147869,
-      "grad_norm": 1361.9794921875,
-      "learning_rate": 4.075470196586755e-05,
-      "loss": 81.0406,
-      "step": 88670
-    },
-    {
-      "epoch": 0.35827842128015447,
-      "grad_norm": 390.9439697265625,
-      "learning_rate": 4.075199151662101e-05,
-      "loss": 67.3965,
-      "step": 88680
-    },
-    {
-      "epoch": 0.3583188225455221,
-      "grad_norm": 433.46759033203125,
-      "learning_rate": 4.074928076028128e-05,
-      "loss": 59.471,
-      "step": 88690
-    },
-    {
-      "epoch": 0.35835922381088975,
-      "grad_norm": 3093.1083984375,
-      "learning_rate": 4.074656969690122e-05,
-      "loss": 67.4171,
-      "step": 88700
-    },
-    {
-      "epoch": 0.3583996250762574,
-      "grad_norm": 650.4321899414062,
-      "learning_rate": 4.0743858326533674e-05,
-      "loss": 57.6206,
-      "step": 88710
-    },
-    {
-      "epoch": 0.35844002634162503,
-      "grad_norm": 693.6144409179688,
-      "learning_rate": 4.0741146649231504e-05,
-      "loss": 69.4233,
-      "step": 88720
-    },
-    {
-      "epoch": 0.35848042760699267,
-      "grad_norm": 1305.1888427734375,
-      "learning_rate": 4.0738434665047575e-05,
-      "loss": 66.7869,
-      "step": 88730
-    },
-    {
-      "epoch": 0.3585208288723603,
-      "grad_norm": 671.9458618164062,
-      "learning_rate": 4.0735722374034764e-05,
-      "loss": 94.4328,
-      "step": 88740
-    },
-    {
-      "epoch": 0.3585612301377279,
-      "grad_norm": 782.94873046875,
-      "learning_rate": 4.073300977624594e-05,
-      "loss": 59.1198,
-      "step": 88750
-    },
-    {
-      "epoch": 0.35860163140309553,
-      "grad_norm": 808.3102416992188,
-      "learning_rate": 4.073029687173399e-05,
-      "loss": 65.3728,
-      "step": 88760
-    },
-    {
-      "epoch": 0.3586420326684632,
-      "grad_norm": 2808.516357421875,
-      "learning_rate": 4.0727583660551806e-05,
-      "loss": 57.111,
-      "step": 88770
-    },
-    {
-      "epoch": 0.3586824339338308,
-      "grad_norm": 905.2903442382812,
-      "learning_rate": 4.0724870142752284e-05,
-      "loss": 80.5446,
-      "step": 88780
-    },
-    {
-      "epoch": 0.35872283519919845,
-      "grad_norm": 245.458251953125,
-      "learning_rate": 4.0722156318388315e-05,
-      "loss": 81.0946,
-      "step": 88790
-    },
-    {
-      "epoch": 0.3587632364645661,
-      "grad_norm": 1166.0023193359375,
-      "learning_rate": 4.071944218751282e-05,
-      "loss": 62.5315,
-      "step": 88800
-    },
-    {
-      "epoch": 0.3588036377299337,
-      "grad_norm": 1779.5706787109375,
-      "learning_rate": 4.0716727750178704e-05,
-      "loss": 73.8024,
-      "step": 88810
-    },
-    {
-      "epoch": 0.3588440389953013,
-      "grad_norm": 3553.527587890625,
-      "learning_rate": 4.071401300643889e-05,
-      "loss": 107.6503,
-      "step": 88820
-    },
-    {
-      "epoch": 0.35888444026066896,
-      "grad_norm": 990.9931030273438,
-      "learning_rate": 4.0711297956346306e-05,
-      "loss": 51.3191,
-      "step": 88830
-    },
-    {
-      "epoch": 0.3589248415260366,
-      "grad_norm": 567.8832397460938,
-      "learning_rate": 4.070858259995387e-05,
-      "loss": 85.3655,
-      "step": 88840
-    },
-    {
-      "epoch": 0.35896524279140424,
-      "grad_norm": 978.09130859375,
-      "learning_rate": 4.070586693731454e-05,
-      "loss": 86.0509,
-      "step": 88850
-    },
-    {
-      "epoch": 0.3590056440567719,
-      "grad_norm": 654.3468017578125,
-      "learning_rate": 4.0703150968481246e-05,
-      "loss": 80.2084,
-      "step": 88860
-    },
-    {
-      "epoch": 0.35904604532213946,
-      "grad_norm": 219.86631774902344,
-      "learning_rate": 4.070043469350694e-05,
-      "loss": 61.1951,
-      "step": 88870
-    },
-    {
-      "epoch": 0.3590864465875071,
-      "grad_norm": 804.8790283203125,
-      "learning_rate": 4.069771811244457e-05,
-      "loss": 75.843,
-      "step": 88880
-    },
-    {
-      "epoch": 0.35912684785287474,
-      "grad_norm": 546.8237915039062,
-      "learning_rate": 4.0695001225347104e-05,
-      "loss": 51.4436,
-      "step": 88890
-    },
-    {
-      "epoch": 0.3591672491182424,
-      "grad_norm": 584.9393920898438,
-      "learning_rate": 4.0692284032267516e-05,
-      "loss": 50.8256,
-      "step": 88900
-    },
-    {
-      "epoch": 0.35920765038361,
-      "grad_norm": 2287.378173828125,
-      "learning_rate": 4.0689566533258765e-05,
-      "loss": 85.8929,
-      "step": 88910
-    },
-    {
-      "epoch": 0.35924805164897766,
-      "grad_norm": 798.82470703125,
-      "learning_rate": 4.068684872837384e-05,
-      "loss": 54.3008,
-      "step": 88920
-    },
-    {
-      "epoch": 0.3592884529143453,
-      "grad_norm": 631.5094604492188,
-      "learning_rate": 4.068413061766572e-05,
-      "loss": 67.0688,
-      "step": 88930
-    },
-    {
-      "epoch": 0.3593288541797129,
-      "grad_norm": 788.1356811523438,
-      "learning_rate": 4.068141220118741e-05,
-      "loss": 63.0645,
-      "step": 88940
-    },
-    {
-      "epoch": 0.3593692554450805,
-      "grad_norm": 424.5524597167969,
-      "learning_rate": 4.067869347899188e-05,
-      "loss": 54.8084,
-      "step": 88950
-    },
-    {
-      "epoch": 0.35940965671044817,
-      "grad_norm": 653.3056640625,
-      "learning_rate": 4.067597445113216e-05,
-      "loss": 50.6989,
-      "step": 88960
-    },
-    {
-      "epoch": 0.3594500579758158,
-      "grad_norm": 343.56512451171875,
-      "learning_rate": 4.067325511766124e-05,
-      "loss": 82.852,
-      "step": 88970
-    },
-    {
-      "epoch": 0.35949045924118345,
-      "grad_norm": 750.8988647460938,
-      "learning_rate": 4.067053547863215e-05,
-      "loss": 85.4746,
-      "step": 88980
-    },
-    {
-      "epoch": 0.3595308605065511,
-      "grad_norm": 1303.02197265625,
-      "learning_rate": 4.06678155340979e-05,
-      "loss": 50.7246,
-      "step": 88990
-    },
-    {
-      "epoch": 0.35957126177191867,
-      "grad_norm": 777.9240112304688,
-      "learning_rate": 4.066509528411152e-05,
-      "loss": 62.3671,
-      "step": 89000
-    },
-    {
-      "epoch": 0.3596116630372863,
-      "grad_norm": 304.01275634765625,
-      "learning_rate": 4.066237472872604e-05,
-      "loss": 68.3459,
-      "step": 89010
-    },
-    {
-      "epoch": 0.35965206430265395,
-      "grad_norm": 1499.3465576171875,
-      "learning_rate": 4.0659653867994496e-05,
-      "loss": 69.0432,
-      "step": 89020
-    },
-    {
-      "epoch": 0.3596924655680216,
-      "grad_norm": 1114.2425537109375,
-      "learning_rate": 4.065693270196995e-05,
-      "loss": 68.1484,
-      "step": 89030
-    },
-    {
-      "epoch": 0.35973286683338923,
-      "grad_norm": 738.8199462890625,
-      "learning_rate": 4.065421123070543e-05,
-      "loss": 59.836,
-      "step": 89040
-    },
-    {
-      "epoch": 0.35977326809875687,
-      "grad_norm": 616.7838134765625,
-      "learning_rate": 4.065148945425401e-05,
-      "loss": 61.9072,
-      "step": 89050
-    },
-    {
-      "epoch": 0.3598136693641245,
-      "grad_norm": 215.50350952148438,
-      "learning_rate": 4.064876737266874e-05,
-      "loss": 81.0354,
-      "step": 89060
-    },
-    {
-      "epoch": 0.3598540706294921,
-      "grad_norm": 344.2589111328125,
-      "learning_rate": 4.06460449860027e-05,
-      "loss": 61.0841,
-      "step": 89070
-    },
-    {
-      "epoch": 0.35989447189485974,
-      "grad_norm": 740.9758911132812,
-      "learning_rate": 4.064332229430895e-05,
-      "loss": 53.366,
-      "step": 89080
-    },
-    {
-      "epoch": 0.3599348731602274,
-      "grad_norm": 717.7742919921875,
-      "learning_rate": 4.0640599297640584e-05,
-      "loss": 42.3575,
-      "step": 89090
-    },
-    {
-      "epoch": 0.359975274425595,
-      "grad_norm": 1074.0352783203125,
-      "learning_rate": 4.063787599605068e-05,
-      "loss": 79.3194,
-      "step": 89100
-    },
-    {
-      "epoch": 0.36001567569096266,
-      "grad_norm": 1248.140869140625,
-      "learning_rate": 4.063515238959233e-05,
-      "loss": 60.0265,
-      "step": 89110
-    },
-    {
-      "epoch": 0.3600560769563303,
-      "grad_norm": 400.5279235839844,
-      "learning_rate": 4.063242847831864e-05,
-      "loss": 56.0902,
-      "step": 89120
-    },
-    {
-      "epoch": 0.3600964782216979,
-      "grad_norm": 4143.07568359375,
-      "learning_rate": 4.06297042622827e-05,
-      "loss": 77.4453,
-      "step": 89130
-    },
-    {
-      "epoch": 0.3601368794870655,
-      "grad_norm": 520.5917358398438,
-      "learning_rate": 4.062697974153764e-05,
-      "loss": 56.0878,
-      "step": 89140
-    },
-    {
-      "epoch": 0.36017728075243316,
-      "grad_norm": 1186.0679931640625,
-      "learning_rate": 4.062425491613656e-05,
-      "loss": 70.4133,
-      "step": 89150
-    },
-    {
-      "epoch": 0.3602176820178008,
-      "grad_norm": 972.8978271484375,
-      "learning_rate": 4.062152978613258e-05,
-      "loss": 47.6216,
-      "step": 89160
-    },
-    {
-      "epoch": 0.36025808328316844,
-      "grad_norm": 1175.881591796875,
-      "learning_rate": 4.061880435157884e-05,
-      "loss": 52.5779,
-      "step": 89170
-    },
-    {
-      "epoch": 0.3602984845485361,
-      "grad_norm": 1707.835205078125,
-      "learning_rate": 4.061607861252847e-05,
-      "loss": 49.4881,
-      "step": 89180
-    },
-    {
-      "epoch": 0.36033888581390366,
-      "grad_norm": 898.6934814453125,
-      "learning_rate": 4.0613352569034615e-05,
-      "loss": 83.9531,
-      "step": 89190
-    },
-    {
-      "epoch": 0.3603792870792713,
-      "grad_norm": 791.3284912109375,
-      "learning_rate": 4.0610626221150394e-05,
-      "loss": 56.8294,
-      "step": 89200
-    },
-    {
-      "epoch": 0.36041968834463894,
-      "grad_norm": 439.3869323730469,
-      "learning_rate": 4.060789956892899e-05,
-      "loss": 78.4056,
-      "step": 89210
-    },
-    {
-      "epoch": 0.3604600896100066,
-      "grad_norm": 306.3957214355469,
-      "learning_rate": 4.060517261242355e-05,
-      "loss": 53.2591,
-      "step": 89220
-    },
-    {
-      "epoch": 0.3605004908753742,
-      "grad_norm": 862.3554077148438,
-      "learning_rate": 4.060244535168723e-05,
-      "loss": 57.5533,
-      "step": 89230
-    },
-    {
-      "epoch": 0.36054089214074186,
-      "grad_norm": 631.755859375,
-      "learning_rate": 4.0599717786773204e-05,
-      "loss": 46.8937,
-      "step": 89240
-    },
-    {
-      "epoch": 0.3605812934061095,
-      "grad_norm": 1259.4813232421875,
-      "learning_rate": 4.059698991773466e-05,
-      "loss": 62.3537,
-      "step": 89250
-    },
-    {
-      "epoch": 0.3606216946714771,
-      "grad_norm": 467.6385192871094,
-      "learning_rate": 4.059426174462476e-05,
-      "loss": 65.602,
-      "step": 89260
-    },
-    {
-      "epoch": 0.36066209593684473,
-      "grad_norm": 492.99609375,
-      "learning_rate": 4.0591533267496694e-05,
-      "loss": 106.2255,
-      "step": 89270
-    },
-    {
-      "epoch": 0.36070249720221237,
-      "grad_norm": 893.2991943359375,
-      "learning_rate": 4.058880448640367e-05,
-      "loss": 96.5574,
-      "step": 89280
-    },
-    {
-      "epoch": 0.36074289846758,
-      "grad_norm": 645.7114868164062,
-      "learning_rate": 4.058607540139887e-05,
-      "loss": 90.8627,
-      "step": 89290
-    },
-    {
-      "epoch": 0.36078329973294765,
-      "grad_norm": 3922.514404296875,
-      "learning_rate": 4.0583346012535506e-05,
-      "loss": 111.6305,
-      "step": 89300
-    },
-    {
-      "epoch": 0.3608237009983153,
-      "grad_norm": 904.625,
-      "learning_rate": 4.058061631986679e-05,
-      "loss": 79.8804,
-      "step": 89310
-    },
-    {
-      "epoch": 0.3608641022636829,
-      "grad_norm": 384.1736755371094,
-      "learning_rate": 4.057788632344593e-05,
-      "loss": 60.1387,
-      "step": 89320
-    },
-    {
-      "epoch": 0.3609045035290505,
-      "grad_norm": 675.1691284179688,
-      "learning_rate": 4.0575156023326166e-05,
-      "loss": 60.848,
-      "step": 89330
-    },
-    {
-      "epoch": 0.36094490479441815,
-      "grad_norm": 963.3970947265625,
-      "learning_rate": 4.0572425419560714e-05,
-      "loss": 57.821,
-      "step": 89340
-    },
-    {
-      "epoch": 0.3609853060597858,
-      "grad_norm": 0.0,
-      "learning_rate": 4.056969451220282e-05,
-      "loss": 72.7515,
-      "step": 89350
-    },
-    {
-      "epoch": 0.36102570732515343,
-      "grad_norm": 610.1162719726562,
-      "learning_rate": 4.0566963301305705e-05,
-      "loss": 50.6369,
-      "step": 89360
-    },
-    {
-      "epoch": 0.3610661085905211,
-      "grad_norm": 643.9298706054688,
-      "learning_rate": 4.056423178692262e-05,
-      "loss": 39.3118,
-      "step": 89370
-    },
-    {
-      "epoch": 0.3611065098558887,
-      "grad_norm": 531.35693359375,
-      "learning_rate": 4.056149996910683e-05,
-      "loss": 75.3905,
-      "step": 89380
-    },
-    {
-      "epoch": 0.3611469111212563,
-      "grad_norm": 503.90777587890625,
-      "learning_rate": 4.05587678479116e-05,
-      "loss": 65.8769,
-      "step": 89390
-    },
-    {
-      "epoch": 0.36118731238662394,
-      "grad_norm": 1145.5985107421875,
-      "learning_rate": 4.055603542339016e-05,
-      "loss": 51.2441,
-      "step": 89400
-    },
-    {
-      "epoch": 0.3612277136519916,
-      "grad_norm": 411.92626953125,
-      "learning_rate": 4.055330269559581e-05,
-      "loss": 35.5195,
-      "step": 89410
-    },
-    {
-      "epoch": 0.3612681149173592,
-      "grad_norm": 745.9331665039062,
-      "learning_rate": 4.055056966458182e-05,
-      "loss": 58.2253,
-      "step": 89420
-    },
-    {
-      "epoch": 0.36130851618272686,
-      "grad_norm": 5621.34814453125,
-      "learning_rate": 4.054783633040146e-05,
-      "loss": 78.9581,
-      "step": 89430
-    },
-    {
-      "epoch": 0.3613489174480945,
-      "grad_norm": 5587.95263671875,
-      "learning_rate": 4.054510269310803e-05,
-      "loss": 83.4821,
-      "step": 89440
-    },
-    {
-      "epoch": 0.3613893187134621,
-      "grad_norm": 402.0248107910156,
-      "learning_rate": 4.0542368752754825e-05,
-      "loss": 57.0781,
-      "step": 89450
-    },
-    {
-      "epoch": 0.3614297199788297,
-      "grad_norm": 1605.563720703125,
-      "learning_rate": 4.053963450939513e-05,
-      "loss": 90.5012,
-      "step": 89460
-    },
-    {
-      "epoch": 0.36147012124419736,
-      "grad_norm": 471.9956359863281,
-      "learning_rate": 4.053689996308227e-05,
-      "loss": 71.2476,
-      "step": 89470
-    },
-    {
-      "epoch": 0.361510522509565,
-      "grad_norm": 0.0,
-      "learning_rate": 4.053416511386954e-05,
-      "loss": 45.8093,
-      "step": 89480
-    },
-    {
-      "epoch": 0.36155092377493264,
-      "grad_norm": 800.1768188476562,
-      "learning_rate": 4.0531429961810264e-05,
-      "loss": 46.3764,
-      "step": 89490
-    },
-    {
-      "epoch": 0.3615913250403003,
-      "grad_norm": 731.527587890625,
-      "learning_rate": 4.052869450695776e-05,
-      "loss": 78.3263,
-      "step": 89500
-    },
-    {
-      "epoch": 0.36163172630566787,
-      "grad_norm": 1999.2166748046875,
-      "learning_rate": 4.052595874936537e-05,
-      "loss": 70.2021,
-      "step": 89510
-    },
-    {
-      "epoch": 0.3616721275710355,
-      "grad_norm": 0.0,
-      "learning_rate": 4.0523222689086414e-05,
-      "loss": 75.0036,
-      "step": 89520
-    },
-    {
-      "epoch": 0.36171252883640315,
-      "grad_norm": 896.73388671875,
-      "learning_rate": 4.052048632617424e-05,
-      "loss": 73.4701,
-      "step": 89530
-    },
-    {
-      "epoch": 0.3617529301017708,
-      "grad_norm": 1283.2412109375,
-      "learning_rate": 4.05177496606822e-05,
-      "loss": 50.1073,
-      "step": 89540
-    },
-    {
-      "epoch": 0.3617933313671384,
-      "grad_norm": 543.897705078125,
-      "learning_rate": 4.0515012692663646e-05,
-      "loss": 52.8131,
-      "step": 89550
-    },
-    {
-      "epoch": 0.36183373263250607,
-      "grad_norm": 1975.94384765625,
-      "learning_rate": 4.051227542217192e-05,
-      "loss": 92.001,
-      "step": 89560
-    },
-    {
-      "epoch": 0.3618741338978737,
-      "grad_norm": 247.90350341796875,
-      "learning_rate": 4.0509537849260404e-05,
-      "loss": 31.5658,
-      "step": 89570
-    },
-    {
-      "epoch": 0.3619145351632413,
-      "grad_norm": 859.4539184570312,
-      "learning_rate": 4.0506799973982465e-05,
-      "loss": 55.6403,
-      "step": 89580
-    },
-    {
-      "epoch": 0.36195493642860893,
-      "grad_norm": 1263.11474609375,
-      "learning_rate": 4.0504061796391474e-05,
-      "loss": 59.762,
-      "step": 89590
-    },
-    {
-      "epoch": 0.36199533769397657,
-      "grad_norm": 263.7984313964844,
-      "learning_rate": 4.050132331654082e-05,
-      "loss": 36.6857,
-      "step": 89600
-    },
-    {
-      "epoch": 0.3620357389593442,
-      "grad_norm": 599.0194702148438,
-      "learning_rate": 4.0498584534483877e-05,
-      "loss": 101.6762,
-      "step": 89610
-    },
-    {
-      "epoch": 0.36207614022471185,
-      "grad_norm": 1161.4105224609375,
-      "learning_rate": 4.0495845450274064e-05,
-      "loss": 73.1191,
-      "step": 89620
-    },
-    {
-      "epoch": 0.3621165414900795,
-      "grad_norm": 977.6044921875,
-      "learning_rate": 4.0493106063964754e-05,
-      "loss": 63.5236,
-      "step": 89630
-    },
-    {
-      "epoch": 0.3621569427554471,
-      "grad_norm": 653.26708984375,
-      "learning_rate": 4.0490366375609376e-05,
-      "loss": 99.8486,
-      "step": 89640
-    },
-    {
-      "epoch": 0.3621973440208147,
-      "grad_norm": 1127.0865478515625,
-      "learning_rate": 4.048762638526132e-05,
-      "loss": 51.8638,
-      "step": 89650
-    },
-    {
-      "epoch": 0.36223774528618236,
-      "grad_norm": 565.7835693359375,
-      "learning_rate": 4.048488609297402e-05,
-      "loss": 61.2772,
-      "step": 89660
-    },
-    {
-      "epoch": 0.36227814655155,
-      "grad_norm": 774.4669799804688,
-      "learning_rate": 4.0482145498800884e-05,
-      "loss": 61.0465,
-      "step": 89670
-    },
-    {
-      "epoch": 0.36231854781691764,
-      "grad_norm": 2059.27880859375,
-      "learning_rate": 4.047940460279537e-05,
-      "loss": 80.3147,
-      "step": 89680
-    },
-    {
-      "epoch": 0.3623589490822853,
-      "grad_norm": 890.2815551757812,
-      "learning_rate": 4.0476663405010874e-05,
-      "loss": 85.4335,
-      "step": 89690
-    },
-    {
-      "epoch": 0.3623993503476529,
-      "grad_norm": 1310.5477294921875,
-      "learning_rate": 4.047392190550087e-05,
-      "loss": 51.1065,
-      "step": 89700
-    },
-    {
-      "epoch": 0.3624397516130205,
-      "grad_norm": 640.0499877929688,
-      "learning_rate": 4.047118010431879e-05,
-      "loss": 79.5663,
-      "step": 89710
-    },
-    {
-      "epoch": 0.36248015287838814,
-      "grad_norm": 773.0057373046875,
-      "learning_rate": 4.0468438001518084e-05,
-      "loss": 56.4114,
-      "step": 89720
-    },
-    {
-      "epoch": 0.3625205541437558,
-      "grad_norm": 652.1478881835938,
-      "learning_rate": 4.046569559715221e-05,
-      "loss": 52.9334,
-      "step": 89730
-    },
-    {
-      "epoch": 0.3625609554091234,
-      "grad_norm": 1459.3680419921875,
-      "learning_rate": 4.0462952891274655e-05,
-      "loss": 71.04,
-      "step": 89740
-    },
-    {
-      "epoch": 0.36260135667449106,
-      "grad_norm": 1920.9996337890625,
-      "learning_rate": 4.046020988393885e-05,
-      "loss": 61.0197,
-      "step": 89750
-    },
-    {
-      "epoch": 0.3626417579398587,
-      "grad_norm": 1488.5030517578125,
-      "learning_rate": 4.045746657519831e-05,
-      "loss": 76.9832,
-      "step": 89760
-    },
-    {
-      "epoch": 0.3626821592052263,
-      "grad_norm": 852.4432983398438,
-      "learning_rate": 4.04547229651065e-05,
-      "loss": 61.3612,
-      "step": 89770
-    },
-    {
-      "epoch": 0.3627225604705939,
-      "grad_norm": 748.500244140625,
-      "learning_rate": 4.0451979053716906e-05,
-      "loss": 61.8753,
-      "step": 89780
-    },
-    {
-      "epoch": 0.36276296173596156,
-      "grad_norm": 988.9141235351562,
-      "learning_rate": 4.044923484108303e-05,
-      "loss": 53.8847,
-      "step": 89790
-    },
-    {
-      "epoch": 0.3628033630013292,
-      "grad_norm": 1280.90673828125,
-      "learning_rate": 4.044649032725836e-05,
-      "loss": 63.5749,
-      "step": 89800
-    },
-    {
-      "epoch": 0.36284376426669684,
-      "grad_norm": 1100.82177734375,
-      "learning_rate": 4.044374551229641e-05,
-      "loss": 87.8433,
-      "step": 89810
-    },
-    {
-      "epoch": 0.3628841655320645,
-      "grad_norm": 936.5944213867188,
-      "learning_rate": 4.0441000396250694e-05,
-      "loss": 62.9258,
-      "step": 89820
-    },
-    {
-      "epoch": 0.36292456679743207,
-      "grad_norm": 701.0770874023438,
-      "learning_rate": 4.0438254979174725e-05,
-      "loss": 50.3677,
-      "step": 89830
-    },
-    {
-      "epoch": 0.3629649680627997,
-      "grad_norm": 0.0,
-      "learning_rate": 4.043550926112203e-05,
-      "loss": 60.1964,
-      "step": 89840
-    },
-    {
-      "epoch": 0.36300536932816735,
-      "grad_norm": 546.4686889648438,
-      "learning_rate": 4.043276324214613e-05,
-      "loss": 53.8509,
-      "step": 89850
-    },
-    {
-      "epoch": 0.363045770593535,
-      "grad_norm": 1106.510009765625,
-      "learning_rate": 4.043001692230056e-05,
-      "loss": 69.3517,
-      "step": 89860
-    },
-    {
-      "epoch": 0.36308617185890263,
-      "grad_norm": 665.1596069335938,
-      "learning_rate": 4.042727030163888e-05,
-      "loss": 53.5625,
-      "step": 89870
-    },
-    {
-      "epoch": 0.36312657312427027,
-      "grad_norm": 640.2094116210938,
-      "learning_rate": 4.042452338021461e-05,
-      "loss": 90.0176,
-      "step": 89880
-    },
-    {
-      "epoch": 0.3631669743896379,
-      "grad_norm": 552.5271606445312,
-      "learning_rate": 4.0421776158081326e-05,
-      "loss": 39.7842,
-      "step": 89890
-    },
-    {
-      "epoch": 0.3632073756550055,
-      "grad_norm": 609.4852905273438,
-      "learning_rate": 4.041902863529256e-05,
-      "loss": 98.3096,
-      "step": 89900
-    },
-    {
-      "epoch": 0.36324777692037313,
-      "grad_norm": 1036.63671875,
-      "learning_rate": 4.041628081190191e-05,
-      "loss": 74.554,
-      "step": 89910
-    },
-    {
-      "epoch": 0.3632881781857408,
-      "grad_norm": 296.92547607421875,
-      "learning_rate": 4.041353268796293e-05,
-      "loss": 75.188,
-      "step": 89920
-    },
-    {
-      "epoch": 0.3633285794511084,
-      "grad_norm": 404.4271240234375,
-      "learning_rate": 4.041078426352918e-05,
-      "loss": 43.2436,
-      "step": 89930
-    },
-    {
-      "epoch": 0.36336898071647605,
-      "grad_norm": 1445.6839599609375,
-      "learning_rate": 4.0408035538654264e-05,
-      "loss": 48.7288,
-      "step": 89940
-    },
-    {
-      "epoch": 0.3634093819818437,
-      "grad_norm": 745.0489501953125,
-      "learning_rate": 4.040528651339176e-05,
-      "loss": 77.1038,
-      "step": 89950
-    },
-    {
-      "epoch": 0.3634497832472113,
-      "grad_norm": 543.2232055664062,
-      "learning_rate": 4.0402537187795274e-05,
-      "loss": 56.8202,
-      "step": 89960
-    },
-    {
-      "epoch": 0.3634901845125789,
-      "grad_norm": 557.49072265625,
-      "learning_rate": 4.039978756191839e-05,
-      "loss": 62.1664,
-      "step": 89970
-    },
-    {
-      "epoch": 0.36353058577794656,
-      "grad_norm": 612.9962768554688,
-      "learning_rate": 4.039703763581472e-05,
-      "loss": 82.6836,
-      "step": 89980
-    },
-    {
-      "epoch": 0.3635709870433142,
-      "grad_norm": 921.2637939453125,
-      "learning_rate": 4.039428740953787e-05,
-      "loss": 50.0055,
-      "step": 89990
-    },
-    {
-      "epoch": 0.36361138830868184,
-      "grad_norm": 1612.236328125,
-      "learning_rate": 4.039153688314145e-05,
-      "loss": 85.7117,
-      "step": 90000
-    },
-    {
-      "epoch": 0.3636517895740495,
-      "grad_norm": 656.6378173828125,
-      "learning_rate": 4.038878605667912e-05,
-      "loss": 84.2719,
-      "step": 90010
-    },
-    {
-      "epoch": 0.3636921908394171,
-      "grad_norm": 1089.3055419921875,
-      "learning_rate": 4.038603493020447e-05,
-      "loss": 76.3893,
-      "step": 90020
-    },
-    {
-      "epoch": 0.3637325921047847,
-      "grad_norm": 954.1974487304688,
-      "learning_rate": 4.038328350377115e-05,
-      "loss": 111.787,
-      "step": 90030
-    },
-    {
-      "epoch": 0.36377299337015234,
-      "grad_norm": 725.2201538085938,
-      "learning_rate": 4.0380531777432794e-05,
-      "loss": 60.9843,
-      "step": 90040
-    },
-    {
-      "epoch": 0.36381339463552,
-      "grad_norm": 1051.9312744140625,
-      "learning_rate": 4.037777975124306e-05,
-      "loss": 75.9995,
-      "step": 90050
-    },
-    {
-      "epoch": 0.3638537959008876,
-      "grad_norm": 1385.2197265625,
-      "learning_rate": 4.037502742525559e-05,
-      "loss": 85.4595,
-      "step": 90060
-    },
-    {
-      "epoch": 0.36389419716625526,
-      "grad_norm": 505.546875,
-      "learning_rate": 4.037227479952404e-05,
-      "loss": 56.1493,
-      "step": 90070
-    },
-    {
-      "epoch": 0.3639345984316229,
-      "grad_norm": 749.1838989257812,
-      "learning_rate": 4.036952187410208e-05,
-      "loss": 63.6978,
-      "step": 90080
-    },
-    {
-      "epoch": 0.3639749996969905,
-      "grad_norm": 534.1312866210938,
-      "learning_rate": 4.036676864904338e-05,
-      "loss": 57.1218,
-      "step": 90090
-    },
-    {
-      "epoch": 0.3640154009623581,
-      "grad_norm": 921.06005859375,
-      "learning_rate": 4.036401512440161e-05,
-      "loss": 58.3411,
-      "step": 90100
-    },
-    {
-      "epoch": 0.36405580222772577,
-      "grad_norm": 679.614990234375,
-      "learning_rate": 4.0361261300230465e-05,
-      "loss": 66.4312,
-      "step": 90110
-    },
-    {
-      "epoch": 0.3640962034930934,
-      "grad_norm": 511.14501953125,
-      "learning_rate": 4.035850717658362e-05,
-      "loss": 106.1777,
-      "step": 90120
-    },
-    {
-      "epoch": 0.36413660475846105,
-      "grad_norm": 1034.344482421875,
-      "learning_rate": 4.035575275351476e-05,
-      "loss": 78.6351,
-      "step": 90130
-    },
-    {
-      "epoch": 0.3641770060238287,
-      "grad_norm": 880.7720947265625,
-      "learning_rate": 4.0352998031077604e-05,
-      "loss": 67.1489,
-      "step": 90140
-    },
-    {
-      "epoch": 0.36421740728919627,
-      "grad_norm": 1041.2879638671875,
-      "learning_rate": 4.035024300932584e-05,
-      "loss": 59.1619,
-      "step": 90150
-    },
-    {
-      "epoch": 0.3642578085545639,
-      "grad_norm": 414.6109924316406,
-      "learning_rate": 4.0347487688313194e-05,
-      "loss": 57.6462,
-      "step": 90160
-    },
-    {
-      "epoch": 0.36429820981993155,
-      "grad_norm": 803.5387573242188,
-      "learning_rate": 4.034473206809337e-05,
-      "loss": 62.8699,
-      "step": 90170
-    },
-    {
-      "epoch": 0.3643386110852992,
-      "grad_norm": 982.3478393554688,
-      "learning_rate": 4.0341976148720095e-05,
-      "loss": 97.5246,
-      "step": 90180
-    },
-    {
-      "epoch": 0.36437901235066683,
-      "grad_norm": 951.2240600585938,
-      "learning_rate": 4.03392199302471e-05,
-      "loss": 51.8766,
-      "step": 90190
-    },
-    {
-      "epoch": 0.36441941361603447,
-      "grad_norm": 754.9638061523438,
-      "learning_rate": 4.033646341272811e-05,
-      "loss": 59.1747,
-      "step": 90200
-    },
-    {
-      "epoch": 0.3644598148814021,
-      "grad_norm": 1095.536376953125,
-      "learning_rate": 4.033370659621687e-05,
-      "loss": 43.2328,
-      "step": 90210
-    },
-    {
-      "epoch": 0.3645002161467697,
-      "grad_norm": 915.0740966796875,
-      "learning_rate": 4.033094948076713e-05,
-      "loss": 77.3148,
-      "step": 90220
-    },
-    {
-      "epoch": 0.36454061741213734,
-      "grad_norm": 502.77545166015625,
-      "learning_rate": 4.032819206643263e-05,
-      "loss": 41.8404,
-      "step": 90230
-    },
-    {
-      "epoch": 0.364581018677505,
-      "grad_norm": 352.51165771484375,
-      "learning_rate": 4.032543435326714e-05,
-      "loss": 55.2335,
-      "step": 90240
-    },
-    {
-      "epoch": 0.3646214199428726,
-      "grad_norm": 1066.30615234375,
-      "learning_rate": 4.0322676341324415e-05,
-      "loss": 87.1459,
-      "step": 90250
-    },
-    {
-      "epoch": 0.36466182120824026,
-      "grad_norm": 486.5985107421875,
-      "learning_rate": 4.0319918030658225e-05,
-      "loss": 67.0294,
-      "step": 90260
-    },
-    {
-      "epoch": 0.3647022224736079,
-      "grad_norm": 637.4968872070312,
-      "learning_rate": 4.031715942132235e-05,
-      "loss": 67.5639,
-      "step": 90270
-    },
-    {
-      "epoch": 0.3647426237389755,
-      "grad_norm": 919.53759765625,
-      "learning_rate": 4.031440051337056e-05,
-      "loss": 88.9865,
-      "step": 90280
-    },
-    {
-      "epoch": 0.3647830250043431,
-      "grad_norm": 551.5164184570312,
-      "learning_rate": 4.031164130685665e-05,
-      "loss": 68.1998,
-      "step": 90290
-    },
-    {
-      "epoch": 0.36482342626971076,
-      "grad_norm": 1867.7833251953125,
-      "learning_rate": 4.030888180183441e-05,
-      "loss": 80.087,
-      "step": 90300
-    },
-    {
-      "epoch": 0.3648638275350784,
-      "grad_norm": 731.1893920898438,
-      "learning_rate": 4.030612199835764e-05,
-      "loss": 49.1685,
-      "step": 90310
-    },
-    {
-      "epoch": 0.36490422880044604,
-      "grad_norm": 1434.2115478515625,
-      "learning_rate": 4.030336189648014e-05,
-      "loss": 79.8181,
-      "step": 90320
-    },
-    {
-      "epoch": 0.3649446300658137,
-      "grad_norm": 5654.36669921875,
-      "learning_rate": 4.030060149625573e-05,
-      "loss": 106.5813,
-      "step": 90330
-    },
-    {
-      "epoch": 0.3649850313311813,
-      "grad_norm": 1383.1339111328125,
-      "learning_rate": 4.02978407977382e-05,
-      "loss": 81.1142,
-      "step": 90340
-    },
-    {
-      "epoch": 0.3650254325965489,
-      "grad_norm": 1352.9986572265625,
-      "learning_rate": 4.0295079800981395e-05,
-      "loss": 75.8624,
-      "step": 90350
-    },
-    {
-      "epoch": 0.36506583386191654,
-      "grad_norm": 821.6564331054688,
-      "learning_rate": 4.029231850603914e-05,
-      "loss": 87.5254,
-      "step": 90360
-    },
-    {
-      "epoch": 0.3651062351272842,
-      "grad_norm": 375.5804138183594,
-      "learning_rate": 4.028955691296526e-05,
-      "loss": 43.4082,
-      "step": 90370
-    },
-    {
-      "epoch": 0.3651466363926518,
-      "grad_norm": 2169.375244140625,
-      "learning_rate": 4.0286795021813594e-05,
-      "loss": 67.1946,
-      "step": 90380
-    },
-    {
-      "epoch": 0.36518703765801946,
-      "grad_norm": 7678.23828125,
-      "learning_rate": 4.0284032832637985e-05,
-      "loss": 96.252,
-      "step": 90390
-    },
-    {
-      "epoch": 0.3652274389233871,
-      "grad_norm": 1672.2906494140625,
-      "learning_rate": 4.028127034549229e-05,
-      "loss": 76.2553,
-      "step": 90400
-    },
-    {
-      "epoch": 0.3652678401887547,
-      "grad_norm": 959.1575927734375,
-      "learning_rate": 4.027850756043037e-05,
-      "loss": 55.6581,
-      "step": 90410
-    },
-    {
-      "epoch": 0.36530824145412233,
-      "grad_norm": 646.5990600585938,
-      "learning_rate": 4.0275744477506074e-05,
-      "loss": 79.772,
-      "step": 90420
-    },
-    {
-      "epoch": 0.36534864271948997,
-      "grad_norm": 905.00439453125,
-      "learning_rate": 4.027298109677327e-05,
-      "loss": 40.6796,
-      "step": 90430
-    },
-    {
-      "epoch": 0.3653890439848576,
-      "grad_norm": 800.4560546875,
-      "learning_rate": 4.027021741828584e-05,
-      "loss": 52.6281,
-      "step": 90440
-    },
-    {
-      "epoch": 0.36542944525022525,
-      "grad_norm": 507.7906188964844,
-      "learning_rate": 4.0267453442097664e-05,
-      "loss": 72.2229,
-      "step": 90450
-    },
-    {
-      "epoch": 0.3654698465155929,
-      "grad_norm": 371.3627624511719,
-      "learning_rate": 4.026468916826262e-05,
-      "loss": 65.535,
-      "step": 90460
-    },
-    {
-      "epoch": 0.3655102477809605,
-      "grad_norm": 393.8462219238281,
-      "learning_rate": 4.02619245968346e-05,
-      "loss": 75.8037,
-      "step": 90470
-    },
-    {
-      "epoch": 0.3655506490463281,
-      "grad_norm": 460.1830749511719,
-      "learning_rate": 4.0259159727867504e-05,
-      "loss": 82.6051,
-      "step": 90480
-    },
-    {
-      "epoch": 0.36559105031169575,
-      "grad_norm": 521.975341796875,
-      "learning_rate": 4.025639456141523e-05,
-      "loss": 61.9708,
-      "step": 90490
-    },
-    {
-      "epoch": 0.3656314515770634,
-      "grad_norm": 1583.7652587890625,
-      "learning_rate": 4.02536290975317e-05,
-      "loss": 64.9353,
-      "step": 90500
-    },
-    {
-      "epoch": 0.36567185284243103,
-      "grad_norm": 465.25244140625,
-      "learning_rate": 4.02508633362708e-05,
-      "loss": 65.8302,
-      "step": 90510
-    },
-    {
-      "epoch": 0.3657122541077987,
-      "grad_norm": 293.57965087890625,
-      "learning_rate": 4.024809727768648e-05,
-      "loss": 45.0099,
-      "step": 90520
-    },
-    {
-      "epoch": 0.3657526553731663,
-      "grad_norm": 536.3772583007812,
-      "learning_rate": 4.024533092183266e-05,
-      "loss": 74.4069,
-      "step": 90530
-    },
-    {
-      "epoch": 0.3657930566385339,
-      "grad_norm": 479.0234680175781,
-      "learning_rate": 4.024256426876325e-05,
-      "loss": 51.6421,
-      "step": 90540
-    },
-    {
-      "epoch": 0.36583345790390154,
-      "grad_norm": 1907.06494140625,
-      "learning_rate": 4.02397973185322e-05,
-      "loss": 77.2531,
-      "step": 90550
-    },
-    {
-      "epoch": 0.3658738591692692,
-      "grad_norm": 1270.7117919921875,
-      "learning_rate": 4.023703007119347e-05,
-      "loss": 59.475,
-      "step": 90560
-    },
-    {
-      "epoch": 0.3659142604346368,
-      "grad_norm": 380.741943359375,
-      "learning_rate": 4.023426252680098e-05,
-      "loss": 59.4132,
-      "step": 90570
-    },
-    {
-      "epoch": 0.36595466170000446,
-      "grad_norm": 1124.605224609375,
-      "learning_rate": 4.023149468540871e-05,
-      "loss": 67.3577,
-      "step": 90580
-    },
-    {
-      "epoch": 0.3659950629653721,
-      "grad_norm": 1151.210205078125,
-      "learning_rate": 4.02287265470706e-05,
-      "loss": 63.5007,
-      "step": 90590
-    },
-    {
-      "epoch": 0.3660354642307397,
-      "grad_norm": 0.0,
-      "learning_rate": 4.022595811184064e-05,
-      "loss": 38.7889,
-      "step": 90600
-    },
-    {
-      "epoch": 0.3660758654961073,
-      "grad_norm": 531.6884155273438,
-      "learning_rate": 4.022318937977277e-05,
-      "loss": 70.17,
-      "step": 90610
-    },
-    {
-      "epoch": 0.36611626676147496,
-      "grad_norm": 928.3363647460938,
-      "learning_rate": 4.022042035092101e-05,
-      "loss": 63.7609,
-      "step": 90620
-    },
-    {
-      "epoch": 0.3661566680268426,
-      "grad_norm": 1050.515380859375,
-      "learning_rate": 4.02176510253393e-05,
-      "loss": 62.498,
-      "step": 90630
-    },
-    {
-      "epoch": 0.36619706929221024,
-      "grad_norm": 1077.6502685546875,
-      "learning_rate": 4.021488140308165e-05,
-      "loss": 90.2313,
-      "step": 90640
-    },
-    {
-      "epoch": 0.3662374705575779,
-      "grad_norm": 1099.2191162109375,
-      "learning_rate": 4.021211148420205e-05,
-      "loss": 65.9301,
-      "step": 90650
-    },
-    {
-      "epoch": 0.3662778718229455,
-      "grad_norm": 633.58935546875,
-      "learning_rate": 4.020934126875452e-05,
-      "loss": 41.3643,
-      "step": 90660
-    },
-    {
-      "epoch": 0.3663182730883131,
-      "grad_norm": 1048.8428955078125,
-      "learning_rate": 4.0206570756793046e-05,
-      "loss": 63.0074,
-      "step": 90670
-    },
-    {
-      "epoch": 0.36635867435368075,
-      "grad_norm": 369.955810546875,
-      "learning_rate": 4.020379994837164e-05,
-      "loss": 76.6917,
-      "step": 90680
-    },
-    {
-      "epoch": 0.3663990756190484,
-      "grad_norm": 654.5360717773438,
-      "learning_rate": 4.020102884354433e-05,
-      "loss": 71.9556,
-      "step": 90690
-    },
-    {
-      "epoch": 0.366439476884416,
-      "grad_norm": 689.0057983398438,
-      "learning_rate": 4.019825744236514e-05,
-      "loss": 68.2348,
-      "step": 90700
-    },
-    {
-      "epoch": 0.36647987814978367,
-      "grad_norm": 408.1641540527344,
-      "learning_rate": 4.0195485744888096e-05,
-      "loss": 99.4495,
-      "step": 90710
-    },
-    {
-      "epoch": 0.3665202794151513,
-      "grad_norm": 817.236328125,
-      "learning_rate": 4.019271375116722e-05,
-      "loss": 72.3637,
-      "step": 90720
-    },
-    {
-      "epoch": 0.3665606806805189,
-      "grad_norm": 1630.006103515625,
-      "learning_rate": 4.018994146125659e-05,
-      "loss": 85.6407,
-      "step": 90730
-    },
-    {
-      "epoch": 0.36660108194588653,
-      "grad_norm": 1074.2464599609375,
-      "learning_rate": 4.0187168875210216e-05,
-      "loss": 59.7042,
-      "step": 90740
-    },
-    {
-      "epoch": 0.36664148321125417,
-      "grad_norm": 380.92657470703125,
-      "learning_rate": 4.018439599308217e-05,
-      "loss": 81.9409,
-      "step": 90750
-    },
-    {
-      "epoch": 0.3666818844766218,
-      "grad_norm": 0.0,
-      "learning_rate": 4.0181622814926504e-05,
-      "loss": 81.8555,
-      "step": 90760
-    },
-    {
-      "epoch": 0.36672228574198945,
-      "grad_norm": 703.6498413085938,
-      "learning_rate": 4.0178849340797285e-05,
-      "loss": 67.2053,
-      "step": 90770
-    },
-    {
-      "epoch": 0.3667626870073571,
-      "grad_norm": 706.4108276367188,
-      "learning_rate": 4.0176075570748596e-05,
-      "loss": 49.0458,
-      "step": 90780
-    },
-    {
-      "epoch": 0.3668030882727247,
-      "grad_norm": 2707.36572265625,
-      "learning_rate": 4.017330150483449e-05,
-      "loss": 53.4966,
-      "step": 90790
-    },
-    {
-      "epoch": 0.3668434895380923,
-      "grad_norm": 472.0765075683594,
-      "learning_rate": 4.017052714310906e-05,
-      "loss": 38.695,
-      "step": 90800
-    },
-    {
-      "epoch": 0.36688389080345996,
-      "grad_norm": 786.6532592773438,
-      "learning_rate": 4.0167752485626385e-05,
-      "loss": 52.5584,
-      "step": 90810
-    },
-    {
-      "epoch": 0.3669242920688276,
-      "grad_norm": 680.0988159179688,
-      "learning_rate": 4.0164977532440584e-05,
-      "loss": 55.2876,
-      "step": 90820
-    },
-    {
-      "epoch": 0.36696469333419524,
-      "grad_norm": 1092.18359375,
-      "learning_rate": 4.0162202283605725e-05,
-      "loss": 55.0247,
-      "step": 90830
-    },
-    {
-      "epoch": 0.3670050945995629,
-      "grad_norm": 944.074951171875,
-      "learning_rate": 4.015942673917593e-05,
-      "loss": 72.8051,
-      "step": 90840
-    },
-    {
-      "epoch": 0.3670454958649305,
-      "grad_norm": 1217.025634765625,
-      "learning_rate": 4.015665089920531e-05,
-      "loss": 74.4885,
-      "step": 90850
-    },
-    {
-      "epoch": 0.3670858971302981,
-      "grad_norm": 710.3255004882812,
-      "learning_rate": 4.0153874763747976e-05,
-      "loss": 72.1298,
-      "step": 90860
-    },
-    {
-      "epoch": 0.36712629839566574,
-      "grad_norm": 494.13946533203125,
-      "learning_rate": 4.015109833285805e-05,
-      "loss": 66.7202,
-      "step": 90870
-    },
-    {
-      "epoch": 0.3671666996610334,
-      "grad_norm": 693.8568115234375,
-      "learning_rate": 4.0148321606589656e-05,
-      "loss": 60.7179,
-      "step": 90880
-    },
-    {
-      "epoch": 0.367207100926401,
-      "grad_norm": 548.3522338867188,
-      "learning_rate": 4.014554458499694e-05,
-      "loss": 47.7584,
-      "step": 90890
-    },
-    {
-      "epoch": 0.36724750219176866,
-      "grad_norm": 885.053466796875,
-      "learning_rate": 4.014276726813404e-05,
-      "loss": 26.558,
-      "step": 90900
-    },
-    {
-      "epoch": 0.3672879034571363,
-      "grad_norm": 1183.4205322265625,
-      "learning_rate": 4.013998965605509e-05,
-      "loss": 71.821,
-      "step": 90910
-    },
-    {
-      "epoch": 0.3673283047225039,
-      "grad_norm": 1207.58203125,
-      "learning_rate": 4.013721174881425e-05,
-      "loss": 49.7049,
-      "step": 90920
-    },
-    {
-      "epoch": 0.3673687059878715,
-      "grad_norm": 1142.948486328125,
-      "learning_rate": 4.013443354646567e-05,
-      "loss": 74.3073,
-      "step": 90930
-    },
-    {
-      "epoch": 0.36740910725323916,
-      "grad_norm": 0.0,
-      "learning_rate": 4.0131655049063514e-05,
-      "loss": 50.2877,
-      "step": 90940
-    },
-    {
-      "epoch": 0.3674495085186068,
-      "grad_norm": 2414.29541015625,
-      "learning_rate": 4.012887625666195e-05,
-      "loss": 76.8267,
-      "step": 90950
-    },
-    {
-      "epoch": 0.36748990978397444,
-      "grad_norm": 682.6073608398438,
-      "learning_rate": 4.012609716931517e-05,
-      "loss": 46.8848,
-      "step": 90960
-    },
-    {
-      "epoch": 0.3675303110493421,
-      "grad_norm": 648.0629272460938,
-      "learning_rate": 4.012331778707732e-05,
-      "loss": 57.106,
-      "step": 90970
-    },
-    {
-      "epoch": 0.3675707123147097,
-      "grad_norm": 976.9451293945312,
-      "learning_rate": 4.012053811000262e-05,
-      "loss": 56.0116,
-      "step": 90980
-    },
-    {
-      "epoch": 0.3676111135800773,
-      "grad_norm": 712.3170166015625,
-      "learning_rate": 4.0117758138145235e-05,
-      "loss": 72.1889,
-      "step": 90990
-    },
-    {
-      "epoch": 0.36765151484544495,
-      "grad_norm": 579.1261596679688,
-      "learning_rate": 4.011497787155938e-05,
-      "loss": 54.2764,
-      "step": 91000
-    },
-    {
-      "epoch": 0.3676919161108126,
-      "grad_norm": 885.5346069335938,
-      "learning_rate": 4.0112197310299235e-05,
-      "loss": 78.5092,
-      "step": 91010
-    },
-    {
-      "epoch": 0.36773231737618023,
-      "grad_norm": 661.86572265625,
-      "learning_rate": 4.010941645441904e-05,
-      "loss": 68.8041,
-      "step": 91020
-    },
-    {
-      "epoch": 0.36777271864154787,
-      "grad_norm": 474.86328125,
-      "learning_rate": 4.010663530397298e-05,
-      "loss": 62.9406,
-      "step": 91030
-    },
-    {
-      "epoch": 0.3678131199069155,
-      "grad_norm": 476.5716857910156,
-      "learning_rate": 4.01038538590153e-05,
-      "loss": 102.7171,
-      "step": 91040
-    },
-    {
-      "epoch": 0.3678535211722831,
-      "grad_norm": 696.08837890625,
-      "learning_rate": 4.0101072119600196e-05,
-      "loss": 97.6508,
-      "step": 91050
-    },
-    {
-      "epoch": 0.36789392243765073,
-      "grad_norm": 751.9807739257812,
-      "learning_rate": 4.009829008578192e-05,
-      "loss": 85.6849,
-      "step": 91060
-    },
-    {
-      "epoch": 0.3679343237030184,
-      "grad_norm": 505.8662109375,
-      "learning_rate": 4.0095507757614717e-05,
-      "loss": 54.5299,
-      "step": 91070
-    },
-    {
-      "epoch": 0.367974724968386,
-      "grad_norm": 711.6263427734375,
-      "learning_rate": 4.009272513515281e-05,
-      "loss": 58.3999,
-      "step": 91080
-    },
-    {
-      "epoch": 0.36801512623375365,
-      "grad_norm": 623.3084106445312,
-      "learning_rate": 4.008994221845046e-05,
-      "loss": 65.1364,
-      "step": 91090
-    },
-    {
-      "epoch": 0.3680555274991213,
-      "grad_norm": 479.5476379394531,
-      "learning_rate": 4.0087159007561916e-05,
-      "loss": 40.831,
-      "step": 91100
-    },
-    {
-      "epoch": 0.3680959287644889,
-      "grad_norm": 969.3098754882812,
-      "learning_rate": 4.0084375502541446e-05,
-      "loss": 64.9479,
-      "step": 91110
-    },
-    {
-      "epoch": 0.3681363300298565,
-      "grad_norm": 267.7486572265625,
-      "learning_rate": 4.00815917034433e-05,
-      "loss": 107.4876,
-      "step": 91120
-    },
-    {
-      "epoch": 0.36817673129522416,
-      "grad_norm": 1258.0924072265625,
-      "learning_rate": 4.007880761032177e-05,
-      "loss": 63.7383,
-      "step": 91130
-    },
-    {
-      "epoch": 0.3682171325605918,
-      "grad_norm": 664.6405029296875,
-      "learning_rate": 4.0076023223231105e-05,
-      "loss": 65.6308,
-      "step": 91140
-    },
-    {
-      "epoch": 0.36825753382595944,
-      "grad_norm": 869.2908325195312,
-      "learning_rate": 4.007323854222562e-05,
-      "loss": 48.834,
-      "step": 91150
-    },
-    {
-      "epoch": 0.3682979350913271,
-      "grad_norm": 359.8714599609375,
-      "learning_rate": 4.007045356735959e-05,
-      "loss": 51.4937,
-      "step": 91160
-    },
-    {
-      "epoch": 0.3683383363566947,
-      "grad_norm": 454.6327209472656,
-      "learning_rate": 4.00676682986873e-05,
-      "loss": 80.0302,
-      "step": 91170
-    },
-    {
-      "epoch": 0.3683787376220623,
-      "grad_norm": 582.725341796875,
-      "learning_rate": 4.006488273626307e-05,
-      "loss": 81.7921,
-      "step": 91180
-    },
-    {
-      "epoch": 0.36841913888742994,
-      "grad_norm": 998.3577880859375,
-      "learning_rate": 4.006209688014119e-05,
-      "loss": 40.4945,
-      "step": 91190
-    },
-    {
-      "epoch": 0.3684595401527976,
-      "grad_norm": 951.7125854492188,
-      "learning_rate": 4.005931073037596e-05,
-      "loss": 54.5525,
-      "step": 91200
-    },
-    {
-      "epoch": 0.3684999414181652,
-      "grad_norm": 517.168701171875,
-      "learning_rate": 4.005652428702173e-05,
-      "loss": 52.3269,
-      "step": 91210
-    },
-    {
-      "epoch": 0.36854034268353286,
-      "grad_norm": 982.7772216796875,
-      "learning_rate": 4.0053737550132816e-05,
-      "loss": 65.8004,
-      "step": 91220
-    },
-    {
-      "epoch": 0.3685807439489005,
-      "grad_norm": 0.0,
-      "learning_rate": 4.005095051976353e-05,
-      "loss": 58.1448,
-      "step": 91230
-    },
-    {
-      "epoch": 0.3686211452142681,
-      "grad_norm": 1081.756103515625,
-      "learning_rate": 4.0048163195968214e-05,
-      "loss": 51.2047,
-      "step": 91240
-    },
-    {
-      "epoch": 0.3686615464796357,
-      "grad_norm": 837.7747802734375,
-      "learning_rate": 4.0045375578801214e-05,
-      "loss": 66.0523,
-      "step": 91250
-    },
-    {
-      "epoch": 0.36870194774500337,
-      "grad_norm": 316.6372375488281,
-      "learning_rate": 4.004258766831686e-05,
-      "loss": 64.1896,
-      "step": 91260
-    },
-    {
-      "epoch": 0.368742349010371,
-      "grad_norm": 820.6416015625,
-      "learning_rate": 4.0039799464569524e-05,
-      "loss": 83.5785,
-      "step": 91270
-    },
-    {
-      "epoch": 0.36878275027573865,
-      "grad_norm": 1047.304443359375,
-      "learning_rate": 4.003701096761355e-05,
-      "loss": 64.335,
-      "step": 91280
-    },
-    {
-      "epoch": 0.3688231515411063,
-      "grad_norm": 882.9609375,
-      "learning_rate": 4.0034222177503314e-05,
-      "loss": 74.856,
-      "step": 91290
-    },
-    {
-      "epoch": 0.36886355280647387,
-      "grad_norm": 758.5153198242188,
-      "learning_rate": 4.003143309429317e-05,
-      "loss": 57.2044,
-      "step": 91300
-    },
-    {
-      "epoch": 0.3689039540718415,
-      "grad_norm": 637.7432861328125,
-      "learning_rate": 4.0028643718037496e-05,
-      "loss": 67.143,
-      "step": 91310
-    },
-    {
-      "epoch": 0.36894435533720915,
-      "grad_norm": 982.8681640625,
-      "learning_rate": 4.0025854048790677e-05,
-      "loss": 56.672,
-      "step": 91320
-    },
-    {
-      "epoch": 0.3689847566025768,
-      "grad_norm": 975.6199951171875,
-      "learning_rate": 4.00230640866071e-05,
-      "loss": 77.8675,
-      "step": 91330
-    },
-    {
-      "epoch": 0.36902515786794443,
-      "grad_norm": 463.59063720703125,
-      "learning_rate": 4.0020273831541155e-05,
-      "loss": 47.0041,
-      "step": 91340
-    },
-    {
-      "epoch": 0.36906555913331207,
-      "grad_norm": 1927.79541015625,
-      "learning_rate": 4.001748328364724e-05,
-      "loss": 100.6176,
-      "step": 91350
-    },
-    {
-      "epoch": 0.3691059603986797,
-      "grad_norm": 614.8819580078125,
-      "learning_rate": 4.001469244297975e-05,
-      "loss": 63.3884,
-      "step": 91360
-    },
-    {
-      "epoch": 0.3691463616640473,
-      "grad_norm": 578.9991455078125,
-      "learning_rate": 4.00119013095931e-05,
-      "loss": 38.7527,
-      "step": 91370
-    },
-    {
-      "epoch": 0.36918676292941494,
-      "grad_norm": 424.689697265625,
-      "learning_rate": 4.0009109883541715e-05,
-      "loss": 71.5254,
-      "step": 91380
-    },
-    {
-      "epoch": 0.3692271641947826,
-      "grad_norm": 592.7898559570312,
-      "learning_rate": 4.000631816488001e-05,
-      "loss": 88.0966,
-      "step": 91390
-    },
-    {
-      "epoch": 0.3692675654601502,
-      "grad_norm": 699.6017456054688,
-      "learning_rate": 4.000352615366239e-05,
-      "loss": 65.126,
-      "step": 91400
-    },
-    {
-      "epoch": 0.36930796672551786,
-      "grad_norm": 3042.03662109375,
-      "learning_rate": 4.0000733849943313e-05,
-      "loss": 60.7142,
-      "step": 91410
-    },
-    {
-      "epoch": 0.3693483679908855,
-      "grad_norm": 799.2810668945312,
-      "learning_rate": 3.999794125377721e-05,
-      "loss": 36.586,
-      "step": 91420
-    },
-    {
-      "epoch": 0.3693887692562531,
-      "grad_norm": 968.9716796875,
-      "learning_rate": 3.999514836521851e-05,
-      "loss": 69.6768,
-      "step": 91430
-    },
-    {
-      "epoch": 0.3694291705216207,
-      "grad_norm": 1178.7864990234375,
-      "learning_rate": 3.999235518432168e-05,
-      "loss": 61.2758,
-      "step": 91440
-    },
-    {
-      "epoch": 0.36946957178698836,
-      "grad_norm": 490.9415283203125,
-      "learning_rate": 3.998956171114116e-05,
-      "loss": 92.8936,
-      "step": 91450
-    },
-    {
-      "epoch": 0.369509973052356,
-      "grad_norm": 429.26019287109375,
-      "learning_rate": 3.998676794573142e-05,
-      "loss": 54.8671,
-      "step": 91460
-    },
-    {
-      "epoch": 0.36955037431772364,
-      "grad_norm": 556.7578125,
-      "learning_rate": 3.998397388814693e-05,
-      "loss": 50.8339,
-      "step": 91470
-    },
-    {
-      "epoch": 0.3695907755830913,
-      "grad_norm": 745.5537719726562,
-      "learning_rate": 3.9981179538442146e-05,
-      "loss": 75.3068,
-      "step": 91480
-    },
-    {
-      "epoch": 0.3696311768484589,
-      "grad_norm": 1237.29638671875,
-      "learning_rate": 3.9978384896671564e-05,
-      "loss": 71.3501,
-      "step": 91490
-    },
-    {
-      "epoch": 0.3696715781138265,
-      "grad_norm": 846.2611083984375,
-      "learning_rate": 3.997558996288965e-05,
-      "loss": 68.3017,
-      "step": 91500
-    },
-    {
-      "epoch": 0.36971197937919414,
-      "grad_norm": 1087.683837890625,
-      "learning_rate": 3.9972794737150895e-05,
-      "loss": 89.0834,
-      "step": 91510
-    },
-    {
-      "epoch": 0.3697523806445618,
-      "grad_norm": 462.5960998535156,
-      "learning_rate": 3.996999921950981e-05,
-      "loss": 72.2417,
-      "step": 91520
-    },
-    {
-      "epoch": 0.3697927819099294,
-      "grad_norm": 623.9275512695312,
-      "learning_rate": 3.9967203410020875e-05,
-      "loss": 79.5964,
-      "step": 91530
-    },
-    {
-      "epoch": 0.36983318317529706,
-      "grad_norm": 2321.183837890625,
-      "learning_rate": 3.99644073087386e-05,
-      "loss": 100.5441,
-      "step": 91540
-    },
-    {
-      "epoch": 0.3698735844406647,
-      "grad_norm": 931.0929565429688,
-      "learning_rate": 3.9961610915717515e-05,
-      "loss": 69.4745,
-      "step": 91550
-    },
-    {
-      "epoch": 0.3699139857060323,
-      "grad_norm": 459.2430419921875,
-      "learning_rate": 3.9958814231012115e-05,
-      "loss": 48.2317,
-      "step": 91560
-    },
-    {
-      "epoch": 0.36995438697139993,
-      "grad_norm": 2218.796630859375,
-      "learning_rate": 3.9956017254676923e-05,
-      "loss": 73.4239,
-      "step": 91570
-    },
-    {
-      "epoch": 0.36999478823676757,
-      "grad_norm": 579.8865966796875,
-      "learning_rate": 3.995321998676648e-05,
-      "loss": 69.1222,
-      "step": 91580
-    },
-    {
-      "epoch": 0.3700351895021352,
-      "grad_norm": 805.66943359375,
-      "learning_rate": 3.995042242733532e-05,
-      "loss": 40.6279,
-      "step": 91590
-    },
-    {
-      "epoch": 0.37007559076750285,
-      "grad_norm": 1477.2010498046875,
-      "learning_rate": 3.9947624576437975e-05,
-      "loss": 54.183,
-      "step": 91600
-    },
-    {
-      "epoch": 0.3701159920328705,
-      "grad_norm": 1422.080078125,
-      "learning_rate": 3.994482643412899e-05,
-      "loss": 44.8291,
-      "step": 91610
-    },
-    {
-      "epoch": 0.3701563932982381,
-      "grad_norm": 596.1489868164062,
-      "learning_rate": 3.994202800046292e-05,
-      "loss": 50.5651,
-      "step": 91620
-    },
-    {
-      "epoch": 0.3701967945636057,
-      "grad_norm": 697.937744140625,
-      "learning_rate": 3.993922927549432e-05,
-      "loss": 51.1245,
-      "step": 91630
-    },
-    {
-      "epoch": 0.37023719582897335,
-      "grad_norm": 660.627197265625,
-      "learning_rate": 3.993643025927776e-05,
-      "loss": 73.9116,
-      "step": 91640
-    },
-    {
-      "epoch": 0.370277597094341,
-      "grad_norm": 735.249267578125,
-      "learning_rate": 3.9933630951867805e-05,
-      "loss": 41.1959,
-      "step": 91650
-    },
-    {
-      "epoch": 0.37031799835970863,
-      "grad_norm": 1717.015380859375,
-      "learning_rate": 3.9930831353319023e-05,
-      "loss": 66.5677,
-      "step": 91660
-    },
-    {
-      "epoch": 0.3703583996250763,
-      "grad_norm": 628.260009765625,
-      "learning_rate": 3.992803146368599e-05,
-      "loss": 50.1199,
-      "step": 91670
-    },
-    {
-      "epoch": 0.3703988008904439,
-      "grad_norm": 1197.982666015625,
-      "learning_rate": 3.99252312830233e-05,
-      "loss": 53.4161,
-      "step": 91680
-    },
-    {
-      "epoch": 0.3704392021558115,
-      "grad_norm": 759.8359375,
-      "learning_rate": 3.992243081138555e-05,
-      "loss": 49.2217,
-      "step": 91690
-    },
-    {
-      "epoch": 0.37047960342117914,
-      "grad_norm": 881.2600708007812,
-      "learning_rate": 3.9919630048827314e-05,
-      "loss": 75.7815,
-      "step": 91700
-    },
-    {
-      "epoch": 0.3705200046865468,
-      "grad_norm": 356.75958251953125,
-      "learning_rate": 3.991682899540322e-05,
-      "loss": 67.6897,
-      "step": 91710
-    },
-    {
-      "epoch": 0.3705604059519144,
-      "grad_norm": 1126.136474609375,
-      "learning_rate": 3.9914027651167866e-05,
-      "loss": 67.9846,
-      "step": 91720
-    },
-    {
-      "epoch": 0.37060080721728206,
-      "grad_norm": 467.5633544921875,
-      "learning_rate": 3.9911226016175866e-05,
-      "loss": 77.1551,
-      "step": 91730
-    },
-    {
-      "epoch": 0.3706412084826497,
-      "grad_norm": 590.6211547851562,
-      "learning_rate": 3.990842409048183e-05,
-      "loss": 52.3267,
-      "step": 91740
-    },
-    {
-      "epoch": 0.3706816097480173,
-      "grad_norm": 2134.7158203125,
-      "learning_rate": 3.99056218741404e-05,
-      "loss": 89.5028,
-      "step": 91750
-    },
-    {
-      "epoch": 0.3707220110133849,
-      "grad_norm": 545.798583984375,
-      "learning_rate": 3.990281936720619e-05,
-      "loss": 59.62,
-      "step": 91760
-    },
-    {
-      "epoch": 0.37076241227875256,
-      "grad_norm": 2029.9058837890625,
-      "learning_rate": 3.990001656973385e-05,
-      "loss": 59.6643,
-      "step": 91770
-    },
-    {
-      "epoch": 0.3708028135441202,
-      "grad_norm": 413.22491455078125,
-      "learning_rate": 3.9897213481778006e-05,
-      "loss": 48.1997,
-      "step": 91780
-    },
-    {
-      "epoch": 0.37084321480948784,
-      "grad_norm": 1251.12744140625,
-      "learning_rate": 3.9894410103393323e-05,
-      "loss": 59.8171,
-      "step": 91790
-    },
-    {
-      "epoch": 0.3708836160748555,
-      "grad_norm": 520.3812866210938,
-      "learning_rate": 3.989160643463445e-05,
-      "loss": 71.2307,
-      "step": 91800
-    },
-    {
-      "epoch": 0.3709240173402231,
-      "grad_norm": 866.8963012695312,
-      "learning_rate": 3.988880247555604e-05,
-      "loss": 102.0608,
-      "step": 91810
-    },
-    {
-      "epoch": 0.3709644186055907,
-      "grad_norm": 546.447998046875,
-      "learning_rate": 3.9885998226212764e-05,
-      "loss": 48.1409,
-      "step": 91820
-    },
-    {
-      "epoch": 0.37100481987095835,
-      "grad_norm": 1045.4818115234375,
-      "learning_rate": 3.988319368665928e-05,
-      "loss": 96.8548,
-      "step": 91830
-    },
-    {
-      "epoch": 0.371045221136326,
-      "grad_norm": 1263.72314453125,
-      "learning_rate": 3.988038885695028e-05,
-      "loss": 51.555,
-      "step": 91840
-    },
-    {
-      "epoch": 0.3710856224016936,
-      "grad_norm": 929.8193359375,
-      "learning_rate": 3.987758373714044e-05,
-      "loss": 48.9912,
-      "step": 91850
-    },
-    {
-      "epoch": 0.37112602366706127,
-      "grad_norm": 1874.9107666015625,
-      "learning_rate": 3.987477832728444e-05,
-      "loss": 48.1193,
-      "step": 91860
-    },
-    {
-      "epoch": 0.3711664249324289,
-      "grad_norm": 449.92388916015625,
-      "learning_rate": 3.987197262743697e-05,
-      "loss": 60.4172,
-      "step": 91870
-    },
-    {
-      "epoch": 0.3712068261977965,
-      "grad_norm": 3075.2314453125,
-      "learning_rate": 3.986916663765275e-05,
-      "loss": 66.7121,
-      "step": 91880
-    },
-    {
-      "epoch": 0.37124722746316413,
-      "grad_norm": 805.412841796875,
-      "learning_rate": 3.9866360357986467e-05,
-      "loss": 60.1465,
-      "step": 91890
-    },
-    {
-      "epoch": 0.37128762872853177,
-      "grad_norm": 925.1229248046875,
-      "learning_rate": 3.9863553788492834e-05,
-      "loss": 40.3686,
-      "step": 91900
-    },
-    {
-      "epoch": 0.3713280299938994,
-      "grad_norm": 600.6539306640625,
-      "learning_rate": 3.9860746929226567e-05,
-      "loss": 66.7164,
-      "step": 91910
-    },
-    {
-      "epoch": 0.37136843125926705,
-      "grad_norm": 1517.1585693359375,
-      "learning_rate": 3.985793978024239e-05,
-      "loss": 42.1975,
-      "step": 91920
-    },
-    {
-      "epoch": 0.3714088325246347,
-      "grad_norm": 1709.787353515625,
-      "learning_rate": 3.985513234159502e-05,
-      "loss": 61.1718,
-      "step": 91930
-    },
-    {
-      "epoch": 0.3714492337900023,
-      "grad_norm": 610.5380249023438,
-      "learning_rate": 3.985232461333921e-05,
-      "loss": 66.061,
-      "step": 91940
-    },
-    {
-      "epoch": 0.3714896350553699,
-      "grad_norm": 1254.192138671875,
-      "learning_rate": 3.984951659552968e-05,
-      "loss": 61.3981,
-      "step": 91950
-    },
-    {
-      "epoch": 0.37153003632073756,
-      "grad_norm": 725.9743041992188,
-      "learning_rate": 3.984670828822118e-05,
-      "loss": 50.754,
-      "step": 91960
-    },
-    {
-      "epoch": 0.3715704375861052,
-      "grad_norm": 898.7955322265625,
-      "learning_rate": 3.984389969146846e-05,
-      "loss": 80.66,
-      "step": 91970
-    },
-    {
-      "epoch": 0.37161083885147284,
-      "grad_norm": 1205.29345703125,
-      "learning_rate": 3.9841090805326264e-05,
-      "loss": 88.857,
-      "step": 91980
-    },
-    {
-      "epoch": 0.3716512401168405,
-      "grad_norm": 304.8878479003906,
-      "learning_rate": 3.983828162984937e-05,
-      "loss": 39.1823,
-      "step": 91990
-    },
-    {
-      "epoch": 0.3716916413822081,
-      "grad_norm": 474.86700439453125,
-      "learning_rate": 3.983547216509254e-05,
-      "loss": 55.8331,
-      "step": 92000
-    },
-    {
-      "epoch": 0.3717320426475757,
-      "grad_norm": 401.64044189453125,
-      "learning_rate": 3.9832662411110536e-05,
-      "loss": 69.07,
-      "step": 92010
-    },
-    {
-      "epoch": 0.37177244391294334,
-      "grad_norm": 338.5098571777344,
-      "learning_rate": 3.9829852367958144e-05,
-      "loss": 49.6506,
-      "step": 92020
-    },
-    {
-      "epoch": 0.371812845178311,
-      "grad_norm": 616.5515747070312,
-      "learning_rate": 3.9827042035690145e-05,
-      "loss": 55.922,
-      "step": 92030
-    },
-    {
-      "epoch": 0.3718532464436786,
-      "grad_norm": 696.16357421875,
-      "learning_rate": 3.9824231414361324e-05,
-      "loss": 56.3399,
-      "step": 92040
-    },
-    {
-      "epoch": 0.37189364770904626,
-      "grad_norm": 555.5895385742188,
-      "learning_rate": 3.982142050402649e-05,
-      "loss": 57.3369,
-      "step": 92050
-    },
-    {
-      "epoch": 0.3719340489744139,
-      "grad_norm": 777.5509033203125,
-      "learning_rate": 3.9818609304740414e-05,
-      "loss": 72.6481,
-      "step": 92060
-    },
-    {
-      "epoch": 0.3719744502397815,
-      "grad_norm": 752.47705078125,
-      "learning_rate": 3.981579781655794e-05,
-      "loss": 71.8169,
-      "step": 92070
-    },
-    {
-      "epoch": 0.3720148515051491,
-      "grad_norm": 973.05419921875,
-      "learning_rate": 3.981298603953385e-05,
-      "loss": 84.1057,
-      "step": 92080
-    },
-    {
-      "epoch": 0.37205525277051676,
-      "grad_norm": 956.1757202148438,
-      "learning_rate": 3.9810173973722974e-05,
-      "loss": 57.5237,
-      "step": 92090
-    },
-    {
-      "epoch": 0.3720956540358844,
-      "grad_norm": 606.1538696289062,
-      "learning_rate": 3.980736161918013e-05,
-      "loss": 51.6493,
-      "step": 92100
-    },
-    {
-      "epoch": 0.37213605530125204,
-      "grad_norm": 401.66314697265625,
-      "learning_rate": 3.980454897596014e-05,
-      "loss": 74.039,
-      "step": 92110
-    },
-    {
-      "epoch": 0.3721764565666197,
-      "grad_norm": 469.2101745605469,
-      "learning_rate": 3.980173604411786e-05,
-      "loss": 62.7403,
-      "step": 92120
-    },
-    {
-      "epoch": 0.3722168578319873,
-      "grad_norm": 814.7521362304688,
-      "learning_rate": 3.979892282370811e-05,
-      "loss": 93.9603,
-      "step": 92130
-    },
-    {
-      "epoch": 0.3722572590973549,
-      "grad_norm": 1050.54296875,
-      "learning_rate": 3.979610931478574e-05,
-      "loss": 79.703,
-      "step": 92140
-    },
-    {
-      "epoch": 0.37229766036272255,
-      "grad_norm": 426.4966735839844,
-      "learning_rate": 3.97932955174056e-05,
-      "loss": 74.3453,
-      "step": 92150
-    },
-    {
-      "epoch": 0.3723380616280902,
-      "grad_norm": 587.7415161132812,
-      "learning_rate": 3.979048143162255e-05,
-      "loss": 58.7905,
-      "step": 92160
-    },
-    {
-      "epoch": 0.37237846289345783,
-      "grad_norm": 1028.9864501953125,
-      "learning_rate": 3.978766705749145e-05,
-      "loss": 49.0045,
-      "step": 92170
-    },
-    {
-      "epoch": 0.37241886415882547,
-      "grad_norm": 722.3172607421875,
-      "learning_rate": 3.9784852395067166e-05,
-      "loss": 50.1978,
-      "step": 92180
-    },
-    {
-      "epoch": 0.3724592654241931,
-      "grad_norm": 1028.6400146484375,
-      "learning_rate": 3.978203744440457e-05,
-      "loss": 65.8073,
-      "step": 92190
-    },
-    {
-      "epoch": 0.3724996666895607,
-      "grad_norm": 350.545654296875,
-      "learning_rate": 3.977922220555855e-05,
-      "loss": 66.0355,
-      "step": 92200
-    },
-    {
-      "epoch": 0.37254006795492833,
-      "grad_norm": 888.4639282226562,
-      "learning_rate": 3.977640667858398e-05,
-      "loss": 61.8959,
-      "step": 92210
-    },
-    {
-      "epoch": 0.372580469220296,
-      "grad_norm": 938.3932495117188,
-      "learning_rate": 3.977359086353576e-05,
-      "loss": 69.4164,
-      "step": 92220
-    },
-    {
-      "epoch": 0.3726208704856636,
-      "grad_norm": 1005.269775390625,
-      "learning_rate": 3.977077476046877e-05,
-      "loss": 65.6192,
-      "step": 92230
-    },
-    {
-      "epoch": 0.37266127175103125,
-      "grad_norm": 557.5337524414062,
-      "learning_rate": 3.976795836943793e-05,
-      "loss": 59.2249,
-      "step": 92240
-    },
-    {
-      "epoch": 0.3727016730163989,
-      "grad_norm": 2857.20703125,
-      "learning_rate": 3.976514169049814e-05,
-      "loss": 68.5465,
-      "step": 92250
-    },
-    {
-      "epoch": 0.3727420742817665,
-      "grad_norm": 1169.2440185546875,
-      "learning_rate": 3.97623247237043e-05,
-      "loss": 46.9372,
-      "step": 92260
-    },
-    {
-      "epoch": 0.3727824755471341,
-      "grad_norm": 561.6387329101562,
-      "learning_rate": 3.9759507469111346e-05,
-      "loss": 59.3901,
-      "step": 92270
-    },
-    {
-      "epoch": 0.37282287681250176,
-      "grad_norm": 835.4718017578125,
-      "learning_rate": 3.9756689926774196e-05,
-      "loss": 85.8709,
-      "step": 92280
-    },
-    {
-      "epoch": 0.3728632780778694,
-      "grad_norm": 789.12841796875,
-      "learning_rate": 3.975387209674778e-05,
-      "loss": 71.775,
-      "step": 92290
-    },
-    {
-      "epoch": 0.37290367934323704,
-      "grad_norm": 0.0,
-      "learning_rate": 3.9751053979087035e-05,
-      "loss": 47.8555,
-      "step": 92300
-    },
-    {
-      "epoch": 0.3729440806086047,
-      "grad_norm": 776.5151977539062,
-      "learning_rate": 3.9748235573846894e-05,
-      "loss": 70.7797,
-      "step": 92310
-    },
-    {
-      "epoch": 0.3729844818739723,
-      "grad_norm": 987.219482421875,
-      "learning_rate": 3.97454168810823e-05,
-      "loss": 74.3305,
-      "step": 92320
-    },
-    {
-      "epoch": 0.3730248831393399,
-      "grad_norm": 508.5974426269531,
-      "learning_rate": 3.974259790084822e-05,
-      "loss": 80.4925,
-      "step": 92330
-    },
-    {
-      "epoch": 0.37306528440470754,
-      "grad_norm": 784.2496948242188,
-      "learning_rate": 3.973977863319961e-05,
-      "loss": 75.617,
-      "step": 92340
-    },
-    {
-      "epoch": 0.3731056856700752,
-      "grad_norm": 418.7388916015625,
-      "learning_rate": 3.973695907819142e-05,
-      "loss": 49.6771,
-      "step": 92350
-    },
-    {
-      "epoch": 0.3731460869354428,
-      "grad_norm": 1385.4676513671875,
-      "learning_rate": 3.973413923587862e-05,
-      "loss": 54.8952,
-      "step": 92360
-    },
-    {
-      "epoch": 0.37318648820081046,
-      "grad_norm": 1004.26611328125,
-      "learning_rate": 3.97313191063162e-05,
-      "loss": 66.2516,
-      "step": 92370
-    },
-    {
-      "epoch": 0.3732268894661781,
-      "grad_norm": 331.77972412109375,
-      "learning_rate": 3.9728498689559126e-05,
-      "loss": 62.8471,
-      "step": 92380
-    },
-    {
-      "epoch": 0.3732672907315457,
-      "grad_norm": 799.3560180664062,
-      "learning_rate": 3.972567798566238e-05,
-      "loss": 76.5567,
-      "step": 92390
-    },
-    {
-      "epoch": 0.3733076919969133,
-      "grad_norm": 733.6048583984375,
-      "learning_rate": 3.9722856994680966e-05,
-      "loss": 60.599,
-      "step": 92400
-    },
-    {
-      "epoch": 0.37334809326228097,
-      "grad_norm": 610.0665893554688,
-      "learning_rate": 3.9720035716669876e-05,
-      "loss": 58.7812,
-      "step": 92410
-    },
-    {
-      "epoch": 0.3733884945276486,
-      "grad_norm": 757.6383666992188,
-      "learning_rate": 3.971721415168411e-05,
-      "loss": 72.6537,
-      "step": 92420
-    },
-    {
-      "epoch": 0.37342889579301625,
-      "grad_norm": 628.03515625,
-      "learning_rate": 3.971439229977869e-05,
-      "loss": 58.028,
-      "step": 92430
-    },
-    {
-      "epoch": 0.3734692970583839,
-      "grad_norm": 567.295654296875,
-      "learning_rate": 3.9711570161008596e-05,
-      "loss": 54.5905,
-      "step": 92440
-    },
-    {
-      "epoch": 0.3735096983237515,
-      "grad_norm": 453.08258056640625,
-      "learning_rate": 3.9708747735428886e-05,
-      "loss": 53.7484,
-      "step": 92450
-    },
-    {
-      "epoch": 0.3735500995891191,
-      "grad_norm": 426.8596496582031,
-      "learning_rate": 3.9705925023094554e-05,
-      "loss": 73.5093,
-      "step": 92460
-    },
-    {
-      "epoch": 0.37359050085448675,
-      "grad_norm": 372.20684814453125,
-      "learning_rate": 3.970310202406064e-05,
-      "loss": 44.4281,
-      "step": 92470
-    },
-    {
-      "epoch": 0.3736309021198544,
-      "grad_norm": 714.4765014648438,
-      "learning_rate": 3.970027873838219e-05,
-      "loss": 78.6218,
-      "step": 92480
-    },
-    {
-      "epoch": 0.37367130338522203,
-      "grad_norm": 709.8063354492188,
-      "learning_rate": 3.969745516611424e-05,
-      "loss": 55.8991,
-      "step": 92490
-    },
-    {
-      "epoch": 0.37371170465058967,
-      "grad_norm": 840.8602294921875,
-      "learning_rate": 3.969463130731183e-05,
-      "loss": 68.27,
-      "step": 92500
-    },
-    {
-      "epoch": 0.3737521059159573,
-      "grad_norm": 1879.0968017578125,
-      "learning_rate": 3.969180716203002e-05,
-      "loss": 98.6429,
-      "step": 92510
-    },
-    {
-      "epoch": 0.3737925071813249,
-      "grad_norm": 925.2473754882812,
-      "learning_rate": 3.9688982730323865e-05,
-      "loss": 74.3842,
-      "step": 92520
-    },
-    {
-      "epoch": 0.37383290844669254,
-      "grad_norm": 1487.61962890625,
-      "learning_rate": 3.968615801224843e-05,
-      "loss": 81.2509,
-      "step": 92530
-    },
-    {
-      "epoch": 0.3738733097120602,
-      "grad_norm": 585.5298461914062,
-      "learning_rate": 3.968333300785878e-05,
-      "loss": 89.9086,
-      "step": 92540
-    },
-    {
-      "epoch": 0.3739137109774278,
-      "grad_norm": 598.8243408203125,
-      "learning_rate": 3.968050771720999e-05,
-      "loss": 52.4405,
-      "step": 92550
-    },
-    {
-      "epoch": 0.37395411224279546,
-      "grad_norm": 451.022705078125,
-      "learning_rate": 3.967768214035715e-05,
-      "loss": 71.7542,
-      "step": 92560
-    },
-    {
-      "epoch": 0.3739945135081631,
-      "grad_norm": 768.9769287109375,
-      "learning_rate": 3.967485627735534e-05,
-      "loss": 92.9239,
-      "step": 92570
-    },
-    {
-      "epoch": 0.3740349147735307,
-      "grad_norm": 854.9164428710938,
-      "learning_rate": 3.967203012825965e-05,
-      "loss": 71.4999,
-      "step": 92580
-    },
-    {
-      "epoch": 0.3740753160388983,
-      "grad_norm": 471.7916259765625,
-      "learning_rate": 3.966920369312518e-05,
-      "loss": 47.2625,
-      "step": 92590
-    },
-    {
-      "epoch": 0.37411571730426596,
-      "grad_norm": 478.0005187988281,
-      "learning_rate": 3.966637697200703e-05,
-      "loss": 56.9574,
-      "step": 92600
-    },
-    {
-      "epoch": 0.3741561185696336,
-      "grad_norm": 968.0185546875,
-      "learning_rate": 3.9663549964960314e-05,
-      "loss": 62.3624,
-      "step": 92610
-    },
-    {
-      "epoch": 0.37419651983500124,
-      "grad_norm": 867.2205810546875,
-      "learning_rate": 3.966072267204014e-05,
-      "loss": 33.5701,
-      "step": 92620
-    },
-    {
-      "epoch": 0.3742369211003689,
-      "grad_norm": 941.1777954101562,
-      "learning_rate": 3.965789509330163e-05,
-      "loss": 78.525,
-      "step": 92630
-    },
-    {
-      "epoch": 0.3742773223657365,
-      "grad_norm": 612.6570434570312,
-      "learning_rate": 3.965506722879991e-05,
-      "loss": 40.913,
-      "step": 92640
-    },
-    {
-      "epoch": 0.3743177236311041,
-      "grad_norm": 127.48741149902344,
-      "learning_rate": 3.965223907859011e-05,
-      "loss": 63.8331,
-      "step": 92650
-    },
-    {
-      "epoch": 0.37435812489647174,
-      "grad_norm": 443.2701721191406,
-      "learning_rate": 3.964941064272736e-05,
-      "loss": 71.8824,
-      "step": 92660
-    },
-    {
-      "epoch": 0.3743985261618394,
-      "grad_norm": 431.3937072753906,
-      "learning_rate": 3.9646581921266815e-05,
-      "loss": 74.9664,
-      "step": 92670
-    },
-    {
-      "epoch": 0.374438927427207,
-      "grad_norm": 852.2191772460938,
-      "learning_rate": 3.964375291426361e-05,
-      "loss": 51.0597,
-      "step": 92680
-    },
-    {
-      "epoch": 0.37447932869257466,
-      "grad_norm": 493.4492492675781,
-      "learning_rate": 3.96409236217729e-05,
-      "loss": 39.8483,
-      "step": 92690
-    },
-    {
-      "epoch": 0.3745197299579423,
-      "grad_norm": 363.88836669921875,
-      "learning_rate": 3.963809404384985e-05,
-      "loss": 44.7268,
-      "step": 92700
-    },
-    {
-      "epoch": 0.3745601312233099,
-      "grad_norm": 2506.282470703125,
-      "learning_rate": 3.9635264180549624e-05,
-      "loss": 79.4526,
-      "step": 92710
-    },
-    {
-      "epoch": 0.37460053248867753,
-      "grad_norm": 1399.570556640625,
-      "learning_rate": 3.963243403192739e-05,
-      "loss": 62.9623,
-      "step": 92720
-    },
-    {
-      "epoch": 0.37464093375404517,
-      "grad_norm": 502.1692810058594,
-      "learning_rate": 3.9629603598038314e-05,
-      "loss": 73.1174,
-      "step": 92730
-    },
-    {
-      "epoch": 0.3746813350194128,
-      "grad_norm": 478.8690185546875,
-      "learning_rate": 3.962677287893758e-05,
-      "loss": 66.815,
-      "step": 92740
-    },
-    {
-      "epoch": 0.37472173628478045,
-      "grad_norm": 563.9537963867188,
-      "learning_rate": 3.962394187468039e-05,
-      "loss": 72.0108,
-      "step": 92750
-    },
-    {
-      "epoch": 0.3747621375501481,
-      "grad_norm": 761.1668090820312,
-      "learning_rate": 3.962111058532192e-05,
-      "loss": 42.6949,
-      "step": 92760
-    },
-    {
-      "epoch": 0.37480253881551573,
-      "grad_norm": 795.1704711914062,
-      "learning_rate": 3.961827901091737e-05,
-      "loss": 58.1634,
-      "step": 92770
-    },
-    {
-      "epoch": 0.3748429400808833,
-      "grad_norm": 811.5784301757812,
-      "learning_rate": 3.9615447151521945e-05,
-      "loss": 75.2723,
-      "step": 92780
-    },
-    {
-      "epoch": 0.37488334134625095,
-      "grad_norm": 444.902099609375,
-      "learning_rate": 3.961261500719085e-05,
-      "loss": 62.2803,
-      "step": 92790
-    },
-    {
-      "epoch": 0.3749237426116186,
-      "grad_norm": 530.8676147460938,
-      "learning_rate": 3.960978257797931e-05,
-      "loss": 62.1482,
-      "step": 92800
-    },
-    {
-      "epoch": 0.37496414387698623,
-      "grad_norm": 230.55746459960938,
-      "learning_rate": 3.9606949863942526e-05,
-      "loss": 60.5643,
-      "step": 92810
-    },
-    {
-      "epoch": 0.3750045451423539,
-      "grad_norm": 656.6939697265625,
-      "learning_rate": 3.960411686513574e-05,
-      "loss": 49.9396,
-      "step": 92820
-    },
-    {
-      "epoch": 0.3750449464077215,
-      "grad_norm": 1161.34765625,
-      "learning_rate": 3.960128358161418e-05,
-      "loss": 50.9719,
-      "step": 92830
-    },
-    {
-      "epoch": 0.3750853476730891,
-      "grad_norm": 768.6620483398438,
-      "learning_rate": 3.9598450013433075e-05,
-      "loss": 68.8406,
-      "step": 92840
-    },
-    {
-      "epoch": 0.37512574893845674,
-      "grad_norm": 512.8631591796875,
-      "learning_rate": 3.9595616160647674e-05,
-      "loss": 61.0759,
-      "step": 92850
-    },
-    {
-      "epoch": 0.3751661502038244,
-      "grad_norm": 341.27178955078125,
-      "learning_rate": 3.959278202331322e-05,
-      "loss": 50.0951,
-      "step": 92860
-    },
-    {
-      "epoch": 0.375206551469192,
-      "grad_norm": 1139.0494384765625,
-      "learning_rate": 3.9589947601484974e-05,
-      "loss": 72.3378,
-      "step": 92870
-    },
-    {
-      "epoch": 0.37524695273455966,
-      "grad_norm": 703.4913940429688,
-      "learning_rate": 3.9587112895218184e-05,
-      "loss": 57.7523,
-      "step": 92880
-    },
-    {
-      "epoch": 0.3752873539999273,
-      "grad_norm": 1080.4427490234375,
-      "learning_rate": 3.958427790456811e-05,
-      "loss": 68.2021,
-      "step": 92890
-    },
-    {
-      "epoch": 0.3753277552652949,
-      "grad_norm": 669.9296264648438,
-      "learning_rate": 3.958144262959004e-05,
-      "loss": 71.9688,
-      "step": 92900
-    },
-    {
-      "epoch": 0.3753681565306625,
-      "grad_norm": 945.6260986328125,
-      "learning_rate": 3.9578607070339235e-05,
-      "loss": 72.3592,
-      "step": 92910
-    },
-    {
-      "epoch": 0.37540855779603016,
-      "grad_norm": 310.77203369140625,
-      "learning_rate": 3.957577122687098e-05,
-      "loss": 53.5487,
-      "step": 92920
-    },
-    {
-      "epoch": 0.3754489590613978,
-      "grad_norm": 602.9820556640625,
-      "learning_rate": 3.957293509924056e-05,
-      "loss": 61.1102,
-      "step": 92930
-    },
-    {
-      "epoch": 0.37548936032676544,
-      "grad_norm": 257.7418518066406,
-      "learning_rate": 3.9570098687503274e-05,
-      "loss": 56.4356,
-      "step": 92940
-    },
-    {
-      "epoch": 0.3755297615921331,
-      "grad_norm": 896.652587890625,
-      "learning_rate": 3.9567261991714404e-05,
-      "loss": 63.6779,
-      "step": 92950
-    },
-    {
-      "epoch": 0.3755701628575007,
-      "grad_norm": 1012.0924072265625,
-      "learning_rate": 3.9564425011929265e-05,
-      "loss": 68.4286,
-      "step": 92960
-    },
-    {
-      "epoch": 0.3756105641228683,
-      "grad_norm": 525.5440673828125,
-      "learning_rate": 3.956158774820316e-05,
-      "loss": 53.852,
-      "step": 92970
-    },
-    {
-      "epoch": 0.37565096538823595,
-      "grad_norm": 657.0477905273438,
-      "learning_rate": 3.955875020059141e-05,
-      "loss": 79.606,
-      "step": 92980
-    },
-    {
-      "epoch": 0.3756913666536036,
-      "grad_norm": 1086.502197265625,
-      "learning_rate": 3.955591236914933e-05,
-      "loss": 46.6891,
-      "step": 92990
-    },
-    {
-      "epoch": 0.3757317679189712,
-      "grad_norm": 1276.398193359375,
-      "learning_rate": 3.955307425393224e-05,
-      "loss": 83.8669,
-      "step": 93000
-    },
-    {
-      "epoch": 0.37577216918433887,
-      "grad_norm": 1241.0797119140625,
-      "learning_rate": 3.955023585499547e-05,
-      "loss": 65.6787,
-      "step": 93010
-    },
-    {
-      "epoch": 0.3758125704497065,
-      "grad_norm": 516.9462890625,
-      "learning_rate": 3.954739717239437e-05,
-      "loss": 80.84,
-      "step": 93020
-    },
-    {
-      "epoch": 0.3758529717150741,
-      "grad_norm": 475.4250183105469,
-      "learning_rate": 3.954455820618427e-05,
-      "loss": 72.3098,
-      "step": 93030
-    },
-    {
-      "epoch": 0.37589337298044173,
-      "grad_norm": 950.8383178710938,
-      "learning_rate": 3.954171895642052e-05,
-      "loss": 76.2437,
-      "step": 93040
-    },
-    {
-      "epoch": 0.37593377424580937,
-      "grad_norm": 631.9634399414062,
-      "learning_rate": 3.953887942315847e-05,
-      "loss": 42.5692,
-      "step": 93050
-    },
-    {
-      "epoch": 0.375974175511177,
-      "grad_norm": 942.7606201171875,
-      "learning_rate": 3.953603960645349e-05,
-      "loss": 79.1041,
-      "step": 93060
-    },
-    {
-      "epoch": 0.37601457677654465,
-      "grad_norm": 1219.8829345703125,
-      "learning_rate": 3.953319950636092e-05,
-      "loss": 107.0019,
-      "step": 93070
-    },
-    {
-      "epoch": 0.3760549780419123,
-      "grad_norm": 688.3053588867188,
-      "learning_rate": 3.953035912293616e-05,
-      "loss": 64.1869,
-      "step": 93080
-    },
-    {
-      "epoch": 0.37609537930727993,
-      "grad_norm": 632.4610595703125,
-      "learning_rate": 3.9527518456234544e-05,
-      "loss": 57.2202,
-      "step": 93090
-    },
-    {
-      "epoch": 0.3761357805726475,
-      "grad_norm": 323.7148742675781,
-      "learning_rate": 3.95246775063115e-05,
-      "loss": 45.6848,
-      "step": 93100
-    },
-    {
-      "epoch": 0.37617618183801516,
-      "grad_norm": 789.0226440429688,
-      "learning_rate": 3.952183627322238e-05,
-      "loss": 76.8052,
-      "step": 93110
-    },
-    {
-      "epoch": 0.3762165831033828,
-      "grad_norm": 631.6780395507812,
-      "learning_rate": 3.951899475702259e-05,
-      "loss": 57.1333,
-      "step": 93120
-    },
-    {
-      "epoch": 0.37625698436875044,
-      "grad_norm": 1251.869140625,
-      "learning_rate": 3.951615295776752e-05,
-      "loss": 120.3193,
-      "step": 93130
-    },
-    {
-      "epoch": 0.3762973856341181,
-      "grad_norm": 525.0264892578125,
-      "learning_rate": 3.951331087551257e-05,
-      "loss": 77.0742,
-      "step": 93140
-    },
-    {
-      "epoch": 0.3763377868994857,
-      "grad_norm": 927.7319946289062,
-      "learning_rate": 3.951046851031315e-05,
-      "loss": 68.461,
-      "step": 93150
-    },
-    {
-      "epoch": 0.3763781881648533,
-      "grad_norm": 388.8170471191406,
-      "learning_rate": 3.950762586222468e-05,
-      "loss": 33.0051,
-      "step": 93160
-    },
-    {
-      "epoch": 0.37641858943022094,
-      "grad_norm": 462.57025146484375,
-      "learning_rate": 3.950478293130258e-05,
-      "loss": 57.6939,
-      "step": 93170
-    },
-    {
-      "epoch": 0.3764589906955886,
-      "grad_norm": 818.913818359375,
-      "learning_rate": 3.950193971760226e-05,
-      "loss": 47.7335,
-      "step": 93180
-    },
-    {
-      "epoch": 0.3764993919609562,
-      "grad_norm": 2200.9609375,
-      "learning_rate": 3.949909622117918e-05,
-      "loss": 64.2052,
-      "step": 93190
-    },
-    {
-      "epoch": 0.37653979322632386,
-      "grad_norm": 908.902099609375,
-      "learning_rate": 3.9496252442088733e-05,
-      "loss": 36.8062,
-      "step": 93200
-    },
-    {
-      "epoch": 0.3765801944916915,
-      "grad_norm": 740.1572265625,
-      "learning_rate": 3.949340838038639e-05,
-      "loss": 68.5629,
-      "step": 93210
-    },
-    {
-      "epoch": 0.3766205957570591,
-      "grad_norm": 1135.758544921875,
-      "learning_rate": 3.949056403612758e-05,
-      "loss": 62.5657,
-      "step": 93220
-    },
-    {
-      "epoch": 0.3766609970224267,
-      "grad_norm": 2026.0989990234375,
-      "learning_rate": 3.9487719409367774e-05,
-      "loss": 74.2239,
-      "step": 93230
-    },
-    {
-      "epoch": 0.37670139828779436,
-      "grad_norm": 864.1495361328125,
-      "learning_rate": 3.948487450016242e-05,
-      "loss": 92.9084,
-      "step": 93240
-    },
-    {
-      "epoch": 0.376741799553162,
-      "grad_norm": 810.767333984375,
-      "learning_rate": 3.948202930856697e-05,
-      "loss": 80.3417,
-      "step": 93250
-    },
-    {
-      "epoch": 0.37678220081852964,
-      "grad_norm": 170.60748291015625,
-      "learning_rate": 3.947918383463691e-05,
-      "loss": 58.1501,
-      "step": 93260
-    },
-    {
-      "epoch": 0.3768226020838973,
-      "grad_norm": 521.3902587890625,
-      "learning_rate": 3.947633807842771e-05,
-      "loss": 24.7328,
-      "step": 93270
-    },
-    {
-      "epoch": 0.3768630033492649,
-      "grad_norm": 615.25048828125,
-      "learning_rate": 3.947349203999484e-05,
-      "loss": 69.8992,
-      "step": 93280
-    },
-    {
-      "epoch": 0.3769034046146325,
-      "grad_norm": 616.8837280273438,
-      "learning_rate": 3.9470645719393794e-05,
-      "loss": 48.8142,
-      "step": 93290
-    },
-    {
-      "epoch": 0.37694380588000015,
-      "grad_norm": 1704.8671875,
-      "learning_rate": 3.946779911668006e-05,
-      "loss": 62.0936,
-      "step": 93300
-    },
-    {
-      "epoch": 0.3769842071453678,
-      "grad_norm": 816.0055541992188,
-      "learning_rate": 3.9464952231909135e-05,
-      "loss": 56.5077,
-      "step": 93310
-    },
-    {
-      "epoch": 0.37702460841073543,
-      "grad_norm": 501.01641845703125,
-      "learning_rate": 3.946210506513651e-05,
-      "loss": 51.1529,
-      "step": 93320
-    },
-    {
-      "epoch": 0.37706500967610307,
-      "grad_norm": 1191.9462890625,
-      "learning_rate": 3.945925761641771e-05,
-      "loss": 56.0396,
-      "step": 93330
-    },
-    {
-      "epoch": 0.3771054109414707,
-      "grad_norm": 910.802734375,
-      "learning_rate": 3.945640988580824e-05,
-      "loss": 83.711,
-      "step": 93340
-    },
-    {
-      "epoch": 0.3771458122068383,
-      "grad_norm": 1067.3336181640625,
-      "learning_rate": 3.9453561873363615e-05,
-      "loss": 57.1342,
-      "step": 93350
-    },
-    {
-      "epoch": 0.37718621347220593,
-      "grad_norm": 894.1754150390625,
-      "learning_rate": 3.945071357913935e-05,
-      "loss": 53.6304,
-      "step": 93360
-    },
-    {
-      "epoch": 0.3772266147375736,
-      "grad_norm": 760.7733764648438,
-      "learning_rate": 3.9447865003191e-05,
-      "loss": 67.4268,
-      "step": 93370
-    },
-    {
-      "epoch": 0.3772670160029412,
-      "grad_norm": 771.8887939453125,
-      "learning_rate": 3.9445016145574074e-05,
-      "loss": 69.5092,
-      "step": 93380
-    },
-    {
-      "epoch": 0.37730741726830885,
-      "grad_norm": 459.5794372558594,
-      "learning_rate": 3.9442167006344124e-05,
-      "loss": 46.507,
-      "step": 93390
-    },
-    {
-      "epoch": 0.3773478185336765,
-      "grad_norm": 2548.902099609375,
-      "learning_rate": 3.943931758555669e-05,
-      "loss": 73.9357,
-      "step": 93400
-    },
-    {
-      "epoch": 0.37738821979904413,
-      "grad_norm": 1343.5716552734375,
-      "learning_rate": 3.9436467883267334e-05,
-      "loss": 52.8962,
-      "step": 93410
-    },
-    {
-      "epoch": 0.3774286210644117,
-      "grad_norm": 1176.486083984375,
-      "learning_rate": 3.9433617899531597e-05,
-      "loss": 71.4715,
-      "step": 93420
-    },
-    {
-      "epoch": 0.37746902232977936,
-      "grad_norm": 3818.821533203125,
-      "learning_rate": 3.943076763440505e-05,
-      "loss": 73.5444,
-      "step": 93430
-    },
-    {
-      "epoch": 0.377509423595147,
-      "grad_norm": 1363.684814453125,
-      "learning_rate": 3.942791708794326e-05,
-      "loss": 45.8224,
-      "step": 93440
-    },
-    {
-      "epoch": 0.37754982486051464,
-      "grad_norm": 602.17138671875,
-      "learning_rate": 3.9425066260201796e-05,
-      "loss": 56.5135,
-      "step": 93450
-    },
-    {
-      "epoch": 0.3775902261258823,
-      "grad_norm": 533.1290893554688,
-      "learning_rate": 3.942221515123623e-05,
-      "loss": 47.6091,
-      "step": 93460
-    },
-    {
-      "epoch": 0.3776306273912499,
-      "grad_norm": 868.53125,
-      "learning_rate": 3.941936376110217e-05,
-      "loss": 60.5942,
-      "step": 93470
-    },
-    {
-      "epoch": 0.3776710286566175,
-      "grad_norm": 581.7073364257812,
-      "learning_rate": 3.9416512089855184e-05,
-      "loss": 52.1139,
-      "step": 93480
-    },
-    {
-      "epoch": 0.37771142992198514,
-      "grad_norm": 983.8074951171875,
-      "learning_rate": 3.941366013755087e-05,
-      "loss": 46.983,
-      "step": 93490
-    },
-    {
-      "epoch": 0.3777518311873528,
-      "grad_norm": 1439.6392822265625,
-      "learning_rate": 3.941080790424484e-05,
-      "loss": 56.4244,
-      "step": 93500
-    },
-    {
-      "epoch": 0.3777922324527204,
-      "grad_norm": 1822.520751953125,
-      "learning_rate": 3.940795538999268e-05,
-      "loss": 77.834,
-      "step": 93510
-    },
-    {
-      "epoch": 0.37783263371808806,
-      "grad_norm": 437.8319396972656,
-      "learning_rate": 3.940510259485002e-05,
-      "loss": 47.9851,
-      "step": 93520
-    },
-    {
-      "epoch": 0.3778730349834557,
-      "grad_norm": 1477.3594970703125,
-      "learning_rate": 3.9402249518872456e-05,
-      "loss": 104.0979,
-      "step": 93530
-    },
-    {
-      "epoch": 0.3779134362488233,
-      "grad_norm": 517.4141235351562,
-      "learning_rate": 3.939939616211563e-05,
-      "loss": 46.596,
-      "step": 93540
-    },
-    {
-      "epoch": 0.3779538375141909,
-      "grad_norm": 673.5197143554688,
-      "learning_rate": 3.9396542524635175e-05,
-      "loss": 61.4171,
-      "step": 93550
-    },
-    {
-      "epoch": 0.37799423877955857,
-      "grad_norm": 679.6529541015625,
-      "learning_rate": 3.939368860648669e-05,
-      "loss": 50.3042,
-      "step": 93560
-    },
-    {
-      "epoch": 0.3780346400449262,
-      "grad_norm": 1323.72314453125,
-      "learning_rate": 3.939083440772585e-05,
-      "loss": 52.2302,
-      "step": 93570
-    },
-    {
-      "epoch": 0.37807504131029385,
-      "grad_norm": 1001.907958984375,
-      "learning_rate": 3.938797992840828e-05,
-      "loss": 66.2964,
-      "step": 93580
-    },
-    {
-      "epoch": 0.3781154425756615,
-      "grad_norm": 615.739990234375,
-      "learning_rate": 3.9385125168589635e-05,
-      "loss": 72.0375,
-      "step": 93590
-    },
-    {
-      "epoch": 0.3781558438410291,
-      "grad_norm": 190.5300750732422,
-      "learning_rate": 3.938227012832557e-05,
-      "loss": 38.2591,
-      "step": 93600
-    },
-    {
-      "epoch": 0.3781962451063967,
-      "grad_norm": 4424.119140625,
-      "learning_rate": 3.9379414807671736e-05,
-      "loss": 86.606,
-      "step": 93610
-    },
-    {
-      "epoch": 0.37823664637176435,
-      "grad_norm": 1783.222900390625,
-      "learning_rate": 3.937655920668382e-05,
-      "loss": 45.5719,
-      "step": 93620
-    },
-    {
-      "epoch": 0.378277047637132,
-      "grad_norm": 508.1633605957031,
-      "learning_rate": 3.937370332541747e-05,
-      "loss": 46.5244,
-      "step": 93630
-    },
-    {
-      "epoch": 0.37831744890249963,
-      "grad_norm": 1679.250732421875,
-      "learning_rate": 3.937084716392838e-05,
-      "loss": 105.5736,
-      "step": 93640
-    },
-    {
-      "epoch": 0.37835785016786727,
-      "grad_norm": 992.9168090820312,
-      "learning_rate": 3.936799072227222e-05,
-      "loss": 66.7025,
-      "step": 93650
-    },
-    {
-      "epoch": 0.3783982514332349,
-      "grad_norm": 709.3222045898438,
-      "learning_rate": 3.936513400050469e-05,
-      "loss": 49.8676,
-      "step": 93660
-    },
-    {
-      "epoch": 0.3784386526986025,
-      "grad_norm": 1248.9063720703125,
-      "learning_rate": 3.936227699868147e-05,
-      "loss": 67.6602,
-      "step": 93670
-    },
-    {
-      "epoch": 0.37847905396397014,
-      "grad_norm": 494.0985107421875,
-      "learning_rate": 3.9359419716858274e-05,
-      "loss": 52.3046,
-      "step": 93680
-    },
-    {
-      "epoch": 0.3785194552293378,
-      "grad_norm": 317.8280334472656,
-      "learning_rate": 3.9356562155090795e-05,
-      "loss": 37.2544,
-      "step": 93690
-    },
-    {
-      "epoch": 0.3785598564947054,
-      "grad_norm": 2666.53662109375,
-      "learning_rate": 3.935370431343475e-05,
-      "loss": 82.7156,
-      "step": 93700
-    },
-    {
-      "epoch": 0.37860025776007306,
-      "grad_norm": 1855.157958984375,
-      "learning_rate": 3.935084619194584e-05,
-      "loss": 79.1015,
-      "step": 93710
-    },
-    {
-      "epoch": 0.3786406590254407,
-      "grad_norm": 621.322265625,
-      "learning_rate": 3.93479877906798e-05,
-      "loss": 57.4279,
-      "step": 93720
-    },
-    {
-      "epoch": 0.37868106029080834,
-      "grad_norm": 785.6000366210938,
-      "learning_rate": 3.934512910969235e-05,
-      "loss": 75.8258,
-      "step": 93730
-    },
-    {
-      "epoch": 0.3787214615561759,
-      "grad_norm": 507.967041015625,
-      "learning_rate": 3.934227014903922e-05,
-      "loss": 63.3973,
-      "step": 93740
-    },
-    {
-      "epoch": 0.37876186282154356,
-      "grad_norm": 1563.5101318359375,
-      "learning_rate": 3.933941090877615e-05,
-      "loss": 52.865,
-      "step": 93750
-    },
-    {
-      "epoch": 0.3788022640869112,
-      "grad_norm": 571.6914672851562,
-      "learning_rate": 3.933655138895889e-05,
-      "loss": 61.8429,
-      "step": 93760
-    },
-    {
-      "epoch": 0.37884266535227884,
-      "grad_norm": 927.9751586914062,
-      "learning_rate": 3.9333691589643177e-05,
-      "loss": 51.3117,
-      "step": 93770
-    },
-    {
-      "epoch": 0.3788830666176465,
-      "grad_norm": 582.5885009765625,
-      "learning_rate": 3.9330831510884755e-05,
-      "loss": 85.4374,
-      "step": 93780
-    },
-    {
-      "epoch": 0.3789234678830141,
-      "grad_norm": 681.2293701171875,
-      "learning_rate": 3.932797115273941e-05,
-      "loss": 75.4987,
-      "step": 93790
-    },
-    {
-      "epoch": 0.3789638691483817,
-      "grad_norm": 720.52392578125,
-      "learning_rate": 3.932511051526289e-05,
-      "loss": 83.3338,
-      "step": 93800
-    },
-    {
-      "epoch": 0.37900427041374934,
-      "grad_norm": 888.4859008789062,
-      "learning_rate": 3.9322249598510955e-05,
-      "loss": 67.421,
-      "step": 93810
-    },
-    {
-      "epoch": 0.379044671679117,
-      "grad_norm": 1525.38134765625,
-      "learning_rate": 3.93193884025394e-05,
-      "loss": 85.0812,
-      "step": 93820
-    },
-    {
-      "epoch": 0.3790850729444846,
-      "grad_norm": 705.2191772460938,
-      "learning_rate": 3.931652692740399e-05,
-      "loss": 67.6637,
-      "step": 93830
-    },
-    {
-      "epoch": 0.37912547420985226,
-      "grad_norm": 1749.32568359375,
-      "learning_rate": 3.931366517316052e-05,
-      "loss": 68.8921,
-      "step": 93840
-    },
-    {
-      "epoch": 0.3791658754752199,
-      "grad_norm": 751.9481201171875,
-      "learning_rate": 3.9310803139864775e-05,
-      "loss": 123.6256,
-      "step": 93850
-    },
-    {
-      "epoch": 0.3792062767405875,
-      "grad_norm": 476.33636474609375,
-      "learning_rate": 3.9307940827572555e-05,
-      "loss": 50.916,
-      "step": 93860
-    },
-    {
-      "epoch": 0.37924667800595513,
-      "grad_norm": 267.0869140625,
-      "learning_rate": 3.9305078236339666e-05,
-      "loss": 57.5847,
-      "step": 93870
-    },
-    {
-      "epoch": 0.37928707927132277,
-      "grad_norm": 953.2073364257812,
-      "learning_rate": 3.930221536622191e-05,
-      "loss": 61.0358,
-      "step": 93880
-    },
-    {
-      "epoch": 0.3793274805366904,
-      "grad_norm": 630.3345336914062,
-      "learning_rate": 3.9299352217275105e-05,
-      "loss": 70.866,
-      "step": 93890
-    },
-    {
-      "epoch": 0.37936788180205805,
-      "grad_norm": 356.209228515625,
-      "learning_rate": 3.9296488789555066e-05,
-      "loss": 45.0516,
-      "step": 93900
-    },
-    {
-      "epoch": 0.3794082830674257,
-      "grad_norm": 410.72320556640625,
-      "learning_rate": 3.9293625083117616e-05,
-      "loss": 73.401,
-      "step": 93910
-    },
-    {
-      "epoch": 0.37944868433279333,
-      "grad_norm": 855.3485107421875,
-      "learning_rate": 3.9290761098018585e-05,
-      "loss": 55.5836,
-      "step": 93920
-    },
-    {
-      "epoch": 0.3794890855981609,
-      "grad_norm": 787.3507690429688,
-      "learning_rate": 3.928789683431381e-05,
-      "loss": 55.8925,
-      "step": 93930
-    },
-    {
-      "epoch": 0.37952948686352855,
-      "grad_norm": 600.7017822265625,
-      "learning_rate": 3.928503229205913e-05,
-      "loss": 60.9399,
-      "step": 93940
-    },
-    {
-      "epoch": 0.3795698881288962,
-      "grad_norm": 561.0108032226562,
-      "learning_rate": 3.928216747131039e-05,
-      "loss": 69.8876,
-      "step": 93950
-    },
-    {
-      "epoch": 0.37961028939426383,
-      "grad_norm": 1152.1119384765625,
-      "learning_rate": 3.927930237212345e-05,
-      "loss": 76.2245,
-      "step": 93960
-    },
-    {
-      "epoch": 0.3796506906596315,
-      "grad_norm": 743.5120239257812,
-      "learning_rate": 3.9276436994554144e-05,
-      "loss": 57.824,
-      "step": 93970
-    },
-    {
-      "epoch": 0.3796910919249991,
-      "grad_norm": 733.8156127929688,
-      "learning_rate": 3.927357133865836e-05,
-      "loss": 42.9731,
-      "step": 93980
-    },
-    {
-      "epoch": 0.3797314931903667,
-      "grad_norm": 431.87158203125,
-      "learning_rate": 3.927070540449195e-05,
-      "loss": 63.5863,
-      "step": 93990
-    },
-    {
-      "epoch": 0.37977189445573434,
-      "grad_norm": 751.9428100585938,
-      "learning_rate": 3.92678391921108e-05,
-      "loss": 78.9025,
-      "step": 94000
-    },
-    {
-      "epoch": 0.379812295721102,
-      "grad_norm": 767.0701293945312,
-      "learning_rate": 3.926497270157077e-05,
-      "loss": 42.9222,
-      "step": 94010
-    },
-    {
-      "epoch": 0.3798526969864696,
-      "grad_norm": 768.52197265625,
-      "learning_rate": 3.926210593292775e-05,
-      "loss": 48.6413,
-      "step": 94020
-    },
-    {
-      "epoch": 0.37989309825183726,
-      "grad_norm": 618.26806640625,
-      "learning_rate": 3.925923888623764e-05,
-      "loss": 54.2861,
-      "step": 94030
-    },
-    {
-      "epoch": 0.3799334995172049,
-      "grad_norm": 647.1188354492188,
-      "learning_rate": 3.925637156155633e-05,
-      "loss": 70.6201,
-      "step": 94040
-    },
-    {
-      "epoch": 0.37997390078257254,
-      "grad_norm": 886.6661987304688,
-      "learning_rate": 3.925350395893971e-05,
-      "loss": 85.2603,
-      "step": 94050
-    },
-    {
-      "epoch": 0.3800143020479401,
-      "grad_norm": 1664.85693359375,
-      "learning_rate": 3.925063607844369e-05,
-      "loss": 55.2438,
-      "step": 94060
-    },
-    {
-      "epoch": 0.38005470331330776,
-      "grad_norm": 788.0062866210938,
-      "learning_rate": 3.9247767920124176e-05,
-      "loss": 73.1715,
-      "step": 94070
-    },
-    {
-      "epoch": 0.3800951045786754,
-      "grad_norm": 435.3290710449219,
-      "learning_rate": 3.924489948403711e-05,
-      "loss": 70.93,
-      "step": 94080
-    },
-    {
-      "epoch": 0.38013550584404304,
-      "grad_norm": 143.54469299316406,
-      "learning_rate": 3.924203077023839e-05,
-      "loss": 31.3308,
-      "step": 94090
-    },
-    {
-      "epoch": 0.3801759071094107,
-      "grad_norm": 1035.6220703125,
-      "learning_rate": 3.923916177878394e-05,
-      "loss": 59.6959,
-      "step": 94100
-    },
-    {
-      "epoch": 0.3802163083747783,
-      "grad_norm": 1633.117431640625,
-      "learning_rate": 3.9236292509729697e-05,
-      "loss": 77.1727,
-      "step": 94110
-    },
-    {
-      "epoch": 0.3802567096401459,
-      "grad_norm": 421.79974365234375,
-      "learning_rate": 3.9233422963131616e-05,
-      "loss": 45.4822,
-      "step": 94120
-    },
-    {
-      "epoch": 0.38029711090551355,
-      "grad_norm": 953.0307006835938,
-      "learning_rate": 3.9230553139045617e-05,
-      "loss": 98.6893,
-      "step": 94130
-    },
-    {
-      "epoch": 0.3803375121708812,
-      "grad_norm": 1626.166015625,
-      "learning_rate": 3.922768303752766e-05,
-      "loss": 73.0103,
-      "step": 94140
-    },
-    {
-      "epoch": 0.3803779134362488,
-      "grad_norm": 1201.143310546875,
-      "learning_rate": 3.92248126586337e-05,
-      "loss": 49.0338,
-      "step": 94150
-    },
-    {
-      "epoch": 0.38041831470161647,
-      "grad_norm": 1199.5015869140625,
-      "learning_rate": 3.922194200241969e-05,
-      "loss": 49.4027,
-      "step": 94160
-    },
-    {
-      "epoch": 0.3804587159669841,
-      "grad_norm": 948.0160522460938,
-      "learning_rate": 3.92190710689416e-05,
-      "loss": 47.4217,
-      "step": 94170
-    },
-    {
-      "epoch": 0.3804991172323517,
-      "grad_norm": 1361.0826416015625,
-      "learning_rate": 3.92161998582554e-05,
-      "loss": 68.6793,
-      "step": 94180
-    },
-    {
-      "epoch": 0.38053951849771933,
-      "grad_norm": 691.0305786132812,
-      "learning_rate": 3.9213328370417065e-05,
-      "loss": 86.5888,
-      "step": 94190
-    },
-    {
-      "epoch": 0.38057991976308697,
-      "grad_norm": 448.7333984375,
-      "learning_rate": 3.9210456605482576e-05,
-      "loss": 45.7264,
-      "step": 94200
-    },
-    {
-      "epoch": 0.3806203210284546,
-      "grad_norm": 926.4880981445312,
-      "learning_rate": 3.920758456350792e-05,
-      "loss": 73.8021,
-      "step": 94210
-    },
-    {
-      "epoch": 0.38066072229382225,
-      "grad_norm": 542.78662109375,
-      "learning_rate": 3.9204712244549085e-05,
-      "loss": 65.7361,
-      "step": 94220
-    },
-    {
-      "epoch": 0.3807011235591899,
-      "grad_norm": 3470.885986328125,
-      "learning_rate": 3.9201839648662074e-05,
-      "loss": 63.5717,
-      "step": 94230
-    },
-    {
-      "epoch": 0.38074152482455753,
-      "grad_norm": 596.5708618164062,
-      "learning_rate": 3.919896677590289e-05,
-      "loss": 69.1542,
-      "step": 94240
-    },
-    {
-      "epoch": 0.3807819260899251,
-      "grad_norm": 797.0637817382812,
-      "learning_rate": 3.919609362632753e-05,
-      "loss": 106.799,
-      "step": 94250
-    },
-    {
-      "epoch": 0.38082232735529276,
-      "grad_norm": 641.5678100585938,
-      "learning_rate": 3.9193220199992025e-05,
-      "loss": 42.721,
-      "step": 94260
-    },
-    {
-      "epoch": 0.3808627286206604,
-      "grad_norm": 370.0324401855469,
-      "learning_rate": 3.919034649695238e-05,
-      "loss": 47.1425,
-      "step": 94270
-    },
-    {
-      "epoch": 0.38090312988602804,
-      "grad_norm": 1143.82861328125,
-      "learning_rate": 3.918747251726463e-05,
-      "loss": 51.9892,
-      "step": 94280
-    },
-    {
-      "epoch": 0.3809435311513957,
-      "grad_norm": 859.6338500976562,
-      "learning_rate": 3.9184598260984795e-05,
-      "loss": 64.1427,
-      "step": 94290
-    },
-    {
-      "epoch": 0.3809839324167633,
-      "grad_norm": 423.8070068359375,
-      "learning_rate": 3.9181723728168916e-05,
-      "loss": 75.3932,
-      "step": 94300
-    },
-    {
-      "epoch": 0.3810243336821309,
-      "grad_norm": 945.8095703125,
-      "learning_rate": 3.9178848918873027e-05,
-      "loss": 47.6868,
-      "step": 94310
-    },
-    {
-      "epoch": 0.38106473494749854,
-      "grad_norm": 1396.765625,
-      "learning_rate": 3.9175973833153186e-05,
-      "loss": 53.5173,
-      "step": 94320
-    },
-    {
-      "epoch": 0.3811051362128662,
-      "grad_norm": 1253.411376953125,
-      "learning_rate": 3.9173098471065434e-05,
-      "loss": 48.6057,
-      "step": 94330
-    },
-    {
-      "epoch": 0.3811455374782338,
-      "grad_norm": 627.4695434570312,
-      "learning_rate": 3.9170222832665825e-05,
-      "loss": 68.0112,
-      "step": 94340
-    },
-    {
-      "epoch": 0.38118593874360146,
-      "grad_norm": 918.7919311523438,
-      "learning_rate": 3.9167346918010425e-05,
-      "loss": 102.3109,
-      "step": 94350
-    },
-    {
-      "epoch": 0.3812263400089691,
-      "grad_norm": 1036.47265625,
-      "learning_rate": 3.9164470727155314e-05,
-      "loss": 77.9767,
-      "step": 94360
-    },
-    {
-      "epoch": 0.3812667412743367,
-      "grad_norm": 1388.8292236328125,
-      "learning_rate": 3.916159426015655e-05,
-      "loss": 74.0313,
-      "step": 94370
-    },
-    {
-      "epoch": 0.3813071425397043,
-      "grad_norm": 761.83203125,
-      "learning_rate": 3.9158717517070214e-05,
-      "loss": 58.2245,
-      "step": 94380
-    },
-    {
-      "epoch": 0.38134754380507196,
-      "grad_norm": 416.04351806640625,
-      "learning_rate": 3.915584049795239e-05,
-      "loss": 50.2299,
-      "step": 94390
-    },
-    {
-      "epoch": 0.3813879450704396,
-      "grad_norm": 1973.710205078125,
-      "learning_rate": 3.915296320285917e-05,
-      "loss": 56.33,
-      "step": 94400
-    },
-    {
-      "epoch": 0.38142834633580724,
-      "grad_norm": 895.8132934570312,
-      "learning_rate": 3.915008563184664e-05,
-      "loss": 54.553,
-      "step": 94410
-    },
-    {
-      "epoch": 0.3814687476011749,
-      "grad_norm": 521.8209838867188,
-      "learning_rate": 3.9147207784970914e-05,
-      "loss": 69.8902,
-      "step": 94420
-    },
-    {
-      "epoch": 0.3815091488665425,
-      "grad_norm": 1060.228515625,
-      "learning_rate": 3.914432966228808e-05,
-      "loss": 72.3957,
-      "step": 94430
-    },
-    {
-      "epoch": 0.3815495501319101,
-      "grad_norm": 478.51885986328125,
-      "learning_rate": 3.914145126385426e-05,
-      "loss": 68.0388,
-      "step": 94440
-    },
-    {
-      "epoch": 0.38158995139727775,
-      "grad_norm": 855.6203002929688,
-      "learning_rate": 3.9138572589725576e-05,
-      "loss": 52.0944,
-      "step": 94450
-    },
-    {
-      "epoch": 0.3816303526626454,
-      "grad_norm": 2904.56298828125,
-      "learning_rate": 3.9135693639958125e-05,
-      "loss": 50.8609,
-      "step": 94460
-    },
-    {
-      "epoch": 0.38167075392801303,
-      "grad_norm": 1457.2967529296875,
-      "learning_rate": 3.913281441460806e-05,
-      "loss": 86.8889,
-      "step": 94470
-    },
-    {
-      "epoch": 0.38171115519338067,
-      "grad_norm": 745.7070922851562,
-      "learning_rate": 3.91299349137315e-05,
-      "loss": 63.1558,
-      "step": 94480
-    },
-    {
-      "epoch": 0.3817515564587483,
-      "grad_norm": 832.9826049804688,
-      "learning_rate": 3.912705513738458e-05,
-      "loss": 59.3702,
-      "step": 94490
-    },
-    {
-      "epoch": 0.3817919577241159,
-      "grad_norm": 768.7559204101562,
-      "learning_rate": 3.912417508562345e-05,
-      "loss": 65.5276,
-      "step": 94500
-    },
-    {
-      "epoch": 0.38183235898948353,
-      "grad_norm": 657.02734375,
-      "learning_rate": 3.912129475850426e-05,
-      "loss": 71.6129,
-      "step": 94510
-    },
-    {
-      "epoch": 0.3818727602548512,
-      "grad_norm": 1496.1007080078125,
-      "learning_rate": 3.911841415608315e-05,
-      "loss": 83.0548,
-      "step": 94520
-    },
-    {
-      "epoch": 0.3819131615202188,
-      "grad_norm": 1321.64697265625,
-      "learning_rate": 3.911553327841629e-05,
-      "loss": 42.8167,
-      "step": 94530
-    },
-    {
-      "epoch": 0.38195356278558645,
-      "grad_norm": 434.47039794921875,
-      "learning_rate": 3.9112652125559845e-05,
-      "loss": 28.5604,
-      "step": 94540
-    },
-    {
-      "epoch": 0.3819939640509541,
-      "grad_norm": 1304.2777099609375,
-      "learning_rate": 3.910977069756998e-05,
-      "loss": 54.2573,
-      "step": 94550
-    },
-    {
-      "epoch": 0.38203436531632173,
-      "grad_norm": 1068.87939453125,
-      "learning_rate": 3.9106888994502864e-05,
-      "loss": 34.1891,
-      "step": 94560
-    },
-    {
-      "epoch": 0.3820747665816893,
-      "grad_norm": 4277.7587890625,
-      "learning_rate": 3.9104007016414695e-05,
-      "loss": 87.1632,
-      "step": 94570
-    },
-    {
-      "epoch": 0.38211516784705696,
-      "grad_norm": 96.375244140625,
-      "learning_rate": 3.910112476336164e-05,
-      "loss": 76.1936,
-      "step": 94580
-    },
-    {
-      "epoch": 0.3821555691124246,
-      "grad_norm": 775.3291015625,
-      "learning_rate": 3.90982422353999e-05,
-      "loss": 74.7171,
-      "step": 94590
-    },
-    {
-      "epoch": 0.38219597037779224,
-      "grad_norm": 696.175537109375,
-      "learning_rate": 3.909535943258567e-05,
-      "loss": 54.9566,
-      "step": 94600
-    },
-    {
-      "epoch": 0.3822363716431599,
-      "grad_norm": 724.5578002929688,
-      "learning_rate": 3.909247635497516e-05,
-      "loss": 77.6065,
-      "step": 94610
-    },
-    {
-      "epoch": 0.3822767729085275,
-      "grad_norm": 393.4891662597656,
-      "learning_rate": 3.9089593002624555e-05,
-      "loss": 65.3597,
-      "step": 94620
-    },
-    {
-      "epoch": 0.3823171741738951,
-      "grad_norm": 1047.53662109375,
-      "learning_rate": 3.908670937559008e-05,
-      "loss": 63.4875,
-      "step": 94630
-    },
-    {
-      "epoch": 0.38235757543926274,
-      "grad_norm": 1051.636962890625,
-      "learning_rate": 3.908382547392796e-05,
-      "loss": 39.8985,
-      "step": 94640
-    },
-    {
-      "epoch": 0.3823979767046304,
-      "grad_norm": 705.0889892578125,
-      "learning_rate": 3.908094129769442e-05,
-      "loss": 52.4874,
-      "step": 94650
-    },
-    {
-      "epoch": 0.382438377969998,
-      "grad_norm": 1593.9310302734375,
-      "learning_rate": 3.907805684694566e-05,
-      "loss": 82.5799,
-      "step": 94660
-    },
-    {
-      "epoch": 0.38247877923536566,
-      "grad_norm": 462.7720947265625,
-      "learning_rate": 3.9075172121737945e-05,
-      "loss": 48.7856,
-      "step": 94670
-    },
-    {
-      "epoch": 0.3825191805007333,
-      "grad_norm": 1333.7330322265625,
-      "learning_rate": 3.907228712212751e-05,
-      "loss": 88.1809,
-      "step": 94680
-    },
-    {
-      "epoch": 0.3825595817661009,
-      "grad_norm": 606.8469848632812,
-      "learning_rate": 3.906940184817057e-05,
-      "loss": 63.5642,
-      "step": 94690
-    },
-    {
-      "epoch": 0.3825999830314685,
-      "grad_norm": 280.5736389160156,
-      "learning_rate": 3.906651629992342e-05,
-      "loss": 54.9645,
-      "step": 94700
-    },
-    {
-      "epoch": 0.38264038429683617,
-      "grad_norm": 1871.5345458984375,
-      "learning_rate": 3.906363047744229e-05,
-      "loss": 162.7685,
-      "step": 94710
-    },
-    {
-      "epoch": 0.3826807855622038,
-      "grad_norm": 605.5426025390625,
-      "learning_rate": 3.9060744380783435e-05,
-      "loss": 43.6138,
-      "step": 94720
-    },
-    {
-      "epoch": 0.38272118682757145,
-      "grad_norm": 2624.923583984375,
-      "learning_rate": 3.9057858010003137e-05,
-      "loss": 108.0628,
-      "step": 94730
-    },
-    {
-      "epoch": 0.3827615880929391,
-      "grad_norm": 978.3139038085938,
-      "learning_rate": 3.905497136515766e-05,
-      "loss": 51.3885,
-      "step": 94740
-    },
-    {
-      "epoch": 0.3828019893583067,
-      "grad_norm": 2022.71435546875,
-      "learning_rate": 3.905208444630327e-05,
-      "loss": 58.5064,
-      "step": 94750
-    },
-    {
-      "epoch": 0.3828423906236743,
-      "grad_norm": 395.371826171875,
-      "learning_rate": 3.9049197253496264e-05,
-      "loss": 60.7483,
-      "step": 94760
-    },
-    {
-      "epoch": 0.38288279188904195,
-      "grad_norm": 848.9832153320312,
-      "learning_rate": 3.904630978679292e-05,
-      "loss": 57.199,
-      "step": 94770
-    },
-    {
-      "epoch": 0.3829231931544096,
-      "grad_norm": 1061.2220458984375,
-      "learning_rate": 3.9043422046249544e-05,
-      "loss": 64.7823,
-      "step": 94780
-    },
-    {
-      "epoch": 0.38296359441977723,
-      "grad_norm": 824.2564086914062,
-      "learning_rate": 3.904053403192242e-05,
-      "loss": 58.263,
-      "step": 94790
-    },
-    {
-      "epoch": 0.38300399568514487,
-      "grad_norm": 1173.652099609375,
-      "learning_rate": 3.903764574386786e-05,
-      "loss": 80.1155,
-      "step": 94800
-    },
-    {
-      "epoch": 0.3830443969505125,
-      "grad_norm": 596.0165405273438,
-      "learning_rate": 3.903475718214217e-05,
-      "loss": 64.6889,
-      "step": 94810
-    },
-    {
-      "epoch": 0.3830847982158801,
-      "grad_norm": 544.8338012695312,
-      "learning_rate": 3.9031868346801656e-05,
-      "loss": 90.0342,
-      "step": 94820
-    },
-    {
-      "epoch": 0.38312519948124774,
-      "grad_norm": 582.4503173828125,
-      "learning_rate": 3.902897923790265e-05,
-      "loss": 67.4698,
-      "step": 94830
-    },
-    {
-      "epoch": 0.3831656007466154,
-      "grad_norm": 1065.5283203125,
-      "learning_rate": 3.902608985550147e-05,
-      "loss": 64.2469,
-      "step": 94840
-    },
-    {
-      "epoch": 0.383206002011983,
-      "grad_norm": 718.9613037109375,
-      "learning_rate": 3.902320019965445e-05,
-      "loss": 50.7413,
-      "step": 94850
-    },
-    {
-      "epoch": 0.38324640327735066,
-      "grad_norm": 497.2271423339844,
-      "learning_rate": 3.902031027041793e-05,
-      "loss": 46.6418,
-      "step": 94860
-    },
-    {
-      "epoch": 0.3832868045427183,
-      "grad_norm": 767.0538330078125,
-      "learning_rate": 3.901742006784822e-05,
-      "loss": 86.6283,
-      "step": 94870
-    },
-    {
-      "epoch": 0.38332720580808594,
-      "grad_norm": 101.88063049316406,
-      "learning_rate": 3.9014529592001705e-05,
-      "loss": 80.5774,
-      "step": 94880
-    },
-    {
-      "epoch": 0.3833676070734535,
-      "grad_norm": 479.095947265625,
-      "learning_rate": 3.901163884293472e-05,
-      "loss": 44.5601,
-      "step": 94890
-    },
-    {
-      "epoch": 0.38340800833882116,
-      "grad_norm": 560.9264526367188,
-      "learning_rate": 3.900874782070362e-05,
-      "loss": 68.0652,
-      "step": 94900
-    },
-    {
-      "epoch": 0.3834484096041888,
-      "grad_norm": 530.4874877929688,
-      "learning_rate": 3.900585652536477e-05,
-      "loss": 87.2366,
-      "step": 94910
-    },
-    {
-      "epoch": 0.38348881086955644,
-      "grad_norm": 779.4183959960938,
-      "learning_rate": 3.900296495697453e-05,
-      "loss": 70.1493,
-      "step": 94920
-    },
-    {
-      "epoch": 0.3835292121349241,
-      "grad_norm": 0.0,
-      "learning_rate": 3.9000073115589286e-05,
-      "loss": 59.7547,
-      "step": 94930
-    },
-    {
-      "epoch": 0.3835696134002917,
-      "grad_norm": 516.5095825195312,
-      "learning_rate": 3.899718100126541e-05,
-      "loss": 50.7623,
-      "step": 94940
-    },
-    {
-      "epoch": 0.3836100146656593,
-      "grad_norm": 569.387939453125,
-      "learning_rate": 3.899428861405928e-05,
-      "loss": 83.759,
-      "step": 94950
-    },
-    {
-      "epoch": 0.38365041593102694,
-      "grad_norm": 586.040771484375,
-      "learning_rate": 3.899139595402729e-05,
-      "loss": 42.7253,
-      "step": 94960
-    },
-    {
-      "epoch": 0.3836908171963946,
-      "grad_norm": 2201.12646484375,
-      "learning_rate": 3.898850302122583e-05,
-      "loss": 56.9618,
-      "step": 94970
-    },
-    {
-      "epoch": 0.3837312184617622,
-      "grad_norm": 511.3911437988281,
-      "learning_rate": 3.898560981571131e-05,
-      "loss": 55.5062,
-      "step": 94980
-    },
-    {
-      "epoch": 0.38377161972712986,
-      "grad_norm": 383.2709045410156,
-      "learning_rate": 3.8982716337540115e-05,
-      "loss": 60.895,
-      "step": 94990
-    },
-    {
-      "epoch": 0.3838120209924975,
-      "grad_norm": 768.9598388671875,
-      "learning_rate": 3.897982258676867e-05,
-      "loss": 59.8323,
-      "step": 95000
-    },
-    {
-      "epoch": 0.3838524222578651,
-      "grad_norm": 1006.3447875976562,
-      "learning_rate": 3.897692856345339e-05,
-      "loss": 43.6415,
-      "step": 95010
-    },
-    {
-      "epoch": 0.38389282352323273,
-      "grad_norm": 774.2186889648438,
-      "learning_rate": 3.897403426765069e-05,
-      "loss": 101.709,
-      "step": 95020
-    },
-    {
-      "epoch": 0.38393322478860037,
-      "grad_norm": 1448.4361572265625,
-      "learning_rate": 3.8971139699417e-05,
-      "loss": 91.2065,
-      "step": 95030
-    },
-    {
-      "epoch": 0.383973626053968,
-      "grad_norm": 771.1356811523438,
-      "learning_rate": 3.896824485880874e-05,
-      "loss": 60.503,
-      "step": 95040
-    },
-    {
-      "epoch": 0.38401402731933565,
-      "grad_norm": 726.0272827148438,
-      "learning_rate": 3.8965349745882365e-05,
-      "loss": 70.719,
-      "step": 95050
-    },
-    {
-      "epoch": 0.3840544285847033,
-      "grad_norm": 791.23193359375,
-      "learning_rate": 3.896245436069431e-05,
-      "loss": 83.7272,
-      "step": 95060
-    },
-    {
-      "epoch": 0.38409482985007093,
-      "grad_norm": 805.988037109375,
-      "learning_rate": 3.8959558703301015e-05,
-      "loss": 62.7216,
-      "step": 95070
-    },
-    {
-      "epoch": 0.3841352311154385,
-      "grad_norm": 318.8636169433594,
-      "learning_rate": 3.895666277375892e-05,
-      "loss": 59.8062,
-      "step": 95080
-    },
-    {
-      "epoch": 0.38417563238080615,
-      "grad_norm": 0.0,
-      "learning_rate": 3.8953766572124515e-05,
-      "loss": 46.7181,
-      "step": 95090
-    },
-    {
-      "epoch": 0.3842160336461738,
-      "grad_norm": 616.468017578125,
-      "learning_rate": 3.895087009845425e-05,
-      "loss": 55.656,
-      "step": 95100
-    },
-    {
-      "epoch": 0.38425643491154143,
-      "grad_norm": 872.3570556640625,
-      "learning_rate": 3.8947973352804584e-05,
-      "loss": 78.4981,
-      "step": 95110
-    },
-    {
-      "epoch": 0.3842968361769091,
-      "grad_norm": 765.6026000976562,
-      "learning_rate": 3.894507633523199e-05,
-      "loss": 54.1042,
-      "step": 95120
-    },
-    {
-      "epoch": 0.3843372374422767,
-      "grad_norm": 903.8804931640625,
-      "learning_rate": 3.894217904579296e-05,
-      "loss": 79.6192,
-      "step": 95130
-    },
-    {
-      "epoch": 0.3843776387076443,
-      "grad_norm": 725.60498046875,
-      "learning_rate": 3.8939281484543974e-05,
-      "loss": 84.4251,
-      "step": 95140
-    },
-    {
-      "epoch": 0.38441803997301194,
-      "grad_norm": 1229.2586669921875,
-      "learning_rate": 3.893638365154152e-05,
-      "loss": 125.7261,
-      "step": 95150
-    },
-    {
-      "epoch": 0.3844584412383796,
-      "grad_norm": 727.9944458007812,
-      "learning_rate": 3.8933485546842094e-05,
-      "loss": 55.8314,
-      "step": 95160
-    },
-    {
-      "epoch": 0.3844988425037472,
-      "grad_norm": 732.46337890625,
-      "learning_rate": 3.893058717050218e-05,
-      "loss": 43.0125,
-      "step": 95170
-    },
-    {
-      "epoch": 0.38453924376911486,
-      "grad_norm": 948.4493408203125,
-      "learning_rate": 3.892768852257831e-05,
-      "loss": 48.8631,
-      "step": 95180
-    },
-    {
-      "epoch": 0.3845796450344825,
-      "grad_norm": 395.610595703125,
-      "learning_rate": 3.892478960312698e-05,
-      "loss": 97.7245,
-      "step": 95190
-    },
-    {
-      "epoch": 0.38462004629985014,
-      "grad_norm": 480.57672119140625,
-      "learning_rate": 3.8921890412204705e-05,
-      "loss": 67.3274,
-      "step": 95200
-    },
-    {
-      "epoch": 0.3846604475652177,
-      "grad_norm": 601.1005859375,
-      "learning_rate": 3.891899094986801e-05,
-      "loss": 76.2722,
-      "step": 95210
-    },
-    {
-      "epoch": 0.38470084883058536,
-      "grad_norm": 1527.212646484375,
-      "learning_rate": 3.891609121617342e-05,
-      "loss": 30.4016,
-      "step": 95220
-    },
-    {
-      "epoch": 0.384741250095953,
-      "grad_norm": 598.6000366210938,
-      "learning_rate": 3.8913191211177464e-05,
-      "loss": 38.4764,
-      "step": 95230
-    },
-    {
-      "epoch": 0.38478165136132064,
-      "grad_norm": 1061.5889892578125,
-      "learning_rate": 3.891029093493669e-05,
-      "loss": 57.333,
-      "step": 95240
-    },
-    {
-      "epoch": 0.3848220526266883,
-      "grad_norm": 719.52392578125,
-      "learning_rate": 3.8907390387507625e-05,
-      "loss": 66.065,
-      "step": 95250
-    },
-    {
-      "epoch": 0.3848624538920559,
-      "grad_norm": 723.9736328125,
-      "learning_rate": 3.890448956894682e-05,
-      "loss": 41.1686,
-      "step": 95260
-    },
-    {
-      "epoch": 0.3849028551574235,
-      "grad_norm": 789.1625366210938,
-      "learning_rate": 3.8901588479310846e-05,
-      "loss": 112.5225,
-      "step": 95270
-    },
-    {
-      "epoch": 0.38494325642279115,
-      "grad_norm": 321.8541259765625,
-      "learning_rate": 3.889868711865624e-05,
-      "loss": 43.6828,
-      "step": 95280
-    },
-    {
-      "epoch": 0.3849836576881588,
-      "grad_norm": 449.05401611328125,
-      "learning_rate": 3.8895785487039574e-05,
-      "loss": 82.2265,
-      "step": 95290
-    },
-    {
-      "epoch": 0.3850240589535264,
-      "grad_norm": 677.4368286132812,
-      "learning_rate": 3.8892883584517415e-05,
-      "loss": 52.7545,
-      "step": 95300
-    },
-    {
-      "epoch": 0.38506446021889407,
-      "grad_norm": 960.1495361328125,
-      "learning_rate": 3.888998141114634e-05,
-      "loss": 51.1408,
-      "step": 95310
-    },
-    {
-      "epoch": 0.3851048614842617,
-      "grad_norm": 3223.043701171875,
-      "learning_rate": 3.8887078966982925e-05,
-      "loss": 75.4026,
-      "step": 95320
-    },
-    {
-      "epoch": 0.3851452627496293,
-      "grad_norm": 1701.433837890625,
-      "learning_rate": 3.888417625208376e-05,
-      "loss": 65.9606,
-      "step": 95330
-    },
-    {
-      "epoch": 0.38518566401499693,
-      "grad_norm": 1047.423583984375,
-      "learning_rate": 3.888127326650542e-05,
-      "loss": 53.6032,
-      "step": 95340
-    },
-    {
-      "epoch": 0.38522606528036457,
-      "grad_norm": 591.1990356445312,
-      "learning_rate": 3.887837001030452e-05,
-      "loss": 67.6553,
-      "step": 95350
-    },
-    {
-      "epoch": 0.3852664665457322,
-      "grad_norm": 1093.99365234375,
-      "learning_rate": 3.887546648353765e-05,
-      "loss": 44.322,
-      "step": 95360
-    },
-    {
-      "epoch": 0.38530686781109985,
-      "grad_norm": 545.5051879882812,
-      "learning_rate": 3.887256268626142e-05,
-      "loss": 76.8624,
-      "step": 95370
-    },
-    {
-      "epoch": 0.3853472690764675,
-      "grad_norm": 1276.1842041015625,
-      "learning_rate": 3.886965861853244e-05,
-      "loss": 48.7556,
-      "step": 95380
-    },
-    {
-      "epoch": 0.38538767034183513,
-      "grad_norm": 739.0816650390625,
-      "learning_rate": 3.886675428040732e-05,
-      "loss": 62.3025,
-      "step": 95390
-    },
-    {
-      "epoch": 0.3854280716072027,
-      "grad_norm": 655.6229858398438,
-      "learning_rate": 3.8863849671942685e-05,
-      "loss": 51.2894,
-      "step": 95400
-    },
-    {
-      "epoch": 0.38546847287257036,
-      "grad_norm": 835.5565795898438,
-      "learning_rate": 3.886094479319517e-05,
-      "loss": 63.1423,
-      "step": 95410
-    },
-    {
-      "epoch": 0.385508874137938,
-      "grad_norm": 983.1041259765625,
-      "learning_rate": 3.885803964422139e-05,
-      "loss": 65.168,
-      "step": 95420
-    },
-    {
-      "epoch": 0.38554927540330564,
-      "grad_norm": 928.8274536132812,
-      "learning_rate": 3.885513422507799e-05,
-      "loss": 54.2019,
-      "step": 95430
-    },
-    {
-      "epoch": 0.3855896766686733,
-      "grad_norm": 716.9531860351562,
-      "learning_rate": 3.885222853582163e-05,
-      "loss": 69.1313,
-      "step": 95440
-    },
-    {
-      "epoch": 0.3856300779340409,
-      "grad_norm": 1002.2490234375,
-      "learning_rate": 3.8849322576508934e-05,
-      "loss": 57.0536,
-      "step": 95450
-    },
-    {
-      "epoch": 0.3856704791994085,
-      "grad_norm": 721.2386474609375,
-      "learning_rate": 3.884641634719657e-05,
-      "loss": 75.6969,
-      "step": 95460
-    },
-    {
-      "epoch": 0.38571088046477614,
-      "grad_norm": 673.9929809570312,
-      "learning_rate": 3.884350984794118e-05,
-      "loss": 68.2813,
-      "step": 95470
-    },
-    {
-      "epoch": 0.3857512817301438,
-      "grad_norm": 728.1707153320312,
-      "learning_rate": 3.8840603078799445e-05,
-      "loss": 53.9483,
-      "step": 95480
-    },
-    {
-      "epoch": 0.3857916829955114,
-      "grad_norm": 823.33740234375,
-      "learning_rate": 3.883769603982803e-05,
-      "loss": 51.6572,
-      "step": 95490
-    },
-    {
-      "epoch": 0.38583208426087906,
-      "grad_norm": 1103.5313720703125,
-      "learning_rate": 3.883478873108361e-05,
-      "loss": 49.7478,
-      "step": 95500
-    },
-    {
-      "epoch": 0.3858724855262467,
-      "grad_norm": 1030.14404296875,
-      "learning_rate": 3.883188115262285e-05,
-      "loss": 52.5702,
-      "step": 95510
-    },
-    {
-      "epoch": 0.38591288679161434,
-      "grad_norm": 937.8312377929688,
-      "learning_rate": 3.8828973304502446e-05,
-      "loss": 66.6772,
-      "step": 95520
-    },
-    {
-      "epoch": 0.3859532880569819,
-      "grad_norm": 1410.5428466796875,
-      "learning_rate": 3.88260651867791e-05,
-      "loss": 55.7531,
-      "step": 95530
-    },
-    {
-      "epoch": 0.38599368932234956,
-      "grad_norm": 480.3839111328125,
-      "learning_rate": 3.8823156799509484e-05,
-      "loss": 61.0244,
-      "step": 95540
-    },
-    {
-      "epoch": 0.3860340905877172,
-      "grad_norm": 471.7522888183594,
-      "learning_rate": 3.8820248142750316e-05,
-      "loss": 86.2028,
-      "step": 95550
-    },
-    {
-      "epoch": 0.38607449185308484,
-      "grad_norm": 531.3958129882812,
-      "learning_rate": 3.881733921655829e-05,
-      "loss": 88.4188,
-      "step": 95560
-    },
-    {
-      "epoch": 0.3861148931184525,
-      "grad_norm": 878.5484008789062,
-      "learning_rate": 3.881443002099012e-05,
-      "loss": 64.7605,
-      "step": 95570
-    },
-    {
-      "epoch": 0.3861552943838201,
-      "grad_norm": 371.7911682128906,
-      "learning_rate": 3.8811520556102535e-05,
-      "loss": 51.486,
-      "step": 95580
-    },
-    {
-      "epoch": 0.3861956956491877,
-      "grad_norm": 529.8397216796875,
-      "learning_rate": 3.880861082195224e-05,
-      "loss": 72.443,
-      "step": 95590
-    },
-    {
-      "epoch": 0.38623609691455535,
-      "grad_norm": 988.7250366210938,
-      "learning_rate": 3.880570081859597e-05,
-      "loss": 83.2871,
-      "step": 95600
-    },
-    {
-      "epoch": 0.386276498179923,
-      "grad_norm": 433.3272705078125,
-      "learning_rate": 3.880279054609045e-05,
-      "loss": 45.9184,
-      "step": 95610
-    },
-    {
-      "epoch": 0.38631689944529063,
-      "grad_norm": 1444.31787109375,
-      "learning_rate": 3.8799880004492425e-05,
-      "loss": 67.2129,
-      "step": 95620
-    },
-    {
-      "epoch": 0.38635730071065827,
-      "grad_norm": 507.1636047363281,
-      "learning_rate": 3.879696919385864e-05,
-      "loss": 44.7271,
-      "step": 95630
-    },
-    {
-      "epoch": 0.3863977019760259,
-      "grad_norm": 1155.373046875,
-      "learning_rate": 3.879405811424583e-05,
-      "loss": 66.5983,
-      "step": 95640
-    },
-    {
-      "epoch": 0.3864381032413935,
-      "grad_norm": 490.9678955078125,
-      "learning_rate": 3.879114676571076e-05,
-      "loss": 60.9269,
-      "step": 95650
-    },
-    {
-      "epoch": 0.38647850450676113,
-      "grad_norm": 0.0,
-      "learning_rate": 3.878823514831018e-05,
-      "loss": 56.5793,
-      "step": 95660
-    },
-    {
-      "epoch": 0.3865189057721288,
-      "grad_norm": 1740.7720947265625,
-      "learning_rate": 3.878532326210086e-05,
-      "loss": 74.9253,
-      "step": 95670
-    },
-    {
-      "epoch": 0.3865593070374964,
-      "grad_norm": 437.5474853515625,
-      "learning_rate": 3.8782411107139564e-05,
-      "loss": 280.9561,
-      "step": 95680
-    },
-    {
-      "epoch": 0.38659970830286405,
-      "grad_norm": 1296.4512939453125,
-      "learning_rate": 3.877949868348307e-05,
-      "loss": 73.3886,
-      "step": 95690
-    },
-    {
-      "epoch": 0.3866401095682317,
-      "grad_norm": 372.0828552246094,
-      "learning_rate": 3.877658599118815e-05,
-      "loss": 70.0043,
-      "step": 95700
-    },
-    {
-      "epoch": 0.38668051083359933,
-      "grad_norm": 462.0464782714844,
-      "learning_rate": 3.87736730303116e-05,
-      "loss": 69.3731,
-      "step": 95710
-    },
-    {
-      "epoch": 0.3867209120989669,
-      "grad_norm": 674.0639038085938,
-      "learning_rate": 3.87707598009102e-05,
-      "loss": 82.2932,
-      "step": 95720
-    },
-    {
-      "epoch": 0.38676131336433456,
-      "grad_norm": 281.2317810058594,
-      "learning_rate": 3.8767846303040746e-05,
-      "loss": 66.2907,
-      "step": 95730
-    },
-    {
-      "epoch": 0.3868017146297022,
-      "grad_norm": 531.0217895507812,
-      "learning_rate": 3.876493253676004e-05,
-      "loss": 68.4629,
-      "step": 95740
-    },
-    {
-      "epoch": 0.38684211589506984,
-      "grad_norm": 871.7710571289062,
-      "learning_rate": 3.8762018502124894e-05,
-      "loss": 61.8916,
-      "step": 95750
-    },
-    {
-      "epoch": 0.3868825171604375,
-      "grad_norm": 600.0103759765625,
-      "learning_rate": 3.875910419919211e-05,
-      "loss": 86.7091,
-      "step": 95760
-    },
-    {
-      "epoch": 0.3869229184258051,
-      "grad_norm": 631.7959594726562,
-      "learning_rate": 3.87561896280185e-05,
-      "loss": 57.5459,
-      "step": 95770
-    },
-    {
-      "epoch": 0.3869633196911727,
-      "grad_norm": 459.5212097167969,
-      "learning_rate": 3.8753274788660894e-05,
-      "loss": 38.7924,
-      "step": 95780
-    },
-    {
-      "epoch": 0.38700372095654034,
-      "grad_norm": 722.750244140625,
-      "learning_rate": 3.875035968117612e-05,
-      "loss": 61.4237,
-      "step": 95790
-    },
-    {
-      "epoch": 0.387044122221908,
-      "grad_norm": 1434.1351318359375,
-      "learning_rate": 3.8747444305621e-05,
-      "loss": 45.202,
-      "step": 95800
-    },
-    {
-      "epoch": 0.3870845234872756,
-      "grad_norm": 1343.0447998046875,
-      "learning_rate": 3.874452866205237e-05,
-      "loss": 74.4229,
-      "step": 95810
-    },
-    {
-      "epoch": 0.38712492475264326,
-      "grad_norm": 851.3370971679688,
-      "learning_rate": 3.874161275052709e-05,
-      "loss": 57.2729,
-      "step": 95820
-    },
-    {
-      "epoch": 0.3871653260180109,
-      "grad_norm": 519.22216796875,
-      "learning_rate": 3.873869657110198e-05,
-      "loss": 52.2334,
-      "step": 95830
-    },
-    {
-      "epoch": 0.38720572728337854,
-      "grad_norm": 1355.063720703125,
-      "learning_rate": 3.873578012383393e-05,
-      "loss": 80.8646,
-      "step": 95840
-    },
-    {
-      "epoch": 0.3872461285487461,
-      "grad_norm": 615.1976318359375,
-      "learning_rate": 3.873286340877975e-05,
-      "loss": 58.7473,
-      "step": 95850
-    },
-    {
-      "epoch": 0.38728652981411377,
-      "grad_norm": 780.4457397460938,
-      "learning_rate": 3.8729946425996345e-05,
-      "loss": 72.4027,
-      "step": 95860
-    },
-    {
-      "epoch": 0.3873269310794814,
-      "grad_norm": 345.5774841308594,
-      "learning_rate": 3.8727029175540554e-05,
-      "loss": 61.2218,
-      "step": 95870
-    },
-    {
-      "epoch": 0.38736733234484905,
-      "grad_norm": 546.84228515625,
-      "learning_rate": 3.872411165746927e-05,
-      "loss": 47.3672,
-      "step": 95880
-    },
-    {
-      "epoch": 0.3874077336102167,
-      "grad_norm": 1022.2073364257812,
-      "learning_rate": 3.872119387183936e-05,
-      "loss": 74.9494,
-      "step": 95890
-    },
-    {
-      "epoch": 0.3874481348755843,
-      "grad_norm": 789.552978515625,
-      "learning_rate": 3.8718275818707715e-05,
-      "loss": 64.7529,
-      "step": 95900
-    },
-    {
-      "epoch": 0.3874885361409519,
-      "grad_norm": 446.9588623046875,
-      "learning_rate": 3.8715357498131214e-05,
-      "loss": 66.2357,
-      "step": 95910
-    },
-    {
-      "epoch": 0.38752893740631955,
-      "grad_norm": 432.23651123046875,
-      "learning_rate": 3.871243891016676e-05,
-      "loss": 65.1847,
-      "step": 95920
-    },
-    {
-      "epoch": 0.3875693386716872,
-      "grad_norm": 382.0699462890625,
-      "learning_rate": 3.870952005487125e-05,
-      "loss": 46.396,
-      "step": 95930
-    },
-    {
-      "epoch": 0.38760973993705483,
-      "grad_norm": 553.678955078125,
-      "learning_rate": 3.870660093230159e-05,
-      "loss": 59.2287,
-      "step": 95940
-    },
-    {
-      "epoch": 0.38765014120242247,
-      "grad_norm": 397.8702392578125,
-      "learning_rate": 3.870368154251469e-05,
-      "loss": 57.8399,
-      "step": 95950
-    },
-    {
-      "epoch": 0.3876905424677901,
-      "grad_norm": 776.6229248046875,
-      "learning_rate": 3.870076188556746e-05,
-      "loss": 49.7638,
-      "step": 95960
-    },
-    {
-      "epoch": 0.3877309437331577,
-      "grad_norm": 940.7283325195312,
-      "learning_rate": 3.869784196151682e-05,
-      "loss": 69.5812,
-      "step": 95970
-    },
-    {
-      "epoch": 0.38777134499852534,
-      "grad_norm": 1250.880859375,
-      "learning_rate": 3.869492177041971e-05,
-      "loss": 86.8493,
-      "step": 95980
-    },
-    {
-      "epoch": 0.387811746263893,
-      "grad_norm": 718.7807006835938,
-      "learning_rate": 3.8692001312333036e-05,
-      "loss": 64.5188,
-      "step": 95990
-    },
-    {
-      "epoch": 0.3878521475292606,
-      "grad_norm": 1685.44921875,
-      "learning_rate": 3.868908058731376e-05,
-      "loss": 81.74,
-      "step": 96000
-    },
-    {
-      "epoch": 0.38789254879462826,
-      "grad_norm": 1482.5496826171875,
-      "learning_rate": 3.8686159595418805e-05,
-      "loss": 85.64,
-      "step": 96010
-    },
-    {
-      "epoch": 0.3879329500599959,
-      "grad_norm": 494.4794006347656,
-      "learning_rate": 3.868323833670512e-05,
-      "loss": 75.5731,
-      "step": 96020
-    },
-    {
-      "epoch": 0.38797335132536354,
-      "grad_norm": 1048.5169677734375,
-      "learning_rate": 3.868031681122966e-05,
-      "loss": 100.7712,
-      "step": 96030
-    },
-    {
-      "epoch": 0.3880137525907311,
-      "grad_norm": 431.1014099121094,
-      "learning_rate": 3.867739501904938e-05,
-      "loss": 41.6055,
-      "step": 96040
-    },
-    {
-      "epoch": 0.38805415385609876,
-      "grad_norm": 670.1493530273438,
-      "learning_rate": 3.867447296022124e-05,
-      "loss": 55.9565,
-      "step": 96050
-    },
-    {
-      "epoch": 0.3880945551214664,
-      "grad_norm": 826.1647338867188,
-      "learning_rate": 3.8671550634802216e-05,
-      "loss": 64.8045,
-      "step": 96060
-    },
-    {
-      "epoch": 0.38813495638683404,
-      "grad_norm": 408.30645751953125,
-      "learning_rate": 3.866862804284928e-05,
-      "loss": 48.8189,
-      "step": 96070
-    },
-    {
-      "epoch": 0.3881753576522017,
-      "grad_norm": 523.0665283203125,
-      "learning_rate": 3.8665705184419386e-05,
-      "loss": 39.3677,
-      "step": 96080
-    },
-    {
-      "epoch": 0.3882157589175693,
-      "grad_norm": 1171.9281005859375,
-      "learning_rate": 3.8662782059569546e-05,
-      "loss": 76.4366,
-      "step": 96090
-    },
-    {
-      "epoch": 0.3882561601829369,
-      "grad_norm": 688.5985717773438,
-      "learning_rate": 3.865985866835673e-05,
-      "loss": 54.2586,
-      "step": 96100
-    },
-    {
-      "epoch": 0.38829656144830454,
-      "grad_norm": 703.115234375,
-      "learning_rate": 3.865693501083794e-05,
-      "loss": 41.2508,
-      "step": 96110
-    },
-    {
-      "epoch": 0.3883369627136722,
-      "grad_norm": 764.5994262695312,
-      "learning_rate": 3.865401108707017e-05,
-      "loss": 70.6168,
-      "step": 96120
-    },
-    {
-      "epoch": 0.3883773639790398,
-      "grad_norm": 653.46240234375,
-      "learning_rate": 3.8651086897110424e-05,
-      "loss": 95.4823,
-      "step": 96130
-    },
-    {
-      "epoch": 0.38841776524440746,
-      "grad_norm": 1796.30029296875,
-      "learning_rate": 3.864816244101571e-05,
-      "loss": 83.386,
-      "step": 96140
-    },
-    {
-      "epoch": 0.3884581665097751,
-      "grad_norm": 552.94677734375,
-      "learning_rate": 3.8645237718843044e-05,
-      "loss": 56.03,
-      "step": 96150
-    },
-    {
-      "epoch": 0.38849856777514274,
-      "grad_norm": 776.851806640625,
-      "learning_rate": 3.864231273064944e-05,
-      "loss": 82.2406,
-      "step": 96160
-    },
-    {
-      "epoch": 0.38853896904051033,
-      "grad_norm": 465.9934997558594,
-      "learning_rate": 3.8639387476491926e-05,
-      "loss": 68.4747,
-      "step": 96170
-    },
-    {
-      "epoch": 0.38857937030587797,
-      "grad_norm": 2382.02685546875,
-      "learning_rate": 3.863646195642754e-05,
-      "loss": 96.2178,
-      "step": 96180
-    },
-    {
-      "epoch": 0.3886197715712456,
-      "grad_norm": 781.694580078125,
-      "learning_rate": 3.8633536170513296e-05,
-      "loss": 69.0198,
-      "step": 96190
-    },
-    {
-      "epoch": 0.38866017283661325,
-      "grad_norm": 822.1322631835938,
-      "learning_rate": 3.8630610118806254e-05,
-      "loss": 56.8671,
-      "step": 96200
-    },
-    {
-      "epoch": 0.3887005741019809,
-      "grad_norm": 518.6736450195312,
-      "learning_rate": 3.862768380136345e-05,
-      "loss": 43.1214,
-      "step": 96210
-    },
-    {
-      "epoch": 0.38874097536734853,
-      "grad_norm": 1137.239013671875,
-      "learning_rate": 3.862475721824193e-05,
-      "loss": 77.5319,
-      "step": 96220
-    },
-    {
-      "epoch": 0.3887813766327161,
-      "grad_norm": 113.21827697753906,
-      "learning_rate": 3.862183036949875e-05,
-      "loss": 64.3933,
-      "step": 96230
-    },
-    {
-      "epoch": 0.38882177789808375,
-      "grad_norm": 1088.9544677734375,
-      "learning_rate": 3.861890325519098e-05,
-      "loss": 69.0374,
-      "step": 96240
-    },
-    {
-      "epoch": 0.3888621791634514,
-      "grad_norm": 1370.9381103515625,
-      "learning_rate": 3.861597587537568e-05,
-      "loss": 59.8645,
-      "step": 96250
-    },
-    {
-      "epoch": 0.38890258042881903,
-      "grad_norm": 523.1607055664062,
-      "learning_rate": 3.861304823010991e-05,
-      "loss": 71.4097,
-      "step": 96260
-    },
-    {
-      "epoch": 0.3889429816941867,
-      "grad_norm": 959.0990600585938,
-      "learning_rate": 3.861012031945077e-05,
-      "loss": 42.5465,
-      "step": 96270
-    },
-    {
-      "epoch": 0.3889833829595543,
-      "grad_norm": 660.4682006835938,
-      "learning_rate": 3.8607192143455326e-05,
-      "loss": 55.936,
-      "step": 96280
-    },
-    {
-      "epoch": 0.3890237842249219,
-      "grad_norm": 748.14990234375,
-      "learning_rate": 3.860426370218067e-05,
-      "loss": 43.6322,
-      "step": 96290
-    },
-    {
-      "epoch": 0.38906418549028954,
-      "grad_norm": 584.9511108398438,
-      "learning_rate": 3.860133499568387e-05,
-      "loss": 70.2999,
-      "step": 96300
-    },
-    {
-      "epoch": 0.3891045867556572,
-      "grad_norm": 1015.5634765625,
-      "learning_rate": 3.859840602402206e-05,
-      "loss": 109.6341,
-      "step": 96310
-    },
-    {
-      "epoch": 0.3891449880210248,
-      "grad_norm": 1177.64697265625,
-      "learning_rate": 3.859547678725231e-05,
-      "loss": 63.7178,
-      "step": 96320
-    },
-    {
-      "epoch": 0.38918538928639246,
-      "grad_norm": 480.65618896484375,
-      "learning_rate": 3.859254728543175e-05,
-      "loss": 62.31,
-      "step": 96330
-    },
-    {
-      "epoch": 0.3892257905517601,
-      "grad_norm": 1283.1507568359375,
-      "learning_rate": 3.8589617518617485e-05,
-      "loss": 61.4974,
-      "step": 96340
-    },
-    {
-      "epoch": 0.38926619181712774,
-      "grad_norm": 1167.147216796875,
-      "learning_rate": 3.858668748686662e-05,
-      "loss": 48.8035,
-      "step": 96350
-    },
-    {
-      "epoch": 0.3893065930824953,
-      "grad_norm": 793.9154052734375,
-      "learning_rate": 3.858375719023629e-05,
-      "loss": 68.4192,
-      "step": 96360
-    },
-    {
-      "epoch": 0.38934699434786296,
-      "grad_norm": 838.4378051757812,
-      "learning_rate": 3.8580826628783625e-05,
-      "loss": 86.5221,
-      "step": 96370
-    },
-    {
-      "epoch": 0.3893873956132306,
-      "grad_norm": 797.8261108398438,
-      "learning_rate": 3.857789580256575e-05,
-      "loss": 63.2042,
-      "step": 96380
-    },
-    {
-      "epoch": 0.38942779687859824,
-      "grad_norm": 601.6024169921875,
-      "learning_rate": 3.857496471163981e-05,
-      "loss": 52.494,
-      "step": 96390
-    },
-    {
-      "epoch": 0.3894681981439659,
-      "grad_norm": 693.2849731445312,
-      "learning_rate": 3.8572033356062943e-05,
-      "loss": 88.3605,
-      "step": 96400
-    },
-    {
-      "epoch": 0.3895085994093335,
-      "grad_norm": 1151.2178955078125,
-      "learning_rate": 3.8569101735892296e-05,
-      "loss": 47.6969,
-      "step": 96410
-    },
-    {
-      "epoch": 0.3895490006747011,
-      "grad_norm": 534.7440185546875,
-      "learning_rate": 3.856616985118502e-05,
-      "loss": 56.3229,
-      "step": 96420
-    },
-    {
-      "epoch": 0.38958940194006875,
-      "grad_norm": 883.9851684570312,
-      "learning_rate": 3.8563237701998286e-05,
-      "loss": 44.7732,
-      "step": 96430
-    },
-    {
-      "epoch": 0.3896298032054364,
-      "grad_norm": 903.25341796875,
-      "learning_rate": 3.856030528838925e-05,
-      "loss": 49.8167,
-      "step": 96440
-    },
-    {
-      "epoch": 0.389670204470804,
-      "grad_norm": 1717.411376953125,
-      "learning_rate": 3.8557372610415074e-05,
-      "loss": 56.202,
-      "step": 96450
-    },
-    {
-      "epoch": 0.38971060573617167,
-      "grad_norm": 1218.4908447265625,
-      "learning_rate": 3.8554439668132946e-05,
-      "loss": 48.3608,
-      "step": 96460
-    },
-    {
-      "epoch": 0.3897510070015393,
-      "grad_norm": 884.7052001953125,
-      "learning_rate": 3.855150646160003e-05,
-      "loss": 46.3305,
-      "step": 96470
-    },
-    {
-      "epoch": 0.38979140826690695,
-      "grad_norm": 360.12017822265625,
-      "learning_rate": 3.854857299087353e-05,
-      "loss": 37.3111,
-      "step": 96480
-    },
-    {
-      "epoch": 0.38983180953227453,
-      "grad_norm": 1448.187744140625,
-      "learning_rate": 3.8545639256010625e-05,
-      "loss": 71.6807,
-      "step": 96490
-    },
-    {
-      "epoch": 0.38987221079764217,
-      "grad_norm": 935.8035278320312,
-      "learning_rate": 3.85427052570685e-05,
-      "loss": 60.7646,
-      "step": 96500
-    },
-    {
-      "epoch": 0.3899126120630098,
-      "grad_norm": 805.13037109375,
-      "learning_rate": 3.853977099410436e-05,
-      "loss": 46.3225,
-      "step": 96510
-    },
-    {
-      "epoch": 0.38995301332837745,
-      "grad_norm": 562.0592041015625,
-      "learning_rate": 3.853683646717543e-05,
-      "loss": 62.2224,
-      "step": 96520
-    },
-    {
-      "epoch": 0.3899934145937451,
-      "grad_norm": 1327.85498046875,
-      "learning_rate": 3.853390167633889e-05,
-      "loss": 66.2915,
-      "step": 96530
-    },
-    {
-      "epoch": 0.39003381585911273,
-      "grad_norm": 1803.8189697265625,
-      "learning_rate": 3.8530966621651976e-05,
-      "loss": 89.2004,
-      "step": 96540
-    },
-    {
-      "epoch": 0.3900742171244803,
-      "grad_norm": 363.34783935546875,
-      "learning_rate": 3.8528031303171895e-05,
-      "loss": 38.5132,
-      "step": 96550
-    },
-    {
-      "epoch": 0.39011461838984796,
-      "grad_norm": 1082.673095703125,
-      "learning_rate": 3.852509572095588e-05,
-      "loss": 58.43,
-      "step": 96560
-    },
-    {
-      "epoch": 0.3901550196552156,
-      "grad_norm": 1270.4439697265625,
-      "learning_rate": 3.852215987506117e-05,
-      "loss": 55.3348,
-      "step": 96570
-    },
-    {
-      "epoch": 0.39019542092058324,
-      "grad_norm": 478.66815185546875,
-      "learning_rate": 3.851922376554499e-05,
-      "loss": 59.5862,
-      "step": 96580
-    },
-    {
-      "epoch": 0.3902358221859509,
-      "grad_norm": 772.7987670898438,
-      "learning_rate": 3.851628739246457e-05,
-      "loss": 53.7853,
-      "step": 96590
-    },
-    {
-      "epoch": 0.3902762234513185,
-      "grad_norm": 353.50360107421875,
-      "learning_rate": 3.851335075587718e-05,
-      "loss": 68.1417,
-      "step": 96600
-    },
-    {
-      "epoch": 0.3903166247166861,
-      "grad_norm": 3968.27587890625,
-      "learning_rate": 3.8510413855840056e-05,
-      "loss": 92.6137,
-      "step": 96610
-    },
-    {
-      "epoch": 0.39035702598205374,
-      "grad_norm": 624.1312255859375,
-      "learning_rate": 3.850747669241046e-05,
-      "loss": 74.6902,
-      "step": 96620
-    },
-    {
-      "epoch": 0.3903974272474214,
-      "grad_norm": 1058.5302734375,
-      "learning_rate": 3.850453926564565e-05,
-      "loss": 67.8454,
-      "step": 96630
-    },
-    {
-      "epoch": 0.390437828512789,
-      "grad_norm": 1056.72021484375,
-      "learning_rate": 3.85016015756029e-05,
-      "loss": 88.1807,
-      "step": 96640
-    },
-    {
-      "epoch": 0.39047822977815666,
-      "grad_norm": 1824.6705322265625,
-      "learning_rate": 3.849866362233947e-05,
-      "loss": 71.9311,
-      "step": 96650
-    },
-    {
-      "epoch": 0.3905186310435243,
-      "grad_norm": 608.2755126953125,
-      "learning_rate": 3.849572540591264e-05,
-      "loss": 45.0875,
-      "step": 96660
-    },
-    {
-      "epoch": 0.39055903230889194,
-      "grad_norm": 665.3267822265625,
-      "learning_rate": 3.84927869263797e-05,
-      "loss": 67.2472,
-      "step": 96670
-    },
-    {
-      "epoch": 0.3905994335742595,
-      "grad_norm": 752.25927734375,
-      "learning_rate": 3.848984818379793e-05,
-      "loss": 108.7115,
-      "step": 96680
-    },
-    {
-      "epoch": 0.39063983483962716,
-      "grad_norm": 393.086181640625,
-      "learning_rate": 3.848690917822463e-05,
-      "loss": 61.483,
-      "step": 96690
-    },
-    {
-      "epoch": 0.3906802361049948,
-      "grad_norm": 1095.4296875,
-      "learning_rate": 3.8483969909717087e-05,
-      "loss": 57.5574,
-      "step": 96700
-    },
-    {
-      "epoch": 0.39072063737036244,
-      "grad_norm": 841.97705078125,
-      "learning_rate": 3.8481030378332614e-05,
-      "loss": 68.264,
-      "step": 96710
-    },
-    {
-      "epoch": 0.3907610386357301,
-      "grad_norm": 410.4032897949219,
-      "learning_rate": 3.84780905841285e-05,
-      "loss": 69.9679,
-      "step": 96720
-    },
-    {
-      "epoch": 0.3908014399010977,
-      "grad_norm": 513.5309448242188,
-      "learning_rate": 3.8475150527162085e-05,
-      "loss": 80.5341,
-      "step": 96730
-    },
-    {
-      "epoch": 0.3908418411664653,
-      "grad_norm": 451.75445556640625,
-      "learning_rate": 3.847221020749067e-05,
-      "loss": 45.8152,
-      "step": 96740
-    },
-    {
-      "epoch": 0.39088224243183295,
-      "grad_norm": 1032.3883056640625,
-      "learning_rate": 3.8469269625171576e-05,
-      "loss": 84.8716,
-      "step": 96750
-    },
-    {
-      "epoch": 0.3909226436972006,
-      "grad_norm": 472.958251953125,
-      "learning_rate": 3.846632878026214e-05,
-      "loss": 49.6663,
-      "step": 96760
-    },
-    {
-      "epoch": 0.39096304496256823,
-      "grad_norm": 969.470703125,
-      "learning_rate": 3.8463387672819696e-05,
-      "loss": 63.3029,
-      "step": 96770
-    },
-    {
-      "epoch": 0.39100344622793587,
-      "grad_norm": 1277.8431396484375,
-      "learning_rate": 3.846044630290158e-05,
-      "loss": 56.5187,
-      "step": 96780
-    },
-    {
-      "epoch": 0.3910438474933035,
-      "grad_norm": 364.2802734375,
-      "learning_rate": 3.845750467056511e-05,
-      "loss": 59.6623,
-      "step": 96790
-    },
-    {
-      "epoch": 0.39108424875867115,
-      "grad_norm": 982.3886108398438,
-      "learning_rate": 3.8454562775867684e-05,
-      "loss": 85.9636,
-      "step": 96800
-    },
-    {
-      "epoch": 0.39112465002403873,
-      "grad_norm": 416.16619873046875,
-      "learning_rate": 3.8451620618866616e-05,
-      "loss": 84.5701,
-      "step": 96810
-    },
-    {
-      "epoch": 0.3911650512894064,
-      "grad_norm": 627.9135131835938,
-      "learning_rate": 3.844867819961928e-05,
-      "loss": 44.0341,
-      "step": 96820
-    },
-    {
-      "epoch": 0.391205452554774,
-      "grad_norm": 1361.487060546875,
-      "learning_rate": 3.8445735518183043e-05,
-      "loss": 48.6817,
-      "step": 96830
-    },
-    {
-      "epoch": 0.39124585382014165,
-      "grad_norm": 1403.5098876953125,
-      "learning_rate": 3.8442792574615275e-05,
-      "loss": 49.1378,
-      "step": 96840
-    },
-    {
-      "epoch": 0.3912862550855093,
-      "grad_norm": 1468.5533447265625,
-      "learning_rate": 3.843984936897334e-05,
-      "loss": 71.1556,
-      "step": 96850
-    },
-    {
-      "epoch": 0.39132665635087693,
-      "grad_norm": 2194.382080078125,
-      "learning_rate": 3.843690590131462e-05,
-      "loss": 89.6682,
-      "step": 96860
-    },
-    {
-      "epoch": 0.3913670576162445,
-      "grad_norm": 899.1420288085938,
-      "learning_rate": 3.84339621716965e-05,
-      "loss": 59.679,
-      "step": 96870
-    },
-    {
-      "epoch": 0.39140745888161216,
-      "grad_norm": 726.7706298828125,
-      "learning_rate": 3.843101818017637e-05,
-      "loss": 81.8773,
-      "step": 96880
-    },
-    {
-      "epoch": 0.3914478601469798,
-      "grad_norm": 441.8895568847656,
-      "learning_rate": 3.8428073926811625e-05,
-      "loss": 87.1567,
-      "step": 96890
-    },
-    {
-      "epoch": 0.39148826141234744,
-      "grad_norm": 788.19921875,
-      "learning_rate": 3.842512941165968e-05,
-      "loss": 81.7942,
-      "step": 96900
-    },
-    {
-      "epoch": 0.3915286626777151,
-      "grad_norm": 0.0,
-      "learning_rate": 3.842218463477791e-05,
-      "loss": 45.6026,
-      "step": 96910
-    },
-    {
-      "epoch": 0.3915690639430827,
-      "grad_norm": 864.9554443359375,
-      "learning_rate": 3.841923959622375e-05,
-      "loss": 49.6541,
-      "step": 96920
-    },
-    {
-      "epoch": 0.3916094652084503,
-      "grad_norm": 903.1978149414062,
-      "learning_rate": 3.84162942960546e-05,
-      "loss": 76.4913,
-      "step": 96930
-    },
-    {
-      "epoch": 0.39164986647381794,
-      "grad_norm": 848.7103271484375,
-      "learning_rate": 3.841334873432789e-05,
-      "loss": 67.0842,
-      "step": 96940
-    },
-    {
-      "epoch": 0.3916902677391856,
-      "grad_norm": 898.5715942382812,
-      "learning_rate": 3.841040291110103e-05,
-      "loss": 40.5335,
-      "step": 96950
-    },
-    {
-      "epoch": 0.3917306690045532,
-      "grad_norm": 422.8970031738281,
-      "learning_rate": 3.840745682643147e-05,
-      "loss": 62.1978,
-      "step": 96960
-    },
-    {
-      "epoch": 0.39177107026992086,
-      "grad_norm": 404.5113220214844,
-      "learning_rate": 3.840451048037663e-05,
-      "loss": 56.2681,
-      "step": 96970
-    },
-    {
-      "epoch": 0.3918114715352885,
-      "grad_norm": 1311.5926513671875,
-      "learning_rate": 3.8401563872993966e-05,
-      "loss": 63.1407,
-      "step": 96980
-    },
-    {
-      "epoch": 0.39185187280065614,
-      "grad_norm": 477.2160339355469,
-      "learning_rate": 3.839861700434091e-05,
-      "loss": 51.8024,
-      "step": 96990
-    },
-    {
-      "epoch": 0.3918922740660237,
-      "grad_norm": 835.2833862304688,
-      "learning_rate": 3.8395669874474915e-05,
-      "loss": 48.2391,
-      "step": 97000
-    },
-    {
-      "epoch": 0.39193267533139137,
-      "grad_norm": 906.5626831054688,
-      "learning_rate": 3.839272248345344e-05,
-      "loss": 61.7122,
-      "step": 97010
-    },
-    {
-      "epoch": 0.391973076596759,
-      "grad_norm": 321.434326171875,
-      "learning_rate": 3.838977483133395e-05,
-      "loss": 66.5921,
-      "step": 97020
-    },
-    {
-      "epoch": 0.39201347786212665,
-      "grad_norm": 408.02197265625,
-      "learning_rate": 3.838682691817391e-05,
-      "loss": 47.1528,
-      "step": 97030
-    },
-    {
-      "epoch": 0.3920538791274943,
-      "grad_norm": 953.2274780273438,
-      "learning_rate": 3.8383878744030776e-05,
-      "loss": 47.6821,
-      "step": 97040
-    },
-    {
-      "epoch": 0.3920942803928619,
-      "grad_norm": 541.6514282226562,
-      "learning_rate": 3.8380930308962036e-05,
-      "loss": 60.5129,
-      "step": 97050
-    },
-    {
-      "epoch": 0.3921346816582295,
-      "grad_norm": 443.59527587890625,
-      "learning_rate": 3.837798161302518e-05,
-      "loss": 67.278,
-      "step": 97060
-    },
-    {
-      "epoch": 0.39217508292359715,
-      "grad_norm": 1681.9339599609375,
-      "learning_rate": 3.8375032656277684e-05,
-      "loss": 59.9378,
-      "step": 97070
-    },
-    {
-      "epoch": 0.3922154841889648,
-      "grad_norm": 696.7039184570312,
-      "learning_rate": 3.837208343877703e-05,
-      "loss": 49.9206,
-      "step": 97080
-    },
-    {
-      "epoch": 0.39225588545433243,
-      "grad_norm": 885.4489135742188,
-      "learning_rate": 3.8369133960580724e-05,
-      "loss": 69.6553,
-      "step": 97090
-    },
-    {
-      "epoch": 0.39229628671970007,
-      "grad_norm": 688.5819702148438,
-      "learning_rate": 3.836618422174628e-05,
-      "loss": 80.361,
-      "step": 97100
-    },
-    {
-      "epoch": 0.3923366879850677,
-      "grad_norm": 757.7618408203125,
-      "learning_rate": 3.83632342223312e-05,
-      "loss": 49.0418,
-      "step": 97110
-    },
-    {
-      "epoch": 0.39237708925043535,
-      "grad_norm": 435.7521057128906,
-      "learning_rate": 3.836028396239297e-05,
-      "loss": 52.3195,
-      "step": 97120
-    },
-    {
-      "epoch": 0.39241749051580294,
-      "grad_norm": 622.2557983398438,
-      "learning_rate": 3.8357333441989134e-05,
-      "loss": 69.8973,
-      "step": 97130
-    },
-    {
-      "epoch": 0.3924578917811706,
-      "grad_norm": 2009.5267333984375,
-      "learning_rate": 3.835438266117721e-05,
-      "loss": 63.6782,
-      "step": 97140
-    },
-    {
-      "epoch": 0.3924982930465382,
-      "grad_norm": 851.5496826171875,
-      "learning_rate": 3.835143162001472e-05,
-      "loss": 79.4912,
-      "step": 97150
-    },
-    {
-      "epoch": 0.39253869431190586,
-      "grad_norm": 496.26544189453125,
-      "learning_rate": 3.834848031855919e-05,
-      "loss": 53.252,
-      "step": 97160
-    },
-    {
-      "epoch": 0.3925790955772735,
-      "grad_norm": 664.895263671875,
-      "learning_rate": 3.8345528756868164e-05,
-      "loss": 51.9402,
-      "step": 97170
-    },
-    {
-      "epoch": 0.39261949684264114,
-      "grad_norm": 963.7388305664062,
-      "learning_rate": 3.8342576934999184e-05,
-      "loss": 57.3356,
-      "step": 97180
-    },
-    {
-      "epoch": 0.3926598981080087,
-      "grad_norm": 706.6784057617188,
-      "learning_rate": 3.83396248530098e-05,
-      "loss": 74.5728,
-      "step": 97190
-    },
-    {
-      "epoch": 0.39270029937337636,
-      "grad_norm": 486.8099365234375,
-      "learning_rate": 3.8336672510957574e-05,
-      "loss": 64.1127,
-      "step": 97200
-    },
-    {
-      "epoch": 0.392740700638744,
-      "grad_norm": 563.300048828125,
-      "learning_rate": 3.833371990890003e-05,
-      "loss": 53.4073,
-      "step": 97210
-    },
-    {
-      "epoch": 0.39278110190411164,
-      "grad_norm": 382.8632507324219,
-      "learning_rate": 3.8330767046894765e-05,
-      "loss": 44.1544,
-      "step": 97220
-    },
-    {
-      "epoch": 0.3928215031694793,
-      "grad_norm": 944.963134765625,
-      "learning_rate": 3.8327813924999326e-05,
-      "loss": 64.7648,
-      "step": 97230
-    },
-    {
-      "epoch": 0.3928619044348469,
-      "grad_norm": 428.9286193847656,
-      "learning_rate": 3.83248605432713e-05,
-      "loss": 57.617,
-      "step": 97240
-    },
-    {
-      "epoch": 0.3929023057002145,
-      "grad_norm": 742.8692016601562,
-      "learning_rate": 3.832190690176825e-05,
-      "loss": 44.4798,
-      "step": 97250
-    },
-    {
-      "epoch": 0.39294270696558214,
-      "grad_norm": 139.9200439453125,
-      "learning_rate": 3.831895300054777e-05,
-      "loss": 73.2557,
-      "step": 97260
-    },
-    {
-      "epoch": 0.3929831082309498,
-      "grad_norm": 2734.72802734375,
-      "learning_rate": 3.8315998839667445e-05,
-      "loss": 73.5291,
-      "step": 97270
-    },
-    {
-      "epoch": 0.3930235094963174,
-      "grad_norm": 1699.1915283203125,
-      "learning_rate": 3.8313044419184873e-05,
-      "loss": 92.5831,
-      "step": 97280
-    },
-    {
-      "epoch": 0.39306391076168506,
-      "grad_norm": 1669.2913818359375,
-      "learning_rate": 3.831008973915764e-05,
-      "loss": 54.4022,
-      "step": 97290
-    },
-    {
-      "epoch": 0.3931043120270527,
-      "grad_norm": 1141.6053466796875,
-      "learning_rate": 3.830713479964335e-05,
-      "loss": 85.7646,
-      "step": 97300
-    },
-    {
-      "epoch": 0.39314471329242034,
-      "grad_norm": 1941.885986328125,
-      "learning_rate": 3.8304179600699626e-05,
-      "loss": 74.6156,
-      "step": 97310
-    },
-    {
-      "epoch": 0.39318511455778793,
-      "grad_norm": 877.8158569335938,
-      "learning_rate": 3.830122414238406e-05,
-      "loss": 56.3947,
-      "step": 97320
-    },
-    {
-      "epoch": 0.39322551582315557,
-      "grad_norm": 600.1071166992188,
-      "learning_rate": 3.829826842475429e-05,
-      "loss": 56.1246,
-      "step": 97330
-    },
-    {
-      "epoch": 0.3932659170885232,
-      "grad_norm": 1540.666259765625,
-      "learning_rate": 3.8295312447867924e-05,
-      "loss": 58.5337,
-      "step": 97340
-    },
-    {
-      "epoch": 0.39330631835389085,
-      "grad_norm": 620.038330078125,
-      "learning_rate": 3.82923562117826e-05,
-      "loss": 40.1906,
-      "step": 97350
-    },
-    {
-      "epoch": 0.3933467196192585,
-      "grad_norm": 649.1076049804688,
-      "learning_rate": 3.828939971655595e-05,
-      "loss": 86.4342,
-      "step": 97360
-    },
-    {
-      "epoch": 0.39338712088462613,
-      "grad_norm": 1530.02001953125,
-      "learning_rate": 3.828644296224562e-05,
-      "loss": 61.6796,
-      "step": 97370
-    },
-    {
-      "epoch": 0.3934275221499937,
-      "grad_norm": 1663.7215576171875,
-      "learning_rate": 3.8283485948909224e-05,
-      "loss": 59.6131,
-      "step": 97380
-    },
-    {
-      "epoch": 0.39346792341536135,
-      "grad_norm": 509.4885559082031,
-      "learning_rate": 3.828052867660445e-05,
-      "loss": 60.7145,
-      "step": 97390
-    },
-    {
-      "epoch": 0.393508324680729,
-      "grad_norm": 586.28955078125,
-      "learning_rate": 3.827757114538892e-05,
-      "loss": 95.243,
-      "step": 97400
-    },
-    {
-      "epoch": 0.39354872594609663,
-      "grad_norm": 808.3604736328125,
-      "learning_rate": 3.82746133553203e-05,
-      "loss": 66.3657,
-      "step": 97410
-    },
-    {
-      "epoch": 0.3935891272114643,
-      "grad_norm": 307.7069396972656,
-      "learning_rate": 3.827165530645627e-05,
-      "loss": 69.5492,
-      "step": 97420
-    },
-    {
-      "epoch": 0.3936295284768319,
-      "grad_norm": 1287.857666015625,
-      "learning_rate": 3.8268696998854486e-05,
-      "loss": 74.0201,
-      "step": 97430
-    },
-    {
-      "epoch": 0.3936699297421995,
-      "grad_norm": 1741.658935546875,
-      "learning_rate": 3.826573843257262e-05,
-      "loss": 95.2432,
-      "step": 97440
-    },
-    {
-      "epoch": 0.39371033100756714,
-      "grad_norm": 530.0614624023438,
-      "learning_rate": 3.826277960766835e-05,
-      "loss": 76.4804,
-      "step": 97450
-    },
-    {
-      "epoch": 0.3937507322729348,
-      "grad_norm": 449.9317932128906,
-      "learning_rate": 3.8259820524199374e-05,
-      "loss": 58.2512,
-      "step": 97460
-    },
-    {
-      "epoch": 0.3937911335383024,
-      "grad_norm": 273.4659729003906,
-      "learning_rate": 3.8256861182223366e-05,
-      "loss": 50.8024,
-      "step": 97470
-    },
-    {
-      "epoch": 0.39383153480367006,
-      "grad_norm": 1517.554931640625,
-      "learning_rate": 3.8253901581798016e-05,
-      "loss": 65.5294,
-      "step": 97480
-    },
-    {
-      "epoch": 0.3938719360690377,
-      "grad_norm": 286.39971923828125,
-      "learning_rate": 3.825094172298104e-05,
-      "loss": 41.5933,
-      "step": 97490
-    },
-    {
-      "epoch": 0.39391233733440534,
-      "grad_norm": 1205.6583251953125,
-      "learning_rate": 3.824798160583012e-05,
-      "loss": 59.1793,
-      "step": 97500
-    },
-    {
-      "epoch": 0.3939527385997729,
-      "grad_norm": 1654.855224609375,
-      "learning_rate": 3.824502123040299e-05,
-      "loss": 71.1765,
-      "step": 97510
-    },
-    {
-      "epoch": 0.39399313986514056,
-      "grad_norm": 636.5582275390625,
-      "learning_rate": 3.824206059675736e-05,
-      "loss": 76.6066,
-      "step": 97520
-    },
-    {
-      "epoch": 0.3940335411305082,
-      "grad_norm": 878.0634765625,
-      "learning_rate": 3.823909970495092e-05,
-      "loss": 66.1705,
-      "step": 97530
-    },
-    {
-      "epoch": 0.39407394239587584,
-      "grad_norm": 733.1356811523438,
-      "learning_rate": 3.8236138555041434e-05,
-      "loss": 86.5831,
-      "step": 97540
-    },
-    {
-      "epoch": 0.3941143436612435,
-      "grad_norm": 372.1419982910156,
-      "learning_rate": 3.823317714708661e-05,
-      "loss": 52.7755,
-      "step": 97550
-    },
-    {
-      "epoch": 0.3941547449266111,
-      "grad_norm": 978.9861450195312,
-      "learning_rate": 3.823021548114417e-05,
-      "loss": 55.6883,
-      "step": 97560
-    },
-    {
-      "epoch": 0.3941951461919787,
-      "grad_norm": 339.8622741699219,
-      "learning_rate": 3.822725355727188e-05,
-      "loss": 51.7447,
-      "step": 97570
-    },
-    {
-      "epoch": 0.39423554745734635,
-      "grad_norm": 610.7191162109375,
-      "learning_rate": 3.8224291375527464e-05,
-      "loss": 49.7207,
-      "step": 97580
-    },
-    {
-      "epoch": 0.394275948722714,
-      "grad_norm": 1472.842041015625,
-      "learning_rate": 3.822132893596869e-05,
-      "loss": 55.3662,
-      "step": 97590
-    },
-    {
-      "epoch": 0.3943163499880816,
-      "grad_norm": 558.3397827148438,
-      "learning_rate": 3.821836623865329e-05,
-      "loss": 91.5254,
-      "step": 97600
-    },
-    {
-      "epoch": 0.39435675125344927,
-      "grad_norm": 770.9660034179688,
-      "learning_rate": 3.821540328363905e-05,
-      "loss": 58.9525,
-      "step": 97610
-    },
-    {
-      "epoch": 0.3943971525188169,
-      "grad_norm": 900.81982421875,
-      "learning_rate": 3.821244007098371e-05,
-      "loss": 49.8541,
-      "step": 97620
-    },
-    {
-      "epoch": 0.39443755378418455,
-      "grad_norm": 472.5091247558594,
-      "learning_rate": 3.820947660074504e-05,
-      "loss": 70.776,
-      "step": 97630
-    },
-    {
-      "epoch": 0.39447795504955213,
-      "grad_norm": 1419.308349609375,
-      "learning_rate": 3.820651287298084e-05,
-      "loss": 62.8083,
-      "step": 97640
-    },
-    {
-      "epoch": 0.39451835631491977,
-      "grad_norm": 1251.2393798828125,
-      "learning_rate": 3.8203548887748865e-05,
-      "loss": 98.8956,
-      "step": 97650
-    },
-    {
-      "epoch": 0.3945587575802874,
-      "grad_norm": 568.4578857421875,
-      "learning_rate": 3.8200584645106904e-05,
-      "loss": 52.2117,
-      "step": 97660
-    },
-    {
-      "epoch": 0.39459915884565505,
-      "grad_norm": 1025.6453857421875,
-      "learning_rate": 3.819762014511275e-05,
-      "loss": 77.8451,
-      "step": 97670
-    },
-    {
-      "epoch": 0.3946395601110227,
-      "grad_norm": 1376.6019287109375,
-      "learning_rate": 3.81946553878242e-05,
-      "loss": 63.783,
-      "step": 97680
-    },
-    {
-      "epoch": 0.39467996137639033,
-      "grad_norm": 1392.58544921875,
-      "learning_rate": 3.819169037329905e-05,
-      "loss": 51.8016,
-      "step": 97690
-    },
-    {
-      "epoch": 0.3947203626417579,
-      "grad_norm": 737.3711547851562,
-      "learning_rate": 3.8188725101595094e-05,
-      "loss": 44.1351,
-      "step": 97700
-    },
-    {
-      "epoch": 0.39476076390712556,
-      "grad_norm": 675.7130737304688,
-      "learning_rate": 3.818575957277016e-05,
-      "loss": 70.7781,
-      "step": 97710
-    },
-    {
-      "epoch": 0.3948011651724932,
-      "grad_norm": 522.35302734375,
-      "learning_rate": 3.8182793786882065e-05,
-      "loss": 54.7861,
-      "step": 97720
-    },
-    {
-      "epoch": 0.39484156643786084,
-      "grad_norm": 1283.5611572265625,
-      "learning_rate": 3.817982774398861e-05,
-      "loss": 86.0404,
-      "step": 97730
-    },
-    {
-      "epoch": 0.3948819677032285,
-      "grad_norm": 837.8106079101562,
-      "learning_rate": 3.817686144414762e-05,
-      "loss": 72.8473,
-      "step": 97740
-    },
-    {
-      "epoch": 0.3949223689685961,
-      "grad_norm": 2973.33740234375,
-      "learning_rate": 3.8173894887416945e-05,
-      "loss": 72.2817,
-      "step": 97750
-    },
-    {
-      "epoch": 0.3949627702339637,
-      "grad_norm": 433.107421875,
-      "learning_rate": 3.8170928073854396e-05,
-      "loss": 71.6905,
-      "step": 97760
-    },
-    {
-      "epoch": 0.39500317149933134,
-      "grad_norm": 559.8467407226562,
-      "learning_rate": 3.816796100351783e-05,
-      "loss": 59.3247,
-      "step": 97770
-    },
-    {
-      "epoch": 0.395043572764699,
-      "grad_norm": 676.9853515625,
-      "learning_rate": 3.8164993676465074e-05,
-      "loss": 59.816,
-      "step": 97780
-    },
-    {
-      "epoch": 0.3950839740300666,
-      "grad_norm": 1209.71923828125,
-      "learning_rate": 3.816202609275401e-05,
-      "loss": 78.3244,
-      "step": 97790
-    },
-    {
-      "epoch": 0.39512437529543426,
-      "grad_norm": 407.7916259765625,
-      "learning_rate": 3.8159058252442446e-05,
-      "loss": 40.6203,
-      "step": 97800
-    },
-    {
-      "epoch": 0.3951647765608019,
-      "grad_norm": 0.0,
-      "learning_rate": 3.815609015558829e-05,
-      "loss": 68.3833,
-      "step": 97810
-    },
-    {
-      "epoch": 0.39520517782616954,
-      "grad_norm": 235.8328094482422,
-      "learning_rate": 3.815312180224937e-05,
-      "loss": 47.2443,
-      "step": 97820
-    },
-    {
-      "epoch": 0.3952455790915371,
-      "grad_norm": 620.359619140625,
-      "learning_rate": 3.8150153192483566e-05,
-      "loss": 53.1402,
-      "step": 97830
-    },
-    {
-      "epoch": 0.39528598035690476,
-      "grad_norm": 1514.925537109375,
-      "learning_rate": 3.814718432634876e-05,
-      "loss": 83.4849,
-      "step": 97840
-    },
-    {
-      "epoch": 0.3953263816222724,
-      "grad_norm": 195.32774353027344,
-      "learning_rate": 3.8144215203902834e-05,
-      "loss": 58.4113,
-      "step": 97850
-    },
-    {
-      "epoch": 0.39536678288764004,
-      "grad_norm": 610.07568359375,
-      "learning_rate": 3.814124582520365e-05,
-      "loss": 47.7628,
-      "step": 97860
-    },
-    {
-      "epoch": 0.3954071841530077,
-      "grad_norm": 1019.016845703125,
-      "learning_rate": 3.813827619030913e-05,
-      "loss": 66.7314,
-      "step": 97870
-    },
-    {
-      "epoch": 0.3954475854183753,
-      "grad_norm": 340.5806884765625,
-      "learning_rate": 3.813530629927714e-05,
-      "loss": 37.2535,
-      "step": 97880
-    },
-    {
-      "epoch": 0.3954879866837429,
-      "grad_norm": 705.3890991210938,
-      "learning_rate": 3.81323361521656e-05,
-      "loss": 63.5229,
-      "step": 97890
-    },
-    {
-      "epoch": 0.39552838794911055,
-      "grad_norm": 911.9359741210938,
-      "learning_rate": 3.81293657490324e-05,
-      "loss": 107.5619,
-      "step": 97900
-    },
-    {
-      "epoch": 0.3955687892144782,
-      "grad_norm": 448.5742492675781,
-      "learning_rate": 3.812639508993545e-05,
-      "loss": 67.1163,
-      "step": 97910
-    },
-    {
-      "epoch": 0.39560919047984583,
-      "grad_norm": 292.3797607421875,
-      "learning_rate": 3.8123424174932674e-05,
-      "loss": 46.8763,
-      "step": 97920
-    },
-    {
-      "epoch": 0.39564959174521347,
-      "grad_norm": 516.19287109375,
-      "learning_rate": 3.812045300408199e-05,
-      "loss": 60.135,
-      "step": 97930
-    },
-    {
-      "epoch": 0.3956899930105811,
-      "grad_norm": 1726.6441650390625,
-      "learning_rate": 3.811748157744132e-05,
-      "loss": 111.1981,
-      "step": 97940
-    },
-    {
-      "epoch": 0.39573039427594875,
-      "grad_norm": 617.226806640625,
-      "learning_rate": 3.8114509895068586e-05,
-      "loss": 40.3687,
-      "step": 97950
-    },
-    {
-      "epoch": 0.39577079554131633,
-      "grad_norm": 1564.2078857421875,
-      "learning_rate": 3.811153795702174e-05,
-      "loss": 57.8685,
-      "step": 97960
-    },
-    {
-      "epoch": 0.395811196806684,
-      "grad_norm": 543.6961669921875,
-      "learning_rate": 3.81085657633587e-05,
-      "loss": 63.6772,
-      "step": 97970
-    },
-    {
-      "epoch": 0.3958515980720516,
-      "grad_norm": 619.5729370117188,
-      "learning_rate": 3.810559331413743e-05,
-      "loss": 53.4549,
-      "step": 97980
-    },
-    {
-      "epoch": 0.39589199933741925,
-      "grad_norm": 1818.6915283203125,
-      "learning_rate": 3.810262060941587e-05,
-      "loss": 46.7142,
-      "step": 97990
-    },
-    {
-      "epoch": 0.3959324006027869,
-      "grad_norm": 423.151611328125,
-      "learning_rate": 3.8099647649251986e-05,
-      "loss": 68.5805,
-      "step": 98000
-    },
-    {
-      "epoch": 0.39597280186815453,
-      "grad_norm": 797.4796752929688,
-      "learning_rate": 3.809667443370372e-05,
-      "loss": 43.375,
-      "step": 98010
-    },
-    {
-      "epoch": 0.3960132031335221,
-      "grad_norm": 478.58349609375,
-      "learning_rate": 3.809370096282902e-05,
-      "loss": 63.259,
-      "step": 98020
-    },
-    {
-      "epoch": 0.39605360439888976,
-      "grad_norm": 477.88470458984375,
-      "learning_rate": 3.8090727236685906e-05,
-      "loss": 85.8146,
-      "step": 98030
-    },
-    {
-      "epoch": 0.3960940056642574,
-      "grad_norm": 358.82342529296875,
-      "learning_rate": 3.808775325533232e-05,
-      "loss": 46.4715,
-      "step": 98040
-    },
-    {
-      "epoch": 0.39613440692962504,
-      "grad_norm": 1070.005615234375,
-      "learning_rate": 3.808477901882624e-05,
-      "loss": 50.1758,
-      "step": 98050
-    },
-    {
-      "epoch": 0.3961748081949927,
-      "grad_norm": 568.8511352539062,
-      "learning_rate": 3.808180452722566e-05,
-      "loss": 53.4005,
-      "step": 98060
-    },
-    {
-      "epoch": 0.3962152094603603,
-      "grad_norm": 523.6063232421875,
-      "learning_rate": 3.8078829780588564e-05,
-      "loss": 65.6613,
-      "step": 98070
-    },
-    {
-      "epoch": 0.3962556107257279,
-      "grad_norm": 1423.616943359375,
-      "learning_rate": 3.8075854778972955e-05,
-      "loss": 88.5826,
-      "step": 98080
-    },
-    {
-      "epoch": 0.39629601199109554,
-      "grad_norm": 821.159912109375,
-      "learning_rate": 3.807287952243682e-05,
-      "loss": 52.51,
-      "step": 98090
-    },
-    {
-      "epoch": 0.3963364132564632,
-      "grad_norm": 1087.4752197265625,
-      "learning_rate": 3.8069904011038165e-05,
-      "loss": 62.1011,
-      "step": 98100
-    },
-    {
-      "epoch": 0.3963768145218308,
-      "grad_norm": 981.5172119140625,
-      "learning_rate": 3.806692824483501e-05,
-      "loss": 43.4204,
-      "step": 98110
-    },
-    {
-      "epoch": 0.39641721578719846,
-      "grad_norm": 1140.0291748046875,
-      "learning_rate": 3.806395222388536e-05,
-      "loss": 67.1618,
-      "step": 98120
-    },
-    {
-      "epoch": 0.3964576170525661,
-      "grad_norm": 863.1648559570312,
-      "learning_rate": 3.8060975948247223e-05,
-      "loss": 71.7122,
-      "step": 98130
-    },
-    {
-      "epoch": 0.39649801831793374,
-      "grad_norm": 631.3798217773438,
-      "learning_rate": 3.805799941797865e-05,
-      "loss": 42.8692,
-      "step": 98140
-    },
-    {
-      "epoch": 0.3965384195833013,
-      "grad_norm": 1376.2213134765625,
-      "learning_rate": 3.805502263313765e-05,
-      "loss": 55.8479,
-      "step": 98150
-    },
-    {
-      "epoch": 0.39657882084866897,
-      "grad_norm": 476.51446533203125,
-      "learning_rate": 3.805204559378227e-05,
-      "loss": 57.2946,
-      "step": 98160
-    },
-    {
-      "epoch": 0.3966192221140366,
-      "grad_norm": 941.9556884765625,
-      "learning_rate": 3.804906829997053e-05,
-      "loss": 71.0839,
-      "step": 98170
-    },
-    {
-      "epoch": 0.39665962337940425,
-      "grad_norm": 764.8499755859375,
-      "learning_rate": 3.804609075176049e-05,
-      "loss": 61.4329,
-      "step": 98180
-    },
-    {
-      "epoch": 0.3967000246447719,
-      "grad_norm": 804.7611694335938,
-      "learning_rate": 3.8043112949210194e-05,
-      "loss": 50.3516,
-      "step": 98190
-    },
-    {
-      "epoch": 0.3967404259101395,
-      "grad_norm": 757.18798828125,
-      "learning_rate": 3.80401348923777e-05,
-      "loss": 65.6766,
-      "step": 98200
-    },
-    {
-      "epoch": 0.3967808271755071,
-      "grad_norm": 802.1757202148438,
-      "learning_rate": 3.803715658132105e-05,
-      "loss": 51.4916,
-      "step": 98210
-    },
-    {
-      "epoch": 0.39682122844087475,
-      "grad_norm": 677.951904296875,
-      "learning_rate": 3.803417801609833e-05,
-      "loss": 40.4412,
-      "step": 98220
-    },
-    {
-      "epoch": 0.3968616297062424,
-      "grad_norm": 960.4522094726562,
-      "learning_rate": 3.803119919676761e-05,
-      "loss": 53.576,
-      "step": 98230
-    },
-    {
-      "epoch": 0.39690203097161003,
-      "grad_norm": 1985.3116455078125,
-      "learning_rate": 3.802822012338694e-05,
-      "loss": 62.2902,
-      "step": 98240
-    },
-    {
-      "epoch": 0.39694243223697767,
-      "grad_norm": 1060.7294921875,
-      "learning_rate": 3.802524079601442e-05,
-      "loss": 59.7074,
-      "step": 98250
-    },
-    {
-      "epoch": 0.3969828335023453,
-      "grad_norm": 926.2281494140625,
-      "learning_rate": 3.802226121470811e-05,
-      "loss": 67.3111,
-      "step": 98260
-    },
-    {
-      "epoch": 0.39702323476771295,
-      "grad_norm": 558.5460205078125,
-      "learning_rate": 3.8019281379526114e-05,
-      "loss": 52.2698,
-      "step": 98270
-    },
-    {
-      "epoch": 0.39706363603308054,
-      "grad_norm": 1760.0323486328125,
-      "learning_rate": 3.8016301290526534e-05,
-      "loss": 63.7224,
-      "step": 98280
-    },
-    {
-      "epoch": 0.3971040372984482,
-      "grad_norm": 717.9991455078125,
-      "learning_rate": 3.8013320947767464e-05,
-      "loss": 61.1942,
-      "step": 98290
-    },
-    {
-      "epoch": 0.3971444385638158,
-      "grad_norm": 302.7607421875,
-      "learning_rate": 3.8010340351306997e-05,
-      "loss": 52.8783,
-      "step": 98300
-    },
-    {
-      "epoch": 0.39718483982918346,
-      "grad_norm": 603.1386108398438,
-      "learning_rate": 3.800735950120324e-05,
-      "loss": 61.837,
-      "step": 98310
-    },
-    {
-      "epoch": 0.3972252410945511,
-      "grad_norm": 714.36474609375,
-      "learning_rate": 3.8004378397514315e-05,
-      "loss": 57.4805,
-      "step": 98320
-    },
-    {
-      "epoch": 0.39726564235991874,
-      "grad_norm": 578.4430541992188,
-      "learning_rate": 3.800139704029835e-05,
-      "loss": 44.0688,
-      "step": 98330
-    },
-    {
-      "epoch": 0.3973060436252863,
-      "grad_norm": 556.3411254882812,
-      "learning_rate": 3.7998415429613444e-05,
-      "loss": 58.5663,
-      "step": 98340
-    },
-    {
-      "epoch": 0.39734644489065396,
-      "grad_norm": 402.5372314453125,
-      "learning_rate": 3.7995433565517735e-05,
-      "loss": 47.3463,
-      "step": 98350
-    },
-    {
-      "epoch": 0.3973868461560216,
-      "grad_norm": 652.9655151367188,
-      "learning_rate": 3.799245144806937e-05,
-      "loss": 77.4669,
-      "step": 98360
-    },
-    {
-      "epoch": 0.39742724742138924,
-      "grad_norm": 351.4555358886719,
-      "learning_rate": 3.7989469077326466e-05,
-      "loss": 55.3536,
-      "step": 98370
-    },
-    {
-      "epoch": 0.3974676486867569,
-      "grad_norm": 530.9819946289062,
-      "learning_rate": 3.798648645334718e-05,
-      "loss": 48.7565,
-      "step": 98380
-    },
-    {
-      "epoch": 0.3975080499521245,
-      "grad_norm": 997.8034057617188,
-      "learning_rate": 3.798350357618965e-05,
-      "loss": 69.4074,
-      "step": 98390
-    },
-    {
-      "epoch": 0.3975484512174921,
-      "grad_norm": 1051.143310546875,
-      "learning_rate": 3.798052044591204e-05,
-      "loss": 68.8657,
-      "step": 98400
-    },
-    {
-      "epoch": 0.39758885248285974,
-      "grad_norm": 763.1336059570312,
-      "learning_rate": 3.79775370625725e-05,
-      "loss": 65.6765,
-      "step": 98410
-    },
-    {
-      "epoch": 0.3976292537482274,
-      "grad_norm": 727.5631713867188,
-      "learning_rate": 3.797455342622919e-05,
-      "loss": 55.0416,
-      "step": 98420
-    },
-    {
-      "epoch": 0.397669655013595,
-      "grad_norm": 747.4799194335938,
-      "learning_rate": 3.797156953694028e-05,
-      "loss": 62.9093,
-      "step": 98430
-    },
-    {
-      "epoch": 0.39771005627896266,
-      "grad_norm": 568.0397338867188,
-      "learning_rate": 3.796858539476394e-05,
-      "loss": 57.0158,
-      "step": 98440
-    },
-    {
-      "epoch": 0.3977504575443303,
-      "grad_norm": 1187.3994140625,
-      "learning_rate": 3.7965600999758356e-05,
-      "loss": 75.5707,
-      "step": 98450
-    },
-    {
-      "epoch": 0.39779085880969794,
-      "grad_norm": 1377.0982666015625,
-      "learning_rate": 3.796261635198171e-05,
-      "loss": 110.1248,
-      "step": 98460
-    },
-    {
-      "epoch": 0.39783126007506553,
-      "grad_norm": 541.7080078125,
-      "learning_rate": 3.7959631451492176e-05,
-      "loss": 113.9989,
-      "step": 98470
-    },
-    {
-      "epoch": 0.39787166134043317,
-      "grad_norm": 1125.3931884765625,
-      "learning_rate": 3.7956646298347956e-05,
-      "loss": 53.4357,
-      "step": 98480
-    },
-    {
-      "epoch": 0.3979120626058008,
-      "grad_norm": 1457.5439453125,
-      "learning_rate": 3.795366089260725e-05,
-      "loss": 47.1348,
-      "step": 98490
-    },
-    {
-      "epoch": 0.39795246387116845,
-      "grad_norm": 639.0778198242188,
-      "learning_rate": 3.795067523432826e-05,
-      "loss": 53.8182,
-      "step": 98500
-    },
-    {
-      "epoch": 0.3979928651365361,
-      "grad_norm": 473.65032958984375,
-      "learning_rate": 3.794768932356918e-05,
-      "loss": 63.5976,
-      "step": 98510
-    },
-    {
-      "epoch": 0.39803326640190373,
-      "grad_norm": 1148.6119384765625,
-      "learning_rate": 3.7944703160388234e-05,
-      "loss": 43.6071,
-      "step": 98520
-    },
-    {
-      "epoch": 0.3980736676672713,
-      "grad_norm": 2093.45849609375,
-      "learning_rate": 3.794171674484363e-05,
-      "loss": 60.3479,
-      "step": 98530
-    },
-    {
-      "epoch": 0.39811406893263895,
-      "grad_norm": 1044.666748046875,
-      "learning_rate": 3.793873007699361e-05,
-      "loss": 88.364,
-      "step": 98540
-    },
-    {
-      "epoch": 0.3981544701980066,
-      "grad_norm": 597.9827270507812,
-      "learning_rate": 3.7935743156896375e-05,
-      "loss": 54.5251,
-      "step": 98550
-    },
-    {
-      "epoch": 0.39819487146337423,
-      "grad_norm": 353.5838928222656,
-      "learning_rate": 3.793275598461017e-05,
-      "loss": 66.4365,
-      "step": 98560
-    },
-    {
-      "epoch": 0.3982352727287419,
-      "grad_norm": 399.634521484375,
-      "learning_rate": 3.792976856019323e-05,
-      "loss": 31.9752,
-      "step": 98570
-    },
-    {
-      "epoch": 0.3982756739941095,
-      "grad_norm": 915.1165771484375,
-      "learning_rate": 3.792678088370379e-05,
-      "loss": 70.9123,
-      "step": 98580
-    },
-    {
-      "epoch": 0.39831607525947715,
-      "grad_norm": 863.623046875,
-      "learning_rate": 3.792379295520011e-05,
-      "loss": 69.3178,
-      "step": 98590
-    },
-    {
-      "epoch": 0.39835647652484474,
-      "grad_norm": 773.9990844726562,
-      "learning_rate": 3.792080477474043e-05,
-      "loss": 75.029,
-      "step": 98600
-    },
-    {
-      "epoch": 0.3983968777902124,
-      "grad_norm": 889.9473876953125,
-      "learning_rate": 3.7917816342383005e-05,
-      "loss": 80.714,
-      "step": 98610
-    },
-    {
-      "epoch": 0.39843727905558,
-      "grad_norm": 551.6591186523438,
-      "learning_rate": 3.7914827658186103e-05,
-      "loss": 59.7947,
-      "step": 98620
-    },
-    {
-      "epoch": 0.39847768032094766,
-      "grad_norm": 457.0266418457031,
-      "learning_rate": 3.791183872220798e-05,
-      "loss": 64.1721,
-      "step": 98630
-    },
-    {
-      "epoch": 0.3985180815863153,
-      "grad_norm": 778.4760131835938,
-      "learning_rate": 3.790884953450692e-05,
-      "loss": 53.5698,
-      "step": 98640
-    },
-    {
-      "epoch": 0.39855848285168294,
-      "grad_norm": 628.935791015625,
-      "learning_rate": 3.790586009514119e-05,
-      "loss": 45.5346,
-      "step": 98650
-    },
-    {
-      "epoch": 0.3985988841170505,
-      "grad_norm": 390.69134521484375,
-      "learning_rate": 3.790287040416908e-05,
-      "loss": 80.0498,
-      "step": 98660
-    },
-    {
-      "epoch": 0.39863928538241816,
-      "grad_norm": 585.2010498046875,
-      "learning_rate": 3.7899880461648865e-05,
-      "loss": 45.9213,
-      "step": 98670
-    },
-    {
-      "epoch": 0.3986796866477858,
-      "grad_norm": 537.3295288085938,
-      "learning_rate": 3.789689026763883e-05,
-      "loss": 52.588,
-      "step": 98680
-    },
-    {
-      "epoch": 0.39872008791315344,
-      "grad_norm": 1386.0994873046875,
-      "learning_rate": 3.789389982219729e-05,
-      "loss": 69.4656,
-      "step": 98690
-    },
-    {
-      "epoch": 0.3987604891785211,
-      "grad_norm": 471.1151123046875,
-      "learning_rate": 3.789090912538253e-05,
-      "loss": 52.704,
-      "step": 98700
-    },
-    {
-      "epoch": 0.3988008904438887,
-      "grad_norm": 407.30584716796875,
-      "learning_rate": 3.7887918177252855e-05,
-      "loss": 86.2156,
-      "step": 98710
-    },
-    {
-      "epoch": 0.3988412917092563,
-      "grad_norm": 563.2343139648438,
-      "learning_rate": 3.788492697786658e-05,
-      "loss": 79.5212,
-      "step": 98720
-    },
-    {
-      "epoch": 0.39888169297462395,
-      "grad_norm": 3062.3173828125,
-      "learning_rate": 3.788193552728204e-05,
-      "loss": 67.4379,
-      "step": 98730
-    },
-    {
-      "epoch": 0.3989220942399916,
-      "grad_norm": 581.1060791015625,
-      "learning_rate": 3.7878943825557516e-05,
-      "loss": 53.4574,
-      "step": 98740
-    },
-    {
-      "epoch": 0.3989624955053592,
-      "grad_norm": 495.8964538574219,
-      "learning_rate": 3.787595187275136e-05,
-      "loss": 58.9977,
-      "step": 98750
-    },
-    {
-      "epoch": 0.39900289677072687,
-      "grad_norm": 576.1397705078125,
-      "learning_rate": 3.7872959668921884e-05,
-      "loss": 87.9499,
-      "step": 98760
-    },
-    {
-      "epoch": 0.3990432980360945,
-      "grad_norm": 429.25439453125,
-      "learning_rate": 3.786996721412745e-05,
-      "loss": 42.8542,
-      "step": 98770
-    },
-    {
-      "epoch": 0.39908369930146215,
-      "grad_norm": 403.6635437011719,
-      "learning_rate": 3.7866974508426354e-05,
-      "loss": 43.5901,
-      "step": 98780
-    },
-    {
-      "epoch": 0.39912410056682973,
-      "grad_norm": 498.6553039550781,
-      "learning_rate": 3.786398155187698e-05,
-      "loss": 81.6215,
-      "step": 98790
-    },
-    {
-      "epoch": 0.39916450183219737,
-      "grad_norm": 779.7023315429688,
-      "learning_rate": 3.786098834453766e-05,
-      "loss": 47.5752,
-      "step": 98800
-    },
-    {
-      "epoch": 0.399204903097565,
-      "grad_norm": 615.2477416992188,
-      "learning_rate": 3.7857994886466755e-05,
-      "loss": 53.3112,
-      "step": 98810
-    },
-    {
-      "epoch": 0.39924530436293265,
-      "grad_norm": 2094.180908203125,
-      "learning_rate": 3.7855001177722615e-05,
-      "loss": 62.0935,
-      "step": 98820
-    },
-    {
-      "epoch": 0.3992857056283003,
-      "grad_norm": 477.36724853515625,
-      "learning_rate": 3.785200721836361e-05,
-      "loss": 75.4263,
-      "step": 98830
-    },
-    {
-      "epoch": 0.39932610689366793,
-      "grad_norm": 2158.408447265625,
-      "learning_rate": 3.7849013008448115e-05,
-      "loss": 99.4542,
-      "step": 98840
-    },
-    {
-      "epoch": 0.3993665081590355,
-      "grad_norm": 1072.518310546875,
-      "learning_rate": 3.784601854803449e-05,
-      "loss": 77.213,
-      "step": 98850
-    },
-    {
-      "epoch": 0.39940690942440316,
-      "grad_norm": 796.7136840820312,
-      "learning_rate": 3.784302383718113e-05,
-      "loss": 62.9405,
-      "step": 98860
-    },
-    {
-      "epoch": 0.3994473106897708,
-      "grad_norm": 716.7407836914062,
-      "learning_rate": 3.784002887594639e-05,
-      "loss": 66.7786,
-      "step": 98870
-    },
-    {
-      "epoch": 0.39948771195513844,
-      "grad_norm": 526.4265747070312,
-      "learning_rate": 3.783703366438868e-05,
-      "loss": 47.3223,
-      "step": 98880
-    },
-    {
-      "epoch": 0.3995281132205061,
-      "grad_norm": 420.97637939453125,
-      "learning_rate": 3.783403820256639e-05,
-      "loss": 43.0085,
-      "step": 98890
-    },
-    {
-      "epoch": 0.3995685144858737,
-      "grad_norm": 931.0668334960938,
-      "learning_rate": 3.783104249053793e-05,
-      "loss": 84.5414,
-      "step": 98900
-    },
-    {
-      "epoch": 0.39960891575124136,
-      "grad_norm": 729.2132568359375,
-      "learning_rate": 3.782804652836168e-05,
-      "loss": 59.9172,
-      "step": 98910
-    },
-    {
-      "epoch": 0.39964931701660894,
-      "grad_norm": 518.9829711914062,
-      "learning_rate": 3.782505031609607e-05,
-      "loss": 64.4889,
-      "step": 98920
-    },
-    {
-      "epoch": 0.3996897182819766,
-      "grad_norm": 756.1053466796875,
-      "learning_rate": 3.782205385379948e-05,
-      "loss": 73.3509,
-      "step": 98930
-    },
-    {
-      "epoch": 0.3997301195473442,
-      "grad_norm": 1349.8555908203125,
-      "learning_rate": 3.781905714153037e-05,
-      "loss": 70.6158,
-      "step": 98940
-    },
-    {
-      "epoch": 0.39977052081271186,
-      "grad_norm": 573.3903198242188,
-      "learning_rate": 3.781606017934713e-05,
-      "loss": 48.6632,
-      "step": 98950
-    },
-    {
-      "epoch": 0.3998109220780795,
-      "grad_norm": 510.5514221191406,
-      "learning_rate": 3.78130629673082e-05,
-      "loss": 65.3422,
-      "step": 98960
-    },
-    {
-      "epoch": 0.39985132334344714,
-      "grad_norm": 676.6386108398438,
-      "learning_rate": 3.781006550547202e-05,
-      "loss": 49.1923,
-      "step": 98970
-    },
-    {
-      "epoch": 0.3998917246088147,
-      "grad_norm": 1945.87353515625,
-      "learning_rate": 3.780706779389701e-05,
-      "loss": 109.869,
-      "step": 98980
-    },
-    {
-      "epoch": 0.39993212587418236,
-      "grad_norm": 607.2948608398438,
-      "learning_rate": 3.7804069832641615e-05,
-      "loss": 59.3958,
-      "step": 98990
-    },
-    {
-      "epoch": 0.39997252713955,
-      "grad_norm": 3488.2783203125,
-      "learning_rate": 3.780107162176429e-05,
-      "loss": 71.8975,
-      "step": 99000
-    },
-    {
-      "epoch": 0.40001292840491764,
-      "grad_norm": 869.2059326171875,
-      "learning_rate": 3.779807316132349e-05,
-      "loss": 71.5209,
-      "step": 99010
-    },
-    {
-      "epoch": 0.4000533296702853,
-      "grad_norm": 1111.3924560546875,
-      "learning_rate": 3.779507445137766e-05,
-      "loss": 72.5968,
-      "step": 99020
-    },
-    {
-      "epoch": 0.4000937309356529,
-      "grad_norm": 590.8203125,
-      "learning_rate": 3.779207549198527e-05,
-      "loss": 68.4518,
-      "step": 99030
-    },
-    {
-      "epoch": 0.4001341322010205,
-      "grad_norm": 1010.8024291992188,
-      "learning_rate": 3.778907628320477e-05,
-      "loss": 55.1932,
-      "step": 99040
-    },
-    {
-      "epoch": 0.40017453346638815,
-      "grad_norm": 1730.03271484375,
-      "learning_rate": 3.778607682509465e-05,
-      "loss": 72.5133,
-      "step": 99050
-    },
-    {
-      "epoch": 0.4002149347317558,
-      "grad_norm": 706.1375122070312,
-      "learning_rate": 3.7783077117713386e-05,
-      "loss": 51.3278,
-      "step": 99060
-    },
-    {
-      "epoch": 0.40025533599712343,
-      "grad_norm": 491.18145751953125,
-      "learning_rate": 3.778007716111945e-05,
-      "loss": 57.2659,
-      "step": 99070
-    },
-    {
-      "epoch": 0.40029573726249107,
-      "grad_norm": 967.4299926757812,
-      "learning_rate": 3.777707695537133e-05,
-      "loss": 54.0169,
-      "step": 99080
-    },
-    {
-      "epoch": 0.4003361385278587,
-      "grad_norm": 586.3837280273438,
-      "learning_rate": 3.777407650052751e-05,
-      "loss": 72.3556,
-      "step": 99090
-    },
-    {
-      "epoch": 0.40037653979322635,
-      "grad_norm": 618.3201904296875,
-      "learning_rate": 3.77710757966465e-05,
-      "loss": 57.029,
-      "step": 99100
-    },
-    {
-      "epoch": 0.40041694105859393,
-      "grad_norm": 753.272705078125,
-      "learning_rate": 3.7768074843786796e-05,
-      "loss": 61.8438,
-      "step": 99110
-    },
-    {
-      "epoch": 0.4004573423239616,
-      "grad_norm": 249.07408142089844,
-      "learning_rate": 3.776507364200689e-05,
-      "loss": 66.9332,
-      "step": 99120
-    },
-    {
-      "epoch": 0.4004977435893292,
-      "grad_norm": 371.7924499511719,
-      "learning_rate": 3.77620721913653e-05,
-      "loss": 57.5363,
-      "step": 99130
-    },
-    {
-      "epoch": 0.40053814485469685,
-      "grad_norm": 1456.956787109375,
-      "learning_rate": 3.7759070491920544e-05,
-      "loss": 66.1014,
-      "step": 99140
-    },
-    {
-      "epoch": 0.4005785461200645,
-      "grad_norm": 700.562744140625,
-      "learning_rate": 3.775606854373115e-05,
-      "loss": 60.8896,
-      "step": 99150
-    },
-    {
-      "epoch": 0.40061894738543213,
-      "grad_norm": 712.740966796875,
-      "learning_rate": 3.775306634685562e-05,
-      "loss": 57.9598,
-      "step": 99160
-    },
-    {
-      "epoch": 0.4006593486507997,
-      "grad_norm": 2491.177490234375,
-      "learning_rate": 3.7750063901352494e-05,
-      "loss": 66.2804,
-      "step": 99170
-    },
-    {
-      "epoch": 0.40069974991616736,
-      "grad_norm": 930.8803100585938,
-      "learning_rate": 3.774706120728032e-05,
-      "loss": 61.9584,
-      "step": 99180
-    },
-    {
-      "epoch": 0.400740151181535,
-      "grad_norm": 488.9775085449219,
-      "learning_rate": 3.774405826469762e-05,
-      "loss": 51.8706,
-      "step": 99190
-    },
-    {
-      "epoch": 0.40078055244690264,
-      "grad_norm": 1185.0855712890625,
-      "learning_rate": 3.7741055073662946e-05,
-      "loss": 59.6434,
-      "step": 99200
-    },
-    {
-      "epoch": 0.4008209537122703,
-      "grad_norm": 1065.01025390625,
-      "learning_rate": 3.773805163423484e-05,
-      "loss": 90.2759,
-      "step": 99210
-    },
-    {
-      "epoch": 0.4008613549776379,
-      "grad_norm": 1346.4959716796875,
-      "learning_rate": 3.773504794647187e-05,
-      "loss": 47.388,
-      "step": 99220
-    },
-    {
-      "epoch": 0.40090175624300556,
-      "grad_norm": 672.511962890625,
-      "learning_rate": 3.7732044010432564e-05,
-      "loss": 43.0481,
-      "step": 99230
-    },
-    {
-      "epoch": 0.40094215750837314,
-      "grad_norm": 655.6102294921875,
-      "learning_rate": 3.772903982617552e-05,
-      "loss": 57.7771,
-      "step": 99240
-    },
-    {
-      "epoch": 0.4009825587737408,
-      "grad_norm": 1324.095703125,
-      "learning_rate": 3.7726035393759285e-05,
-      "loss": 57.1472,
-      "step": 99250
-    },
-    {
-      "epoch": 0.4010229600391084,
-      "grad_norm": 1157.129150390625,
-      "learning_rate": 3.772303071324244e-05,
-      "loss": 60.1481,
-      "step": 99260
-    },
-    {
-      "epoch": 0.40106336130447606,
-      "grad_norm": 784.1769409179688,
-      "learning_rate": 3.772002578468356e-05,
-      "loss": 101.2776,
-      "step": 99270
-    },
-    {
-      "epoch": 0.4011037625698437,
-      "grad_norm": 682.7789306640625,
-      "learning_rate": 3.771702060814123e-05,
-      "loss": 86.0453,
-      "step": 99280
-    },
-    {
-      "epoch": 0.40114416383521134,
-      "grad_norm": 3423.19921875,
-      "learning_rate": 3.771401518367403e-05,
-      "loss": 74.2687,
-      "step": 99290
-    },
-    {
-      "epoch": 0.4011845651005789,
-      "grad_norm": 672.9019775390625,
-      "learning_rate": 3.771100951134057e-05,
-      "loss": 55.2397,
-      "step": 99300
-    },
-    {
-      "epoch": 0.40122496636594657,
-      "grad_norm": 642.078857421875,
-      "learning_rate": 3.770800359119943e-05,
-      "loss": 34.1067,
-      "step": 99310
-    },
-    {
-      "epoch": 0.4012653676313142,
-      "grad_norm": 986.0284423828125,
-      "learning_rate": 3.770499742330922e-05,
-      "loss": 64.7369,
-      "step": 99320
-    },
-    {
-      "epoch": 0.40130576889668185,
-      "grad_norm": 416.7942199707031,
-      "learning_rate": 3.770199100772853e-05,
-      "loss": 63.9787,
-      "step": 99330
-    },
-    {
-      "epoch": 0.4013461701620495,
-      "grad_norm": 1163.7266845703125,
-      "learning_rate": 3.7698984344515997e-05,
-      "loss": 55.3137,
-      "step": 99340
-    },
-    {
-      "epoch": 0.4013865714274171,
-      "grad_norm": 591.773681640625,
-      "learning_rate": 3.769597743373023e-05,
-      "loss": 54.0791,
-      "step": 99350
-    },
-    {
-      "epoch": 0.4014269726927847,
-      "grad_norm": 1694.82373046875,
-      "learning_rate": 3.769297027542985e-05,
-      "loss": 46.7332,
-      "step": 99360
-    },
-    {
-      "epoch": 0.40146737395815235,
-      "grad_norm": 717.4148559570312,
-      "learning_rate": 3.768996286967347e-05,
-      "loss": 55.8938,
-      "step": 99370
-    },
-    {
-      "epoch": 0.40150777522352,
-      "grad_norm": 591.4462890625,
-      "learning_rate": 3.768695521651973e-05,
-      "loss": 46.4734,
-      "step": 99380
-    },
-    {
-      "epoch": 0.40154817648888763,
-      "grad_norm": 1100.54150390625,
-      "learning_rate": 3.7683947316027276e-05,
-      "loss": 63.5928,
-      "step": 99390
-    },
-    {
-      "epoch": 0.40158857775425527,
-      "grad_norm": 421.6627197265625,
-      "learning_rate": 3.7680939168254733e-05,
-      "loss": 59.3417,
-      "step": 99400
-    },
-    {
-      "epoch": 0.4016289790196229,
-      "grad_norm": 307.5633544921875,
-      "learning_rate": 3.767793077326075e-05,
-      "loss": 94.9799,
-      "step": 99410
-    },
-    {
-      "epoch": 0.40166938028499055,
-      "grad_norm": 743.0738525390625,
-      "learning_rate": 3.767492213110397e-05,
-      "loss": 58.2898,
-      "step": 99420
-    },
-    {
-      "epoch": 0.40170978155035814,
-      "grad_norm": 724.935791015625,
-      "learning_rate": 3.767191324184308e-05,
-      "loss": 54.7535,
-      "step": 99430
-    },
-    {
-      "epoch": 0.4017501828157258,
-      "grad_norm": 1314.9210205078125,
-      "learning_rate": 3.7668904105536706e-05,
-      "loss": 76.4102,
-      "step": 99440
-    },
-    {
-      "epoch": 0.4017905840810934,
-      "grad_norm": 920.00537109375,
-      "learning_rate": 3.7665894722243525e-05,
-      "loss": 58.301,
-      "step": 99450
-    },
-    {
-      "epoch": 0.40183098534646106,
-      "grad_norm": 1673.596435546875,
-      "learning_rate": 3.76628850920222e-05,
-      "loss": 57.4322,
-      "step": 99460
-    },
-    {
-      "epoch": 0.4018713866118287,
-      "grad_norm": 826.5586547851562,
-      "learning_rate": 3.7659875214931426e-05,
-      "loss": 69.376,
-      "step": 99470
-    },
-    {
-      "epoch": 0.40191178787719634,
-      "grad_norm": 581.40283203125,
-      "learning_rate": 3.765686509102985e-05,
-      "loss": 36.2128,
-      "step": 99480
-    },
-    {
-      "epoch": 0.4019521891425639,
-      "grad_norm": 326.0263977050781,
-      "learning_rate": 3.765385472037618e-05,
-      "loss": 64.3921,
-      "step": 99490
-    },
-    {
-      "epoch": 0.40199259040793156,
-      "grad_norm": 729.6510009765625,
-      "learning_rate": 3.765084410302909e-05,
-      "loss": 63.8964,
-      "step": 99500
-    },
-    {
-      "epoch": 0.4020329916732992,
-      "grad_norm": 920.1556396484375,
-      "learning_rate": 3.76478332390473e-05,
-      "loss": 41.994,
-      "step": 99510
-    },
-    {
-      "epoch": 0.40207339293866684,
-      "grad_norm": 888.444091796875,
-      "learning_rate": 3.764482212848948e-05,
-      "loss": 70.7583,
-      "step": 99520
-    },
-    {
-      "epoch": 0.4021137942040345,
-      "grad_norm": 909.1885375976562,
-      "learning_rate": 3.7641810771414335e-05,
-      "loss": 67.3313,
-      "step": 99530
-    },
-    {
-      "epoch": 0.4021541954694021,
-      "grad_norm": 1571.4930419921875,
-      "learning_rate": 3.763879916788059e-05,
-      "loss": 66.0156,
-      "step": 99540
-    },
-    {
-      "epoch": 0.40219459673476976,
-      "grad_norm": 724.2689208984375,
-      "learning_rate": 3.763578731794695e-05,
-      "loss": 66.495,
-      "step": 99550
-    },
-    {
-      "epoch": 0.40223499800013734,
-      "grad_norm": 734.593505859375,
-      "learning_rate": 3.7632775221672115e-05,
-      "loss": 41.2429,
-      "step": 99560
-    },
-    {
-      "epoch": 0.402275399265505,
-      "grad_norm": 1102.271728515625,
-      "learning_rate": 3.7629762879114835e-05,
-      "loss": 71.2054,
-      "step": 99570
-    },
-    {
-      "epoch": 0.4023158005308726,
-      "grad_norm": 1053.3270263671875,
-      "learning_rate": 3.7626750290333824e-05,
-      "loss": 49.4548,
-      "step": 99580
-    },
-    {
-      "epoch": 0.40235620179624026,
-      "grad_norm": 774.8560180664062,
-      "learning_rate": 3.7623737455387814e-05,
-      "loss": 53.1317,
-      "step": 99590
-    },
-    {
-      "epoch": 0.4023966030616079,
-      "grad_norm": 792.8966064453125,
-      "learning_rate": 3.762072437433555e-05,
-      "loss": 63.3754,
-      "step": 99600
-    },
-    {
-      "epoch": 0.40243700432697554,
-      "grad_norm": 648.0517578125,
-      "learning_rate": 3.761771104723576e-05,
-      "loss": 41.7011,
-      "step": 99610
-    },
-    {
-      "epoch": 0.40247740559234313,
-      "grad_norm": 816.7391967773438,
-      "learning_rate": 3.76146974741472e-05,
-      "loss": 55.6581,
-      "step": 99620
-    },
-    {
-      "epoch": 0.40251780685771077,
-      "grad_norm": 2065.635498046875,
-      "learning_rate": 3.761168365512862e-05,
-      "loss": 76.0005,
-      "step": 99630
-    },
-    {
-      "epoch": 0.4025582081230784,
-      "grad_norm": 1137.087890625,
-      "learning_rate": 3.760866959023877e-05,
-      "loss": 58.8005,
-      "step": 99640
-    },
-    {
-      "epoch": 0.40259860938844605,
-      "grad_norm": 1612.660400390625,
-      "learning_rate": 3.760565527953641e-05,
-      "loss": 52.3195,
-      "step": 99650
-    },
-    {
-      "epoch": 0.4026390106538137,
-      "grad_norm": 1130.4898681640625,
-      "learning_rate": 3.7602640723080315e-05,
-      "loss": 58.5023,
-      "step": 99660
-    },
-    {
-      "epoch": 0.40267941191918133,
-      "grad_norm": 1282.1605224609375,
-      "learning_rate": 3.7599625920929254e-05,
-      "loss": 60.2157,
-      "step": 99670
-    },
-    {
-      "epoch": 0.4027198131845489,
-      "grad_norm": 626.7401733398438,
-      "learning_rate": 3.759661087314199e-05,
-      "loss": 61.224,
-      "step": 99680
-    },
-    {
-      "epoch": 0.40276021444991655,
-      "grad_norm": 403.7524108886719,
-      "learning_rate": 3.759359557977732e-05,
-      "loss": 43.2286,
-      "step": 99690
-    },
-    {
-      "epoch": 0.4028006157152842,
-      "grad_norm": 459.4110107421875,
-      "learning_rate": 3.759058004089402e-05,
-      "loss": 63.8654,
-      "step": 99700
-    },
-    {
-      "epoch": 0.40284101698065183,
-      "grad_norm": 689.9656982421875,
-      "learning_rate": 3.758756425655089e-05,
-      "loss": 55.9281,
-      "step": 99710
-    },
-    {
-      "epoch": 0.4028814182460195,
-      "grad_norm": 493.5693664550781,
-      "learning_rate": 3.7584548226806696e-05,
-      "loss": 46.4668,
-      "step": 99720
-    },
-    {
-      "epoch": 0.4029218195113871,
-      "grad_norm": 568.452880859375,
-      "learning_rate": 3.758153195172026e-05,
-      "loss": 85.0147,
-      "step": 99730
-    },
-    {
-      "epoch": 0.40296222077675475,
-      "grad_norm": 775.391845703125,
-      "learning_rate": 3.7578515431350384e-05,
-      "loss": 62.4054,
-      "step": 99740
-    },
-    {
-      "epoch": 0.40300262204212234,
-      "grad_norm": 1037.2498779296875,
-      "learning_rate": 3.757549866575588e-05,
-      "loss": 58.9337,
-      "step": 99750
-    },
-    {
-      "epoch": 0.40304302330749,
-      "grad_norm": 487.4396667480469,
-      "learning_rate": 3.757248165499555e-05,
-      "loss": 78.7348,
-      "step": 99760
-    },
-    {
-      "epoch": 0.4030834245728576,
-      "grad_norm": 1064.0037841796875,
-      "learning_rate": 3.7569464399128215e-05,
-      "loss": 43.4251,
-      "step": 99770
-    },
-    {
-      "epoch": 0.40312382583822526,
-      "grad_norm": 417.7305603027344,
-      "learning_rate": 3.75664468982127e-05,
-      "loss": 43.3443,
-      "step": 99780
-    },
-    {
-      "epoch": 0.4031642271035929,
-      "grad_norm": 1124.982177734375,
-      "learning_rate": 3.756342915230784e-05,
-      "loss": 72.6339,
-      "step": 99790
-    },
-    {
-      "epoch": 0.40320462836896054,
-      "grad_norm": 1895.08935546875,
-      "learning_rate": 3.7560411161472456e-05,
-      "loss": 87.87,
-      "step": 99800
-    },
-    {
-      "epoch": 0.4032450296343281,
-      "grad_norm": 885.612060546875,
-      "learning_rate": 3.755739292576539e-05,
-      "loss": 68.6611,
-      "step": 99810
-    },
-    {
-      "epoch": 0.40328543089969576,
-      "grad_norm": 1059.715576171875,
-      "learning_rate": 3.7554374445245474e-05,
-      "loss": 50.6557,
-      "step": 99820
-    },
-    {
-      "epoch": 0.4033258321650634,
-      "grad_norm": 1289.69677734375,
-      "learning_rate": 3.755135571997158e-05,
-      "loss": 55.3921,
-      "step": 99830
-    },
-    {
-      "epoch": 0.40336623343043104,
-      "grad_norm": 948.4808959960938,
-      "learning_rate": 3.7548336750002544e-05,
-      "loss": 52.4533,
-      "step": 99840
-    },
-    {
-      "epoch": 0.4034066346957987,
-      "grad_norm": 768.9962768554688,
-      "learning_rate": 3.7545317535397214e-05,
-      "loss": 52.6331,
-      "step": 99850
-    },
-    {
-      "epoch": 0.4034470359611663,
-      "grad_norm": 806.0327758789062,
-      "learning_rate": 3.754229807621446e-05,
-      "loss": 58.2938,
-      "step": 99860
-    },
-    {
-      "epoch": 0.40348743722653396,
-      "grad_norm": 631.0634155273438,
-      "learning_rate": 3.753927837251315e-05,
-      "loss": 68.7876,
-      "step": 99870
-    },
-    {
-      "epoch": 0.40352783849190155,
-      "grad_norm": 574.1565551757812,
-      "learning_rate": 3.753625842435216e-05,
-      "loss": 45.7764,
-      "step": 99880
-    },
-    {
-      "epoch": 0.4035682397572692,
-      "grad_norm": 1697.2900390625,
-      "learning_rate": 3.753323823179035e-05,
-      "loss": 65.6436,
-      "step": 99890
-    },
-    {
-      "epoch": 0.4036086410226368,
-      "grad_norm": 260.11083984375,
-      "learning_rate": 3.7530217794886606e-05,
-      "loss": 71.5105,
-      "step": 99900
-    },
-    {
-      "epoch": 0.40364904228800447,
-      "grad_norm": 932.4937133789062,
-      "learning_rate": 3.752719711369982e-05,
-      "loss": 61.679,
-      "step": 99910
-    },
-    {
-      "epoch": 0.4036894435533721,
-      "grad_norm": 676.4361572265625,
-      "learning_rate": 3.752417618828888e-05,
-      "loss": 90.9519,
-      "step": 99920
-    },
-    {
-      "epoch": 0.40372984481873975,
-      "grad_norm": 6873.72802734375,
-      "learning_rate": 3.752115501871267e-05,
-      "loss": 99.3489,
-      "step": 99930
-    },
-    {
-      "epoch": 0.40377024608410733,
-      "grad_norm": 0.0,
-      "learning_rate": 3.75181336050301e-05,
-      "loss": 58.3508,
-      "step": 99940
-    },
-    {
-      "epoch": 0.40381064734947497,
-      "grad_norm": 619.0031127929688,
-      "learning_rate": 3.751511194730007e-05,
-      "loss": 58.6171,
-      "step": 99950
-    },
-    {
-      "epoch": 0.4038510486148426,
-      "grad_norm": 963.3147583007812,
-      "learning_rate": 3.751209004558149e-05,
-      "loss": 60.5256,
-      "step": 99960
-    },
-    {
-      "epoch": 0.40389144988021025,
-      "grad_norm": 720.6400756835938,
-      "learning_rate": 3.750906789993327e-05,
-      "loss": 59.5601,
-      "step": 99970
-    },
-    {
-      "epoch": 0.4039318511455779,
-      "grad_norm": 1478.5609130859375,
-      "learning_rate": 3.7506045510414335e-05,
-      "loss": 46.86,
-      "step": 99980
-    },
-    {
-      "epoch": 0.40397225241094553,
-      "grad_norm": 427.1527099609375,
-      "learning_rate": 3.7503022877083606e-05,
-      "loss": 57.4738,
-      "step": 99990
-    },
-    {
-      "epoch": 0.4040126536763131,
-      "grad_norm": 967.2838134765625,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 77.0952,
-      "step": 100000
-    },
-    {
-      "epoch": 0.40405305494168076,
-      "grad_norm": 896.5092163085938,
-      "learning_rate": 3.749697687922247e-05,
-      "loss": 49.8017,
-      "step": 100010
-    },
-    {
-      "epoch": 0.4040934562070484,
-      "grad_norm": 663.4006958007812,
-      "learning_rate": 3.749395351480993e-05,
-      "loss": 45.9336,
-      "step": 100020
-    },
-    {
-      "epoch": 0.40413385747241604,
-      "grad_norm": 471.2986145019531,
-      "learning_rate": 3.749092990682134e-05,
-      "loss": 62.4995,
-      "step": 100030
-    },
-    {
-      "epoch": 0.4041742587377837,
-      "grad_norm": 956.7597045898438,
-      "learning_rate": 3.748790605531565e-05,
-      "loss": 62.3297,
-      "step": 100040
-    },
-    {
-      "epoch": 0.4042146600031513,
-      "grad_norm": 2078.440185546875,
-      "learning_rate": 3.748488196035179e-05,
-      "loss": 129.3817,
-      "step": 100050
-    },
-    {
-      "epoch": 0.40425506126851896,
-      "grad_norm": 4852.7265625,
-      "learning_rate": 3.748185762198873e-05,
-      "loss": 86.8261,
-      "step": 100060
-    },
-    {
-      "epoch": 0.40429546253388654,
-      "grad_norm": 1042.48828125,
-      "learning_rate": 3.747883304028543e-05,
-      "loss": 51.4865,
-      "step": 100070
-    },
-    {
-      "epoch": 0.4043358637992542,
-      "grad_norm": 758.50048828125,
-      "learning_rate": 3.7475808215300854e-05,
-      "loss": 47.7026,
-      "step": 100080
-    },
-    {
-      "epoch": 0.4043762650646218,
-      "grad_norm": 769.8150024414062,
-      "learning_rate": 3.7472783147093985e-05,
-      "loss": 85.2325,
-      "step": 100090
-    },
-    {
-      "epoch": 0.40441666632998946,
-      "grad_norm": 777.2942504882812,
-      "learning_rate": 3.746975783572377e-05,
-      "loss": 67.7482,
-      "step": 100100
-    },
-    {
-      "epoch": 0.4044570675953571,
-      "grad_norm": 605.9794921875,
-      "learning_rate": 3.746673228124922e-05,
-      "loss": 53.6268,
-      "step": 100110
-    },
-    {
-      "epoch": 0.40449746886072474,
-      "grad_norm": 1080.2708740234375,
-      "learning_rate": 3.7463706483729296e-05,
-      "loss": 69.3192,
-      "step": 100120
-    },
-    {
-      "epoch": 0.4045378701260923,
-      "grad_norm": 725.3931884765625,
-      "learning_rate": 3.7460680443223004e-05,
-      "loss": 42.6117,
-      "step": 100130
-    },
-    {
-      "epoch": 0.40457827139145996,
-      "grad_norm": 629.57470703125,
-      "learning_rate": 3.745765415978933e-05,
-      "loss": 69.1875,
-      "step": 100140
-    },
-    {
-      "epoch": 0.4046186726568276,
-      "grad_norm": 788.4440307617188,
-      "learning_rate": 3.7454627633487274e-05,
-      "loss": 86.3007,
-      "step": 100150
-    },
-    {
-      "epoch": 0.40465907392219524,
-      "grad_norm": 879.8585815429688,
-      "learning_rate": 3.7451600864375844e-05,
-      "loss": 81.9851,
-      "step": 100160
-    },
-    {
-      "epoch": 0.4046994751875629,
-      "grad_norm": 578.5433349609375,
-      "learning_rate": 3.7448573852514035e-05,
-      "loss": 56.7397,
-      "step": 100170
-    },
-    {
-      "epoch": 0.4047398764529305,
-      "grad_norm": 731.2473754882812,
-      "learning_rate": 3.744554659796088e-05,
-      "loss": 54.396,
-      "step": 100180
-    },
-    {
-      "epoch": 0.40478027771829816,
-      "grad_norm": 718.974609375,
-      "learning_rate": 3.744251910077538e-05,
-      "loss": 47.8276,
-      "step": 100190
-    },
-    {
-      "epoch": 0.40482067898366575,
-      "grad_norm": 714.3526000976562,
-      "learning_rate": 3.7439491361016564e-05,
-      "loss": 68.6014,
-      "step": 100200
-    },
-    {
-      "epoch": 0.4048610802490334,
-      "grad_norm": 681.6476440429688,
-      "learning_rate": 3.743646337874346e-05,
-      "loss": 68.5919,
-      "step": 100210
-    },
-    {
-      "epoch": 0.40490148151440103,
-      "grad_norm": 709.8939819335938,
-      "learning_rate": 3.743343515401511e-05,
-      "loss": 60.53,
-      "step": 100220
-    },
-    {
-      "epoch": 0.40494188277976867,
-      "grad_norm": 730.64892578125,
-      "learning_rate": 3.743040668689053e-05,
-      "loss": 91.2893,
-      "step": 100230
-    },
-    {
-      "epoch": 0.4049822840451363,
-      "grad_norm": 1242.37451171875,
-      "learning_rate": 3.742737797742878e-05,
-      "loss": 50.5919,
-      "step": 100240
-    },
-    {
-      "epoch": 0.40502268531050395,
-      "grad_norm": 1279.793701171875,
-      "learning_rate": 3.742434902568889e-05,
-      "loss": 47.2868,
-      "step": 100250
-    },
-    {
-      "epoch": 0.40506308657587153,
-      "grad_norm": 446.11175537109375,
-      "learning_rate": 3.742131983172992e-05,
-      "loss": 43.8764,
-      "step": 100260
-    },
-    {
-      "epoch": 0.4051034878412392,
-      "grad_norm": 478.0585021972656,
-      "learning_rate": 3.741829039561092e-05,
-      "loss": 67.3198,
-      "step": 100270
-    },
-    {
-      "epoch": 0.4051438891066068,
-      "grad_norm": 1781.1514892578125,
-      "learning_rate": 3.741526071739097e-05,
-      "loss": 55.5198,
-      "step": 100280
-    },
-    {
-      "epoch": 0.40518429037197445,
-      "grad_norm": 601.347412109375,
-      "learning_rate": 3.741223079712911e-05,
-      "loss": 50.489,
-      "step": 100290
-    },
-    {
-      "epoch": 0.4052246916373421,
-      "grad_norm": 631.7415161132812,
-      "learning_rate": 3.7409200634884426e-05,
-      "loss": 52.5421,
-      "step": 100300
-    },
-    {
-      "epoch": 0.40526509290270973,
-      "grad_norm": 431.8924255371094,
-      "learning_rate": 3.740617023071598e-05,
-      "loss": 59.2278,
-      "step": 100310
-    },
-    {
-      "epoch": 0.4053054941680773,
-      "grad_norm": 1287.239501953125,
-      "learning_rate": 3.740313958468287e-05,
-      "loss": 63.1438,
-      "step": 100320
-    },
-    {
-      "epoch": 0.40534589543344496,
-      "grad_norm": 769.814208984375,
-      "learning_rate": 3.7400108696844156e-05,
-      "loss": 56.3332,
-      "step": 100330
-    },
-    {
-      "epoch": 0.4053862966988126,
-      "grad_norm": 372.637451171875,
-      "learning_rate": 3.739707756725894e-05,
-      "loss": 61.8536,
-      "step": 100340
-    },
-    {
-      "epoch": 0.40542669796418024,
-      "grad_norm": 537.6292114257812,
-      "learning_rate": 3.739404619598632e-05,
-      "loss": 75.01,
-      "step": 100350
-    },
-    {
-      "epoch": 0.4054670992295479,
-      "grad_norm": 1282.1539306640625,
-      "learning_rate": 3.7391014583085385e-05,
-      "loss": 42.7495,
-      "step": 100360
-    },
-    {
-      "epoch": 0.4055075004949155,
-      "grad_norm": 1255.515869140625,
-      "learning_rate": 3.738798272861525e-05,
-      "loss": 79.5692,
-      "step": 100370
-    },
-    {
-      "epoch": 0.40554790176028316,
-      "grad_norm": 1217.02294921875,
-      "learning_rate": 3.7384950632634995e-05,
-      "loss": 55.6681,
-      "step": 100380
-    },
-    {
-      "epoch": 0.40558830302565074,
-      "grad_norm": 480.9776306152344,
-      "learning_rate": 3.7381918295203774e-05,
-      "loss": 43.6625,
-      "step": 100390
-    },
-    {
-      "epoch": 0.4056287042910184,
-      "grad_norm": 861.3986206054688,
-      "learning_rate": 3.7378885716380664e-05,
-      "loss": 48.5241,
-      "step": 100400
-    },
-    {
-      "epoch": 0.405669105556386,
-      "grad_norm": 490.213134765625,
-      "learning_rate": 3.737585289622482e-05,
-      "loss": 60.6238,
-      "step": 100410
-    },
-    {
-      "epoch": 0.40570950682175366,
-      "grad_norm": 0.0,
-      "learning_rate": 3.7372819834795335e-05,
-      "loss": 82.8279,
-      "step": 100420
-    },
-    {
-      "epoch": 0.4057499080871213,
-      "grad_norm": 3635.81689453125,
-      "learning_rate": 3.736978653215136e-05,
-      "loss": 79.4986,
-      "step": 100430
-    },
-    {
-      "epoch": 0.40579030935248894,
-      "grad_norm": 803.3690795898438,
-      "learning_rate": 3.736675298835203e-05,
-      "loss": 84.2047,
-      "step": 100440
-    },
-    {
-      "epoch": 0.4058307106178565,
-      "grad_norm": 1111.7283935546875,
-      "learning_rate": 3.7363719203456495e-05,
-      "loss": 51.9508,
-      "step": 100450
-    },
-    {
-      "epoch": 0.40587111188322417,
-      "grad_norm": 472.92242431640625,
-      "learning_rate": 3.736068517752388e-05,
-      "loss": 47.1429,
-      "step": 100460
-    },
-    {
-      "epoch": 0.4059115131485918,
-      "grad_norm": 638.3177490234375,
-      "learning_rate": 3.735765091061334e-05,
-      "loss": 74.3898,
-      "step": 100470
-    },
-    {
-      "epoch": 0.40595191441395945,
-      "grad_norm": 877.7601318359375,
-      "learning_rate": 3.7354616402784035e-05,
-      "loss": 63.0587,
-      "step": 100480
-    },
-    {
-      "epoch": 0.4059923156793271,
-      "grad_norm": 625.0929565429688,
-      "learning_rate": 3.735158165409514e-05,
-      "loss": 38.2794,
-      "step": 100490
-    },
-    {
-      "epoch": 0.4060327169446947,
-      "grad_norm": 756.3575439453125,
-      "learning_rate": 3.7348546664605777e-05,
-      "loss": 64.2025,
-      "step": 100500
-    },
-    {
-      "epoch": 0.4060731182100623,
-      "grad_norm": 1217.9346923828125,
-      "learning_rate": 3.7345511434375145e-05,
-      "loss": 68.1389,
-      "step": 100510
-    },
-    {
-      "epoch": 0.40611351947542995,
-      "grad_norm": 873.300048828125,
-      "learning_rate": 3.734247596346242e-05,
-      "loss": 66.9642,
-      "step": 100520
-    },
-    {
-      "epoch": 0.4061539207407976,
-      "grad_norm": 959.7666015625,
-      "learning_rate": 3.733944025192677e-05,
-      "loss": 66.9598,
-      "step": 100530
-    },
-    {
-      "epoch": 0.40619432200616523,
-      "grad_norm": 789.1475830078125,
-      "learning_rate": 3.733640429982738e-05,
-      "loss": 58.7012,
-      "step": 100540
-    },
-    {
-      "epoch": 0.40623472327153287,
-      "grad_norm": 480.3835754394531,
-      "learning_rate": 3.7333368107223424e-05,
-      "loss": 43.2679,
-      "step": 100550
-    },
-    {
-      "epoch": 0.4062751245369005,
-      "grad_norm": 759.8263549804688,
-      "learning_rate": 3.7330331674174125e-05,
-      "loss": 66.052,
-      "step": 100560
-    },
-    {
-      "epoch": 0.40631552580226815,
-      "grad_norm": 859.6227416992188,
-      "learning_rate": 3.732729500073866e-05,
-      "loss": 56.3151,
-      "step": 100570
-    },
-    {
-      "epoch": 0.40635592706763574,
-      "grad_norm": 405.04998779296875,
-      "learning_rate": 3.732425808697622e-05,
-      "loss": 54.2167,
-      "step": 100580
-    },
-    {
-      "epoch": 0.4063963283330034,
-      "grad_norm": 821.4298706054688,
-      "learning_rate": 3.732122093294603e-05,
-      "loss": 67.599,
-      "step": 100590
-    },
-    {
-      "epoch": 0.406436729598371,
-      "grad_norm": 777.3472290039062,
-      "learning_rate": 3.731818353870729e-05,
-      "loss": 58.1797,
-      "step": 100600
-    },
-    {
-      "epoch": 0.40647713086373866,
-      "grad_norm": 1343.93359375,
-      "learning_rate": 3.731514590431922e-05,
-      "loss": 65.9451,
-      "step": 100610
-    },
-    {
-      "epoch": 0.4065175321291063,
-      "grad_norm": 800.9368896484375,
-      "learning_rate": 3.731210802984105e-05,
-      "loss": 67.0606,
-      "step": 100620
-    },
-    {
-      "epoch": 0.40655793339447394,
-      "grad_norm": 1185.2408447265625,
-      "learning_rate": 3.730906991533199e-05,
-      "loss": 76.7648,
-      "step": 100630
-    },
-    {
-      "epoch": 0.4065983346598415,
-      "grad_norm": 545.9066772460938,
-      "learning_rate": 3.7306031560851275e-05,
-      "loss": 42.591,
-      "step": 100640
-    },
-    {
-      "epoch": 0.40663873592520916,
-      "grad_norm": 551.5760498046875,
-      "learning_rate": 3.730299296645814e-05,
-      "loss": 81.1927,
-      "step": 100650
-    },
-    {
-      "epoch": 0.4066791371905768,
-      "grad_norm": 3141.09326171875,
-      "learning_rate": 3.729995413221183e-05,
-      "loss": 85.903,
-      "step": 100660
-    },
-    {
-      "epoch": 0.40671953845594444,
-      "grad_norm": 1117.46728515625,
-      "learning_rate": 3.7296915058171566e-05,
-      "loss": 59.2733,
-      "step": 100670
-    },
-    {
-      "epoch": 0.4067599397213121,
-      "grad_norm": 2471.260986328125,
-      "learning_rate": 3.729387574439662e-05,
-      "loss": 68.098,
-      "step": 100680
-    },
-    {
-      "epoch": 0.4068003409866797,
-      "grad_norm": 768.341552734375,
-      "learning_rate": 3.729083619094624e-05,
-      "loss": 73.1918,
-      "step": 100690
-    },
-    {
-      "epoch": 0.40684074225204736,
-      "grad_norm": 0.0,
-      "learning_rate": 3.7287796397879674e-05,
-      "loss": 28.0081,
-      "step": 100700
-    },
-    {
-      "epoch": 0.40688114351741494,
-      "grad_norm": 1264.1614990234375,
-      "learning_rate": 3.72847563652562e-05,
-      "loss": 45.6677,
-      "step": 100710
-    },
-    {
-      "epoch": 0.4069215447827826,
-      "grad_norm": 287.6547546386719,
-      "learning_rate": 3.7281716093135063e-05,
-      "loss": 60.4336,
-      "step": 100720
-    },
-    {
-      "epoch": 0.4069619460481502,
-      "grad_norm": 565.496826171875,
-      "learning_rate": 3.7278675581575564e-05,
-      "loss": 49.9453,
-      "step": 100730
-    },
-    {
-      "epoch": 0.40700234731351786,
-      "grad_norm": 513.9716796875,
-      "learning_rate": 3.7275634830636957e-05,
-      "loss": 48.138,
-      "step": 100740
-    },
-    {
-      "epoch": 0.4070427485788855,
-      "grad_norm": 798.354248046875,
-      "learning_rate": 3.727259384037852e-05,
-      "loss": 49.801,
-      "step": 100750
-    },
-    {
-      "epoch": 0.40708314984425314,
-      "grad_norm": 901.558837890625,
-      "learning_rate": 3.726955261085956e-05,
-      "loss": 38.1762,
-      "step": 100760
-    },
-    {
-      "epoch": 0.40712355110962073,
-      "grad_norm": 808.7131958007812,
-      "learning_rate": 3.726651114213935e-05,
-      "loss": 63.8376,
-      "step": 100770
-    },
-    {
-      "epoch": 0.40716395237498837,
-      "grad_norm": 726.5311279296875,
-      "learning_rate": 3.726346943427719e-05,
-      "loss": 51.7929,
-      "step": 100780
-    },
-    {
-      "epoch": 0.407204353640356,
-      "grad_norm": 367.88018798828125,
-      "learning_rate": 3.726042748733238e-05,
-      "loss": 85.8663,
-      "step": 100790
-    },
-    {
-      "epoch": 0.40724475490572365,
-      "grad_norm": 368.78338623046875,
-      "learning_rate": 3.725738530136422e-05,
-      "loss": 33.2297,
-      "step": 100800
-    },
-    {
-      "epoch": 0.4072851561710913,
-      "grad_norm": 942.588623046875,
-      "learning_rate": 3.7254342876432026e-05,
-      "loss": 74.4766,
-      "step": 100810
-    },
-    {
-      "epoch": 0.40732555743645893,
-      "grad_norm": 704.6798095703125,
-      "learning_rate": 3.7251300212595106e-05,
-      "loss": 84.4977,
-      "step": 100820
-    },
-    {
-      "epoch": 0.4073659587018265,
-      "grad_norm": 411.1459045410156,
-      "learning_rate": 3.724825730991279e-05,
-      "loss": 67.2837,
-      "step": 100830
-    },
-    {
-      "epoch": 0.40740635996719415,
-      "grad_norm": 969.4332885742188,
-      "learning_rate": 3.7245214168444386e-05,
-      "loss": 73.3558,
-      "step": 100840
-    },
-    {
-      "epoch": 0.4074467612325618,
-      "grad_norm": 857.5194091796875,
-      "learning_rate": 3.724217078824923e-05,
-      "loss": 87.0812,
-      "step": 100850
-    },
-    {
-      "epoch": 0.40748716249792943,
-      "grad_norm": 789.8101196289062,
-      "learning_rate": 3.723912716938665e-05,
-      "loss": 62.384,
-      "step": 100860
-    },
-    {
-      "epoch": 0.4075275637632971,
-      "grad_norm": 963.7822265625,
-      "learning_rate": 3.723608331191598e-05,
-      "loss": 70.5818,
-      "step": 100870
-    },
-    {
-      "epoch": 0.4075679650286647,
-      "grad_norm": 1622.1092529296875,
-      "learning_rate": 3.723303921589657e-05,
-      "loss": 73.844,
-      "step": 100880
-    },
-    {
-      "epoch": 0.40760836629403235,
-      "grad_norm": 4206.56787109375,
-      "learning_rate": 3.722999488138776e-05,
-      "loss": 86.4621,
-      "step": 100890
-    },
-    {
-      "epoch": 0.40764876755939994,
-      "grad_norm": 780.2285766601562,
-      "learning_rate": 3.722695030844891e-05,
-      "loss": 42.5627,
-      "step": 100900
-    },
-    {
-      "epoch": 0.4076891688247676,
-      "grad_norm": 1180.5223388671875,
-      "learning_rate": 3.7223905497139366e-05,
-      "loss": 59.9004,
-      "step": 100910
-    },
-    {
-      "epoch": 0.4077295700901352,
-      "grad_norm": 609.8385009765625,
-      "learning_rate": 3.722086044751849e-05,
-      "loss": 39.532,
-      "step": 100920
-    },
-    {
-      "epoch": 0.40776997135550286,
-      "grad_norm": 0.0,
-      "learning_rate": 3.721781515964565e-05,
-      "loss": 33.3144,
-      "step": 100930
-    },
-    {
-      "epoch": 0.4078103726208705,
-      "grad_norm": 1108.013671875,
-      "learning_rate": 3.721476963358021e-05,
-      "loss": 67.8884,
-      "step": 100940
-    },
-    {
-      "epoch": 0.40785077388623814,
-      "grad_norm": 503.3689270019531,
-      "learning_rate": 3.721172386938155e-05,
-      "loss": 63.0006,
-      "step": 100950
-    },
-    {
-      "epoch": 0.4078911751516057,
-      "grad_norm": 2834.478515625,
-      "learning_rate": 3.720867786710904e-05,
-      "loss": 74.9134,
-      "step": 100960
-    },
-    {
-      "epoch": 0.40793157641697336,
-      "grad_norm": 1194.1221923828125,
-      "learning_rate": 3.7205631626822074e-05,
-      "loss": 47.9969,
-      "step": 100970
-    },
-    {
-      "epoch": 0.407971977682341,
-      "grad_norm": 798.5789794921875,
-      "learning_rate": 3.7202585148580036e-05,
-      "loss": 86.6126,
-      "step": 100980
-    },
-    {
-      "epoch": 0.40801237894770864,
-      "grad_norm": 505.0960998535156,
-      "learning_rate": 3.7199538432442316e-05,
-      "loss": 59.1577,
-      "step": 100990
-    },
-    {
-      "epoch": 0.4080527802130763,
-      "grad_norm": 767.0346069335938,
-      "learning_rate": 3.719649147846832e-05,
-      "loss": 64.2209,
-      "step": 101000
-    },
-    {
-      "epoch": 0.4080931814784439,
-      "grad_norm": 577.536865234375,
-      "learning_rate": 3.7193444286717436e-05,
-      "loss": 64.3384,
-      "step": 101010
-    },
-    {
-      "epoch": 0.40813358274381156,
-      "grad_norm": 668.321533203125,
-      "learning_rate": 3.719039685724909e-05,
-      "loss": 39.2026,
-      "step": 101020
-    },
-    {
-      "epoch": 0.40817398400917915,
-      "grad_norm": 211.69288635253906,
-      "learning_rate": 3.718734919012267e-05,
-      "loss": 51.6079,
-      "step": 101030
-    },
-    {
-      "epoch": 0.4082143852745468,
-      "grad_norm": 391.42852783203125,
-      "learning_rate": 3.71843012853976e-05,
-      "loss": 42.6838,
-      "step": 101040
-    },
-    {
-      "epoch": 0.4082547865399144,
-      "grad_norm": 441.98419189453125,
-      "learning_rate": 3.718125314313331e-05,
-      "loss": 40.6746,
-      "step": 101050
-    },
-    {
-      "epoch": 0.40829518780528207,
-      "grad_norm": 772.721923828125,
-      "learning_rate": 3.7178204763389216e-05,
-      "loss": 68.107,
-      "step": 101060
-    },
-    {
-      "epoch": 0.4083355890706497,
-      "grad_norm": 286.0000305175781,
-      "learning_rate": 3.717515614622476e-05,
-      "loss": 79.75,
-      "step": 101070
-    },
-    {
-      "epoch": 0.40837599033601735,
-      "grad_norm": 966.0552368164062,
-      "learning_rate": 3.717210729169935e-05,
-      "loss": 72.2583,
-      "step": 101080
-    },
-    {
-      "epoch": 0.40841639160138493,
-      "grad_norm": 756.3565673828125,
-      "learning_rate": 3.7169058199872455e-05,
-      "loss": 69.5634,
-      "step": 101090
-    },
-    {
-      "epoch": 0.40845679286675257,
-      "grad_norm": 754.9552001953125,
-      "learning_rate": 3.71660088708035e-05,
-      "loss": 58.8931,
-      "step": 101100
-    },
-    {
-      "epoch": 0.4084971941321202,
-      "grad_norm": 1555.589599609375,
-      "learning_rate": 3.716295930455194e-05,
-      "loss": 79.8807,
-      "step": 101110
-    },
-    {
-      "epoch": 0.40853759539748785,
-      "grad_norm": 1325.0550537109375,
-      "learning_rate": 3.7159909501177226e-05,
-      "loss": 68.7145,
-      "step": 101120
-    },
-    {
-      "epoch": 0.4085779966628555,
-      "grad_norm": 620.8428955078125,
-      "learning_rate": 3.715685946073881e-05,
-      "loss": 55.6151,
-      "step": 101130
-    },
-    {
-      "epoch": 0.40861839792822313,
-      "grad_norm": 955.8616943359375,
-      "learning_rate": 3.7153809183296176e-05,
-      "loss": 66.1763,
-      "step": 101140
-    },
-    {
-      "epoch": 0.4086587991935907,
-      "grad_norm": 599.90234375,
-      "learning_rate": 3.715075866890876e-05,
-      "loss": 65.0991,
-      "step": 101150
-    },
-    {
-      "epoch": 0.40869920045895836,
-      "grad_norm": 3034.8486328125,
-      "learning_rate": 3.7147707917636046e-05,
-      "loss": 69.5837,
-      "step": 101160
-    },
-    {
-      "epoch": 0.408739601724326,
-      "grad_norm": 0.0,
-      "learning_rate": 3.7144656929537524e-05,
-      "loss": 45.1975,
-      "step": 101170
-    },
-    {
-      "epoch": 0.40878000298969364,
-      "grad_norm": 502.8624572753906,
-      "learning_rate": 3.714160570467266e-05,
-      "loss": 73.8093,
-      "step": 101180
-    },
-    {
-      "epoch": 0.4088204042550613,
-      "grad_norm": 0.0,
-      "learning_rate": 3.7138554243100934e-05,
-      "loss": 48.2301,
-      "step": 101190
-    },
-    {
-      "epoch": 0.4088608055204289,
-      "grad_norm": 648.0908203125,
-      "learning_rate": 3.713550254488185e-05,
-      "loss": 103.6744,
-      "step": 101200
-    },
-    {
-      "epoch": 0.40890120678579656,
-      "grad_norm": 1352.32421875,
-      "learning_rate": 3.71324506100749e-05,
-      "loss": 82.1861,
-      "step": 101210
-    },
-    {
-      "epoch": 0.40894160805116414,
-      "grad_norm": 1041.673583984375,
-      "learning_rate": 3.712939843873957e-05,
-      "loss": 40.6213,
-      "step": 101220
-    },
-    {
-      "epoch": 0.4089820093165318,
-      "grad_norm": 763.6616821289062,
-      "learning_rate": 3.7126346030935374e-05,
-      "loss": 55.786,
-      "step": 101230
-    },
-    {
-      "epoch": 0.4090224105818994,
-      "grad_norm": 892.0834350585938,
-      "learning_rate": 3.712329338672182e-05,
-      "loss": 60.7086,
-      "step": 101240
-    },
-    {
-      "epoch": 0.40906281184726706,
-      "grad_norm": 788.8603515625,
-      "learning_rate": 3.712024050615843e-05,
-      "loss": 53.5069,
-      "step": 101250
-    },
-    {
-      "epoch": 0.4091032131126347,
-      "grad_norm": 1101.8909912109375,
-      "learning_rate": 3.71171873893047e-05,
-      "loss": 47.3406,
-      "step": 101260
-    },
-    {
-      "epoch": 0.40914361437800234,
-      "grad_norm": 296.9659729003906,
-      "learning_rate": 3.711413403622017e-05,
-      "loss": 55.5417,
-      "step": 101270
-    },
-    {
-      "epoch": 0.4091840156433699,
-      "grad_norm": 653.2747802734375,
-      "learning_rate": 3.711108044696436e-05,
-      "loss": 71.033,
-      "step": 101280
-    },
-    {
-      "epoch": 0.40922441690873756,
-      "grad_norm": 671.96533203125,
-      "learning_rate": 3.710802662159679e-05,
-      "loss": 58.3862,
-      "step": 101290
-    },
-    {
-      "epoch": 0.4092648181741052,
-      "grad_norm": 519.48095703125,
-      "learning_rate": 3.710497256017702e-05,
-      "loss": 90.0393,
-      "step": 101300
-    },
-    {
-      "epoch": 0.40930521943947284,
-      "grad_norm": 837.6402587890625,
-      "learning_rate": 3.7101918262764576e-05,
-      "loss": 60.9863,
-      "step": 101310
-    },
-    {
-      "epoch": 0.4093456207048405,
-      "grad_norm": 907.7230224609375,
-      "learning_rate": 3.7098863729419e-05,
-      "loss": 80.4925,
-      "step": 101320
-    },
-    {
-      "epoch": 0.4093860219702081,
-      "grad_norm": 2056.220947265625,
-      "learning_rate": 3.709580896019985e-05,
-      "loss": 49.3788,
-      "step": 101330
-    },
-    {
-      "epoch": 0.40942642323557576,
-      "grad_norm": 531.6597900390625,
-      "learning_rate": 3.7092753955166674e-05,
-      "loss": 81.7104,
-      "step": 101340
-    },
-    {
-      "epoch": 0.40946682450094335,
-      "grad_norm": 714.7631225585938,
-      "learning_rate": 3.708969871437904e-05,
-      "loss": 63.8878,
-      "step": 101350
-    },
-    {
-      "epoch": 0.409507225766311,
-      "grad_norm": 1157.5352783203125,
-      "learning_rate": 3.7086643237896504e-05,
-      "loss": 41.3693,
-      "step": 101360
-    },
-    {
-      "epoch": 0.40954762703167863,
-      "grad_norm": 531.5262451171875,
-      "learning_rate": 3.708358752577863e-05,
-      "loss": 71.7268,
-      "step": 101370
-    },
-    {
-      "epoch": 0.40958802829704627,
-      "grad_norm": 447.32330322265625,
-      "learning_rate": 3.7080531578085e-05,
-      "loss": 66.6252,
-      "step": 101380
-    },
-    {
-      "epoch": 0.4096284295624139,
-      "grad_norm": 983.9207153320312,
-      "learning_rate": 3.707747539487519e-05,
-      "loss": 62.158,
-      "step": 101390
-    },
-    {
-      "epoch": 0.40966883082778155,
-      "grad_norm": 294.036376953125,
-      "learning_rate": 3.7074418976208766e-05,
-      "loss": 54.1539,
-      "step": 101400
-    },
-    {
-      "epoch": 0.40970923209314913,
-      "grad_norm": 2036.715087890625,
-      "learning_rate": 3.707136232214534e-05,
-      "loss": 53.9095,
-      "step": 101410
-    },
-    {
-      "epoch": 0.4097496333585168,
-      "grad_norm": 1761.55908203125,
-      "learning_rate": 3.706830543274449e-05,
-      "loss": 57.585,
-      "step": 101420
-    },
-    {
-      "epoch": 0.4097900346238844,
-      "grad_norm": 1144.66064453125,
-      "learning_rate": 3.706524830806581e-05,
-      "loss": 80.4368,
-      "step": 101430
-    },
-    {
-      "epoch": 0.40983043588925205,
-      "grad_norm": 457.09368896484375,
-      "learning_rate": 3.706219094816891e-05,
-      "loss": 80.4445,
-      "step": 101440
-    },
-    {
-      "epoch": 0.4098708371546197,
-      "grad_norm": 451.6939697265625,
-      "learning_rate": 3.705913335311338e-05,
-      "loss": 46.0204,
-      "step": 101450
-    },
-    {
-      "epoch": 0.40991123841998733,
-      "grad_norm": 675.3792724609375,
-      "learning_rate": 3.705607552295883e-05,
-      "loss": 29.4593,
-      "step": 101460
-    },
-    {
-      "epoch": 0.4099516396853549,
-      "grad_norm": 473.2438659667969,
-      "learning_rate": 3.7053017457764895e-05,
-      "loss": 69.6346,
-      "step": 101470
-    },
-    {
-      "epoch": 0.40999204095072256,
-      "grad_norm": 1086.820556640625,
-      "learning_rate": 3.704995915759117e-05,
-      "loss": 52.2497,
-      "step": 101480
-    },
-    {
-      "epoch": 0.4100324422160902,
-      "grad_norm": 1838.8997802734375,
-      "learning_rate": 3.704690062249729e-05,
-      "loss": 67.0797,
-      "step": 101490
-    },
-    {
-      "epoch": 0.41007284348145784,
-      "grad_norm": 1140.804931640625,
-      "learning_rate": 3.704384185254288e-05,
-      "loss": 59.0365,
-      "step": 101500
-    },
-    {
-      "epoch": 0.4101132447468255,
-      "grad_norm": 967.5361938476562,
-      "learning_rate": 3.7040782847787576e-05,
-      "loss": 62.3368,
-      "step": 101510
-    },
-    {
-      "epoch": 0.4101536460121931,
-      "grad_norm": 862.10107421875,
-      "learning_rate": 3.7037723608291015e-05,
-      "loss": 66.9464,
-      "step": 101520
-    },
-    {
-      "epoch": 0.41019404727756076,
-      "grad_norm": 333.78411865234375,
-      "learning_rate": 3.703466413411282e-05,
-      "loss": 57.9339,
-      "step": 101530
-    },
-    {
-      "epoch": 0.41023444854292834,
-      "grad_norm": 1024.932861328125,
-      "learning_rate": 3.703160442531266e-05,
-      "loss": 50.8443,
-      "step": 101540
-    },
-    {
-      "epoch": 0.410274849808296,
-      "grad_norm": 1006.3261108398438,
-      "learning_rate": 3.702854448195019e-05,
-      "loss": 66.2163,
-      "step": 101550
-    },
-    {
-      "epoch": 0.4103152510736636,
-      "grad_norm": 720.7424926757812,
-      "learning_rate": 3.7025484304085034e-05,
-      "loss": 61.6989,
-      "step": 101560
-    },
-    {
-      "epoch": 0.41035565233903126,
-      "grad_norm": 972.9140625,
-      "learning_rate": 3.702242389177687e-05,
-      "loss": 58.963,
-      "step": 101570
-    },
-    {
-      "epoch": 0.4103960536043989,
-      "grad_norm": 819.4244384765625,
-      "learning_rate": 3.701936324508537e-05,
-      "loss": 58.8996,
-      "step": 101580
-    },
-    {
-      "epoch": 0.41043645486976654,
-      "grad_norm": 1408.4876708984375,
-      "learning_rate": 3.7016302364070196e-05,
-      "loss": 104.3692,
-      "step": 101590
-    },
-    {
-      "epoch": 0.4104768561351341,
-      "grad_norm": 1001.0791625976562,
-      "learning_rate": 3.701324124879102e-05,
-      "loss": 76.1169,
-      "step": 101600
-    },
-    {
-      "epoch": 0.41051725740050177,
-      "grad_norm": 1362.98193359375,
-      "learning_rate": 3.701017989930752e-05,
-      "loss": 79.0587,
-      "step": 101610
-    },
-    {
-      "epoch": 0.4105576586658694,
-      "grad_norm": 494.7317199707031,
-      "learning_rate": 3.7007118315679384e-05,
-      "loss": 41.3256,
-      "step": 101620
-    },
-    {
-      "epoch": 0.41059805993123705,
-      "grad_norm": 418.2009582519531,
-      "learning_rate": 3.700405649796629e-05,
-      "loss": 77.2711,
-      "step": 101630
-    },
-    {
-      "epoch": 0.4106384611966047,
-      "grad_norm": 934.8997192382812,
-      "learning_rate": 3.700099444622794e-05,
-      "loss": 64.7147,
-      "step": 101640
-    },
-    {
-      "epoch": 0.4106788624619723,
-      "grad_norm": 829.1802978515625,
-      "learning_rate": 3.699793216052402e-05,
-      "loss": 57.4641,
-      "step": 101650
-    },
-    {
-      "epoch": 0.41071926372733997,
-      "grad_norm": 536.0474853515625,
-      "learning_rate": 3.699486964091423e-05,
-      "loss": 79.0541,
-      "step": 101660
-    },
-    {
-      "epoch": 0.41075966499270755,
-      "grad_norm": 808.2246704101562,
-      "learning_rate": 3.699180688745829e-05,
-      "loss": 50.0221,
-      "step": 101670
-    },
-    {
-      "epoch": 0.4108000662580752,
-      "grad_norm": 558.8209838867188,
-      "learning_rate": 3.6988743900215894e-05,
-      "loss": 52.1435,
-      "step": 101680
-    },
-    {
-      "epoch": 0.41084046752344283,
-      "grad_norm": 1482.892578125,
-      "learning_rate": 3.698568067924677e-05,
-      "loss": 49.8897,
-      "step": 101690
-    },
-    {
-      "epoch": 0.41088086878881047,
-      "grad_norm": 494.3254699707031,
-      "learning_rate": 3.698261722461063e-05,
-      "loss": 64.2092,
-      "step": 101700
-    },
-    {
-      "epoch": 0.4109212700541781,
-      "grad_norm": 455.20489501953125,
-      "learning_rate": 3.6979553536367194e-05,
-      "loss": 38.1696,
-      "step": 101710
-    },
-    {
-      "epoch": 0.41096167131954575,
-      "grad_norm": 528.5197143554688,
-      "learning_rate": 3.69764896145762e-05,
-      "loss": 42.2116,
-      "step": 101720
-    },
-    {
-      "epoch": 0.41100207258491334,
-      "grad_norm": 935.036865234375,
-      "learning_rate": 3.697342545929737e-05,
-      "loss": 76.0732,
-      "step": 101730
-    },
-    {
-      "epoch": 0.411042473850281,
-      "grad_norm": 1118.97216796875,
-      "learning_rate": 3.697036107059044e-05,
-      "loss": 37.8624,
-      "step": 101740
-    },
-    {
-      "epoch": 0.4110828751156486,
-      "grad_norm": 1046.9444580078125,
-      "learning_rate": 3.696729644851518e-05,
-      "loss": 34.6092,
-      "step": 101750
-    },
-    {
-      "epoch": 0.41112327638101626,
-      "grad_norm": 427.7893981933594,
-      "learning_rate": 3.696423159313129e-05,
-      "loss": 67.689,
-      "step": 101760
-    },
-    {
-      "epoch": 0.4111636776463839,
-      "grad_norm": 655.9987182617188,
-      "learning_rate": 3.696116650449856e-05,
-      "loss": 42.4555,
-      "step": 101770
-    },
-    {
-      "epoch": 0.41120407891175154,
-      "grad_norm": 911.2579956054688,
-      "learning_rate": 3.6958101182676726e-05,
-      "loss": 61.5867,
-      "step": 101780
-    },
-    {
-      "epoch": 0.4112444801771191,
-      "grad_norm": 2679.955078125,
-      "learning_rate": 3.6955035627725557e-05,
-      "loss": 62.913,
-      "step": 101790
-    },
-    {
-      "epoch": 0.41128488144248676,
-      "grad_norm": 736.22802734375,
-      "learning_rate": 3.695196983970481e-05,
-      "loss": 54.0211,
-      "step": 101800
-    },
-    {
-      "epoch": 0.4113252827078544,
-      "grad_norm": 1243.8758544921875,
-      "learning_rate": 3.694890381867425e-05,
-      "loss": 58.7315,
-      "step": 101810
-    },
-    {
-      "epoch": 0.41136568397322204,
-      "grad_norm": 828.197265625,
-      "learning_rate": 3.6945837564693666e-05,
-      "loss": 65.7947,
-      "step": 101820
-    },
-    {
-      "epoch": 0.4114060852385897,
-      "grad_norm": 682.7947387695312,
-      "learning_rate": 3.6942771077822835e-05,
-      "loss": 57.709,
-      "step": 101830
-    },
-    {
-      "epoch": 0.4114464865039573,
-      "grad_norm": 1457.8909912109375,
-      "learning_rate": 3.693970435812153e-05,
-      "loss": 53.162,
-      "step": 101840
-    },
-    {
-      "epoch": 0.41148688776932496,
-      "grad_norm": 1344.3275146484375,
-      "learning_rate": 3.693663740564953e-05,
-      "loss": 54.1553,
-      "step": 101850
-    },
-    {
-      "epoch": 0.41152728903469254,
-      "grad_norm": 0.0,
-      "learning_rate": 3.693357022046665e-05,
-      "loss": 47.7167,
-      "step": 101860
-    },
-    {
-      "epoch": 0.4115676903000602,
-      "grad_norm": 927.4996337890625,
-      "learning_rate": 3.693050280263268e-05,
-      "loss": 74.7277,
-      "step": 101870
-    },
-    {
-      "epoch": 0.4116080915654278,
-      "grad_norm": 401.41693115234375,
-      "learning_rate": 3.6927435152207406e-05,
-      "loss": 51.1262,
-      "step": 101880
-    },
-    {
-      "epoch": 0.41164849283079546,
-      "grad_norm": 756.7000122070312,
-      "learning_rate": 3.6924367269250644e-05,
-      "loss": 43.8885,
-      "step": 101890
-    },
-    {
-      "epoch": 0.4116888940961631,
-      "grad_norm": 1407.22412109375,
-      "learning_rate": 3.69212991538222e-05,
-      "loss": 51.1716,
-      "step": 101900
-    },
-    {
-      "epoch": 0.41172929536153074,
-      "grad_norm": 1306.109619140625,
-      "learning_rate": 3.691823080598189e-05,
-      "loss": 56.4997,
-      "step": 101910
-    },
-    {
-      "epoch": 0.41176969662689833,
-      "grad_norm": 890.89013671875,
-      "learning_rate": 3.6915162225789546e-05,
-      "loss": 88.45,
-      "step": 101920
-    },
-    {
-      "epoch": 0.41181009789226597,
-      "grad_norm": 195.86825561523438,
-      "learning_rate": 3.691209341330497e-05,
-      "loss": 70.7165,
-      "step": 101930
-    },
-    {
-      "epoch": 0.4118504991576336,
-      "grad_norm": 1073.44677734375,
-      "learning_rate": 3.690902436858801e-05,
-      "loss": 88.2749,
-      "step": 101940
-    },
-    {
-      "epoch": 0.41189090042300125,
-      "grad_norm": 310.966552734375,
-      "learning_rate": 3.690595509169848e-05,
-      "loss": 101.5226,
-      "step": 101950
-    },
-    {
-      "epoch": 0.4119313016883689,
-      "grad_norm": 515.8529663085938,
-      "learning_rate": 3.690288558269623e-05,
-      "loss": 55.0783,
-      "step": 101960
-    },
-    {
-      "epoch": 0.41197170295373653,
-      "grad_norm": 548.2571411132812,
-      "learning_rate": 3.68998158416411e-05,
-      "loss": 51.7741,
-      "step": 101970
-    },
-    {
-      "epoch": 0.41201210421910417,
-      "grad_norm": 712.4876708984375,
-      "learning_rate": 3.689674586859292e-05,
-      "loss": 55.8182,
-      "step": 101980
-    },
-    {
-      "epoch": 0.41205250548447175,
-      "grad_norm": 1289.3074951171875,
-      "learning_rate": 3.689367566361157e-05,
-      "loss": 83.5634,
-      "step": 101990
-    },
-    {
-      "epoch": 0.4120929067498394,
-      "grad_norm": 378.75164794921875,
-      "learning_rate": 3.689060522675689e-05,
-      "loss": 62.7789,
-      "step": 102000
-    },
-    {
-      "epoch": 0.41213330801520703,
-      "grad_norm": 788.7225341796875,
-      "learning_rate": 3.6887534558088727e-05,
-      "loss": 51.0795,
-      "step": 102010
-    },
-    {
-      "epoch": 0.4121737092805747,
-      "grad_norm": 419.6149597167969,
-      "learning_rate": 3.688446365766696e-05,
-      "loss": 72.9321,
-      "step": 102020
-    },
-    {
-      "epoch": 0.4122141105459423,
-      "grad_norm": 424.8109436035156,
-      "learning_rate": 3.688139252555146e-05,
-      "loss": 51.3233,
-      "step": 102030
-    },
-    {
-      "epoch": 0.41225451181130995,
-      "grad_norm": 340.1546630859375,
-      "learning_rate": 3.6878321161802104e-05,
-      "loss": 99.2203,
-      "step": 102040
-    },
-    {
-      "epoch": 0.41229491307667754,
-      "grad_norm": 572.083984375,
-      "learning_rate": 3.6875249566478745e-05,
-      "loss": 62.0742,
-      "step": 102050
-    },
-    {
-      "epoch": 0.4123353143420452,
-      "grad_norm": 639.38818359375,
-      "learning_rate": 3.687217773964129e-05,
-      "loss": 72.4136,
-      "step": 102060
-    },
-    {
-      "epoch": 0.4123757156074128,
-      "grad_norm": 405.4486389160156,
-      "learning_rate": 3.686910568134962e-05,
-      "loss": 57.4386,
-      "step": 102070
-    },
-    {
-      "epoch": 0.41241611687278046,
-      "grad_norm": 938.0454711914062,
-      "learning_rate": 3.686603339166362e-05,
-      "loss": 64.274,
-      "step": 102080
-    },
-    {
-      "epoch": 0.4124565181381481,
-      "grad_norm": 1430.1524658203125,
-      "learning_rate": 3.686296087064319e-05,
-      "loss": 62.3886,
-      "step": 102090
-    },
-    {
-      "epoch": 0.41249691940351574,
-      "grad_norm": 796.8814086914062,
-      "learning_rate": 3.685988811834823e-05,
-      "loss": 85.7702,
-      "step": 102100
-    },
-    {
-      "epoch": 0.4125373206688833,
-      "grad_norm": 1625.54833984375,
-      "learning_rate": 3.685681513483865e-05,
-      "loss": 86.8409,
-      "step": 102110
-    },
-    {
-      "epoch": 0.41257772193425096,
-      "grad_norm": 754.6946411132812,
-      "learning_rate": 3.685374192017436e-05,
-      "loss": 42.4442,
-      "step": 102120
-    },
-    {
-      "epoch": 0.4126181231996186,
-      "grad_norm": 599.8799438476562,
-      "learning_rate": 3.6850668474415255e-05,
-      "loss": 42.2114,
-      "step": 102130
-    },
-    {
-      "epoch": 0.41265852446498624,
-      "grad_norm": 919.1255493164062,
-      "learning_rate": 3.684759479762127e-05,
-      "loss": 66.864,
-      "step": 102140
-    },
-    {
-      "epoch": 0.4126989257303539,
-      "grad_norm": 352.7879943847656,
-      "learning_rate": 3.684452088985233e-05,
-      "loss": 46.0815,
-      "step": 102150
-    },
-    {
-      "epoch": 0.4127393269957215,
-      "grad_norm": 1374.6343994140625,
-      "learning_rate": 3.6841446751168355e-05,
-      "loss": 69.1762,
-      "step": 102160
-    },
-    {
-      "epoch": 0.41277972826108916,
-      "grad_norm": 597.5450439453125,
-      "learning_rate": 3.683837238162928e-05,
-      "loss": 85.0557,
-      "step": 102170
-    },
-    {
-      "epoch": 0.41282012952645675,
-      "grad_norm": 674.8424072265625,
-      "learning_rate": 3.683529778129503e-05,
-      "loss": 79.6359,
-      "step": 102180
-    },
-    {
-      "epoch": 0.4128605307918244,
-      "grad_norm": 624.8746337890625,
-      "learning_rate": 3.683222295022557e-05,
-      "loss": 39.7842,
-      "step": 102190
-    },
-    {
-      "epoch": 0.412900932057192,
-      "grad_norm": 1277.397705078125,
-      "learning_rate": 3.682914788848083e-05,
-      "loss": 58.8937,
-      "step": 102200
-    },
-    {
-      "epoch": 0.41294133332255967,
-      "grad_norm": 722.8589477539062,
-      "learning_rate": 3.682607259612076e-05,
-      "loss": 80.0508,
-      "step": 102210
-    },
-    {
-      "epoch": 0.4129817345879273,
-      "grad_norm": 824.6448364257812,
-      "learning_rate": 3.682299707320532e-05,
-      "loss": 53.1664,
-      "step": 102220
-    },
-    {
-      "epoch": 0.41302213585329495,
-      "grad_norm": 234.46766662597656,
-      "learning_rate": 3.681992131979446e-05,
-      "loss": 66.3412,
-      "step": 102230
-    },
-    {
-      "epoch": 0.41306253711866253,
-      "grad_norm": 1732.6822509765625,
-      "learning_rate": 3.681684533594815e-05,
-      "loss": 118.9763,
-      "step": 102240
-    },
-    {
-      "epoch": 0.41310293838403017,
-      "grad_norm": 985.6689453125,
-      "learning_rate": 3.681376912172636e-05,
-      "loss": 48.3016,
-      "step": 102250
-    },
-    {
-      "epoch": 0.4131433396493978,
-      "grad_norm": 534.2235107421875,
-      "learning_rate": 3.6810692677189046e-05,
-      "loss": 41.1004,
-      "step": 102260
-    },
-    {
-      "epoch": 0.41318374091476545,
-      "grad_norm": 0.0,
-      "learning_rate": 3.680761600239621e-05,
-      "loss": 47.6259,
-      "step": 102270
-    },
-    {
-      "epoch": 0.4132241421801331,
-      "grad_norm": 2476.089111328125,
-      "learning_rate": 3.680453909740782e-05,
-      "loss": 60.4867,
-      "step": 102280
-    },
-    {
-      "epoch": 0.41326454344550073,
-      "grad_norm": 733.741455078125,
-      "learning_rate": 3.680146196228386e-05,
-      "loss": 58.1537,
-      "step": 102290
-    },
-    {
-      "epoch": 0.41330494471086837,
-      "grad_norm": 543.4186401367188,
-      "learning_rate": 3.6798384597084325e-05,
-      "loss": 42.9503,
-      "step": 102300
-    },
-    {
-      "epoch": 0.41334534597623596,
-      "grad_norm": 535.3422241210938,
-      "learning_rate": 3.67953070018692e-05,
-      "loss": 76.8054,
-      "step": 102310
-    },
-    {
-      "epoch": 0.4133857472416036,
-      "grad_norm": 851.113037109375,
-      "learning_rate": 3.679222917669851e-05,
-      "loss": 53.6934,
-      "step": 102320
-    },
-    {
-      "epoch": 0.41342614850697124,
-      "grad_norm": 3898.60693359375,
-      "learning_rate": 3.6789151121632226e-05,
-      "loss": 58.586,
-      "step": 102330
-    },
-    {
-      "epoch": 0.4134665497723389,
-      "grad_norm": 425.6492919921875,
-      "learning_rate": 3.678607283673037e-05,
-      "loss": 61.4479,
-      "step": 102340
-    },
-    {
-      "epoch": 0.4135069510377065,
-      "grad_norm": 877.3761596679688,
-      "learning_rate": 3.678299432205296e-05,
-      "loss": 73.7707,
-      "step": 102350
-    },
-    {
-      "epoch": 0.41354735230307416,
-      "grad_norm": 763.9284057617188,
-      "learning_rate": 3.6779915577660015e-05,
-      "loss": 56.7363,
-      "step": 102360
-    },
-    {
-      "epoch": 0.41358775356844174,
-      "grad_norm": 1059.4683837890625,
-      "learning_rate": 3.677683660361155e-05,
-      "loss": 52.8749,
-      "step": 102370
-    },
-    {
-      "epoch": 0.4136281548338094,
-      "grad_norm": 408.92071533203125,
-      "learning_rate": 3.677375739996759e-05,
-      "loss": 53.0879,
-      "step": 102380
-    },
-    {
-      "epoch": 0.413668556099177,
-      "grad_norm": 529.644775390625,
-      "learning_rate": 3.677067796678817e-05,
-      "loss": 53.6878,
-      "step": 102390
-    },
-    {
-      "epoch": 0.41370895736454466,
-      "grad_norm": 888.7438354492188,
-      "learning_rate": 3.6767598304133324e-05,
-      "loss": 44.5858,
-      "step": 102400
-    },
-    {
-      "epoch": 0.4137493586299123,
-      "grad_norm": 1067.380615234375,
-      "learning_rate": 3.676451841206308e-05,
-      "loss": 42.4786,
-      "step": 102410
-    },
-    {
-      "epoch": 0.41378975989527994,
-      "grad_norm": 2619.344482421875,
-      "learning_rate": 3.67614382906375e-05,
-      "loss": 59.8869,
-      "step": 102420
-    },
-    {
-      "epoch": 0.4138301611606475,
-      "grad_norm": 730.7283325195312,
-      "learning_rate": 3.675835793991662e-05,
-      "loss": 53.2614,
-      "step": 102430
-    },
-    {
-      "epoch": 0.41387056242601516,
-      "grad_norm": 540.2338256835938,
-      "learning_rate": 3.67552773599605e-05,
-      "loss": 53.5669,
-      "step": 102440
-    },
-    {
-      "epoch": 0.4139109636913828,
-      "grad_norm": 424.4781799316406,
-      "learning_rate": 3.675219655082921e-05,
-      "loss": 40.3122,
-      "step": 102450
-    },
-    {
-      "epoch": 0.41395136495675044,
-      "grad_norm": 1885.5537109375,
-      "learning_rate": 3.6749115512582786e-05,
-      "loss": 59.4535,
-      "step": 102460
-    },
-    {
-      "epoch": 0.4139917662221181,
-      "grad_norm": 790.8214111328125,
-      "learning_rate": 3.674603424528131e-05,
-      "loss": 68.5578,
-      "step": 102470
-    },
-    {
-      "epoch": 0.4140321674874857,
-      "grad_norm": 788.549560546875,
-      "learning_rate": 3.674295274898485e-05,
-      "loss": 115.0106,
-      "step": 102480
-    },
-    {
-      "epoch": 0.41407256875285336,
-      "grad_norm": 3169.2861328125,
-      "learning_rate": 3.673987102375348e-05,
-      "loss": 78.3599,
-      "step": 102490
-    },
-    {
-      "epoch": 0.41411297001822095,
-      "grad_norm": 1180.6651611328125,
-      "learning_rate": 3.673678906964727e-05,
-      "loss": 69.2081,
-      "step": 102500
-    },
-    {
-      "epoch": 0.4141533712835886,
-      "grad_norm": 870.7091674804688,
-      "learning_rate": 3.673370688672632e-05,
-      "loss": 76.4967,
-      "step": 102510
-    },
-    {
-      "epoch": 0.41419377254895623,
-      "grad_norm": 535.1434326171875,
-      "learning_rate": 3.673062447505072e-05,
-      "loss": 37.986,
-      "step": 102520
-    },
-    {
-      "epoch": 0.41423417381432387,
-      "grad_norm": 574.4082641601562,
-      "learning_rate": 3.672754183468055e-05,
-      "loss": 43.2866,
-      "step": 102530
-    },
-    {
-      "epoch": 0.4142745750796915,
-      "grad_norm": 1545.2213134765625,
-      "learning_rate": 3.672445896567592e-05,
-      "loss": 84.1171,
-      "step": 102540
-    },
-    {
-      "epoch": 0.41431497634505915,
-      "grad_norm": 625.9170532226562,
-      "learning_rate": 3.6721375868096924e-05,
-      "loss": 44.562,
-      "step": 102550
-    },
-    {
-      "epoch": 0.41435537761042673,
-      "grad_norm": 948.0220336914062,
-      "learning_rate": 3.6718292542003666e-05,
-      "loss": 68.8121,
-      "step": 102560
-    },
-    {
-      "epoch": 0.4143957788757944,
-      "grad_norm": 400.49859619140625,
-      "learning_rate": 3.671520898745627e-05,
-      "loss": 59.8667,
-      "step": 102570
-    },
-    {
-      "epoch": 0.414436180141162,
-      "grad_norm": 329.8058776855469,
-      "learning_rate": 3.671212520451484e-05,
-      "loss": 76.8248,
-      "step": 102580
-    },
-    {
-      "epoch": 0.41447658140652965,
-      "grad_norm": 916.3284301757812,
-      "learning_rate": 3.670904119323949e-05,
-      "loss": 74.5609,
-      "step": 102590
-    },
-    {
-      "epoch": 0.4145169826718973,
-      "grad_norm": 561.3569946289062,
-      "learning_rate": 3.6705956953690364e-05,
-      "loss": 43.266,
-      "step": 102600
-    },
-    {
-      "epoch": 0.41455738393726493,
-      "grad_norm": 627.5663452148438,
-      "learning_rate": 3.670287248592758e-05,
-      "loss": 49.7998,
-      "step": 102610
-    },
-    {
-      "epoch": 0.4145977852026326,
-      "grad_norm": 919.4302368164062,
-      "learning_rate": 3.669978779001127e-05,
-      "loss": 72.0918,
-      "step": 102620
-    },
-    {
-      "epoch": 0.41463818646800016,
-      "grad_norm": 536.8106079101562,
-      "learning_rate": 3.669670286600157e-05,
-      "loss": 84.2724,
-      "step": 102630
-    },
-    {
-      "epoch": 0.4146785877333678,
-      "grad_norm": 926.4111328125,
-      "learning_rate": 3.6693617713958634e-05,
-      "loss": 49.7705,
-      "step": 102640
-    },
-    {
-      "epoch": 0.41471898899873544,
-      "grad_norm": 511.818359375,
-      "learning_rate": 3.66905323339426e-05,
-      "loss": 46.5629,
-      "step": 102650
-    },
-    {
-      "epoch": 0.4147593902641031,
-      "grad_norm": 1487.813720703125,
-      "learning_rate": 3.668744672601361e-05,
-      "loss": 55.226,
-      "step": 102660
-    },
-    {
-      "epoch": 0.4147997915294707,
-      "grad_norm": 1999.3150634765625,
-      "learning_rate": 3.668436089023184e-05,
-      "loss": 84.6968,
-      "step": 102670
-    },
-    {
-      "epoch": 0.41484019279483836,
-      "grad_norm": 443.5814208984375,
-      "learning_rate": 3.668127482665743e-05,
-      "loss": 59.7227,
-      "step": 102680
-    },
-    {
-      "epoch": 0.41488059406020594,
-      "grad_norm": 798.040283203125,
-      "learning_rate": 3.667818853535056e-05,
-      "loss": 88.5318,
-      "step": 102690
-    },
-    {
-      "epoch": 0.4149209953255736,
-      "grad_norm": 1065.25927734375,
-      "learning_rate": 3.667510201637139e-05,
-      "loss": 46.2704,
-      "step": 102700
-    },
-    {
-      "epoch": 0.4149613965909412,
-      "grad_norm": 213.02117919921875,
-      "learning_rate": 3.667201526978009e-05,
-      "loss": 72.9519,
-      "step": 102710
-    },
-    {
-      "epoch": 0.41500179785630886,
-      "grad_norm": 582.4226684570312,
-      "learning_rate": 3.6668928295636854e-05,
-      "loss": 43.031,
-      "step": 102720
-    },
-    {
-      "epoch": 0.4150421991216765,
-      "grad_norm": 550.1882934570312,
-      "learning_rate": 3.666584109400185e-05,
-      "loss": 56.0368,
-      "step": 102730
-    },
-    {
-      "epoch": 0.41508260038704414,
-      "grad_norm": 563.0902709960938,
-      "learning_rate": 3.666275366493526e-05,
-      "loss": 58.4682,
-      "step": 102740
-    },
-    {
-      "epoch": 0.4151230016524117,
-      "grad_norm": 1416.572998046875,
-      "learning_rate": 3.665966600849728e-05,
-      "loss": 48.4765,
-      "step": 102750
-    },
-    {
-      "epoch": 0.41516340291777937,
-      "grad_norm": 977.8565063476562,
-      "learning_rate": 3.665657812474812e-05,
-      "loss": 99.4884,
-      "step": 102760
-    },
-    {
-      "epoch": 0.415203804183147,
-      "grad_norm": 605.6978149414062,
-      "learning_rate": 3.6653490013747965e-05,
-      "loss": 62.5299,
-      "step": 102770
-    },
-    {
-      "epoch": 0.41524420544851465,
-      "grad_norm": 1297.1224365234375,
-      "learning_rate": 3.665040167555702e-05,
-      "loss": 53.5037,
-      "step": 102780
-    },
-    {
-      "epoch": 0.4152846067138823,
-      "grad_norm": 823.6423950195312,
-      "learning_rate": 3.664731311023549e-05,
-      "loss": 50.0008,
-      "step": 102790
-    },
-    {
-      "epoch": 0.4153250079792499,
-      "grad_norm": 611.2938842773438,
-      "learning_rate": 3.664422431784361e-05,
-      "loss": 50.3032,
-      "step": 102800
-    },
-    {
-      "epoch": 0.41536540924461757,
-      "grad_norm": 713.4339599609375,
-      "learning_rate": 3.6641135298441576e-05,
-      "loss": 65.9606,
-      "step": 102810
-    },
-    {
-      "epoch": 0.41540581050998515,
-      "grad_norm": 908.730712890625,
-      "learning_rate": 3.6638046052089616e-05,
-      "loss": 72.0269,
-      "step": 102820
-    },
-    {
-      "epoch": 0.4154462117753528,
-      "grad_norm": 714.7061767578125,
-      "learning_rate": 3.6634956578847954e-05,
-      "loss": 85.8837,
-      "step": 102830
-    },
-    {
-      "epoch": 0.41548661304072043,
-      "grad_norm": 490.82269287109375,
-      "learning_rate": 3.663186687877682e-05,
-      "loss": 57.8224,
-      "step": 102840
-    },
-    {
-      "epoch": 0.41552701430608807,
-      "grad_norm": 954.6878662109375,
-      "learning_rate": 3.662877695193646e-05,
-      "loss": 58.0162,
-      "step": 102850
-    },
-    {
-      "epoch": 0.4155674155714557,
-      "grad_norm": 220.2834930419922,
-      "learning_rate": 3.6625686798387106e-05,
-      "loss": 67.6011,
-      "step": 102860
-    },
-    {
-      "epoch": 0.41560781683682335,
-      "grad_norm": 1146.68994140625,
-      "learning_rate": 3.6622596418188995e-05,
-      "loss": 86.136,
-      "step": 102870
-    },
-    {
-      "epoch": 0.41564821810219094,
-      "grad_norm": 983.2677001953125,
-      "learning_rate": 3.661950581140239e-05,
-      "loss": 32.4779,
-      "step": 102880
-    },
-    {
-      "epoch": 0.4156886193675586,
-      "grad_norm": 694.6705322265625,
-      "learning_rate": 3.661641497808754e-05,
-      "loss": 52.4579,
-      "step": 102890
-    },
-    {
-      "epoch": 0.4157290206329262,
-      "grad_norm": 500.0966491699219,
-      "learning_rate": 3.66133239183047e-05,
-      "loss": 54.7675,
-      "step": 102900
-    },
-    {
-      "epoch": 0.41576942189829386,
-      "grad_norm": 847.8213500976562,
-      "learning_rate": 3.6610232632114124e-05,
-      "loss": 55.5015,
-      "step": 102910
-    },
-    {
-      "epoch": 0.4158098231636615,
-      "grad_norm": 1668.6177978515625,
-      "learning_rate": 3.6607141119576084e-05,
-      "loss": 98.7418,
-      "step": 102920
-    },
-    {
-      "epoch": 0.41585022442902914,
-      "grad_norm": 397.88885498046875,
-      "learning_rate": 3.6604049380750855e-05,
-      "loss": 49.9987,
-      "step": 102930
-    },
-    {
-      "epoch": 0.4158906256943968,
-      "grad_norm": 378.2391662597656,
-      "learning_rate": 3.660095741569871e-05,
-      "loss": 62.2791,
-      "step": 102940
-    },
-    {
-      "epoch": 0.41593102695976436,
-      "grad_norm": 539.7728271484375,
-      "learning_rate": 3.659786522447993e-05,
-      "loss": 57.2088,
-      "step": 102950
-    },
-    {
-      "epoch": 0.415971428225132,
-      "grad_norm": 925.6470336914062,
-      "learning_rate": 3.659477280715479e-05,
-      "loss": 54.7426,
-      "step": 102960
-    },
-    {
-      "epoch": 0.41601182949049964,
-      "grad_norm": 0.0,
-      "learning_rate": 3.659168016378359e-05,
-      "loss": 54.7942,
-      "step": 102970
-    },
-    {
-      "epoch": 0.4160522307558673,
-      "grad_norm": 1118.713623046875,
-      "learning_rate": 3.658858729442662e-05,
-      "loss": 45.2592,
-      "step": 102980
-    },
-    {
-      "epoch": 0.4160926320212349,
-      "grad_norm": 478.5465087890625,
-      "learning_rate": 3.658549419914417e-05,
-      "loss": 78.469,
-      "step": 102990
-    },
-    {
-      "epoch": 0.41613303328660256,
-      "grad_norm": 556.0774536132812,
-      "learning_rate": 3.6582400877996546e-05,
-      "loss": 56.5781,
-      "step": 103000
-    },
-    {
-      "epoch": 0.41617343455197015,
-      "grad_norm": 360.80340576171875,
-      "learning_rate": 3.6579307331044054e-05,
-      "loss": 52.6853,
-      "step": 103010
-    },
-    {
-      "epoch": 0.4162138358173378,
-      "grad_norm": 601.7576293945312,
-      "learning_rate": 3.657621355834701e-05,
-      "loss": 61.4476,
-      "step": 103020
-    },
-    {
-      "epoch": 0.4162542370827054,
-      "grad_norm": 592.1106567382812,
-      "learning_rate": 3.657311955996571e-05,
-      "loss": 51.4591,
-      "step": 103030
-    },
-    {
-      "epoch": 0.41629463834807306,
-      "grad_norm": 637.8342895507812,
-      "learning_rate": 3.657002533596049e-05,
-      "loss": 61.3845,
-      "step": 103040
-    },
-    {
-      "epoch": 0.4163350396134407,
-      "grad_norm": 371.7774963378906,
-      "learning_rate": 3.6566930886391674e-05,
-      "loss": 74.0396,
-      "step": 103050
-    },
-    {
-      "epoch": 0.41637544087880834,
-      "grad_norm": 375.3711242675781,
-      "learning_rate": 3.656383621131959e-05,
-      "loss": 56.7201,
-      "step": 103060
-    },
-    {
-      "epoch": 0.41641584214417593,
-      "grad_norm": 538.1626586914062,
-      "learning_rate": 3.656074131080457e-05,
-      "loss": 51.9634,
-      "step": 103070
-    },
-    {
-      "epoch": 0.41645624340954357,
-      "grad_norm": 1070.9267578125,
-      "learning_rate": 3.655764618490692e-05,
-      "loss": 92.4124,
-      "step": 103080
-    },
-    {
-      "epoch": 0.4164966446749112,
-      "grad_norm": 656.9863891601562,
-      "learning_rate": 3.655455083368703e-05,
-      "loss": 65.6765,
-      "step": 103090
-    },
-    {
-      "epoch": 0.41653704594027885,
-      "grad_norm": 841.900634765625,
-      "learning_rate": 3.655145525720522e-05,
-      "loss": 68.1387,
-      "step": 103100
-    },
-    {
-      "epoch": 0.4165774472056465,
-      "grad_norm": 409.5385437011719,
-      "learning_rate": 3.6548359455521836e-05,
-      "loss": 41.3072,
-      "step": 103110
-    },
-    {
-      "epoch": 0.41661784847101413,
-      "grad_norm": 598.8934936523438,
-      "learning_rate": 3.654526342869724e-05,
-      "loss": 64.1787,
-      "step": 103120
-    },
-    {
-      "epoch": 0.41665824973638177,
-      "grad_norm": 938.2333374023438,
-      "learning_rate": 3.654216717679179e-05,
-      "loss": 63.3869,
-      "step": 103130
-    },
-    {
-      "epoch": 0.41669865100174935,
-      "grad_norm": 0.0,
-      "learning_rate": 3.6539070699865853e-05,
-      "loss": 45.5103,
-      "step": 103140
-    },
-    {
-      "epoch": 0.416739052267117,
-      "grad_norm": 555.0664672851562,
-      "learning_rate": 3.653597399797979e-05,
-      "loss": 78.243,
-      "step": 103150
-    },
-    {
-      "epoch": 0.41677945353248463,
-      "grad_norm": 200.4738006591797,
-      "learning_rate": 3.6532877071193974e-05,
-      "loss": 45.6943,
-      "step": 103160
-    },
-    {
-      "epoch": 0.4168198547978523,
-      "grad_norm": 773.7369384765625,
-      "learning_rate": 3.652977991956878e-05,
-      "loss": 76.8474,
-      "step": 103170
-    },
-    {
-      "epoch": 0.4168602560632199,
-      "grad_norm": 1079.94775390625,
-      "learning_rate": 3.6526682543164595e-05,
-      "loss": 42.3459,
-      "step": 103180
-    },
-    {
-      "epoch": 0.41690065732858755,
-      "grad_norm": 919.0646362304688,
-      "learning_rate": 3.6523584942041794e-05,
-      "loss": 57.7047,
-      "step": 103190
-    },
-    {
-      "epoch": 0.41694105859395514,
-      "grad_norm": 891.546630859375,
-      "learning_rate": 3.6520487116260776e-05,
-      "loss": 73.0311,
-      "step": 103200
-    },
-    {
-      "epoch": 0.4169814598593228,
-      "grad_norm": 673.0984497070312,
-      "learning_rate": 3.6517389065881925e-05,
-      "loss": 52.7989,
-      "step": 103210
-    },
-    {
-      "epoch": 0.4170218611246904,
-      "grad_norm": 622.553955078125,
-      "learning_rate": 3.651429079096566e-05,
-      "loss": 61.8157,
-      "step": 103220
-    },
-    {
-      "epoch": 0.41706226239005806,
-      "grad_norm": 590.778564453125,
-      "learning_rate": 3.651119229157235e-05,
-      "loss": 42.1001,
-      "step": 103230
-    },
-    {
-      "epoch": 0.4171026636554257,
-      "grad_norm": 783.4470825195312,
-      "learning_rate": 3.650809356776242e-05,
-      "loss": 57.0175,
-      "step": 103240
-    },
-    {
-      "epoch": 0.41714306492079334,
-      "grad_norm": 1166.0804443359375,
-      "learning_rate": 3.6504994619596294e-05,
-      "loss": 72.4459,
-      "step": 103250
-    },
-    {
-      "epoch": 0.4171834661861609,
-      "grad_norm": 981.3162231445312,
-      "learning_rate": 3.650189544713437e-05,
-      "loss": 64.13,
-      "step": 103260
-    },
-    {
-      "epoch": 0.41722386745152856,
-      "grad_norm": 556.0223388671875,
-      "learning_rate": 3.649879605043707e-05,
-      "loss": 61.0497,
-      "step": 103270
-    },
-    {
-      "epoch": 0.4172642687168962,
-      "grad_norm": 1262.341064453125,
-      "learning_rate": 3.6495696429564823e-05,
-      "loss": 66.8657,
-      "step": 103280
-    },
-    {
-      "epoch": 0.41730466998226384,
-      "grad_norm": 322.91082763671875,
-      "learning_rate": 3.649259658457805e-05,
-      "loss": 62.4766,
-      "step": 103290
-    },
-    {
-      "epoch": 0.4173450712476315,
-      "grad_norm": 603.0479736328125,
-      "learning_rate": 3.6489496515537204e-05,
-      "loss": 65.0343,
-      "step": 103300
-    },
-    {
-      "epoch": 0.4173854725129991,
-      "grad_norm": 724.8381958007812,
-      "learning_rate": 3.648639622250269e-05,
-      "loss": 71.2941,
-      "step": 103310
-    },
-    {
-      "epoch": 0.41742587377836676,
-      "grad_norm": 709.285888671875,
-      "learning_rate": 3.648329570553498e-05,
-      "loss": 69.7722,
-      "step": 103320
-    },
-    {
-      "epoch": 0.41746627504373435,
-      "grad_norm": 1008.2940673828125,
-      "learning_rate": 3.648019496469451e-05,
-      "loss": 61.9354,
-      "step": 103330
-    },
-    {
-      "epoch": 0.417506676309102,
-      "grad_norm": 716.356689453125,
-      "learning_rate": 3.647709400004172e-05,
-      "loss": 102.8124,
-      "step": 103340
-    },
-    {
-      "epoch": 0.4175470775744696,
-      "grad_norm": 661.27880859375,
-      "learning_rate": 3.647399281163708e-05,
-      "loss": 54.8302,
-      "step": 103350
-    },
-    {
-      "epoch": 0.41758747883983727,
-      "grad_norm": 1043.45263671875,
-      "learning_rate": 3.647089139954104e-05,
-      "loss": 60.8454,
-      "step": 103360
-    },
-    {
-      "epoch": 0.4176278801052049,
-      "grad_norm": 862.45263671875,
-      "learning_rate": 3.646778976381407e-05,
-      "loss": 50.1279,
-      "step": 103370
-    },
-    {
-      "epoch": 0.41766828137057255,
-      "grad_norm": 924.0723266601562,
-      "learning_rate": 3.646468790451663e-05,
-      "loss": 41.2529,
-      "step": 103380
-    },
-    {
-      "epoch": 0.41770868263594013,
-      "grad_norm": 597.7651977539062,
-      "learning_rate": 3.64615858217092e-05,
-      "loss": 57.1287,
-      "step": 103390
-    },
-    {
-      "epoch": 0.41774908390130777,
-      "grad_norm": 379.25152587890625,
-      "learning_rate": 3.645848351545225e-05,
-      "loss": 62.9141,
-      "step": 103400
-    },
-    {
-      "epoch": 0.4177894851666754,
-      "grad_norm": 953.6447143554688,
-      "learning_rate": 3.645538098580627e-05,
-      "loss": 53.8604,
-      "step": 103410
-    },
-    {
-      "epoch": 0.41782988643204305,
-      "grad_norm": 575.6273193359375,
-      "learning_rate": 3.6452278232831735e-05,
-      "loss": 76.9551,
-      "step": 103420
-    },
-    {
-      "epoch": 0.4178702876974107,
-      "grad_norm": 1073.8568115234375,
-      "learning_rate": 3.644917525658914e-05,
-      "loss": 35.6009,
-      "step": 103430
-    },
-    {
-      "epoch": 0.41791068896277833,
-      "grad_norm": 1554.2855224609375,
-      "learning_rate": 3.644607205713898e-05,
-      "loss": 59.1008,
-      "step": 103440
-    },
-    {
-      "epoch": 0.41795109022814597,
-      "grad_norm": 619.3630981445312,
-      "learning_rate": 3.644296863454175e-05,
-      "loss": 48.0955,
-      "step": 103450
-    },
-    {
-      "epoch": 0.41799149149351356,
-      "grad_norm": 810.5908813476562,
-      "learning_rate": 3.643986498885796e-05,
-      "loss": 66.1217,
-      "step": 103460
-    },
-    {
-      "epoch": 0.4180318927588812,
-      "grad_norm": 2242.524658203125,
-      "learning_rate": 3.643676112014811e-05,
-      "loss": 79.2221,
-      "step": 103470
-    },
-    {
-      "epoch": 0.41807229402424884,
-      "grad_norm": 303.63385009765625,
-      "learning_rate": 3.643365702847272e-05,
-      "loss": 45.5608,
-      "step": 103480
-    },
-    {
-      "epoch": 0.4181126952896165,
-      "grad_norm": 1422.8160400390625,
-      "learning_rate": 3.643055271389229e-05,
-      "loss": 66.0965,
-      "step": 103490
-    },
-    {
-      "epoch": 0.4181530965549841,
-      "grad_norm": 741.1747436523438,
-      "learning_rate": 3.642744817646736e-05,
-      "loss": 45.8928,
-      "step": 103500
-    },
-    {
-      "epoch": 0.41819349782035176,
-      "grad_norm": 506.9162902832031,
-      "learning_rate": 3.642434341625844e-05,
-      "loss": 67.4611,
-      "step": 103510
-    },
-    {
-      "epoch": 0.41823389908571934,
-      "grad_norm": 592.9998168945312,
-      "learning_rate": 3.642123843332606e-05,
-      "loss": 63.5184,
-      "step": 103520
-    },
-    {
-      "epoch": 0.418274300351087,
-      "grad_norm": 303.4530334472656,
-      "learning_rate": 3.641813322773076e-05,
-      "loss": 50.6571,
-      "step": 103530
-    },
-    {
-      "epoch": 0.4183147016164546,
-      "grad_norm": 1942.93115234375,
-      "learning_rate": 3.641502779953307e-05,
-      "loss": 46.2943,
-      "step": 103540
-    },
-    {
-      "epoch": 0.41835510288182226,
-      "grad_norm": 748.298095703125,
-      "learning_rate": 3.6411922148793544e-05,
-      "loss": 54.8428,
-      "step": 103550
-    },
-    {
-      "epoch": 0.4183955041471899,
-      "grad_norm": 902.7544555664062,
-      "learning_rate": 3.640881627557271e-05,
-      "loss": 54.3055,
-      "step": 103560
-    },
-    {
-      "epoch": 0.41843590541255754,
-      "grad_norm": 1064.920654296875,
-      "learning_rate": 3.640571017993113e-05,
-      "loss": 58.3005,
-      "step": 103570
-    },
-    {
-      "epoch": 0.4184763066779251,
-      "grad_norm": 1428.6287841796875,
-      "learning_rate": 3.6402603861929374e-05,
-      "loss": 64.2658,
-      "step": 103580
-    },
-    {
-      "epoch": 0.41851670794329277,
-      "grad_norm": 503.11871337890625,
-      "learning_rate": 3.639949732162797e-05,
-      "loss": 61.6714,
-      "step": 103590
-    },
-    {
-      "epoch": 0.4185571092086604,
-      "grad_norm": 1219.691162109375,
-      "learning_rate": 3.639639055908751e-05,
-      "loss": 66.9562,
-      "step": 103600
-    },
-    {
-      "epoch": 0.41859751047402805,
-      "grad_norm": 223.49790954589844,
-      "learning_rate": 3.639328357436853e-05,
-      "loss": 55.7522,
-      "step": 103610
-    },
-    {
-      "epoch": 0.4186379117393957,
-      "grad_norm": 514.73681640625,
-      "learning_rate": 3.639017636753163e-05,
-      "loss": 71.0878,
-      "step": 103620
-    },
-    {
-      "epoch": 0.4186783130047633,
-      "grad_norm": 383.3266906738281,
-      "learning_rate": 3.638706893863739e-05,
-      "loss": 71.9139,
-      "step": 103630
-    },
-    {
-      "epoch": 0.41871871427013097,
-      "grad_norm": 539.1884765625,
-      "learning_rate": 3.638396128774636e-05,
-      "loss": 76.8819,
-      "step": 103640
-    },
-    {
-      "epoch": 0.41875911553549855,
-      "grad_norm": 1771.3558349609375,
-      "learning_rate": 3.6380853414919144e-05,
-      "loss": 65.5004,
-      "step": 103650
-    },
-    {
-      "epoch": 0.4187995168008662,
-      "grad_norm": 257.1325988769531,
-      "learning_rate": 3.6377745320216346e-05,
-      "loss": 84.9693,
-      "step": 103660
-    },
-    {
-      "epoch": 0.41883991806623383,
-      "grad_norm": 1277.81787109375,
-      "learning_rate": 3.6374637003698536e-05,
-      "loss": 59.5876,
-      "step": 103670
-    },
-    {
-      "epoch": 0.41888031933160147,
-      "grad_norm": 985.5901489257812,
-      "learning_rate": 3.637152846542633e-05,
-      "loss": 47.3202,
-      "step": 103680
-    },
-    {
-      "epoch": 0.4189207205969691,
-      "grad_norm": 943.6722412109375,
-      "learning_rate": 3.636841970546031e-05,
-      "loss": 52.7835,
-      "step": 103690
-    },
-    {
-      "epoch": 0.41896112186233675,
-      "grad_norm": 512.9838256835938,
-      "learning_rate": 3.63653107238611e-05,
-      "loss": 59.9564,
-      "step": 103700
-    },
-    {
-      "epoch": 0.41900152312770433,
-      "grad_norm": 886.2669067382812,
-      "learning_rate": 3.636220152068931e-05,
-      "loss": 69.5346,
-      "step": 103710
-    },
-    {
-      "epoch": 0.419041924393072,
-      "grad_norm": 866.3492431640625,
-      "learning_rate": 3.635909209600555e-05,
-      "loss": 68.2752,
-      "step": 103720
-    },
-    {
-      "epoch": 0.4190823256584396,
-      "grad_norm": 536.70947265625,
-      "learning_rate": 3.635598244987043e-05,
-      "loss": 56.4688,
-      "step": 103730
-    },
-    {
-      "epoch": 0.41912272692380725,
-      "grad_norm": 421.8968200683594,
-      "learning_rate": 3.6352872582344596e-05,
-      "loss": 63.9303,
-      "step": 103740
-    },
-    {
-      "epoch": 0.4191631281891749,
-      "grad_norm": 437.35546875,
-      "learning_rate": 3.634976249348867e-05,
-      "loss": 29.7742,
-      "step": 103750
-    },
-    {
-      "epoch": 0.41920352945454253,
-      "grad_norm": 1521.4986572265625,
-      "learning_rate": 3.634665218336328e-05,
-      "loss": 52.6372,
-      "step": 103760
-    },
-    {
-      "epoch": 0.4192439307199102,
-      "grad_norm": 814.3853759765625,
-      "learning_rate": 3.6343541652029064e-05,
-      "loss": 68.0177,
-      "step": 103770
-    },
-    {
-      "epoch": 0.41928433198527776,
-      "grad_norm": 495.0726318359375,
-      "learning_rate": 3.6340430899546656e-05,
-      "loss": 49.5452,
-      "step": 103780
-    },
-    {
-      "epoch": 0.4193247332506454,
-      "grad_norm": 1200.4290771484375,
-      "learning_rate": 3.633731992597672e-05,
-      "loss": 81.554,
-      "step": 103790
-    },
-    {
-      "epoch": 0.41936513451601304,
-      "grad_norm": 1105.92822265625,
-      "learning_rate": 3.633420873137988e-05,
-      "loss": 59.0107,
-      "step": 103800
-    },
-    {
-      "epoch": 0.4194055357813807,
-      "grad_norm": 965.9473266601562,
-      "learning_rate": 3.633109731581682e-05,
-      "loss": 52.3541,
-      "step": 103810
-    },
-    {
-      "epoch": 0.4194459370467483,
-      "grad_norm": 784.390625,
-      "learning_rate": 3.632798567934817e-05,
-      "loss": 42.5555,
-      "step": 103820
-    },
-    {
-      "epoch": 0.41948633831211596,
-      "grad_norm": 1471.4720458984375,
-      "learning_rate": 3.632487382203462e-05,
-      "loss": 61.9829,
-      "step": 103830
-    },
-    {
-      "epoch": 0.41952673957748354,
-      "grad_norm": 826.1594848632812,
-      "learning_rate": 3.632176174393682e-05,
-      "loss": 53.5257,
-      "step": 103840
-    },
-    {
-      "epoch": 0.4195671408428512,
-      "grad_norm": 526.9407958984375,
-      "learning_rate": 3.631864944511545e-05,
-      "loss": 58.7132,
-      "step": 103850
-    },
-    {
-      "epoch": 0.4196075421082188,
-      "grad_norm": 845.7454833984375,
-      "learning_rate": 3.6315536925631174e-05,
-      "loss": 44.2861,
-      "step": 103860
-    },
-    {
-      "epoch": 0.41964794337358646,
-      "grad_norm": 869.853515625,
-      "learning_rate": 3.631242418554469e-05,
-      "loss": 78.6071,
-      "step": 103870
-    },
-    {
-      "epoch": 0.4196883446389541,
-      "grad_norm": 858.0875244140625,
-      "learning_rate": 3.630931122491666e-05,
-      "loss": 53.3545,
-      "step": 103880
-    },
-    {
-      "epoch": 0.41972874590432174,
-      "grad_norm": 472.39337158203125,
-      "learning_rate": 3.6306198043807795e-05,
-      "loss": 44.7906,
-      "step": 103890
-    },
-    {
-      "epoch": 0.4197691471696893,
-      "grad_norm": 1292.262939453125,
-      "learning_rate": 3.630308464227877e-05,
-      "loss": 46.8033,
-      "step": 103900
-    },
-    {
-      "epoch": 0.41980954843505697,
-      "grad_norm": 667.8062133789062,
-      "learning_rate": 3.62999710203903e-05,
-      "loss": 75.6769,
-      "step": 103910
-    },
-    {
-      "epoch": 0.4198499497004246,
-      "grad_norm": 862.8394775390625,
-      "learning_rate": 3.629685717820307e-05,
-      "loss": 56.4103,
-      "step": 103920
-    },
-    {
-      "epoch": 0.41989035096579225,
-      "grad_norm": 348.3255920410156,
-      "learning_rate": 3.629374311577779e-05,
-      "loss": 44.0832,
-      "step": 103930
-    },
-    {
-      "epoch": 0.4199307522311599,
-      "grad_norm": 1450.09375,
-      "learning_rate": 3.629062883317519e-05,
-      "loss": 68.2507,
-      "step": 103940
-    },
-    {
-      "epoch": 0.4199711534965275,
-      "grad_norm": 537.00537109375,
-      "learning_rate": 3.628751433045596e-05,
-      "loss": 69.9783,
-      "step": 103950
-    },
-    {
-      "epoch": 0.42001155476189517,
-      "grad_norm": 581.5643310546875,
-      "learning_rate": 3.628439960768082e-05,
-      "loss": 63.0526,
-      "step": 103960
-    },
-    {
-      "epoch": 0.42005195602726275,
-      "grad_norm": 664.759521484375,
-      "learning_rate": 3.62812846649105e-05,
-      "loss": 56.3519,
-      "step": 103970
-    },
-    {
-      "epoch": 0.4200923572926304,
-      "grad_norm": 635.3744506835938,
-      "learning_rate": 3.6278169502205736e-05,
-      "loss": 61.1876,
-      "step": 103980
-    },
-    {
-      "epoch": 0.42013275855799803,
-      "grad_norm": 530.5980224609375,
-      "learning_rate": 3.627505411962724e-05,
-      "loss": 66.6494,
-      "step": 103990
-    },
-    {
-      "epoch": 0.42017315982336567,
-      "grad_norm": 758.6454467773438,
-      "learning_rate": 3.627193851723577e-05,
-      "loss": 58.0647,
-      "step": 104000
-    },
-    {
-      "epoch": 0.4202135610887333,
-      "grad_norm": 2163.636474609375,
-      "learning_rate": 3.6268822695092056e-05,
-      "loss": 46.5954,
-      "step": 104010
-    },
-    {
-      "epoch": 0.42025396235410095,
-      "grad_norm": 1194.97119140625,
-      "learning_rate": 3.626570665325684e-05,
-      "loss": 61.3154,
-      "step": 104020
-    },
-    {
-      "epoch": 0.42029436361946854,
-      "grad_norm": 1891.653076171875,
-      "learning_rate": 3.626259039179086e-05,
-      "loss": 59.1667,
-      "step": 104030
-    },
-    {
-      "epoch": 0.4203347648848362,
-      "grad_norm": 609.3426513671875,
-      "learning_rate": 3.6259473910754904e-05,
-      "loss": 41.8143,
-      "step": 104040
-    },
-    {
-      "epoch": 0.4203751661502038,
-      "grad_norm": 1182.2841796875,
-      "learning_rate": 3.625635721020969e-05,
-      "loss": 64.0979,
-      "step": 104050
-    },
-    {
-      "epoch": 0.42041556741557146,
-      "grad_norm": 960.9483642578125,
-      "learning_rate": 3.6253240290216e-05,
-      "loss": 49.4696,
-      "step": 104060
-    },
-    {
-      "epoch": 0.4204559686809391,
-      "grad_norm": 292.2507629394531,
-      "learning_rate": 3.62501231508346e-05,
-      "loss": 71.0735,
-      "step": 104070
-    },
-    {
-      "epoch": 0.42049636994630674,
-      "grad_norm": 782.7526245117188,
-      "learning_rate": 3.624700579212626e-05,
-      "loss": 55.9743,
-      "step": 104080
-    },
-    {
-      "epoch": 0.4205367712116744,
-      "grad_norm": 1657.7484130859375,
-      "learning_rate": 3.624388821415175e-05,
-      "loss": 35.3792,
-      "step": 104090
-    },
-    {
-      "epoch": 0.42057717247704196,
-      "grad_norm": 462.6631774902344,
-      "learning_rate": 3.624077041697185e-05,
-      "loss": 53.0793,
-      "step": 104100
-    },
-    {
-      "epoch": 0.4206175737424096,
-      "grad_norm": 606.4874267578125,
-      "learning_rate": 3.6237652400647345e-05,
-      "loss": 45.0725,
-      "step": 104110
-    },
-    {
-      "epoch": 0.42065797500777724,
-      "grad_norm": 842.7166137695312,
-      "learning_rate": 3.623453416523902e-05,
-      "loss": 67.8834,
-      "step": 104120
-    },
-    {
-      "epoch": 0.4206983762731449,
-      "grad_norm": 249.5082550048828,
-      "learning_rate": 3.623141571080766e-05,
-      "loss": 37.8796,
-      "step": 104130
-    },
-    {
-      "epoch": 0.4207387775385125,
-      "grad_norm": 343.2146911621094,
-      "learning_rate": 3.6228297037414074e-05,
-      "loss": 57.456,
-      "step": 104140
-    },
-    {
-      "epoch": 0.42077917880388016,
-      "grad_norm": 3797.084228515625,
-      "learning_rate": 3.622517814511906e-05,
-      "loss": 60.7318,
-      "step": 104150
-    },
-    {
-      "epoch": 0.42081958006924775,
-      "grad_norm": 617.2786254882812,
-      "learning_rate": 3.622205903398342e-05,
-      "loss": 59.6277,
-      "step": 104160
-    },
-    {
-      "epoch": 0.4208599813346154,
-      "grad_norm": 836.6347045898438,
-      "learning_rate": 3.6218939704067955e-05,
-      "loss": 69.2782,
-      "step": 104170
-    },
-    {
-      "epoch": 0.420900382599983,
-      "grad_norm": 988.6712646484375,
-      "learning_rate": 3.621582015543348e-05,
-      "loss": 71.7698,
-      "step": 104180
-    },
-    {
-      "epoch": 0.42094078386535067,
-      "grad_norm": 377.2164001464844,
-      "learning_rate": 3.621270038814083e-05,
-      "loss": 58.6837,
-      "step": 104190
-    },
-    {
-      "epoch": 0.4209811851307183,
-      "grad_norm": 498.7535095214844,
-      "learning_rate": 3.6209580402250815e-05,
-      "loss": 38.0926,
-      "step": 104200
-    },
-    {
-      "epoch": 0.42102158639608595,
-      "grad_norm": 403.9892883300781,
-      "learning_rate": 3.620646019782425e-05,
-      "loss": 51.1487,
-      "step": 104210
-    },
-    {
-      "epoch": 0.42106198766145353,
-      "grad_norm": 1094.6707763671875,
-      "learning_rate": 3.6203339774921976e-05,
-      "loss": 49.0772,
-      "step": 104220
-    },
-    {
-      "epoch": 0.42110238892682117,
-      "grad_norm": 1082.783935546875,
-      "learning_rate": 3.6200219133604816e-05,
-      "loss": 71.491,
-      "step": 104230
-    },
-    {
-      "epoch": 0.4211427901921888,
-      "grad_norm": 443.6391296386719,
-      "learning_rate": 3.6197098273933634e-05,
-      "loss": 57.1322,
-      "step": 104240
-    },
-    {
-      "epoch": 0.42118319145755645,
-      "grad_norm": 431.85430908203125,
-      "learning_rate": 3.619397719596924e-05,
-      "loss": 51.8768,
-      "step": 104250
-    },
-    {
-      "epoch": 0.4212235927229241,
-      "grad_norm": 2122.891845703125,
-      "learning_rate": 3.619085589977251e-05,
-      "loss": 62.3512,
-      "step": 104260
-    },
-    {
-      "epoch": 0.42126399398829173,
-      "grad_norm": 1347.9971923828125,
-      "learning_rate": 3.618773438540428e-05,
-      "loss": 68.3303,
-      "step": 104270
-    },
-    {
-      "epoch": 0.42130439525365937,
-      "grad_norm": 583.928955078125,
-      "learning_rate": 3.618461265292541e-05,
-      "loss": 52.5091,
-      "step": 104280
-    },
-    {
-      "epoch": 0.42134479651902695,
-      "grad_norm": 523.10546875,
-      "learning_rate": 3.618149070239676e-05,
-      "loss": 55.4588,
-      "step": 104290
-    },
-    {
-      "epoch": 0.4213851977843946,
-      "grad_norm": 891.4425659179688,
-      "learning_rate": 3.617836853387918e-05,
-      "loss": 56.3773,
-      "step": 104300
-    },
-    {
-      "epoch": 0.42142559904976223,
-      "grad_norm": 534.82470703125,
-      "learning_rate": 3.6175246147433563e-05,
-      "loss": 64.5376,
-      "step": 104310
-    },
-    {
-      "epoch": 0.4214660003151299,
-      "grad_norm": 529.0205688476562,
-      "learning_rate": 3.617212354312076e-05,
-      "loss": 50.4187,
-      "step": 104320
-    },
-    {
-      "epoch": 0.4215064015804975,
-      "grad_norm": 303.3257141113281,
-      "learning_rate": 3.616900072100166e-05,
-      "loss": 42.9221,
-      "step": 104330
-    },
-    {
-      "epoch": 0.42154680284586515,
-      "grad_norm": 1264.615478515625,
-      "learning_rate": 3.6165877681137136e-05,
-      "loss": 72.0312,
-      "step": 104340
-    },
-    {
-      "epoch": 0.42158720411123274,
-      "grad_norm": 1191.180908203125,
-      "learning_rate": 3.6162754423588085e-05,
-      "loss": 57.8124,
-      "step": 104350
-    },
-    {
-      "epoch": 0.4216276053766004,
-      "grad_norm": 366.5780944824219,
-      "learning_rate": 3.61596309484154e-05,
-      "loss": 56.5291,
-      "step": 104360
-    },
-    {
-      "epoch": 0.421668006641968,
-      "grad_norm": 512.8042602539062,
-      "learning_rate": 3.615650725567995e-05,
-      "loss": 57.7641,
-      "step": 104370
-    },
-    {
-      "epoch": 0.42170840790733566,
-      "grad_norm": 782.4150390625,
-      "learning_rate": 3.615338334544265e-05,
-      "loss": 41.4072,
-      "step": 104380
-    },
-    {
-      "epoch": 0.4217488091727033,
-      "grad_norm": 1751.27587890625,
-      "learning_rate": 3.615025921776439e-05,
-      "loss": 49.6701,
-      "step": 104390
-    },
-    {
-      "epoch": 0.42178921043807094,
-      "grad_norm": 1079.94189453125,
-      "learning_rate": 3.614713487270611e-05,
-      "loss": 46.7276,
-      "step": 104400
-    },
-    {
-      "epoch": 0.4218296117034386,
-      "grad_norm": 1060.4190673828125,
-      "learning_rate": 3.614401031032867e-05,
-      "loss": 106.2965,
-      "step": 104410
-    },
-    {
-      "epoch": 0.42187001296880616,
-      "grad_norm": 749.7593994140625,
-      "learning_rate": 3.614088553069303e-05,
-      "loss": 44.3926,
-      "step": 104420
-    },
-    {
-      "epoch": 0.4219104142341738,
-      "grad_norm": 628.2740478515625,
-      "learning_rate": 3.6137760533860074e-05,
-      "loss": 88.6748,
-      "step": 104430
-    },
-    {
-      "epoch": 0.42195081549954144,
-      "grad_norm": 461.367431640625,
-      "learning_rate": 3.613463531989076e-05,
-      "loss": 49.6176,
-      "step": 104440
-    },
-    {
-      "epoch": 0.4219912167649091,
-      "grad_norm": 873.3343505859375,
-      "learning_rate": 3.613150988884599e-05,
-      "loss": 61.126,
-      "step": 104450
-    },
-    {
-      "epoch": 0.4220316180302767,
-      "grad_norm": 393.72869873046875,
-      "learning_rate": 3.612838424078671e-05,
-      "loss": 38.0402,
-      "step": 104460
-    },
-    {
-      "epoch": 0.42207201929564436,
-      "grad_norm": 850.967041015625,
-      "learning_rate": 3.612525837577384e-05,
-      "loss": 66.1431,
-      "step": 104470
-    },
-    {
-      "epoch": 0.42211242056101195,
-      "grad_norm": 1090.09814453125,
-      "learning_rate": 3.6122132293868335e-05,
-      "loss": 82.8322,
-      "step": 104480
-    },
-    {
-      "epoch": 0.4221528218263796,
-      "grad_norm": 956.2141723632812,
-      "learning_rate": 3.611900599513114e-05,
-      "loss": 57.5093,
-      "step": 104490
-    },
-    {
-      "epoch": 0.4221932230917472,
-      "grad_norm": 407.1945495605469,
-      "learning_rate": 3.611587947962319e-05,
-      "loss": 50.0537,
-      "step": 104500
-    },
-    {
-      "epoch": 0.42223362435711487,
-      "grad_norm": 486.4323425292969,
-      "learning_rate": 3.6112752747405447e-05,
-      "loss": 52.2107,
-      "step": 104510
-    },
-    {
-      "epoch": 0.4222740256224825,
-      "grad_norm": 864.6810913085938,
-      "learning_rate": 3.6109625798538873e-05,
-      "loss": 50.7976,
-      "step": 104520
-    },
-    {
-      "epoch": 0.42231442688785015,
-      "grad_norm": 534.1397705078125,
-      "learning_rate": 3.6106498633084424e-05,
-      "loss": 48.8516,
-      "step": 104530
-    },
-    {
-      "epoch": 0.42235482815321773,
-      "grad_norm": 636.9712524414062,
-      "learning_rate": 3.610337125110307e-05,
-      "loss": 52.8733,
-      "step": 104540
-    },
-    {
-      "epoch": 0.42239522941858537,
-      "grad_norm": 1049.5821533203125,
-      "learning_rate": 3.610024365265577e-05,
-      "loss": 56.6238,
-      "step": 104550
-    },
-    {
-      "epoch": 0.422435630683953,
-      "grad_norm": 374.7794189453125,
-      "learning_rate": 3.6097115837803505e-05,
-      "loss": 73.0223,
-      "step": 104560
-    },
-    {
-      "epoch": 0.42247603194932065,
-      "grad_norm": 321.7105712890625,
-      "learning_rate": 3.609398780660726e-05,
-      "loss": 51.7502,
-      "step": 104570
-    },
-    {
-      "epoch": 0.4225164332146883,
-      "grad_norm": 833.6190185546875,
-      "learning_rate": 3.6090859559128e-05,
-      "loss": 44.2709,
-      "step": 104580
-    },
-    {
-      "epoch": 0.42255683448005593,
-      "grad_norm": 465.5598449707031,
-      "learning_rate": 3.6087731095426733e-05,
-      "loss": 54.5164,
-      "step": 104590
-    },
-    {
-      "epoch": 0.42259723574542357,
-      "grad_norm": 460.9774169921875,
-      "learning_rate": 3.608460241556443e-05,
-      "loss": 59.3148,
-      "step": 104600
-    },
-    {
-      "epoch": 0.42263763701079116,
-      "grad_norm": 390.5638427734375,
-      "learning_rate": 3.6081473519602105e-05,
-      "loss": 70.8456,
-      "step": 104610
-    },
-    {
-      "epoch": 0.4226780382761588,
-      "grad_norm": 1563.58642578125,
-      "learning_rate": 3.607834440760074e-05,
-      "loss": 57.9671,
-      "step": 104620
-    },
-    {
-      "epoch": 0.42271843954152644,
-      "grad_norm": 1309.08935546875,
-      "learning_rate": 3.607521507962136e-05,
-      "loss": 63.8607,
-      "step": 104630
-    },
-    {
-      "epoch": 0.4227588408068941,
-      "grad_norm": 2113.58447265625,
-      "learning_rate": 3.6072085535724956e-05,
-      "loss": 57.3463,
-      "step": 104640
-    },
-    {
-      "epoch": 0.4227992420722617,
-      "grad_norm": 1004.9712524414062,
-      "learning_rate": 3.606895577597255e-05,
-      "loss": 67.326,
-      "step": 104650
-    },
-    {
-      "epoch": 0.42283964333762936,
-      "grad_norm": 0.0,
-      "learning_rate": 3.606582580042513e-05,
-      "loss": 33.3827,
-      "step": 104660
-    },
-    {
-      "epoch": 0.42288004460299694,
-      "grad_norm": 594.0809936523438,
-      "learning_rate": 3.606269560914376e-05,
-      "loss": 48.728,
-      "step": 104670
-    },
-    {
-      "epoch": 0.4229204458683646,
-      "grad_norm": 891.5621948242188,
-      "learning_rate": 3.6059565202189435e-05,
-      "loss": 63.9774,
-      "step": 104680
-    },
-    {
-      "epoch": 0.4229608471337322,
-      "grad_norm": 622.961669921875,
-      "learning_rate": 3.605643457962319e-05,
-      "loss": 53.7502,
-      "step": 104690
-    },
-    {
-      "epoch": 0.42300124839909986,
-      "grad_norm": 899.34716796875,
-      "learning_rate": 3.605330374150607e-05,
-      "loss": 62.4448,
-      "step": 104700
-    },
-    {
-      "epoch": 0.4230416496644675,
-      "grad_norm": 698.8828735351562,
-      "learning_rate": 3.60501726878991e-05,
-      "loss": 72.48,
-      "step": 104710
-    },
-    {
-      "epoch": 0.42308205092983514,
-      "grad_norm": 591.2288208007812,
-      "learning_rate": 3.604704141886332e-05,
-      "loss": 50.3654,
-      "step": 104720
-    },
-    {
-      "epoch": 0.4231224521952028,
-      "grad_norm": 1411.56640625,
-      "learning_rate": 3.6043909934459785e-05,
-      "loss": 68.7279,
-      "step": 104730
-    },
-    {
-      "epoch": 0.42316285346057037,
-      "grad_norm": 3371.13134765625,
-      "learning_rate": 3.604077823474954e-05,
-      "loss": 72.1287,
-      "step": 104740
-    },
-    {
-      "epoch": 0.423203254725938,
-      "grad_norm": 0.0,
-      "learning_rate": 3.603764631979363e-05,
-      "loss": 61.998,
-      "step": 104750
-    },
-    {
-      "epoch": 0.42324365599130565,
-      "grad_norm": 704.408935546875,
-      "learning_rate": 3.603451418965313e-05,
-      "loss": 27.3794,
-      "step": 104760
-    },
-    {
-      "epoch": 0.4232840572566733,
-      "grad_norm": 973.9273681640625,
-      "learning_rate": 3.60313818443891e-05,
-      "loss": 87.7188,
-      "step": 104770
-    },
-    {
-      "epoch": 0.4233244585220409,
-      "grad_norm": 908.014404296875,
-      "learning_rate": 3.602824928406259e-05,
-      "loss": 51.2298,
-      "step": 104780
-    },
-    {
-      "epoch": 0.42336485978740857,
-      "grad_norm": 693.9036254882812,
-      "learning_rate": 3.602511650873469e-05,
-      "loss": 97.5675,
-      "step": 104790
-    },
-    {
-      "epoch": 0.42340526105277615,
-      "grad_norm": 1070.6943359375,
-      "learning_rate": 3.602198351846647e-05,
-      "loss": 70.0561,
-      "step": 104800
-    },
-    {
-      "epoch": 0.4234456623181438,
-      "grad_norm": 970.8385620117188,
-      "learning_rate": 3.6018850313319e-05,
-      "loss": 71.5289,
-      "step": 104810
-    },
-    {
-      "epoch": 0.42348606358351143,
-      "grad_norm": 1016.22021484375,
-      "learning_rate": 3.6015716893353376e-05,
-      "loss": 53.878,
-      "step": 104820
-    },
-    {
-      "epoch": 0.42352646484887907,
-      "grad_norm": 532.7132568359375,
-      "learning_rate": 3.601258325863067e-05,
-      "loss": 62.0343,
-      "step": 104830
-    },
-    {
-      "epoch": 0.4235668661142467,
-      "grad_norm": 94.07963562011719,
-      "learning_rate": 3.600944940921199e-05,
-      "loss": 67.1461,
-      "step": 104840
-    },
-    {
-      "epoch": 0.42360726737961435,
-      "grad_norm": 921.68017578125,
-      "learning_rate": 3.6006315345158434e-05,
-      "loss": 65.9748,
-      "step": 104850
-    },
-    {
-      "epoch": 0.42364766864498193,
-      "grad_norm": 1337.1324462890625,
-      "learning_rate": 3.600318106653108e-05,
-      "loss": 60.1088,
-      "step": 104860
-    },
-    {
-      "epoch": 0.4236880699103496,
-      "grad_norm": 522.1008911132812,
-      "learning_rate": 3.600004657339105e-05,
-      "loss": 55.0299,
-      "step": 104870
-    },
-    {
-      "epoch": 0.4237284711757172,
-      "grad_norm": 452.1679992675781,
-      "learning_rate": 3.5996911865799454e-05,
-      "loss": 42.4841,
-      "step": 104880
-    },
-    {
-      "epoch": 0.42376887244108485,
-      "grad_norm": 712.0603637695312,
-      "learning_rate": 3.59937769438174e-05,
-      "loss": 57.7705,
-      "step": 104890
-    },
-    {
-      "epoch": 0.4238092737064525,
-      "grad_norm": 486.09613037109375,
-      "learning_rate": 3.5990641807506e-05,
-      "loss": 39.141,
-      "step": 104900
-    },
-    {
-      "epoch": 0.42384967497182013,
-      "grad_norm": 1278.7674560546875,
-      "learning_rate": 3.598750645692638e-05,
-      "loss": 59.3348,
-      "step": 104910
-    },
-    {
-      "epoch": 0.4238900762371878,
-      "grad_norm": 1631.982666015625,
-      "learning_rate": 3.5984370892139666e-05,
-      "loss": 68.6455,
-      "step": 104920
-    },
-    {
-      "epoch": 0.42393047750255536,
-      "grad_norm": 466.52874755859375,
-      "learning_rate": 3.598123511320699e-05,
-      "loss": 37.1398,
-      "step": 104930
-    },
-    {
-      "epoch": 0.423970878767923,
-      "grad_norm": 1008.3381958007812,
-      "learning_rate": 3.597809912018947e-05,
-      "loss": 65.8413,
-      "step": 104940
-    },
-    {
-      "epoch": 0.42401128003329064,
-      "grad_norm": 738.3510131835938,
-      "learning_rate": 3.597496291314827e-05,
-      "loss": 67.1094,
-      "step": 104950
-    },
-    {
-      "epoch": 0.4240516812986583,
-      "grad_norm": 337.72796630859375,
-      "learning_rate": 3.5971826492144504e-05,
-      "loss": 63.2226,
-      "step": 104960
-    },
-    {
-      "epoch": 0.4240920825640259,
-      "grad_norm": 295.6142272949219,
-      "learning_rate": 3.5968689857239345e-05,
-      "loss": 65.1638,
-      "step": 104970
-    },
-    {
-      "epoch": 0.42413248382939356,
-      "grad_norm": 797.707275390625,
-      "learning_rate": 3.596555300849392e-05,
-      "loss": 77.6298,
-      "step": 104980
-    },
-    {
-      "epoch": 0.42417288509476114,
-      "grad_norm": 476.190185546875,
-      "learning_rate": 3.5962415945969405e-05,
-      "loss": 69.4243,
-      "step": 104990
-    },
-    {
-      "epoch": 0.4242132863601288,
-      "grad_norm": 412.16644287109375,
-      "learning_rate": 3.5959278669726935e-05,
-      "loss": 40.4091,
-      "step": 105000
-    },
-    {
-      "epoch": 0.4242536876254964,
-      "grad_norm": 565.7656860351562,
-      "learning_rate": 3.595614117982769e-05,
-      "loss": 68.7301,
-      "step": 105010
-    },
-    {
-      "epoch": 0.42429408889086406,
-      "grad_norm": 733.7787475585938,
-      "learning_rate": 3.5953003476332835e-05,
-      "loss": 54.2594,
-      "step": 105020
-    },
-    {
-      "epoch": 0.4243344901562317,
-      "grad_norm": 842.4686279296875,
-      "learning_rate": 3.5949865559303536e-05,
-      "loss": 65.4532,
-      "step": 105030
-    },
-    {
-      "epoch": 0.42437489142159934,
-      "grad_norm": 1525.9725341796875,
-      "learning_rate": 3.594672742880097e-05,
-      "loss": 51.1515,
-      "step": 105040
-    },
-    {
-      "epoch": 0.424415292686967,
-      "grad_norm": 1515.8177490234375,
-      "learning_rate": 3.594358908488632e-05,
-      "loss": 42.8702,
-      "step": 105050
-    },
-    {
-      "epoch": 0.42445569395233457,
-      "grad_norm": 922.44091796875,
-      "learning_rate": 3.594045052762076e-05,
-      "loss": 57.4078,
-      "step": 105060
-    },
-    {
-      "epoch": 0.4244960952177022,
-      "grad_norm": 1049.3338623046875,
-      "learning_rate": 3.5937311757065494e-05,
-      "loss": 55.3803,
-      "step": 105070
-    },
-    {
-      "epoch": 0.42453649648306985,
-      "grad_norm": 563.2704467773438,
-      "learning_rate": 3.5934172773281696e-05,
-      "loss": 28.5142,
-      "step": 105080
-    },
-    {
-      "epoch": 0.4245768977484375,
-      "grad_norm": 655.0813598632812,
-      "learning_rate": 3.593103357633058e-05,
-      "loss": 41.6843,
-      "step": 105090
-    },
-    {
-      "epoch": 0.4246172990138051,
-      "grad_norm": 684.439453125,
-      "learning_rate": 3.592789416627332e-05,
-      "loss": 50.5486,
-      "step": 105100
-    },
-    {
-      "epoch": 0.42465770027917277,
-      "grad_norm": 765.9920654296875,
-      "learning_rate": 3.592475454317115e-05,
-      "loss": 68.0954,
-      "step": 105110
-    },
-    {
-      "epoch": 0.42469810154454035,
-      "grad_norm": 1379.385009765625,
-      "learning_rate": 3.592161470708526e-05,
-      "loss": 65.1052,
-      "step": 105120
-    },
-    {
-      "epoch": 0.424738502809908,
-      "grad_norm": 296.92437744140625,
-      "learning_rate": 3.591847465807687e-05,
-      "loss": 46.8323,
-      "step": 105130
-    },
-    {
-      "epoch": 0.42477890407527563,
-      "grad_norm": 554.3091430664062,
-      "learning_rate": 3.59153343962072e-05,
-      "loss": 54.8523,
-      "step": 105140
-    },
-    {
-      "epoch": 0.42481930534064327,
-      "grad_norm": 529.84521484375,
-      "learning_rate": 3.5912193921537476e-05,
-      "loss": 44.9619,
-      "step": 105150
-    },
-    {
-      "epoch": 0.4248597066060109,
-      "grad_norm": 571.56591796875,
-      "learning_rate": 3.5909053234128895e-05,
-      "loss": 44.5782,
-      "step": 105160
-    },
-    {
-      "epoch": 0.42490010787137855,
-      "grad_norm": 598.4688110351562,
-      "learning_rate": 3.590591233404271e-05,
-      "loss": 75.1092,
-      "step": 105170
-    },
-    {
-      "epoch": 0.42494050913674614,
-      "grad_norm": 1925.43798828125,
-      "learning_rate": 3.590277122134015e-05,
-      "loss": 52.7716,
-      "step": 105180
-    },
-    {
-      "epoch": 0.4249809104021138,
-      "grad_norm": 435.9007263183594,
-      "learning_rate": 3.5899629896082454e-05,
-      "loss": 35.1308,
-      "step": 105190
-    },
-    {
-      "epoch": 0.4250213116674814,
-      "grad_norm": 492.9730224609375,
-      "learning_rate": 3.5896488358330856e-05,
-      "loss": 47.9135,
-      "step": 105200
-    },
-    {
-      "epoch": 0.42506171293284906,
-      "grad_norm": 1603.56591796875,
-      "learning_rate": 3.5893346608146607e-05,
-      "loss": 62.7425,
-      "step": 105210
-    },
-    {
-      "epoch": 0.4251021141982167,
-      "grad_norm": 715.2278442382812,
-      "learning_rate": 3.5890204645590964e-05,
-      "loss": 127.7718,
-      "step": 105220
-    },
-    {
-      "epoch": 0.42514251546358434,
-      "grad_norm": 572.968017578125,
-      "learning_rate": 3.588706247072518e-05,
-      "loss": 65.1566,
-      "step": 105230
-    },
-    {
-      "epoch": 0.425182916728952,
-      "grad_norm": 859.6871948242188,
-      "learning_rate": 3.588392008361049e-05,
-      "loss": 49.4484,
-      "step": 105240
-    },
-    {
-      "epoch": 0.42522331799431956,
-      "grad_norm": 1073.6341552734375,
-      "learning_rate": 3.588077748430819e-05,
-      "loss": 65.2147,
-      "step": 105250
-    },
-    {
-      "epoch": 0.4252637192596872,
-      "grad_norm": 802.9885864257812,
-      "learning_rate": 3.587763467287953e-05,
-      "loss": 44.3761,
-      "step": 105260
-    },
-    {
-      "epoch": 0.42530412052505484,
-      "grad_norm": 1480.44921875,
-      "learning_rate": 3.587449164938578e-05,
-      "loss": 53.1393,
-      "step": 105270
-    },
-    {
-      "epoch": 0.4253445217904225,
-      "grad_norm": 544.9711303710938,
-      "learning_rate": 3.5871348413888204e-05,
-      "loss": 64.2403,
-      "step": 105280
-    },
-    {
-      "epoch": 0.4253849230557901,
-      "grad_norm": 1001.657958984375,
-      "learning_rate": 3.586820496644811e-05,
-      "loss": 61.8277,
-      "step": 105290
-    },
-    {
-      "epoch": 0.42542532432115776,
-      "grad_norm": 1063.820068359375,
-      "learning_rate": 3.586506130712676e-05,
-      "loss": 47.9627,
-      "step": 105300
-    },
-    {
-      "epoch": 0.42546572558652535,
-      "grad_norm": 1031.9720458984375,
-      "learning_rate": 3.5861917435985445e-05,
-      "loss": 58.5947,
-      "step": 105310
-    },
-    {
-      "epoch": 0.425506126851893,
-      "grad_norm": 376.46630859375,
-      "learning_rate": 3.585877335308546e-05,
-      "loss": 62.4376,
-      "step": 105320
-    },
-    {
-      "epoch": 0.4255465281172606,
-      "grad_norm": 1131.775390625,
-      "learning_rate": 3.5855629058488095e-05,
-      "loss": 71.483,
-      "step": 105330
-    },
-    {
-      "epoch": 0.42558692938262827,
-      "grad_norm": 687.2857666015625,
-      "learning_rate": 3.585248455225466e-05,
-      "loss": 69.5148,
-      "step": 105340
-    },
-    {
-      "epoch": 0.4256273306479959,
-      "grad_norm": 473.1686706542969,
-      "learning_rate": 3.584933983444644e-05,
-      "loss": 89.6963,
-      "step": 105350
-    },
-    {
-      "epoch": 0.42566773191336355,
-      "grad_norm": 742.19873046875,
-      "learning_rate": 3.5846194905124757e-05,
-      "loss": 67.2487,
-      "step": 105360
-    },
-    {
-      "epoch": 0.4257081331787312,
-      "grad_norm": 369.0582275390625,
-      "learning_rate": 3.584304976435092e-05,
-      "loss": 84.1498,
-      "step": 105370
-    },
-    {
-      "epoch": 0.42574853444409877,
-      "grad_norm": 409.1021728515625,
-      "learning_rate": 3.5839904412186256e-05,
-      "loss": 63.912,
-      "step": 105380
-    },
-    {
-      "epoch": 0.4257889357094664,
-      "grad_norm": 777.0023193359375,
-      "learning_rate": 3.583675884869206e-05,
-      "loss": 52.3509,
-      "step": 105390
-    },
-    {
-      "epoch": 0.42582933697483405,
-      "grad_norm": 345.2575988769531,
-      "learning_rate": 3.5833613073929684e-05,
-      "loss": 37.5077,
-      "step": 105400
-    },
-    {
-      "epoch": 0.4258697382402017,
-      "grad_norm": 1329.019287109375,
-      "learning_rate": 3.583046708796043e-05,
-      "loss": 46.7334,
-      "step": 105410
-    },
-    {
-      "epoch": 0.42591013950556933,
-      "grad_norm": 574.7626953125,
-      "learning_rate": 3.582732089084566e-05,
-      "loss": 49.3083,
-      "step": 105420
-    },
-    {
-      "epoch": 0.42595054077093697,
-      "grad_norm": 0.0,
-      "learning_rate": 3.582417448264669e-05,
-      "loss": 53.3416,
-      "step": 105430
-    },
-    {
-      "epoch": 0.42599094203630455,
-      "grad_norm": 1714.5047607421875,
-      "learning_rate": 3.582102786342485e-05,
-      "loss": 67.8292,
-      "step": 105440
-    },
-    {
-      "epoch": 0.4260313433016722,
-      "grad_norm": 1370.195068359375,
-      "learning_rate": 3.581788103324152e-05,
-      "loss": 58.7996,
-      "step": 105450
-    },
-    {
-      "epoch": 0.42607174456703983,
-      "grad_norm": 1272.4110107421875,
-      "learning_rate": 3.581473399215802e-05,
-      "loss": 70.6126,
-      "step": 105460
-    },
-    {
-      "epoch": 0.4261121458324075,
-      "grad_norm": 662.2067260742188,
-      "learning_rate": 3.581158674023572e-05,
-      "loss": 51.143,
-      "step": 105470
-    },
-    {
-      "epoch": 0.4261525470977751,
-      "grad_norm": 803.9454956054688,
-      "learning_rate": 3.5808439277535964e-05,
-      "loss": 38.5995,
-      "step": 105480
-    },
-    {
-      "epoch": 0.42619294836314275,
-      "grad_norm": 507.4480285644531,
-      "learning_rate": 3.580529160412013e-05,
-      "loss": 56.6335,
-      "step": 105490
-    },
-    {
-      "epoch": 0.42623334962851034,
-      "grad_norm": 673.6394653320312,
-      "learning_rate": 3.580214372004956e-05,
-      "loss": 71.4622,
-      "step": 105500
-    },
-    {
-      "epoch": 0.426273750893878,
-      "grad_norm": 506.5180969238281,
-      "learning_rate": 3.579899562538564e-05,
-      "loss": 34.8514,
-      "step": 105510
-    },
-    {
-      "epoch": 0.4263141521592456,
-      "grad_norm": 912.122802734375,
-      "learning_rate": 3.5795847320189746e-05,
-      "loss": 57.6717,
-      "step": 105520
-    },
-    {
-      "epoch": 0.42635455342461326,
-      "grad_norm": 888.0779418945312,
-      "learning_rate": 3.5792698804523245e-05,
-      "loss": 55.5145,
-      "step": 105530
-    },
-    {
-      "epoch": 0.4263949546899809,
-      "grad_norm": 670.391845703125,
-      "learning_rate": 3.5789550078447526e-05,
-      "loss": 78.2021,
-      "step": 105540
-    },
-    {
-      "epoch": 0.42643535595534854,
-      "grad_norm": 418.4927673339844,
-      "learning_rate": 3.5786401142023975e-05,
-      "loss": 73.2309,
-      "step": 105550
-    },
-    {
-      "epoch": 0.4264757572207162,
-      "grad_norm": 579.839599609375,
-      "learning_rate": 3.5783251995313985e-05,
-      "loss": 49.0248,
-      "step": 105560
-    },
-    {
-      "epoch": 0.42651615848608376,
-      "grad_norm": 1221.287841796875,
-      "learning_rate": 3.5780102638378936e-05,
-      "loss": 48.0422,
-      "step": 105570
-    },
-    {
-      "epoch": 0.4265565597514514,
-      "grad_norm": 576.9136962890625,
-      "learning_rate": 3.577695307128024e-05,
-      "loss": 48.1065,
-      "step": 105580
-    },
-    {
-      "epoch": 0.42659696101681904,
-      "grad_norm": 682.1599731445312,
-      "learning_rate": 3.57738032940793e-05,
-      "loss": 80.5689,
-      "step": 105590
-    },
-    {
-      "epoch": 0.4266373622821867,
-      "grad_norm": 952.7507934570312,
-      "learning_rate": 3.577065330683751e-05,
-      "loss": 59.2839,
-      "step": 105600
-    },
-    {
-      "epoch": 0.4266777635475543,
-      "grad_norm": 755.3192749023438,
-      "learning_rate": 3.5767503109616296e-05,
-      "loss": 63.4179,
-      "step": 105610
-    },
-    {
-      "epoch": 0.42671816481292196,
-      "grad_norm": 528.3074340820312,
-      "learning_rate": 3.576435270247706e-05,
-      "loss": 49.0063,
-      "step": 105620
-    },
-    {
-      "epoch": 0.42675856607828955,
-      "grad_norm": 1181.072509765625,
-      "learning_rate": 3.5761202085481235e-05,
-      "loss": 53.1849,
-      "step": 105630
-    },
-    {
-      "epoch": 0.4267989673436572,
-      "grad_norm": 825.5591430664062,
-      "learning_rate": 3.575805125869022e-05,
-      "loss": 54.1307,
-      "step": 105640
-    },
-    {
-      "epoch": 0.4268393686090248,
-      "grad_norm": 1657.726806640625,
-      "learning_rate": 3.5754900222165465e-05,
-      "loss": 84.8316,
-      "step": 105650
-    },
-    {
-      "epoch": 0.42687976987439247,
-      "grad_norm": 707.2351684570312,
-      "learning_rate": 3.5751748975968394e-05,
-      "loss": 71.4824,
-      "step": 105660
-    },
-    {
-      "epoch": 0.4269201711397601,
-      "grad_norm": 514.521484375,
-      "learning_rate": 3.574859752016045e-05,
-      "loss": 60.5219,
-      "step": 105670
-    },
-    {
-      "epoch": 0.42696057240512775,
-      "grad_norm": 661.9301147460938,
-      "learning_rate": 3.574544585480305e-05,
-      "loss": 45.8945,
-      "step": 105680
-    },
-    {
-      "epoch": 0.4270009736704954,
-      "grad_norm": 416.9234924316406,
-      "learning_rate": 3.574229397995765e-05,
-      "loss": 63.908,
-      "step": 105690
-    },
-    {
-      "epoch": 0.42704137493586297,
-      "grad_norm": 835.8658447265625,
-      "learning_rate": 3.573914189568571e-05,
-      "loss": 77.679,
-      "step": 105700
-    },
-    {
-      "epoch": 0.4270817762012306,
-      "grad_norm": 1938.5263671875,
-      "learning_rate": 3.5735989602048665e-05,
-      "loss": 62.1259,
-      "step": 105710
-    },
-    {
-      "epoch": 0.42712217746659825,
-      "grad_norm": 825.2261962890625,
-      "learning_rate": 3.573283709910798e-05,
-      "loss": 50.9862,
-      "step": 105720
-    },
-    {
-      "epoch": 0.4271625787319659,
-      "grad_norm": 1108.54345703125,
-      "learning_rate": 3.572968438692509e-05,
-      "loss": 75.7806,
-      "step": 105730
-    },
-    {
-      "epoch": 0.42720297999733353,
-      "grad_norm": 338.0368347167969,
-      "learning_rate": 3.5726531465561504e-05,
-      "loss": 40.806,
-      "step": 105740
-    },
-    {
-      "epoch": 0.42724338126270117,
-      "grad_norm": 876.4752197265625,
-      "learning_rate": 3.572337833507865e-05,
-      "loss": 63.9452,
-      "step": 105750
-    },
-    {
-      "epoch": 0.42728378252806876,
-      "grad_norm": 429.4320983886719,
-      "learning_rate": 3.572022499553802e-05,
-      "loss": 79.8431,
-      "step": 105760
-    },
-    {
-      "epoch": 0.4273241837934364,
-      "grad_norm": 749.8831787109375,
-      "learning_rate": 3.5717071447001083e-05,
-      "loss": 54.6913,
-      "step": 105770
-    },
-    {
-      "epoch": 0.42736458505880404,
-      "grad_norm": 902.8178100585938,
-      "learning_rate": 3.571391768952932e-05,
-      "loss": 45.6217,
-      "step": 105780
-    },
-    {
-      "epoch": 0.4274049863241717,
-      "grad_norm": 1094.285400390625,
-      "learning_rate": 3.571076372318422e-05,
-      "loss": 76.6032,
-      "step": 105790
-    },
-    {
-      "epoch": 0.4274453875895393,
-      "grad_norm": 1502.871337890625,
-      "learning_rate": 3.570760954802726e-05,
-      "loss": 80.5316,
-      "step": 105800
-    },
-    {
-      "epoch": 0.42748578885490696,
-      "grad_norm": 572.2935180664062,
-      "learning_rate": 3.5704455164119945e-05,
-      "loss": 64.48,
-      "step": 105810
-    },
-    {
-      "epoch": 0.42752619012027454,
-      "grad_norm": 742.1455688476562,
-      "learning_rate": 3.5701300571523755e-05,
-      "loss": 82.0077,
-      "step": 105820
-    },
-    {
-      "epoch": 0.4275665913856422,
-      "grad_norm": 520.0400390625,
-      "learning_rate": 3.569814577030022e-05,
-      "loss": 74.5518,
-      "step": 105830
-    },
-    {
-      "epoch": 0.4276069926510098,
-      "grad_norm": 967.4630126953125,
-      "learning_rate": 3.569499076051081e-05,
-      "loss": 51.6548,
-      "step": 105840
-    },
-    {
-      "epoch": 0.42764739391637746,
-      "grad_norm": 1097.524658203125,
-      "learning_rate": 3.5691835542217054e-05,
-      "loss": 66.9995,
-      "step": 105850
-    },
-    {
-      "epoch": 0.4276877951817451,
-      "grad_norm": 1097.96533203125,
-      "learning_rate": 3.5688680115480455e-05,
-      "loss": 55.4627,
-      "step": 105860
-    },
-    {
-      "epoch": 0.42772819644711274,
-      "grad_norm": 936.7929077148438,
-      "learning_rate": 3.5685524480362543e-05,
-      "loss": 79.1348,
-      "step": 105870
-    },
-    {
-      "epoch": 0.4277685977124804,
-      "grad_norm": 427.792724609375,
-      "learning_rate": 3.568236863692482e-05,
-      "loss": 57.1244,
-      "step": 105880
-    },
-    {
-      "epoch": 0.42780899897784797,
-      "grad_norm": 430.7917785644531,
-      "learning_rate": 3.567921258522883e-05,
-      "loss": 45.7751,
-      "step": 105890
-    },
-    {
-      "epoch": 0.4278494002432156,
-      "grad_norm": 1268.4412841796875,
-      "learning_rate": 3.567605632533608e-05,
-      "loss": 51.2699,
-      "step": 105900
-    },
-    {
-      "epoch": 0.42788980150858325,
-      "grad_norm": 156.56045532226562,
-      "learning_rate": 3.5672899857308134e-05,
-      "loss": 58.0844,
-      "step": 105910
-    },
-    {
-      "epoch": 0.4279302027739509,
-      "grad_norm": 829.1176147460938,
-      "learning_rate": 3.56697431812065e-05,
-      "loss": 55.2921,
-      "step": 105920
-    },
-    {
-      "epoch": 0.4279706040393185,
-      "grad_norm": 610.2461547851562,
-      "learning_rate": 3.566658629709273e-05,
-      "loss": 50.3534,
-      "step": 105930
-    },
-    {
-      "epoch": 0.42801100530468617,
-      "grad_norm": 1424.289306640625,
-      "learning_rate": 3.566342920502837e-05,
-      "loss": 63.0704,
-      "step": 105940
-    },
-    {
-      "epoch": 0.42805140657005375,
-      "grad_norm": 3496.262939453125,
-      "learning_rate": 3.5660271905074974e-05,
-      "loss": 59.1756,
-      "step": 105950
-    },
-    {
-      "epoch": 0.4280918078354214,
-      "grad_norm": 413.6312561035156,
-      "learning_rate": 3.565711439729408e-05,
-      "loss": 55.2588,
-      "step": 105960
-    },
-    {
-      "epoch": 0.42813220910078903,
-      "grad_norm": 632.3339233398438,
-      "learning_rate": 3.565395668174725e-05,
-      "loss": 69.6535,
-      "step": 105970
-    },
-    {
-      "epoch": 0.42817261036615667,
-      "grad_norm": 1112.0599365234375,
-      "learning_rate": 3.565079875849605e-05,
-      "loss": 51.4274,
-      "step": 105980
-    },
-    {
-      "epoch": 0.4282130116315243,
-      "grad_norm": 127.4210433959961,
-      "learning_rate": 3.564764062760205e-05,
-      "loss": 64.7097,
-      "step": 105990
-    },
-    {
-      "epoch": 0.42825341289689195,
-      "grad_norm": 1343.2515869140625,
-      "learning_rate": 3.564448228912682e-05,
-      "loss": 50.9704,
-      "step": 106000
-    },
-    {
-      "epoch": 0.4282938141622596,
-      "grad_norm": 794.98779296875,
-      "learning_rate": 3.564132374313192e-05,
-      "loss": 38.4331,
-      "step": 106010
-    },
-    {
-      "epoch": 0.4283342154276272,
-      "grad_norm": 195.28814697265625,
-      "learning_rate": 3.5638164989678935e-05,
-      "loss": 61.8604,
-      "step": 106020
-    },
-    {
-      "epoch": 0.4283746166929948,
-      "grad_norm": 1086.4222412109375,
-      "learning_rate": 3.563500602882945e-05,
-      "loss": 65.9661,
-      "step": 106030
-    },
-    {
-      "epoch": 0.42841501795836245,
-      "grad_norm": 737.5653076171875,
-      "learning_rate": 3.5631846860645044e-05,
-      "loss": 55.4186,
-      "step": 106040
-    },
-    {
-      "epoch": 0.4284554192237301,
-      "grad_norm": 864.1533203125,
-      "learning_rate": 3.562868748518732e-05,
-      "loss": 82.7763,
-      "step": 106050
-    },
-    {
-      "epoch": 0.42849582048909773,
-      "grad_norm": 1645.748046875,
-      "learning_rate": 3.562552790251785e-05,
-      "loss": 70.2854,
-      "step": 106060
-    },
-    {
-      "epoch": 0.4285362217544654,
-      "grad_norm": 754.142578125,
-      "learning_rate": 3.562236811269824e-05,
-      "loss": 55.0108,
-      "step": 106070
-    },
-    {
-      "epoch": 0.42857662301983296,
-      "grad_norm": 408.4161376953125,
-      "learning_rate": 3.56192081157901e-05,
-      "loss": 69.7249,
-      "step": 106080
-    },
-    {
-      "epoch": 0.4286170242852006,
-      "grad_norm": 308.8442077636719,
-      "learning_rate": 3.561604791185503e-05,
-      "loss": 54.3556,
-      "step": 106090
-    },
-    {
-      "epoch": 0.42865742555056824,
-      "grad_norm": 1148.19677734375,
-      "learning_rate": 3.561288750095465e-05,
-      "loss": 57.7566,
-      "step": 106100
-    },
-    {
-      "epoch": 0.4286978268159359,
-      "grad_norm": 589.8562622070312,
-      "learning_rate": 3.560972688315055e-05,
-      "loss": 65.197,
-      "step": 106110
-    },
-    {
-      "epoch": 0.4287382280813035,
-      "grad_norm": 1979.934326171875,
-      "learning_rate": 3.5606566058504375e-05,
-      "loss": 67.4663,
-      "step": 106120
-    },
-    {
-      "epoch": 0.42877862934667116,
-      "grad_norm": 638.1552124023438,
-      "learning_rate": 3.560340502707773e-05,
-      "loss": 40.0643,
-      "step": 106130
-    },
-    {
-      "epoch": 0.42881903061203874,
-      "grad_norm": 1344.94580078125,
-      "learning_rate": 3.560024378893224e-05,
-      "loss": 71.7906,
-      "step": 106140
-    },
-    {
-      "epoch": 0.4288594318774064,
-      "grad_norm": 541.87353515625,
-      "learning_rate": 3.559708234412954e-05,
-      "loss": 89.126,
-      "step": 106150
-    },
-    {
-      "epoch": 0.428899833142774,
-      "grad_norm": 1129.01708984375,
-      "learning_rate": 3.559392069273127e-05,
-      "loss": 74.1695,
-      "step": 106160
-    },
-    {
-      "epoch": 0.42894023440814166,
-      "grad_norm": 741.240234375,
-      "learning_rate": 3.559075883479906e-05,
-      "loss": 77.3889,
-      "step": 106170
-    },
-    {
-      "epoch": 0.4289806356735093,
-      "grad_norm": 373.0411682128906,
-      "learning_rate": 3.558759677039455e-05,
-      "loss": 85.0204,
-      "step": 106180
-    },
-    {
-      "epoch": 0.42902103693887694,
-      "grad_norm": 705.1157836914062,
-      "learning_rate": 3.558443449957939e-05,
-      "loss": 77.1509,
-      "step": 106190
-    },
-    {
-      "epoch": 0.4290614382042446,
-      "grad_norm": 565.6890258789062,
-      "learning_rate": 3.5581272022415244e-05,
-      "loss": 62.9252,
-      "step": 106200
-    },
-    {
-      "epoch": 0.42910183946961217,
-      "grad_norm": 2649.929443359375,
-      "learning_rate": 3.5578109338963736e-05,
-      "loss": 79.2853,
-      "step": 106210
-    },
-    {
-      "epoch": 0.4291422407349798,
-      "grad_norm": 645.3712768554688,
-      "learning_rate": 3.557494644928654e-05,
-      "loss": 52.6527,
-      "step": 106220
-    },
-    {
-      "epoch": 0.42918264200034745,
-      "grad_norm": 446.864013671875,
-      "learning_rate": 3.5571783353445325e-05,
-      "loss": 54.5988,
-      "step": 106230
-    },
-    {
-      "epoch": 0.4292230432657151,
-      "grad_norm": 1004.2807006835938,
-      "learning_rate": 3.5568620051501756e-05,
-      "loss": 67.0789,
-      "step": 106240
-    },
-    {
-      "epoch": 0.4292634445310827,
-      "grad_norm": 611.4927368164062,
-      "learning_rate": 3.556545654351749e-05,
-      "loss": 38.8004,
-      "step": 106250
-    },
-    {
-      "epoch": 0.42930384579645037,
-      "grad_norm": 868.6104125976562,
-      "learning_rate": 3.556229282955421e-05,
-      "loss": 44.7674,
-      "step": 106260
-    },
-    {
-      "epoch": 0.42934424706181795,
-      "grad_norm": 1725.8865966796875,
-      "learning_rate": 3.5559128909673595e-05,
-      "loss": 71.4952,
-      "step": 106270
-    },
-    {
-      "epoch": 0.4293846483271856,
-      "grad_norm": 654.6998291015625,
-      "learning_rate": 3.555596478393733e-05,
-      "loss": 48.7473,
-      "step": 106280
-    },
-    {
-      "epoch": 0.42942504959255323,
-      "grad_norm": 1723.461669921875,
-      "learning_rate": 3.555280045240709e-05,
-      "loss": 62.5021,
-      "step": 106290
-    },
-    {
-      "epoch": 0.42946545085792087,
-      "grad_norm": 989.3924560546875,
-      "learning_rate": 3.554963591514457e-05,
-      "loss": 64.8296,
-      "step": 106300
-    },
-    {
-      "epoch": 0.4295058521232885,
-      "grad_norm": 767.1257934570312,
-      "learning_rate": 3.554647117221147e-05,
-      "loss": 58.58,
-      "step": 106310
-    },
-    {
-      "epoch": 0.42954625338865615,
-      "grad_norm": 958.0810546875,
-      "learning_rate": 3.554330622366949e-05,
-      "loss": 82.886,
-      "step": 106320
-    },
-    {
-      "epoch": 0.42958665465402374,
-      "grad_norm": 1372.3026123046875,
-      "learning_rate": 3.554014106958032e-05,
-      "loss": 91.1392,
-      "step": 106330
-    },
-    {
-      "epoch": 0.4296270559193914,
-      "grad_norm": 765.9713745117188,
-      "learning_rate": 3.5536975710005677e-05,
-      "loss": 43.6938,
-      "step": 106340
-    },
-    {
-      "epoch": 0.429667457184759,
-      "grad_norm": 2553.733642578125,
-      "learning_rate": 3.553381014500727e-05,
-      "loss": 54.1921,
-      "step": 106350
-    },
-    {
-      "epoch": 0.42970785845012666,
-      "grad_norm": 1080.475341796875,
-      "learning_rate": 3.5530644374646815e-05,
-      "loss": 62.0937,
-      "step": 106360
-    },
-    {
-      "epoch": 0.4297482597154943,
-      "grad_norm": 2564.04638671875,
-      "learning_rate": 3.5527478398986015e-05,
-      "loss": 63.5857,
-      "step": 106370
-    },
-    {
-      "epoch": 0.42978866098086194,
-      "grad_norm": 299.32989501953125,
-      "learning_rate": 3.552431221808661e-05,
-      "loss": 44.5124,
-      "step": 106380
-    },
-    {
-      "epoch": 0.4298290622462296,
-      "grad_norm": 631.6810302734375,
-      "learning_rate": 3.5521145832010314e-05,
-      "loss": 49.5127,
-      "step": 106390
-    },
-    {
-      "epoch": 0.42986946351159716,
-      "grad_norm": 1121.8922119140625,
-      "learning_rate": 3.551797924081887e-05,
-      "loss": 58.7079,
-      "step": 106400
-    },
-    {
-      "epoch": 0.4299098647769648,
-      "grad_norm": 657.3575439453125,
-      "learning_rate": 3.5514812444574004e-05,
-      "loss": 50.7894,
-      "step": 106410
-    },
-    {
-      "epoch": 0.42995026604233244,
-      "grad_norm": 662.6183471679688,
-      "learning_rate": 3.551164544333745e-05,
-      "loss": 79.3807,
-      "step": 106420
-    },
-    {
-      "epoch": 0.4299906673077001,
-      "grad_norm": 593.0071411132812,
-      "learning_rate": 3.550847823717096e-05,
-      "loss": 65.9797,
-      "step": 106430
-    },
-    {
-      "epoch": 0.4300310685730677,
-      "grad_norm": 755.3797607421875,
-      "learning_rate": 3.5505310826136286e-05,
-      "loss": 51.0179,
-      "step": 106440
-    },
-    {
-      "epoch": 0.43007146983843536,
-      "grad_norm": 1839.5137939453125,
-      "learning_rate": 3.5502143210295165e-05,
-      "loss": 62.1264,
-      "step": 106450
-    },
-    {
-      "epoch": 0.43011187110380295,
-      "grad_norm": 722.4067993164062,
-      "learning_rate": 3.549897538970934e-05,
-      "loss": 63.9707,
-      "step": 106460
-    },
-    {
-      "epoch": 0.4301522723691706,
-      "grad_norm": 465.209716796875,
-      "learning_rate": 3.54958073644406e-05,
-      "loss": 51.9835,
-      "step": 106470
-    },
-    {
-      "epoch": 0.4301926736345382,
-      "grad_norm": 348.4991760253906,
-      "learning_rate": 3.5492639134550695e-05,
-      "loss": 84.2273,
-      "step": 106480
-    },
-    {
-      "epoch": 0.43023307489990587,
-      "grad_norm": 1531.664794921875,
-      "learning_rate": 3.548947070010138e-05,
-      "loss": 81.6072,
-      "step": 106490
-    },
-    {
-      "epoch": 0.4302734761652735,
-      "grad_norm": 797.4585571289062,
-      "learning_rate": 3.548630206115443e-05,
-      "loss": 56.6595,
-      "step": 106500
-    },
-    {
-      "epoch": 0.43031387743064115,
-      "grad_norm": 412.25360107421875,
-      "learning_rate": 3.5483133217771625e-05,
-      "loss": 57.9216,
-      "step": 106510
-    },
-    {
-      "epoch": 0.4303542786960088,
-      "grad_norm": 647.94921875,
-      "learning_rate": 3.5479964170014746e-05,
-      "loss": 99.7013,
-      "step": 106520
-    },
-    {
-      "epoch": 0.43039467996137637,
-      "grad_norm": 649.0404052734375,
-      "learning_rate": 3.547679491794557e-05,
-      "loss": 46.1641,
-      "step": 106530
-    },
-    {
-      "epoch": 0.430435081226744,
-      "grad_norm": 434.8841552734375,
-      "learning_rate": 3.547362546162588e-05,
-      "loss": 61.1318,
-      "step": 106540
-    },
-    {
-      "epoch": 0.43047548249211165,
-      "grad_norm": 733.9017333984375,
-      "learning_rate": 3.547045580111746e-05,
-      "loss": 49.06,
-      "step": 106550
-    },
-    {
-      "epoch": 0.4305158837574793,
-      "grad_norm": 721.9252319335938,
-      "learning_rate": 3.546728593648213e-05,
-      "loss": 61.6053,
-      "step": 106560
-    },
-    {
-      "epoch": 0.43055628502284693,
-      "grad_norm": 553.4278564453125,
-      "learning_rate": 3.546411586778167e-05,
-      "loss": 59.98,
-      "step": 106570
-    },
-    {
-      "epoch": 0.43059668628821457,
-      "grad_norm": 410.0889892578125,
-      "learning_rate": 3.546094559507787e-05,
-      "loss": 44.391,
-      "step": 106580
-    },
-    {
-      "epoch": 0.43063708755358215,
-      "grad_norm": 429.70361328125,
-      "learning_rate": 3.5457775118432556e-05,
-      "loss": 90.1694,
-      "step": 106590
-    },
-    {
-      "epoch": 0.4306774888189498,
-      "grad_norm": 899.5925903320312,
-      "learning_rate": 3.545460443790753e-05,
-      "loss": 79.3271,
-      "step": 106600
-    },
-    {
-      "epoch": 0.43071789008431743,
-      "grad_norm": 1410.063232421875,
-      "learning_rate": 3.545143355356462e-05,
-      "loss": 75.9909,
-      "step": 106610
-    },
-    {
-      "epoch": 0.4307582913496851,
-      "grad_norm": 1006.2098999023438,
-      "learning_rate": 3.544826246546563e-05,
-      "loss": 61.4801,
-      "step": 106620
-    },
-    {
-      "epoch": 0.4307986926150527,
-      "grad_norm": 618.147216796875,
-      "learning_rate": 3.544509117367238e-05,
-      "loss": 56.4724,
-      "step": 106630
-    },
-    {
-      "epoch": 0.43083909388042035,
-      "grad_norm": 866.3955078125,
-      "learning_rate": 3.544191967824669e-05,
-      "loss": 47.1831,
-      "step": 106640
-    },
-    {
-      "epoch": 0.43087949514578794,
-      "grad_norm": 802.1903686523438,
-      "learning_rate": 3.543874797925042e-05,
-      "loss": 62.7335,
-      "step": 106650
-    },
-    {
-      "epoch": 0.4309198964111556,
-      "grad_norm": 722.4039306640625,
-      "learning_rate": 3.543557607674537e-05,
-      "loss": 57.5358,
-      "step": 106660
-    },
-    {
-      "epoch": 0.4309602976765232,
-      "grad_norm": 467.41937255859375,
-      "learning_rate": 3.543240397079339e-05,
-      "loss": 42.7253,
-      "step": 106670
-    },
-    {
-      "epoch": 0.43100069894189086,
-      "grad_norm": 421.191162109375,
-      "learning_rate": 3.542923166145633e-05,
-      "loss": 60.0862,
-      "step": 106680
-    },
-    {
-      "epoch": 0.4310411002072585,
-      "grad_norm": 797.8417358398438,
-      "learning_rate": 3.542605914879603e-05,
-      "loss": 69.1785,
-      "step": 106690
-    },
-    {
-      "epoch": 0.43108150147262614,
-      "grad_norm": 1393.1776123046875,
-      "learning_rate": 3.542288643287434e-05,
-      "loss": 42.2223,
-      "step": 106700
-    },
-    {
-      "epoch": 0.4311219027379938,
-      "grad_norm": 699.2664184570312,
-      "learning_rate": 3.5419713513753114e-05,
-      "loss": 71.7247,
-      "step": 106710
-    },
-    {
-      "epoch": 0.43116230400336136,
-      "grad_norm": 1001.225830078125,
-      "learning_rate": 3.54165403914942e-05,
-      "loss": 54.1351,
-      "step": 106720
-    },
-    {
-      "epoch": 0.431202705268729,
-      "grad_norm": 587.420166015625,
-      "learning_rate": 3.541336706615947e-05,
-      "loss": 62.8683,
-      "step": 106730
-    },
-    {
-      "epoch": 0.43124310653409664,
-      "grad_norm": 481.1246643066406,
-      "learning_rate": 3.541019353781079e-05,
-      "loss": 48.2253,
-      "step": 106740
-    },
-    {
-      "epoch": 0.4312835077994643,
-      "grad_norm": 694.5806274414062,
-      "learning_rate": 3.540701980651003e-05,
-      "loss": 50.7718,
-      "step": 106750
-    },
-    {
-      "epoch": 0.4313239090648319,
-      "grad_norm": 1048.5947265625,
-      "learning_rate": 3.540384587231906e-05,
-      "loss": 78.4464,
-      "step": 106760
-    },
-    {
-      "epoch": 0.43136431033019956,
-      "grad_norm": 372.85406494140625,
-      "learning_rate": 3.540067173529976e-05,
-      "loss": 48.4107,
-      "step": 106770
-    },
-    {
-      "epoch": 0.43140471159556715,
-      "grad_norm": 2308.911865234375,
-      "learning_rate": 3.5397497395514004e-05,
-      "loss": 73.4827,
-      "step": 106780
-    },
-    {
-      "epoch": 0.4314451128609348,
-      "grad_norm": 183.2915802001953,
-      "learning_rate": 3.5394322853023694e-05,
-      "loss": 35.4445,
-      "step": 106790
-    },
-    {
-      "epoch": 0.4314855141263024,
-      "grad_norm": 4619.44140625,
-      "learning_rate": 3.53911481078907e-05,
-      "loss": 50.2226,
-      "step": 106800
-    },
-    {
-      "epoch": 0.43152591539167007,
-      "grad_norm": 873.1148071289062,
-      "learning_rate": 3.5387973160176926e-05,
-      "loss": 51.0404,
-      "step": 106810
-    },
-    {
-      "epoch": 0.4315663166570377,
-      "grad_norm": 480.3205871582031,
-      "learning_rate": 3.538479800994426e-05,
-      "loss": 41.806,
-      "step": 106820
-    },
-    {
-      "epoch": 0.43160671792240535,
-      "grad_norm": 560.2127685546875,
-      "learning_rate": 3.538162265725462e-05,
-      "loss": 34.7096,
-      "step": 106830
-    },
-    {
-      "epoch": 0.431647119187773,
-      "grad_norm": 800.5224609375,
-      "learning_rate": 3.5378447102169895e-05,
-      "loss": 83.1514,
-      "step": 106840
-    },
-    {
-      "epoch": 0.43168752045314057,
-      "grad_norm": 403.2615661621094,
-      "learning_rate": 3.537527134475201e-05,
-      "loss": 81.7792,
-      "step": 106850
-    },
-    {
-      "epoch": 0.4317279217185082,
-      "grad_norm": 613.8062744140625,
-      "learning_rate": 3.537209538506286e-05,
-      "loss": 51.7972,
-      "step": 106860
-    },
-    {
-      "epoch": 0.43176832298387585,
-      "grad_norm": 794.0146484375,
-      "learning_rate": 3.5368919223164374e-05,
-      "loss": 78.0091,
-      "step": 106870
-    },
-    {
-      "epoch": 0.4318087242492435,
-      "grad_norm": 877.7191162109375,
-      "learning_rate": 3.536574285911847e-05,
-      "loss": 47.9676,
-      "step": 106880
-    },
-    {
-      "epoch": 0.43184912551461113,
-      "grad_norm": 1015.1839599609375,
-      "learning_rate": 3.5362566292987076e-05,
-      "loss": 56.4783,
-      "step": 106890
-    },
-    {
-      "epoch": 0.43188952677997877,
-      "grad_norm": 519.8533935546875,
-      "learning_rate": 3.535938952483211e-05,
-      "loss": 53.7836,
-      "step": 106900
-    },
-    {
-      "epoch": 0.43192992804534636,
-      "grad_norm": 356.93109130859375,
-      "learning_rate": 3.5356212554715506e-05,
-      "loss": 63.2428,
-      "step": 106910
-    },
-    {
-      "epoch": 0.431970329310714,
-      "grad_norm": 694.5162963867188,
-      "learning_rate": 3.535303538269922e-05,
-      "loss": 33.9129,
-      "step": 106920
-    },
-    {
-      "epoch": 0.43201073057608164,
-      "grad_norm": 796.4173583984375,
-      "learning_rate": 3.534985800884517e-05,
-      "loss": 58.3042,
-      "step": 106930
-    },
-    {
-      "epoch": 0.4320511318414493,
-      "grad_norm": 0.0,
-      "learning_rate": 3.5346680433215316e-05,
-      "loss": 38.0315,
-      "step": 106940
-    },
-    {
-      "epoch": 0.4320915331068169,
-      "grad_norm": 570.5447387695312,
-      "learning_rate": 3.5343502655871594e-05,
-      "loss": 58.1119,
-      "step": 106950
-    },
-    {
-      "epoch": 0.43213193437218456,
-      "grad_norm": 2234.41162109375,
-      "learning_rate": 3.534032467687597e-05,
-      "loss": 50.3924,
-      "step": 106960
-    },
-    {
-      "epoch": 0.43217233563755214,
-      "grad_norm": 470.1455993652344,
-      "learning_rate": 3.533714649629039e-05,
-      "loss": 51.558,
-      "step": 106970
-    },
-    {
-      "epoch": 0.4322127369029198,
-      "grad_norm": 641.6006469726562,
-      "learning_rate": 3.533396811417682e-05,
-      "loss": 46.6653,
-      "step": 106980
-    },
-    {
-      "epoch": 0.4322531381682874,
-      "grad_norm": 1771.727294921875,
-      "learning_rate": 3.533078953059721e-05,
-      "loss": 42.6348,
-      "step": 106990
-    },
-    {
-      "epoch": 0.43229353943365506,
-      "grad_norm": 885.59765625,
-      "learning_rate": 3.532761074561355e-05,
-      "loss": 58.1762,
-      "step": 107000
-    },
-    {
-      "epoch": 0.4323339406990227,
-      "grad_norm": 618.5126953125,
-      "learning_rate": 3.5324431759287796e-05,
-      "loss": 43.6578,
-      "step": 107010
-    },
-    {
-      "epoch": 0.43237434196439034,
-      "grad_norm": 740.6704711914062,
-      "learning_rate": 3.532125257168193e-05,
-      "loss": 49.2011,
-      "step": 107020
-    },
-    {
-      "epoch": 0.432414743229758,
-      "grad_norm": 713.0264892578125,
-      "learning_rate": 3.531807318285793e-05,
-      "loss": 67.6269,
-      "step": 107030
-    },
-    {
-      "epoch": 0.43245514449512557,
-      "grad_norm": 793.6551513671875,
-      "learning_rate": 3.531489359287779e-05,
-      "loss": 78.4188,
-      "step": 107040
-    },
-    {
-      "epoch": 0.4324955457604932,
-      "grad_norm": 1205.0350341796875,
-      "learning_rate": 3.531171380180348e-05,
-      "loss": 77.0156,
-      "step": 107050
-    },
-    {
-      "epoch": 0.43253594702586085,
-      "grad_norm": 1664.7984619140625,
-      "learning_rate": 3.530853380969701e-05,
-      "loss": 51.596,
-      "step": 107060
-    },
-    {
-      "epoch": 0.4325763482912285,
-      "grad_norm": 353.55523681640625,
-      "learning_rate": 3.5305353616620355e-05,
-      "loss": 36.4841,
-      "step": 107070
-    },
-    {
-      "epoch": 0.4326167495565961,
-      "grad_norm": 607.7645263671875,
-      "learning_rate": 3.5302173222635524e-05,
-      "loss": 56.0603,
-      "step": 107080
-    },
-    {
-      "epoch": 0.43265715082196377,
-      "grad_norm": 688.8688354492188,
-      "learning_rate": 3.529899262780453e-05,
-      "loss": 60.969,
-      "step": 107090
-    },
-    {
-      "epoch": 0.43269755208733135,
-      "grad_norm": 431.1716003417969,
-      "learning_rate": 3.529581183218937e-05,
-      "loss": 43.1526,
-      "step": 107100
-    },
-    {
-      "epoch": 0.432737953352699,
-      "grad_norm": 1323.6634521484375,
-      "learning_rate": 3.529263083585206e-05,
-      "loss": 64.5691,
-      "step": 107110
-    },
-    {
-      "epoch": 0.43277835461806663,
-      "grad_norm": 650.7156372070312,
-      "learning_rate": 3.528944963885461e-05,
-      "loss": 24.5781,
-      "step": 107120
-    },
-    {
-      "epoch": 0.43281875588343427,
-      "grad_norm": 1326.6907958984375,
-      "learning_rate": 3.528626824125905e-05,
-      "loss": 55.9819,
-      "step": 107130
-    },
-    {
-      "epoch": 0.4328591571488019,
-      "grad_norm": 727.043701171875,
-      "learning_rate": 3.528308664312739e-05,
-      "loss": 44.3257,
-      "step": 107140
-    },
-    {
-      "epoch": 0.43289955841416955,
-      "grad_norm": 719.6072998046875,
-      "learning_rate": 3.527990484452166e-05,
-      "loss": 46.9241,
-      "step": 107150
-    },
-    {
-      "epoch": 0.4329399596795372,
-      "grad_norm": 1634.4859619140625,
-      "learning_rate": 3.527672284550389e-05,
-      "loss": 87.1335,
-      "step": 107160
-    },
-    {
-      "epoch": 0.4329803609449048,
-      "grad_norm": 281.7586975097656,
-      "learning_rate": 3.527354064613612e-05,
-      "loss": 70.6858,
-      "step": 107170
-    },
-    {
-      "epoch": 0.4330207622102724,
-      "grad_norm": 1065.198974609375,
-      "learning_rate": 3.5270358246480386e-05,
-      "loss": 79.6831,
-      "step": 107180
-    },
-    {
-      "epoch": 0.43306116347564005,
-      "grad_norm": 263.0101623535156,
-      "learning_rate": 3.526717564659873e-05,
-      "loss": 61.8488,
-      "step": 107190
-    },
-    {
-      "epoch": 0.4331015647410077,
-      "grad_norm": 918.0792236328125,
-      "learning_rate": 3.52639928465532e-05,
-      "loss": 56.7985,
-      "step": 107200
-    },
-    {
-      "epoch": 0.43314196600637533,
-      "grad_norm": 454.3590393066406,
-      "learning_rate": 3.526080984640585e-05,
-      "loss": 59.4921,
-      "step": 107210
-    },
-    {
-      "epoch": 0.433182367271743,
-      "grad_norm": 332.873291015625,
-      "learning_rate": 3.525762664621872e-05,
-      "loss": 34.6081,
-      "step": 107220
-    },
-    {
-      "epoch": 0.43322276853711056,
-      "grad_norm": 1424.8919677734375,
-      "learning_rate": 3.5254443246053886e-05,
-      "loss": 54.6382,
-      "step": 107230
-    },
-    {
-      "epoch": 0.4332631698024782,
-      "grad_norm": 1493.1998291015625,
-      "learning_rate": 3.5251259645973394e-05,
-      "loss": 77.3529,
-      "step": 107240
-    },
-    {
-      "epoch": 0.43330357106784584,
-      "grad_norm": 310.91943359375,
-      "learning_rate": 3.524807584603932e-05,
-      "loss": 58.498,
-      "step": 107250
-    },
-    {
-      "epoch": 0.4333439723332135,
-      "grad_norm": 713.1577758789062,
-      "learning_rate": 3.5244891846313736e-05,
-      "loss": 48.7548,
-      "step": 107260
-    },
-    {
-      "epoch": 0.4333843735985811,
-      "grad_norm": 746.4721069335938,
-      "learning_rate": 3.5241707646858703e-05,
-      "loss": 67.363,
-      "step": 107270
-    },
-    {
-      "epoch": 0.43342477486394876,
-      "grad_norm": 1737.3177490234375,
-      "learning_rate": 3.523852324773631e-05,
-      "loss": 58.3859,
-      "step": 107280
-    },
-    {
-      "epoch": 0.43346517612931634,
-      "grad_norm": 240.0654754638672,
-      "learning_rate": 3.523533864900863e-05,
-      "loss": 89.2563,
-      "step": 107290
-    },
-    {
-      "epoch": 0.433505577394684,
-      "grad_norm": 697.5162353515625,
-      "learning_rate": 3.523215385073777e-05,
-      "loss": 66.2658,
-      "step": 107300
-    },
-    {
-      "epoch": 0.4335459786600516,
-      "grad_norm": 527.6311645507812,
-      "learning_rate": 3.52289688529858e-05,
-      "loss": 39.8144,
-      "step": 107310
-    },
-    {
-      "epoch": 0.43358637992541926,
-      "grad_norm": 1482.4761962890625,
-      "learning_rate": 3.5225783655814796e-05,
-      "loss": 56.7785,
-      "step": 107320
-    },
-    {
-      "epoch": 0.4336267811907869,
-      "grad_norm": 1850.587646484375,
-      "learning_rate": 3.522259825928689e-05,
-      "loss": 69.2113,
-      "step": 107330
-    },
-    {
-      "epoch": 0.43366718245615454,
-      "grad_norm": 821.7341918945312,
-      "learning_rate": 3.5219412663464167e-05,
-      "loss": 85.9736,
-      "step": 107340
-    },
-    {
-      "epoch": 0.4337075837215222,
-      "grad_norm": 1092.7030029296875,
-      "learning_rate": 3.521622686840873e-05,
-      "loss": 74.6831,
-      "step": 107350
-    },
-    {
-      "epoch": 0.43374798498688977,
-      "grad_norm": 633.818115234375,
-      "learning_rate": 3.521304087418269e-05,
-      "loss": 64.548,
-      "step": 107360
-    },
-    {
-      "epoch": 0.4337883862522574,
-      "grad_norm": 994.1685791015625,
-      "learning_rate": 3.520985468084816e-05,
-      "loss": 31.1137,
-      "step": 107370
-    },
-    {
-      "epoch": 0.43382878751762505,
-      "grad_norm": 887.774169921875,
-      "learning_rate": 3.520666828846726e-05,
-      "loss": 52.128,
-      "step": 107380
-    },
-    {
-      "epoch": 0.4338691887829927,
-      "grad_norm": 478.0501708984375,
-      "learning_rate": 3.52034816971021e-05,
-      "loss": 60.6667,
-      "step": 107390
-    },
-    {
-      "epoch": 0.4339095900483603,
-      "grad_norm": 1626.81884765625,
-      "learning_rate": 3.5200294906814824e-05,
-      "loss": 108.8241,
-      "step": 107400
-    },
-    {
-      "epoch": 0.43394999131372797,
-      "grad_norm": 482.78741455078125,
-      "learning_rate": 3.519710791766754e-05,
-      "loss": 57.1852,
-      "step": 107410
-    },
-    {
-      "epoch": 0.43399039257909555,
-      "grad_norm": 536.291259765625,
-      "learning_rate": 3.5193920729722384e-05,
-      "loss": 55.354,
-      "step": 107420
-    },
-    {
-      "epoch": 0.4340307938444632,
-      "grad_norm": 395.05181884765625,
-      "learning_rate": 3.51907333430415e-05,
-      "loss": 51.6174,
-      "step": 107430
-    },
-    {
-      "epoch": 0.43407119510983083,
-      "grad_norm": 504.0698547363281,
-      "learning_rate": 3.5187545757687015e-05,
-      "loss": 48.5033,
-      "step": 107440
-    },
-    {
-      "epoch": 0.43411159637519847,
-      "grad_norm": 938.1822509765625,
-      "learning_rate": 3.518435797372109e-05,
-      "loss": 77.9516,
-      "step": 107450
-    },
-    {
-      "epoch": 0.4341519976405661,
-      "grad_norm": 542.053955078125,
-      "learning_rate": 3.5181169991205866e-05,
-      "loss": 50.1242,
-      "step": 107460
-    },
-    {
-      "epoch": 0.43419239890593375,
-      "grad_norm": 945.6251831054688,
-      "learning_rate": 3.517798181020348e-05,
-      "loss": 82.4342,
-      "step": 107470
-    },
-    {
-      "epoch": 0.4342328001713014,
-      "grad_norm": 841.2509765625,
-      "learning_rate": 3.517479343077611e-05,
-      "loss": 48.6936,
-      "step": 107480
-    },
-    {
-      "epoch": 0.434273201436669,
-      "grad_norm": 1135.741943359375,
-      "learning_rate": 3.517160485298589e-05,
-      "loss": 57.5404,
-      "step": 107490
-    },
-    {
-      "epoch": 0.4343136027020366,
-      "grad_norm": 634.0276489257812,
-      "learning_rate": 3.516841607689501e-05,
-      "loss": 57.2011,
-      "step": 107500
-    },
-    {
-      "epoch": 0.43435400396740426,
-      "grad_norm": 684.9427490234375,
-      "learning_rate": 3.516522710256562e-05,
-      "loss": 63.644,
-      "step": 107510
-    },
-    {
-      "epoch": 0.4343944052327719,
-      "grad_norm": 707.3285522460938,
-      "learning_rate": 3.516203793005989e-05,
-      "loss": 44.6688,
-      "step": 107520
-    },
-    {
-      "epoch": 0.43443480649813954,
-      "grad_norm": 765.8327026367188,
-      "learning_rate": 3.515884855944e-05,
-      "loss": 46.0717,
-      "step": 107530
-    },
-    {
-      "epoch": 0.4344752077635072,
-      "grad_norm": 708.8120727539062,
-      "learning_rate": 3.515565899076813e-05,
-      "loss": 47.5356,
-      "step": 107540
-    },
-    {
-      "epoch": 0.43451560902887476,
-      "grad_norm": 779.5784912109375,
-      "learning_rate": 3.5152469224106454e-05,
-      "loss": 50.02,
-      "step": 107550
-    },
-    {
-      "epoch": 0.4345560102942424,
-      "grad_norm": 638.3084106445312,
-      "learning_rate": 3.514927925951717e-05,
-      "loss": 69.4422,
-      "step": 107560
-    },
-    {
-      "epoch": 0.43459641155961004,
-      "grad_norm": 0.0,
-      "learning_rate": 3.5146089097062456e-05,
-      "loss": 48.1131,
-      "step": 107570
-    },
-    {
-      "epoch": 0.4346368128249777,
-      "grad_norm": 613.4315185546875,
-      "learning_rate": 3.514289873680451e-05,
-      "loss": 39.7531,
-      "step": 107580
-    },
-    {
-      "epoch": 0.4346772140903453,
-      "grad_norm": 596.2418212890625,
-      "learning_rate": 3.513970817880554e-05,
-      "loss": 46.1025,
-      "step": 107590
-    },
-    {
-      "epoch": 0.43471761535571296,
-      "grad_norm": 742.696044921875,
-      "learning_rate": 3.513651742312774e-05,
-      "loss": 50.7197,
-      "step": 107600
-    },
-    {
-      "epoch": 0.43475801662108055,
-      "grad_norm": 517.7498168945312,
-      "learning_rate": 3.51333264698333e-05,
-      "loss": 72.2126,
-      "step": 107610
-    },
-    {
-      "epoch": 0.4347984178864482,
-      "grad_norm": 336.8522644042969,
-      "learning_rate": 3.5130135318984456e-05,
-      "loss": 97.5587,
-      "step": 107620
-    },
-    {
-      "epoch": 0.4348388191518158,
-      "grad_norm": 1882.3516845703125,
-      "learning_rate": 3.512694397064341e-05,
-      "loss": 46.2149,
-      "step": 107630
-    },
-    {
-      "epoch": 0.43487922041718347,
-      "grad_norm": 852.6275024414062,
-      "learning_rate": 3.512375242487236e-05,
-      "loss": 59.0864,
-      "step": 107640
-    },
-    {
-      "epoch": 0.4349196216825511,
-      "grad_norm": 1158.6456298828125,
-      "learning_rate": 3.512056068173356e-05,
-      "loss": 86.1055,
-      "step": 107650
-    },
-    {
-      "epoch": 0.43496002294791875,
-      "grad_norm": 1050.2406005859375,
-      "learning_rate": 3.511736874128922e-05,
-      "loss": 47.8806,
-      "step": 107660
-    },
-    {
-      "epoch": 0.4350004242132864,
-      "grad_norm": 576.0054931640625,
-      "learning_rate": 3.5114176603601564e-05,
-      "loss": 44.2818,
-      "step": 107670
-    },
-    {
-      "epoch": 0.43504082547865397,
-      "grad_norm": 3012.95849609375,
-      "learning_rate": 3.511098426873283e-05,
-      "loss": 75.0666,
-      "step": 107680
-    },
-    {
-      "epoch": 0.4350812267440216,
-      "grad_norm": 1107.3419189453125,
-      "learning_rate": 3.5107791736745244e-05,
-      "loss": 47.5885,
-      "step": 107690
-    },
-    {
-      "epoch": 0.43512162800938925,
-      "grad_norm": 486.3786926269531,
-      "learning_rate": 3.5104599007701054e-05,
-      "loss": 52.94,
-      "step": 107700
-    },
-    {
-      "epoch": 0.4351620292747569,
-      "grad_norm": 1440.493896484375,
-      "learning_rate": 3.510140608166251e-05,
-      "loss": 44.7074,
-      "step": 107710
-    },
-    {
-      "epoch": 0.43520243054012453,
-      "grad_norm": 665.9118041992188,
-      "learning_rate": 3.5098212958691854e-05,
-      "loss": 84.7358,
-      "step": 107720
-    },
-    {
-      "epoch": 0.43524283180549217,
-      "grad_norm": 670.6646118164062,
-      "learning_rate": 3.509501963885134e-05,
-      "loss": 47.459,
-      "step": 107730
-    },
-    {
-      "epoch": 0.43528323307085975,
-      "grad_norm": 1127.4244384765625,
-      "learning_rate": 3.509182612220322e-05,
-      "loss": 66.771,
-      "step": 107740
-    },
-    {
-      "epoch": 0.4353236343362274,
-      "grad_norm": 1750.5572509765625,
-      "learning_rate": 3.5088632408809755e-05,
-      "loss": 100.8368,
-      "step": 107750
-    },
-    {
-      "epoch": 0.43536403560159503,
-      "grad_norm": 694.012451171875,
-      "learning_rate": 3.50854384987332e-05,
-      "loss": 67.874,
-      "step": 107760
-    },
-    {
-      "epoch": 0.4354044368669627,
-      "grad_norm": 543.4619750976562,
-      "learning_rate": 3.508224439203583e-05,
-      "loss": 48.8384,
-      "step": 107770
-    },
-    {
-      "epoch": 0.4354448381323303,
-      "grad_norm": 1029.557373046875,
-      "learning_rate": 3.5079050088779926e-05,
-      "loss": 41.3493,
-      "step": 107780
-    },
-    {
-      "epoch": 0.43548523939769795,
-      "grad_norm": 404.4317321777344,
-      "learning_rate": 3.5075855589027746e-05,
-      "loss": 59.9128,
-      "step": 107790
-    },
-    {
-      "epoch": 0.4355256406630656,
-      "grad_norm": 459.72357177734375,
-      "learning_rate": 3.507266089284157e-05,
-      "loss": 49.0815,
-      "step": 107800
-    },
-    {
-      "epoch": 0.4355660419284332,
-      "grad_norm": 1144.3963623046875,
-      "learning_rate": 3.506946600028368e-05,
-      "loss": 93.688,
-      "step": 107810
-    },
-    {
-      "epoch": 0.4356064431938008,
-      "grad_norm": 1337.915771484375,
-      "learning_rate": 3.5066270911416373e-05,
-      "loss": 41.8152,
-      "step": 107820
-    },
-    {
-      "epoch": 0.43564684445916846,
-      "grad_norm": 537.2900390625,
-      "learning_rate": 3.506307562630194e-05,
-      "loss": 33.384,
-      "step": 107830
-    },
-    {
-      "epoch": 0.4356872457245361,
-      "grad_norm": 676.332275390625,
-      "learning_rate": 3.5059880145002654e-05,
-      "loss": 72.9059,
-      "step": 107840
-    },
-    {
-      "epoch": 0.43572764698990374,
-      "grad_norm": 1162.9178466796875,
-      "learning_rate": 3.505668446758083e-05,
-      "loss": 46.9766,
-      "step": 107850
-    },
-    {
-      "epoch": 0.4357680482552714,
-      "grad_norm": 1080.3890380859375,
-      "learning_rate": 3.505348859409876e-05,
-      "loss": 53.5876,
-      "step": 107860
-    },
-    {
-      "epoch": 0.43580844952063896,
-      "grad_norm": 1019.6876220703125,
-      "learning_rate": 3.5050292524618764e-05,
-      "loss": 39.1764,
-      "step": 107870
-    },
-    {
-      "epoch": 0.4358488507860066,
-      "grad_norm": 521.6009521484375,
-      "learning_rate": 3.5047096259203135e-05,
-      "loss": 69.4503,
-      "step": 107880
-    },
-    {
-      "epoch": 0.43588925205137424,
-      "grad_norm": 1152.7193603515625,
-      "learning_rate": 3.5043899797914187e-05,
-      "loss": 74.8687,
-      "step": 107890
-    },
-    {
-      "epoch": 0.4359296533167419,
-      "grad_norm": 993.6572265625,
-      "learning_rate": 3.504070314081425e-05,
-      "loss": 64.1491,
-      "step": 107900
-    },
-    {
-      "epoch": 0.4359700545821095,
-      "grad_norm": 3389.17578125,
-      "learning_rate": 3.503750628796563e-05,
-      "loss": 62.572,
-      "step": 107910
-    },
-    {
-      "epoch": 0.43601045584747716,
-      "grad_norm": 1042.504150390625,
-      "learning_rate": 3.503430923943066e-05,
-      "loss": 51.6757,
-      "step": 107920
-    },
-    {
-      "epoch": 0.43605085711284475,
-      "grad_norm": 1820.093505859375,
-      "learning_rate": 3.503111199527167e-05,
-      "loss": 46.9761,
-      "step": 107930
-    },
-    {
-      "epoch": 0.4360912583782124,
-      "grad_norm": 530.4893798828125,
-      "learning_rate": 3.5027914555550976e-05,
-      "loss": 80.2376,
-      "step": 107940
-    },
-    {
-      "epoch": 0.43613165964358,
-      "grad_norm": 673.11767578125,
-      "learning_rate": 3.502471692033094e-05,
-      "loss": 36.759,
-      "step": 107950
-    },
-    {
-      "epoch": 0.43617206090894767,
-      "grad_norm": 979.2339477539062,
-      "learning_rate": 3.5021519089673876e-05,
-      "loss": 53.1278,
-      "step": 107960
-    },
-    {
-      "epoch": 0.4362124621743153,
-      "grad_norm": 533.0077514648438,
-      "learning_rate": 3.501832106364213e-05,
-      "loss": 44.0872,
-      "step": 107970
-    },
-    {
-      "epoch": 0.43625286343968295,
-      "grad_norm": 2183.581298828125,
-      "learning_rate": 3.501512284229807e-05,
-      "loss": 74.3517,
-      "step": 107980
-    },
-    {
-      "epoch": 0.4362932647050506,
-      "grad_norm": 1532.984375,
-      "learning_rate": 3.5011924425704036e-05,
-      "loss": 55.7305,
-      "step": 107990
-    },
-    {
-      "epoch": 0.43633366597041817,
-      "grad_norm": 794.1177978515625,
-      "learning_rate": 3.5008725813922386e-05,
-      "loss": 39.9372,
-      "step": 108000
-    },
-    {
-      "epoch": 0.4363740672357858,
-      "grad_norm": 489.5924072265625,
-      "learning_rate": 3.5005527007015455e-05,
-      "loss": 51.468,
-      "step": 108010
-    },
-    {
-      "epoch": 0.43641446850115345,
-      "grad_norm": 1078.9996337890625,
-      "learning_rate": 3.500232800504563e-05,
-      "loss": 69.4806,
-      "step": 108020
-    },
-    {
-      "epoch": 0.4364548697665211,
-      "grad_norm": 880.2926025390625,
-      "learning_rate": 3.499912880807528e-05,
-      "loss": 53.5359,
-      "step": 108030
-    },
-    {
-      "epoch": 0.43649527103188873,
-      "grad_norm": 827.0285034179688,
-      "learning_rate": 3.4995929416166756e-05,
-      "loss": 50.6137,
-      "step": 108040
-    },
-    {
-      "epoch": 0.43653567229725637,
-      "grad_norm": 720.4219970703125,
-      "learning_rate": 3.499272982938244e-05,
-      "loss": 35.6677,
-      "step": 108050
-    },
-    {
-      "epoch": 0.43657607356262396,
-      "grad_norm": 739.00927734375,
-      "learning_rate": 3.4989530047784716e-05,
-      "loss": 70.8884,
-      "step": 108060
-    },
-    {
-      "epoch": 0.4366164748279916,
-      "grad_norm": 387.0387878417969,
-      "learning_rate": 3.498633007143596e-05,
-      "loss": 47.103,
-      "step": 108070
-    },
-    {
-      "epoch": 0.43665687609335924,
-      "grad_norm": 912.795654296875,
-      "learning_rate": 3.498312990039856e-05,
-      "loss": 56.658,
-      "step": 108080
-    },
-    {
-      "epoch": 0.4366972773587269,
-      "grad_norm": 1193.864990234375,
-      "learning_rate": 3.497992953473491e-05,
-      "loss": 77.9862,
-      "step": 108090
-    },
-    {
-      "epoch": 0.4367376786240945,
-      "grad_norm": 711.3486938476562,
-      "learning_rate": 3.4976728974507384e-05,
-      "loss": 31.2605,
-      "step": 108100
-    },
-    {
-      "epoch": 0.43677807988946216,
-      "grad_norm": 1228.09033203125,
-      "learning_rate": 3.497352821977839e-05,
-      "loss": 42.339,
-      "step": 108110
-    },
-    {
-      "epoch": 0.4368184811548298,
-      "grad_norm": 524.43798828125,
-      "learning_rate": 3.497032727061034e-05,
-      "loss": 59.911,
-      "step": 108120
-    },
-    {
-      "epoch": 0.4368588824201974,
-      "grad_norm": 255.4713134765625,
-      "learning_rate": 3.496712612706561e-05,
-      "loss": 63.1904,
-      "step": 108130
-    },
-    {
-      "epoch": 0.436899283685565,
-      "grad_norm": 261.4523620605469,
-      "learning_rate": 3.4963924789206636e-05,
-      "loss": 88.5668,
-      "step": 108140
-    },
-    {
-      "epoch": 0.43693968495093266,
-      "grad_norm": 680.0599365234375,
-      "learning_rate": 3.496072325709582e-05,
-      "loss": 45.9959,
-      "step": 108150
-    },
-    {
-      "epoch": 0.4369800862163003,
-      "grad_norm": 402.27239990234375,
-      "learning_rate": 3.495752153079557e-05,
-      "loss": 50.0955,
-      "step": 108160
-    },
-    {
-      "epoch": 0.43702048748166794,
-      "grad_norm": 0.0,
-      "learning_rate": 3.495431961036832e-05,
-      "loss": 44.4594,
-      "step": 108170
-    },
-    {
-      "epoch": 0.4370608887470356,
-      "grad_norm": 455.03326416015625,
-      "learning_rate": 3.495111749587647e-05,
-      "loss": 51.1414,
-      "step": 108180
-    },
-    {
-      "epoch": 0.43710129001240317,
-      "grad_norm": 438.03314208984375,
-      "learning_rate": 3.494791518738247e-05,
-      "loss": 112.6708,
-      "step": 108190
-    },
-    {
-      "epoch": 0.4371416912777708,
-      "grad_norm": 547.6517333984375,
-      "learning_rate": 3.494471268494875e-05,
-      "loss": 31.8104,
-      "step": 108200
-    },
-    {
-      "epoch": 0.43718209254313845,
-      "grad_norm": 541.253173828125,
-      "learning_rate": 3.494150998863772e-05,
-      "loss": 63.8281,
-      "step": 108210
-    },
-    {
-      "epoch": 0.4372224938085061,
-      "grad_norm": 710.3536376953125,
-      "learning_rate": 3.4938307098511846e-05,
-      "loss": 43.2936,
-      "step": 108220
-    },
-    {
-      "epoch": 0.4372628950738737,
-      "grad_norm": 795.8914184570312,
-      "learning_rate": 3.493510401463355e-05,
-      "loss": 48.5519,
-      "step": 108230
-    },
-    {
-      "epoch": 0.43730329633924137,
-      "grad_norm": 1529.8331298828125,
-      "learning_rate": 3.493190073706529e-05,
-      "loss": 42.2938,
-      "step": 108240
-    },
-    {
-      "epoch": 0.43734369760460895,
-      "grad_norm": 1139.19189453125,
-      "learning_rate": 3.4928697265869515e-05,
-      "loss": 43.6591,
-      "step": 108250
-    },
-    {
-      "epoch": 0.4373840988699766,
-      "grad_norm": 427.7257080078125,
-      "learning_rate": 3.492549360110868e-05,
-      "loss": 126.2329,
-      "step": 108260
-    },
-    {
-      "epoch": 0.43742450013534423,
-      "grad_norm": 489.84521484375,
-      "learning_rate": 3.4922289742845224e-05,
-      "loss": 51.3288,
-      "step": 108270
-    },
-    {
-      "epoch": 0.43746490140071187,
-      "grad_norm": 0.0,
-      "learning_rate": 3.491908569114164e-05,
-      "loss": 57.9718,
-      "step": 108280
-    },
-    {
-      "epoch": 0.4375053026660795,
-      "grad_norm": 901.902099609375,
-      "learning_rate": 3.491588144606035e-05,
-      "loss": 58.3395,
-      "step": 108290
-    },
-    {
-      "epoch": 0.43754570393144715,
-      "grad_norm": 549.3409423828125,
-      "learning_rate": 3.491267700766386e-05,
-      "loss": 82.0675,
-      "step": 108300
-    },
-    {
-      "epoch": 0.4375861051968148,
-      "grad_norm": 490.4176025390625,
-      "learning_rate": 3.490947237601462e-05,
-      "loss": 52.1806,
-      "step": 108310
-    },
-    {
-      "epoch": 0.4376265064621824,
-      "grad_norm": 650.8810424804688,
-      "learning_rate": 3.4906267551175124e-05,
-      "loss": 76.2948,
-      "step": 108320
-    },
-    {
-      "epoch": 0.43766690772755,
-      "grad_norm": 959.3955688476562,
-      "learning_rate": 3.4903062533207834e-05,
-      "loss": 76.9859,
-      "step": 108330
-    },
-    {
-      "epoch": 0.43770730899291765,
-      "grad_norm": 609.113037109375,
-      "learning_rate": 3.489985732217525e-05,
-      "loss": 34.4562,
-      "step": 108340
-    },
-    {
-      "epoch": 0.4377477102582853,
-      "grad_norm": 2289.5068359375,
-      "learning_rate": 3.4896651918139845e-05,
-      "loss": 71.8435,
-      "step": 108350
-    },
-    {
-      "epoch": 0.43778811152365293,
-      "grad_norm": 0.0,
-      "learning_rate": 3.489344632116412e-05,
-      "loss": 54.8811,
-      "step": 108360
-    },
-    {
-      "epoch": 0.4378285127890206,
-      "grad_norm": 416.5489196777344,
-      "learning_rate": 3.489024053131056e-05,
-      "loss": 54.7843,
-      "step": 108370
-    },
-    {
-      "epoch": 0.43786891405438816,
-      "grad_norm": 850.0728149414062,
-      "learning_rate": 3.488703454864167e-05,
-      "loss": 83.0363,
-      "step": 108380
-    },
-    {
-      "epoch": 0.4379093153197558,
-      "grad_norm": 580.010498046875,
-      "learning_rate": 3.488382837321995e-05,
-      "loss": 45.165,
-      "step": 108390
-    },
-    {
-      "epoch": 0.43794971658512344,
-      "grad_norm": 684.89111328125,
-      "learning_rate": 3.488062200510791e-05,
-      "loss": 78.0684,
-      "step": 108400
-    },
-    {
-      "epoch": 0.4379901178504911,
-      "grad_norm": 1013.9144287109375,
-      "learning_rate": 3.487741544436806e-05,
-      "loss": 63.4015,
-      "step": 108410
-    },
-    {
-      "epoch": 0.4380305191158587,
-      "grad_norm": 758.853759765625,
-      "learning_rate": 3.48742086910629e-05,
-      "loss": 71.6265,
-      "step": 108420
-    },
-    {
-      "epoch": 0.43807092038122636,
-      "grad_norm": 398.92095947265625,
-      "learning_rate": 3.487100174525498e-05,
-      "loss": 43.4276,
-      "step": 108430
-    },
-    {
-      "epoch": 0.438111321646594,
-      "grad_norm": 313.7893371582031,
-      "learning_rate": 3.4867794607006784e-05,
-      "loss": 48.0121,
-      "step": 108440
-    },
-    {
-      "epoch": 0.4381517229119616,
-      "grad_norm": 1349.3988037109375,
-      "learning_rate": 3.486458727638085e-05,
-      "loss": 70.9725,
-      "step": 108450
-    },
-    {
-      "epoch": 0.4381921241773292,
-      "grad_norm": 433.0930480957031,
-      "learning_rate": 3.486137975343971e-05,
-      "loss": 61.4865,
-      "step": 108460
-    },
-    {
-      "epoch": 0.43823252544269686,
-      "grad_norm": 1083.5059814453125,
-      "learning_rate": 3.48581720382459e-05,
-      "loss": 54.0927,
-      "step": 108470
-    },
-    {
-      "epoch": 0.4382729267080645,
-      "grad_norm": 2715.4462890625,
-      "learning_rate": 3.485496413086195e-05,
-      "loss": 87.5101,
-      "step": 108480
-    },
-    {
-      "epoch": 0.43831332797343214,
-      "grad_norm": 566.393798828125,
-      "learning_rate": 3.4851756031350394e-05,
-      "loss": 56.3198,
-      "step": 108490
-    },
-    {
-      "epoch": 0.4383537292387998,
-      "grad_norm": 815.7106323242188,
-      "learning_rate": 3.484854773977378e-05,
-      "loss": 61.3737,
-      "step": 108500
-    },
-    {
-      "epoch": 0.43839413050416737,
-      "grad_norm": 807.414306640625,
-      "learning_rate": 3.4845339256194666e-05,
-      "loss": 44.4692,
-      "step": 108510
-    },
-    {
-      "epoch": 0.438434531769535,
-      "grad_norm": 984.805419921875,
-      "learning_rate": 3.484213058067559e-05,
-      "loss": 75.2083,
-      "step": 108520
-    },
-    {
-      "epoch": 0.43847493303490265,
-      "grad_norm": 1864.611083984375,
-      "learning_rate": 3.483892171327911e-05,
-      "loss": 86.4765,
-      "step": 108530
-    },
-    {
-      "epoch": 0.4385153343002703,
-      "grad_norm": 1487.40869140625,
-      "learning_rate": 3.4835712654067785e-05,
-      "loss": 76.399,
-      "step": 108540
-    },
-    {
-      "epoch": 0.4385557355656379,
-      "grad_norm": 735.3146362304688,
-      "learning_rate": 3.483250340310418e-05,
-      "loss": 70.6352,
-      "step": 108550
-    },
-    {
-      "epoch": 0.43859613683100557,
-      "grad_norm": 807.2081298828125,
-      "learning_rate": 3.482929396045087e-05,
-      "loss": 48.8148,
-      "step": 108560
-    },
-    {
-      "epoch": 0.43863653809637315,
-      "grad_norm": 632.7271118164062,
-      "learning_rate": 3.48260843261704e-05,
-      "loss": 77.7649,
-      "step": 108570
-    },
-    {
-      "epoch": 0.4386769393617408,
-      "grad_norm": 897.5272216796875,
-      "learning_rate": 3.482287450032536e-05,
-      "loss": 51.7576,
-      "step": 108580
-    },
-    {
-      "epoch": 0.43871734062710843,
-      "grad_norm": 753.651611328125,
-      "learning_rate": 3.4819664482978325e-05,
-      "loss": 71.9576,
-      "step": 108590
-    },
-    {
-      "epoch": 0.43875774189247607,
-      "grad_norm": 524.542724609375,
-      "learning_rate": 3.481645427419188e-05,
-      "loss": 51.1556,
-      "step": 108600
-    },
-    {
-      "epoch": 0.4387981431578437,
-      "grad_norm": 578.78173828125,
-      "learning_rate": 3.48132438740286e-05,
-      "loss": 42.8052,
-      "step": 108610
-    },
-    {
-      "epoch": 0.43883854442321135,
-      "grad_norm": 634.4146118164062,
-      "learning_rate": 3.481003328255108e-05,
-      "loss": 56.4694,
-      "step": 108620
-    },
-    {
-      "epoch": 0.438878945688579,
-      "grad_norm": 823.702392578125,
-      "learning_rate": 3.480682249982191e-05,
-      "loss": 51.0967,
-      "step": 108630
-    },
-    {
-      "epoch": 0.4389193469539466,
-      "grad_norm": 625.2899169921875,
-      "learning_rate": 3.4803611525903685e-05,
-      "loss": 55.0041,
-      "step": 108640
-    },
-    {
-      "epoch": 0.4389597482193142,
-      "grad_norm": 251.00901794433594,
-      "learning_rate": 3.480040036085901e-05,
-      "loss": 52.5765,
-      "step": 108650
-    },
-    {
-      "epoch": 0.43900014948468186,
-      "grad_norm": 1115.88134765625,
-      "learning_rate": 3.479718900475049e-05,
-      "loss": 83.0873,
-      "step": 108660
-    },
-    {
-      "epoch": 0.4390405507500495,
-      "grad_norm": 9209.01953125,
-      "learning_rate": 3.479397745764071e-05,
-      "loss": 142.5869,
-      "step": 108670
-    },
-    {
-      "epoch": 0.43908095201541714,
-      "grad_norm": 704.7980346679688,
-      "learning_rate": 3.479076571959231e-05,
-      "loss": 47.6946,
-      "step": 108680
-    },
-    {
-      "epoch": 0.4391213532807848,
-      "grad_norm": 1096.587890625,
-      "learning_rate": 3.4787553790667896e-05,
-      "loss": 57.3606,
-      "step": 108690
-    },
-    {
-      "epoch": 0.43916175454615236,
-      "grad_norm": 685.4246215820312,
-      "learning_rate": 3.4784341670930065e-05,
-      "loss": 52.4217,
-      "step": 108700
-    },
-    {
-      "epoch": 0.43920215581152,
-      "grad_norm": 385.501220703125,
-      "learning_rate": 3.478112936044146e-05,
-      "loss": 43.921,
-      "step": 108710
-    },
-    {
-      "epoch": 0.43924255707688764,
-      "grad_norm": 648.174560546875,
-      "learning_rate": 3.477791685926471e-05,
-      "loss": 48.4695,
-      "step": 108720
-    },
-    {
-      "epoch": 0.4392829583422553,
-      "grad_norm": 201.7768096923828,
-      "learning_rate": 3.4774704167462434e-05,
-      "loss": 63.3514,
-      "step": 108730
-    },
-    {
-      "epoch": 0.4393233596076229,
-      "grad_norm": 579.0665283203125,
-      "learning_rate": 3.477149128509727e-05,
-      "loss": 59.8261,
-      "step": 108740
-    },
-    {
-      "epoch": 0.43936376087299056,
-      "grad_norm": 727.160888671875,
-      "learning_rate": 3.476827821223184e-05,
-      "loss": 48.0612,
-      "step": 108750
-    },
-    {
-      "epoch": 0.4394041621383582,
-      "grad_norm": 842.6546020507812,
-      "learning_rate": 3.4765064948928814e-05,
-      "loss": 52.5195,
-      "step": 108760
-    },
-    {
-      "epoch": 0.4394445634037258,
-      "grad_norm": 680.0587158203125,
-      "learning_rate": 3.4761851495250816e-05,
-      "loss": 83.2483,
-      "step": 108770
-    },
-    {
-      "epoch": 0.4394849646690934,
-      "grad_norm": 946.002197265625,
-      "learning_rate": 3.475863785126049e-05,
-      "loss": 61.0262,
-      "step": 108780
-    },
-    {
-      "epoch": 0.43952536593446107,
-      "grad_norm": 1292.424072265625,
-      "learning_rate": 3.47554240170205e-05,
-      "loss": 117.6964,
-      "step": 108790
-    },
-    {
-      "epoch": 0.4395657671998287,
-      "grad_norm": 593.932861328125,
-      "learning_rate": 3.475220999259349e-05,
-      "loss": 60.331,
-      "step": 108800
-    },
-    {
-      "epoch": 0.43960616846519635,
-      "grad_norm": 740.8704833984375,
-      "learning_rate": 3.4748995778042136e-05,
-      "loss": 88.3325,
-      "step": 108810
-    },
-    {
-      "epoch": 0.439646569730564,
-      "grad_norm": 495.8048400878906,
-      "learning_rate": 3.474578137342909e-05,
-      "loss": 46.6653,
-      "step": 108820
-    },
-    {
-      "epoch": 0.43968697099593157,
-      "grad_norm": 907.8876342773438,
-      "learning_rate": 3.474256677881701e-05,
-      "loss": 65.5185,
-      "step": 108830
-    },
-    {
-      "epoch": 0.4397273722612992,
-      "grad_norm": 1308.182861328125,
-      "learning_rate": 3.473935199426858e-05,
-      "loss": 109.5993,
-      "step": 108840
-    },
-    {
-      "epoch": 0.43976777352666685,
-      "grad_norm": 538.4534912109375,
-      "learning_rate": 3.4736137019846465e-05,
-      "loss": 47.8979,
-      "step": 108850
-    },
-    {
-      "epoch": 0.4398081747920345,
-      "grad_norm": 302.4723815917969,
-      "learning_rate": 3.4732921855613355e-05,
-      "loss": 60.7093,
-      "step": 108860
-    },
-    {
-      "epoch": 0.43984857605740213,
-      "grad_norm": 1104.8883056640625,
-      "learning_rate": 3.472970650163191e-05,
-      "loss": 61.7136,
-      "step": 108870
-    },
-    {
-      "epoch": 0.43988897732276977,
-      "grad_norm": 1319.9013671875,
-      "learning_rate": 3.4726490957964834e-05,
-      "loss": 71.9735,
-      "step": 108880
-    },
-    {
-      "epoch": 0.43992937858813735,
-      "grad_norm": 3414.708251953125,
-      "learning_rate": 3.472327522467481e-05,
-      "loss": 74.4029,
-      "step": 108890
-    },
-    {
-      "epoch": 0.439969779853505,
-      "grad_norm": 442.96405029296875,
-      "learning_rate": 3.4720059301824525e-05,
-      "loss": 74.8008,
-      "step": 108900
-    },
-    {
-      "epoch": 0.44001018111887263,
-      "grad_norm": 849.4617919921875,
-      "learning_rate": 3.4716843189476687e-05,
-      "loss": 67.8839,
-      "step": 108910
-    },
-    {
-      "epoch": 0.4400505823842403,
-      "grad_norm": 464.7969970703125,
-      "learning_rate": 3.471362688769398e-05,
-      "loss": 55.6521,
-      "step": 108920
-    },
-    {
-      "epoch": 0.4400909836496079,
-      "grad_norm": 426.9241943359375,
-      "learning_rate": 3.471041039653913e-05,
-      "loss": 53.6479,
-      "step": 108930
-    },
-    {
-      "epoch": 0.44013138491497555,
-      "grad_norm": 926.24853515625,
-      "learning_rate": 3.4707193716074816e-05,
-      "loss": 58.2524,
-      "step": 108940
-    },
-    {
-      "epoch": 0.4401717861803432,
-      "grad_norm": 1493.707763671875,
-      "learning_rate": 3.470397684636377e-05,
-      "loss": 60.5845,
-      "step": 108950
-    },
-    {
-      "epoch": 0.4402121874457108,
-      "grad_norm": 715.8504638671875,
-      "learning_rate": 3.4700759787468695e-05,
-      "loss": 61.8466,
-      "step": 108960
-    },
-    {
-      "epoch": 0.4402525887110784,
-      "grad_norm": 539.120849609375,
-      "learning_rate": 3.469754253945232e-05,
-      "loss": 50.327,
-      "step": 108970
-    },
-    {
-      "epoch": 0.44029298997644606,
-      "grad_norm": 644.7427978515625,
-      "learning_rate": 3.4694325102377355e-05,
-      "loss": 74.642,
-      "step": 108980
-    },
-    {
-      "epoch": 0.4403333912418137,
-      "grad_norm": 702.2979736328125,
-      "learning_rate": 3.469110747630653e-05,
-      "loss": 56.1246,
-      "step": 108990
-    },
-    {
-      "epoch": 0.44037379250718134,
-      "grad_norm": 1032.9031982421875,
-      "learning_rate": 3.4687889661302576e-05,
-      "loss": 56.0517,
-      "step": 109000
-    },
-    {
-      "epoch": 0.440414193772549,
-      "grad_norm": 972.0687866210938,
-      "learning_rate": 3.468467165742823e-05,
-      "loss": 56.0222,
-      "step": 109010
-    },
-    {
-      "epoch": 0.44045459503791656,
-      "grad_norm": 598.9658203125,
-      "learning_rate": 3.468145346474622e-05,
-      "loss": 57.363,
-      "step": 109020
-    },
-    {
-      "epoch": 0.4404949963032842,
-      "grad_norm": 616.7764282226562,
-      "learning_rate": 3.4678235083319296e-05,
-      "loss": 82.0413,
-      "step": 109030
-    },
-    {
-      "epoch": 0.44053539756865184,
-      "grad_norm": 318.1781005859375,
-      "learning_rate": 3.467501651321019e-05,
-      "loss": 62.9576,
-      "step": 109040
-    },
-    {
-      "epoch": 0.4405757988340195,
-      "grad_norm": 718.73681640625,
-      "learning_rate": 3.467179775448166e-05,
-      "loss": 66.6093,
-      "step": 109050
-    },
-    {
-      "epoch": 0.4406162000993871,
-      "grad_norm": 973.098388671875,
-      "learning_rate": 3.466857880719645e-05,
-      "loss": 65.0112,
-      "step": 109060
-    },
-    {
-      "epoch": 0.44065660136475476,
-      "grad_norm": 1736.9281005859375,
-      "learning_rate": 3.466535967141732e-05,
-      "loss": 67.5672,
-      "step": 109070
-    },
-    {
-      "epoch": 0.4406970026301224,
-      "grad_norm": 1171.598388671875,
-      "learning_rate": 3.466214034720702e-05,
-      "loss": 62.2778,
-      "step": 109080
-    },
-    {
-      "epoch": 0.44073740389549,
-      "grad_norm": 981.1380004882812,
-      "learning_rate": 3.4658920834628335e-05,
-      "loss": 72.6023,
-      "step": 109090
-    },
-    {
-      "epoch": 0.4407778051608576,
-      "grad_norm": 858.4905395507812,
-      "learning_rate": 3.4655701133744e-05,
-      "loss": 43.7386,
-      "step": 109100
-    },
-    {
-      "epoch": 0.44081820642622527,
-      "grad_norm": 734.6194458007812,
-      "learning_rate": 3.465248124461681e-05,
-      "loss": 41.7183,
-      "step": 109110
-    },
-    {
-      "epoch": 0.4408586076915929,
-      "grad_norm": 2607.31787109375,
-      "learning_rate": 3.4649261167309526e-05,
-      "loss": 105.1528,
-      "step": 109120
-    },
-    {
-      "epoch": 0.44089900895696055,
-      "grad_norm": 675.8436279296875,
-      "learning_rate": 3.464604090188493e-05,
-      "loss": 75.2605,
-      "step": 109130
-    },
-    {
-      "epoch": 0.4409394102223282,
-      "grad_norm": 472.8738098144531,
-      "learning_rate": 3.46428204484058e-05,
-      "loss": 50.3132,
-      "step": 109140
-    },
-    {
-      "epoch": 0.44097981148769577,
-      "grad_norm": 629.1958618164062,
-      "learning_rate": 3.463959980693492e-05,
-      "loss": 47.585,
-      "step": 109150
-    },
-    {
-      "epoch": 0.4410202127530634,
-      "grad_norm": 714.365478515625,
-      "learning_rate": 3.4636378977535075e-05,
-      "loss": 62.9666,
-      "step": 109160
-    },
-    {
-      "epoch": 0.44106061401843105,
-      "grad_norm": 1306.74169921875,
-      "learning_rate": 3.4633157960269056e-05,
-      "loss": 67.3224,
-      "step": 109170
-    },
-    {
-      "epoch": 0.4411010152837987,
-      "grad_norm": 810.0322265625,
-      "learning_rate": 3.462993675519968e-05,
-      "loss": 45.5099,
-      "step": 109180
-    },
-    {
-      "epoch": 0.44114141654916633,
-      "grad_norm": 822.7820434570312,
-      "learning_rate": 3.462671536238972e-05,
-      "loss": 47.5919,
-      "step": 109190
-    },
-    {
-      "epoch": 0.44118181781453397,
-      "grad_norm": 719.748046875,
-      "learning_rate": 3.462349378190199e-05,
-      "loss": 58.4106,
-      "step": 109200
-    },
-    {
-      "epoch": 0.44122221907990156,
-      "grad_norm": 1387.738525390625,
-      "learning_rate": 3.4620272013799286e-05,
-      "loss": 55.8598,
-      "step": 109210
-    },
-    {
-      "epoch": 0.4412626203452692,
-      "grad_norm": 291.135009765625,
-      "learning_rate": 3.461705005814444e-05,
-      "loss": 56.4821,
-      "step": 109220
-    },
-    {
-      "epoch": 0.44130302161063684,
-      "grad_norm": 2201.89453125,
-      "learning_rate": 3.4613827915000244e-05,
-      "loss": 112.0971,
-      "step": 109230
-    },
-    {
-      "epoch": 0.4413434228760045,
-      "grad_norm": 828.1795043945312,
-      "learning_rate": 3.461060558442952e-05,
-      "loss": 52.2993,
-      "step": 109240
-    },
-    {
-      "epoch": 0.4413838241413721,
-      "grad_norm": 493.1712341308594,
-      "learning_rate": 3.460738306649509e-05,
-      "loss": 38.8712,
-      "step": 109250
-    },
-    {
-      "epoch": 0.44142422540673976,
-      "grad_norm": 310.04742431640625,
-      "learning_rate": 3.4604160361259796e-05,
-      "loss": 54.1063,
-      "step": 109260
-    },
-    {
-      "epoch": 0.4414646266721074,
-      "grad_norm": 1105.275634765625,
-      "learning_rate": 3.460093746878644e-05,
-      "loss": 57.73,
-      "step": 109270
-    },
-    {
-      "epoch": 0.441505027937475,
-      "grad_norm": 422.4311828613281,
-      "learning_rate": 3.459771438913787e-05,
-      "loss": 76.8583,
-      "step": 109280
-    },
-    {
-      "epoch": 0.4415454292028426,
-      "grad_norm": 533.1647338867188,
-      "learning_rate": 3.459449112237691e-05,
-      "loss": 70.399,
-      "step": 109290
-    },
-    {
-      "epoch": 0.44158583046821026,
-      "grad_norm": 453.3349304199219,
-      "learning_rate": 3.459126766856641e-05,
-      "loss": 47.0437,
-      "step": 109300
-    },
-    {
-      "epoch": 0.4416262317335779,
-      "grad_norm": 868.0079345703125,
-      "learning_rate": 3.458804402776921e-05,
-      "loss": 49.0908,
-      "step": 109310
-    },
-    {
-      "epoch": 0.44166663299894554,
-      "grad_norm": 986.170166015625,
-      "learning_rate": 3.458482020004815e-05,
-      "loss": 54.6284,
-      "step": 109320
-    },
-    {
-      "epoch": 0.4417070342643132,
-      "grad_norm": 447.4167785644531,
-      "learning_rate": 3.4581596185466094e-05,
-      "loss": 59.609,
-      "step": 109330
-    },
-    {
-      "epoch": 0.44174743552968077,
-      "grad_norm": 641.2802124023438,
-      "learning_rate": 3.457837198408588e-05,
-      "loss": 85.7572,
-      "step": 109340
-    },
-    {
-      "epoch": 0.4417878367950484,
-      "grad_norm": 808.4306640625,
-      "learning_rate": 3.457514759597038e-05,
-      "loss": 64.826,
-      "step": 109350
-    },
-    {
-      "epoch": 0.44182823806041605,
-      "grad_norm": 202.7559051513672,
-      "learning_rate": 3.457192302118244e-05,
-      "loss": 52.6859,
-      "step": 109360
-    },
-    {
-      "epoch": 0.4418686393257837,
-      "grad_norm": 1533.5028076171875,
-      "learning_rate": 3.4568698259784945e-05,
-      "loss": 71.5681,
-      "step": 109370
-    },
-    {
-      "epoch": 0.4419090405911513,
-      "grad_norm": 788.40771484375,
-      "learning_rate": 3.4565473311840735e-05,
-      "loss": 56.7714,
-      "step": 109380
-    },
-    {
-      "epoch": 0.44194944185651897,
-      "grad_norm": 659.1009521484375,
-      "learning_rate": 3.4562248177412715e-05,
-      "loss": 44.7925,
-      "step": 109390
-    },
-    {
-      "epoch": 0.44198984312188655,
-      "grad_norm": 695.70068359375,
-      "learning_rate": 3.455902285656373e-05,
-      "loss": 95.6126,
-      "step": 109400
-    },
-    {
-      "epoch": 0.4420302443872542,
-      "grad_norm": 536.9435424804688,
-      "learning_rate": 3.4555797349356676e-05,
-      "loss": 52.6949,
-      "step": 109410
-    },
-    {
-      "epoch": 0.44207064565262183,
-      "grad_norm": 928.5408935546875,
-      "learning_rate": 3.455257165585444e-05,
-      "loss": 65.1302,
-      "step": 109420
-    },
-    {
-      "epoch": 0.44211104691798947,
-      "grad_norm": 635.9898071289062,
-      "learning_rate": 3.454934577611989e-05,
-      "loss": 88.1777,
-      "step": 109430
-    },
-    {
-      "epoch": 0.4421514481833571,
-      "grad_norm": 1163.3731689453125,
-      "learning_rate": 3.454611971021593e-05,
-      "loss": 75.0465,
-      "step": 109440
-    },
-    {
-      "epoch": 0.44219184944872475,
-      "grad_norm": 1179.76708984375,
-      "learning_rate": 3.454289345820546e-05,
-      "loss": 72.196,
-      "step": 109450
-    },
-    {
-      "epoch": 0.4422322507140924,
-      "grad_norm": 407.4288024902344,
-      "learning_rate": 3.453966702015137e-05,
-      "loss": 37.4781,
-      "step": 109460
-    },
-    {
-      "epoch": 0.44227265197946,
-      "grad_norm": 1263.00732421875,
-      "learning_rate": 3.453644039611656e-05,
-      "loss": 55.1117,
-      "step": 109470
-    },
-    {
-      "epoch": 0.4423130532448276,
-      "grad_norm": 441.8350830078125,
-      "learning_rate": 3.453321358616393e-05,
-      "loss": 34.3389,
-      "step": 109480
-    },
-    {
-      "epoch": 0.44235345451019525,
-      "grad_norm": 813.268798828125,
-      "learning_rate": 3.452998659035639e-05,
-      "loss": 61.2214,
-      "step": 109490
-    },
-    {
-      "epoch": 0.4423938557755629,
-      "grad_norm": 1138.2327880859375,
-      "learning_rate": 3.452675940875686e-05,
-      "loss": 69.0981,
-      "step": 109500
-    },
-    {
-      "epoch": 0.44243425704093053,
-      "grad_norm": 1215.769287109375,
-      "learning_rate": 3.452353204142824e-05,
-      "loss": 52.1015,
-      "step": 109510
-    },
-    {
-      "epoch": 0.4424746583062982,
-      "grad_norm": 169.15260314941406,
-      "learning_rate": 3.452030448843347e-05,
-      "loss": 49.8006,
-      "step": 109520
-    },
-    {
-      "epoch": 0.44251505957166576,
-      "grad_norm": 372.5907287597656,
-      "learning_rate": 3.451707674983546e-05,
-      "loss": 67.884,
-      "step": 109530
-    },
-    {
-      "epoch": 0.4425554608370334,
-      "grad_norm": 1219.9879150390625,
-      "learning_rate": 3.451384882569714e-05,
-      "loss": 48.5552,
-      "step": 109540
-    },
-    {
-      "epoch": 0.44259586210240104,
-      "grad_norm": 1142.11865234375,
-      "learning_rate": 3.4510620716081446e-05,
-      "loss": 50.2349,
-      "step": 109550
-    },
-    {
-      "epoch": 0.4426362633677687,
-      "grad_norm": 1006.7579956054688,
-      "learning_rate": 3.45073924210513e-05,
-      "loss": 100.4007,
-      "step": 109560
-    },
-    {
-      "epoch": 0.4426766646331363,
-      "grad_norm": 714.3269653320312,
-      "learning_rate": 3.4504163940669634e-05,
-      "loss": 65.3141,
-      "step": 109570
-    },
-    {
-      "epoch": 0.44271706589850396,
-      "grad_norm": 1006.07177734375,
-      "learning_rate": 3.4500935274999413e-05,
-      "loss": 98.7038,
-      "step": 109580
-    },
-    {
-      "epoch": 0.4427574671638716,
-      "grad_norm": 1100.7425537109375,
-      "learning_rate": 3.449770642410356e-05,
-      "loss": 58.6847,
-      "step": 109590
-    },
-    {
-      "epoch": 0.4427978684292392,
-      "grad_norm": 1185.401123046875,
-      "learning_rate": 3.4494477388045035e-05,
-      "loss": 56.3779,
-      "step": 109600
-    },
-    {
-      "epoch": 0.4428382696946068,
-      "grad_norm": 1830.9031982421875,
-      "learning_rate": 3.449124816688677e-05,
-      "loss": 71.803,
-      "step": 109610
-    },
-    {
-      "epoch": 0.44287867095997446,
-      "grad_norm": 639.5352172851562,
-      "learning_rate": 3.448801876069176e-05,
-      "loss": 59.7502,
-      "step": 109620
-    },
-    {
-      "epoch": 0.4429190722253421,
-      "grad_norm": 818.8923950195312,
-      "learning_rate": 3.4484789169522927e-05,
-      "loss": 65.0993,
-      "step": 109630
-    },
-    {
-      "epoch": 0.44295947349070974,
-      "grad_norm": 520.5197143554688,
-      "learning_rate": 3.448155939344324e-05,
-      "loss": 58.1046,
-      "step": 109640
-    },
-    {
-      "epoch": 0.4429998747560774,
-      "grad_norm": 2226.27978515625,
-      "learning_rate": 3.4478329432515674e-05,
-      "loss": 49.6613,
-      "step": 109650
-    },
-    {
-      "epoch": 0.44304027602144497,
-      "grad_norm": 3536.548095703125,
-      "learning_rate": 3.44750992868032e-05,
-      "loss": 60.4391,
-      "step": 109660
-    },
-    {
-      "epoch": 0.4430806772868126,
-      "grad_norm": 922.1171875,
-      "learning_rate": 3.447186895636879e-05,
-      "loss": 49.5081,
-      "step": 109670
-    },
-    {
-      "epoch": 0.44312107855218025,
-      "grad_norm": 485.9327697753906,
-      "learning_rate": 3.4468638441275415e-05,
-      "loss": 52.8482,
-      "step": 109680
-    },
-    {
-      "epoch": 0.4431614798175479,
-      "grad_norm": 1202.439208984375,
-      "learning_rate": 3.4465407741586056e-05,
-      "loss": 51.3917,
-      "step": 109690
-    },
-    {
-      "epoch": 0.4432018810829155,
-      "grad_norm": 3666.6962890625,
-      "learning_rate": 3.4462176857363704e-05,
-      "loss": 50.0755,
-      "step": 109700
-    },
-    {
-      "epoch": 0.44324228234828317,
-      "grad_norm": 566.9150390625,
-      "learning_rate": 3.445894578867134e-05,
-      "loss": 77.9221,
-      "step": 109710
-    },
-    {
-      "epoch": 0.44328268361365075,
-      "grad_norm": 695.4548950195312,
-      "learning_rate": 3.445571453557196e-05,
-      "loss": 92.9399,
-      "step": 109720
-    },
-    {
-      "epoch": 0.4433230848790184,
-      "grad_norm": 1101.68212890625,
-      "learning_rate": 3.445248309812856e-05,
-      "loss": 67.3483,
-      "step": 109730
-    },
-    {
-      "epoch": 0.44336348614438603,
-      "grad_norm": 782.0620727539062,
-      "learning_rate": 3.4449251476404135e-05,
-      "loss": 47.7574,
-      "step": 109740
-    },
-    {
-      "epoch": 0.44340388740975367,
-      "grad_norm": 404.2615661621094,
-      "learning_rate": 3.444601967046168e-05,
-      "loss": 68.9839,
-      "step": 109750
-    },
-    {
-      "epoch": 0.4434442886751213,
-      "grad_norm": 1344.5721435546875,
-      "learning_rate": 3.444278768036421e-05,
-      "loss": 51.2115,
-      "step": 109760
-    },
-    {
-      "epoch": 0.44348468994048895,
-      "grad_norm": 453.09136962890625,
-      "learning_rate": 3.443955550617474e-05,
-      "loss": 62.8084,
-      "step": 109770
-    },
-    {
-      "epoch": 0.4435250912058566,
-      "grad_norm": 690.1050415039062,
-      "learning_rate": 3.443632314795627e-05,
-      "loss": 37.6234,
-      "step": 109780
-    },
-    {
-      "epoch": 0.4435654924712242,
-      "grad_norm": 942.2594604492188,
-      "learning_rate": 3.443309060577182e-05,
-      "loss": 57.9371,
-      "step": 109790
-    },
-    {
-      "epoch": 0.4436058937365918,
-      "grad_norm": 2099.522216796875,
-      "learning_rate": 3.442985787968442e-05,
-      "loss": 96.0077,
-      "step": 109800
-    },
-    {
-      "epoch": 0.44364629500195946,
-      "grad_norm": 894.1741333007812,
-      "learning_rate": 3.4426624969757083e-05,
-      "loss": 43.2961,
-      "step": 109810
-    },
-    {
-      "epoch": 0.4436866962673271,
-      "grad_norm": 515.2462158203125,
-      "learning_rate": 3.442339187605283e-05,
-      "loss": 44.3043,
-      "step": 109820
-    },
-    {
-      "epoch": 0.44372709753269474,
-      "grad_norm": 1050.0601806640625,
-      "learning_rate": 3.442015859863472e-05,
-      "loss": 80.5312,
-      "step": 109830
-    },
-    {
-      "epoch": 0.4437674987980624,
-      "grad_norm": 2025.021484375,
-      "learning_rate": 3.4416925137565754e-05,
-      "loss": 60.7582,
-      "step": 109840
-    },
-    {
-      "epoch": 0.44380790006342996,
-      "grad_norm": 1199.0517578125,
-      "learning_rate": 3.4413691492908985e-05,
-      "loss": 62.2759,
-      "step": 109850
-    },
-    {
-      "epoch": 0.4438483013287976,
-      "grad_norm": 781.955078125,
-      "learning_rate": 3.441045766472745e-05,
-      "loss": 55.6269,
-      "step": 109860
-    },
-    {
-      "epoch": 0.44388870259416524,
-      "grad_norm": 1648.44140625,
-      "learning_rate": 3.440722365308421e-05,
-      "loss": 57.5258,
-      "step": 109870
-    },
-    {
-      "epoch": 0.4439291038595329,
-      "grad_norm": 1811.8358154296875,
-      "learning_rate": 3.440398945804229e-05,
-      "loss": 78.8444,
-      "step": 109880
-    },
-    {
-      "epoch": 0.4439695051249005,
-      "grad_norm": 678.7551879882812,
-      "learning_rate": 3.440075507966476e-05,
-      "loss": 58.5628,
-      "step": 109890
-    },
-    {
-      "epoch": 0.44400990639026816,
-      "grad_norm": 1020.2061157226562,
-      "learning_rate": 3.439752051801467e-05,
-      "loss": 41.0844,
-      "step": 109900
-    },
-    {
-      "epoch": 0.4440503076556358,
-      "grad_norm": 1242.7587890625,
-      "learning_rate": 3.439428577315508e-05,
-      "loss": 50.6694,
-      "step": 109910
-    },
-    {
-      "epoch": 0.4440907089210034,
-      "grad_norm": 1669.455810546875,
-      "learning_rate": 3.439105084514905e-05,
-      "loss": 61.132,
-      "step": 109920
-    },
-    {
-      "epoch": 0.444131110186371,
-      "grad_norm": 357.2478942871094,
-      "learning_rate": 3.4387815734059654e-05,
-      "loss": 64.2863,
-      "step": 109930
-    },
-    {
-      "epoch": 0.44417151145173867,
-      "grad_norm": 696.1216430664062,
-      "learning_rate": 3.438458043994995e-05,
-      "loss": 52.3681,
-      "step": 109940
-    },
-    {
-      "epoch": 0.4442119127171063,
-      "grad_norm": 1389.9837646484375,
-      "learning_rate": 3.438134496288302e-05,
-      "loss": 59.619,
-      "step": 109950
-    },
-    {
-      "epoch": 0.44425231398247395,
-      "grad_norm": 819.2481689453125,
-      "learning_rate": 3.437810930292195e-05,
-      "loss": 71.3323,
-      "step": 109960
-    },
-    {
-      "epoch": 0.4442927152478416,
-      "grad_norm": 541.750732421875,
-      "learning_rate": 3.43748734601298e-05,
-      "loss": 73.0371,
-      "step": 109970
-    },
-    {
-      "epoch": 0.44433311651320917,
-      "grad_norm": 923.3477172851562,
-      "learning_rate": 3.437163743456967e-05,
-      "loss": 50.583,
-      "step": 109980
-    },
-    {
-      "epoch": 0.4443735177785768,
-      "grad_norm": 1519.56005859375,
-      "learning_rate": 3.436840122630464e-05,
-      "loss": 63.0997,
-      "step": 109990
-    },
-    {
-      "epoch": 0.44441391904394445,
-      "grad_norm": 1023.0885620117188,
-      "learning_rate": 3.436516483539781e-05,
-      "loss": 35.0977,
-      "step": 110000
-    },
-    {
-      "epoch": 0.4444543203093121,
-      "grad_norm": 787.899658203125,
-      "learning_rate": 3.4361928261912254e-05,
-      "loss": 56.4767,
-      "step": 110010
-    },
-    {
-      "epoch": 0.44449472157467973,
-      "grad_norm": 0.0,
-      "learning_rate": 3.4358691505911104e-05,
-      "loss": 44.5502,
-      "step": 110020
-    },
-    {
-      "epoch": 0.44453512284004737,
-      "grad_norm": 385.0424499511719,
-      "learning_rate": 3.4355454567457445e-05,
-      "loss": 70.914,
-      "step": 110030
-    },
-    {
-      "epoch": 0.44457552410541495,
-      "grad_norm": 967.9429931640625,
-      "learning_rate": 3.435221744661438e-05,
-      "loss": 62.6088,
-      "step": 110040
-    },
-    {
-      "epoch": 0.4446159253707826,
-      "grad_norm": 389.7696533203125,
-      "learning_rate": 3.434898014344501e-05,
-      "loss": 48.413,
-      "step": 110050
-    },
-    {
-      "epoch": 0.44465632663615023,
-      "grad_norm": 1513.93017578125,
-      "learning_rate": 3.434574265801247e-05,
-      "loss": 50.9386,
-      "step": 110060
-    },
-    {
-      "epoch": 0.4446967279015179,
-      "grad_norm": 461.18096923828125,
-      "learning_rate": 3.4342504990379866e-05,
-      "loss": 66.7544,
-      "step": 110070
-    },
-    {
-      "epoch": 0.4447371291668855,
-      "grad_norm": 593.7630615234375,
-      "learning_rate": 3.433926714061032e-05,
-      "loss": 60.6121,
-      "step": 110080
-    },
-    {
-      "epoch": 0.44477753043225315,
-      "grad_norm": 1564.3096923828125,
-      "learning_rate": 3.433602910876694e-05,
-      "loss": 70.2792,
-      "step": 110090
-    },
-    {
-      "epoch": 0.4448179316976208,
-      "grad_norm": 1443.179931640625,
-      "learning_rate": 3.433279089491288e-05,
-      "loss": 36.5048,
-      "step": 110100
-    },
-    {
-      "epoch": 0.4448583329629884,
-      "grad_norm": 822.4376220703125,
-      "learning_rate": 3.432955249911125e-05,
-      "loss": 78.1726,
-      "step": 110110
-    },
-    {
-      "epoch": 0.444898734228356,
-      "grad_norm": 1508.205078125,
-      "learning_rate": 3.432631392142519e-05,
-      "loss": 66.1732,
-      "step": 110120
-    },
-    {
-      "epoch": 0.44493913549372366,
-      "grad_norm": 251.31320190429688,
-      "learning_rate": 3.432307516191783e-05,
-      "loss": 34.4932,
-      "step": 110130
-    },
-    {
-      "epoch": 0.4449795367590913,
-      "grad_norm": 1120.1636962890625,
-      "learning_rate": 3.4319836220652335e-05,
-      "loss": 38.0345,
-      "step": 110140
-    },
-    {
-      "epoch": 0.44501993802445894,
-      "grad_norm": 439.1837158203125,
-      "learning_rate": 3.431659709769183e-05,
-      "loss": 56.9619,
-      "step": 110150
-    },
-    {
-      "epoch": 0.4450603392898266,
-      "grad_norm": 969.8883666992188,
-      "learning_rate": 3.431335779309947e-05,
-      "loss": 37.3155,
-      "step": 110160
-    },
-    {
-      "epoch": 0.44510074055519416,
-      "grad_norm": 813.2791137695312,
-      "learning_rate": 3.43101183069384e-05,
-      "loss": 82.1769,
-      "step": 110170
-    },
-    {
-      "epoch": 0.4451411418205618,
-      "grad_norm": 712.302001953125,
-      "learning_rate": 3.430687863927178e-05,
-      "loss": 53.5758,
-      "step": 110180
-    },
-    {
-      "epoch": 0.44518154308592944,
-      "grad_norm": 730.143310546875,
-      "learning_rate": 3.4303638790162774e-05,
-      "loss": 61.2881,
-      "step": 110190
-    },
-    {
-      "epoch": 0.4452219443512971,
-      "grad_norm": 945.42041015625,
-      "learning_rate": 3.430039875967454e-05,
-      "loss": 62.6643,
-      "step": 110200
-    },
-    {
-      "epoch": 0.4452623456166647,
-      "grad_norm": 887.8884887695312,
-      "learning_rate": 3.429715854787024e-05,
-      "loss": 52.4225,
-      "step": 110210
-    },
-    {
-      "epoch": 0.44530274688203236,
-      "grad_norm": 778.4572143554688,
-      "learning_rate": 3.429391815481305e-05,
-      "loss": 80.6638,
-      "step": 110220
-    },
-    {
-      "epoch": 0.4453431481474,
-      "grad_norm": 701.6489868164062,
-      "learning_rate": 3.429067758056613e-05,
-      "loss": 70.6435,
-      "step": 110230
-    },
-    {
-      "epoch": 0.4453835494127676,
-      "grad_norm": 1139.0123291015625,
-      "learning_rate": 3.428743682519269e-05,
-      "loss": 78.8899,
-      "step": 110240
-    },
-    {
-      "epoch": 0.4454239506781352,
-      "grad_norm": 0.0,
-      "learning_rate": 3.428419588875588e-05,
-      "loss": 51.6693,
-      "step": 110250
-    },
-    {
-      "epoch": 0.44546435194350287,
-      "grad_norm": 919.1795654296875,
-      "learning_rate": 3.428095477131888e-05,
-      "loss": 56.4408,
-      "step": 110260
-    },
-    {
-      "epoch": 0.4455047532088705,
-      "grad_norm": 953.95263671875,
-      "learning_rate": 3.427771347294489e-05,
-      "loss": 72.902,
-      "step": 110270
-    },
-    {
-      "epoch": 0.44554515447423815,
-      "grad_norm": 1921.18359375,
-      "learning_rate": 3.427447199369711e-05,
-      "loss": 43.3139,
-      "step": 110280
-    },
-    {
-      "epoch": 0.4455855557396058,
-      "grad_norm": 790.10791015625,
-      "learning_rate": 3.4271230333638716e-05,
-      "loss": 71.2949,
-      "step": 110290
-    },
-    {
-      "epoch": 0.44562595700497337,
-      "grad_norm": 1008.7927856445312,
-      "learning_rate": 3.426798849283291e-05,
-      "loss": 60.3911,
-      "step": 110300
-    },
-    {
-      "epoch": 0.445666358270341,
-      "grad_norm": 749.1685791015625,
-      "learning_rate": 3.4264746471342905e-05,
-      "loss": 78.0302,
-      "step": 110310
-    },
-    {
-      "epoch": 0.44570675953570865,
-      "grad_norm": 967.6669921875,
-      "learning_rate": 3.4261504269231904e-05,
-      "loss": 49.5635,
-      "step": 110320
-    },
-    {
-      "epoch": 0.4457471608010763,
-      "grad_norm": 1514.6182861328125,
-      "learning_rate": 3.4258261886563104e-05,
-      "loss": 76.0732,
-      "step": 110330
-    },
-    {
-      "epoch": 0.44578756206644393,
-      "grad_norm": 1716.936279296875,
-      "learning_rate": 3.425501932339971e-05,
-      "loss": 133.2402,
-      "step": 110340
-    },
-    {
-      "epoch": 0.44582796333181157,
-      "grad_norm": 959.5086669921875,
-      "learning_rate": 3.425177657980496e-05,
-      "loss": 71.294,
-      "step": 110350
-    },
-    {
-      "epoch": 0.44586836459717916,
-      "grad_norm": 812.0950317382812,
-      "learning_rate": 3.4248533655842066e-05,
-      "loss": 57.06,
-      "step": 110360
-    },
-    {
-      "epoch": 0.4459087658625468,
-      "grad_norm": 267.0820617675781,
-      "learning_rate": 3.4245290551574237e-05,
-      "loss": 44.4798,
-      "step": 110370
-    },
-    {
-      "epoch": 0.44594916712791444,
-      "grad_norm": 1544.18310546875,
-      "learning_rate": 3.4242047267064715e-05,
-      "loss": 57.8316,
-      "step": 110380
-    },
-    {
-      "epoch": 0.4459895683932821,
-      "grad_norm": 695.9027709960938,
-      "learning_rate": 3.4238803802376716e-05,
-      "loss": 56.366,
-      "step": 110390
-    },
-    {
-      "epoch": 0.4460299696586497,
-      "grad_norm": 685.2529296875,
-      "learning_rate": 3.423556015757349e-05,
-      "loss": 57.191,
-      "step": 110400
-    },
-    {
-      "epoch": 0.44607037092401736,
-      "grad_norm": 653.3543090820312,
-      "learning_rate": 3.423231633271826e-05,
-      "loss": 61.4118,
-      "step": 110410
-    },
-    {
-      "epoch": 0.446110772189385,
-      "grad_norm": 1396.4661865234375,
-      "learning_rate": 3.4229072327874274e-05,
-      "loss": 53.9518,
-      "step": 110420
-    },
-    {
-      "epoch": 0.4461511734547526,
-      "grad_norm": 742.2490234375,
-      "learning_rate": 3.422582814310476e-05,
-      "loss": 64.2797,
-      "step": 110430
-    },
-    {
-      "epoch": 0.4461915747201202,
-      "grad_norm": 811.0308837890625,
-      "learning_rate": 3.4222583778472996e-05,
-      "loss": 34.3383,
-      "step": 110440
-    },
-    {
-      "epoch": 0.44623197598548786,
-      "grad_norm": 784.7843017578125,
-      "learning_rate": 3.421933923404219e-05,
-      "loss": 83.5336,
-      "step": 110450
-    },
-    {
-      "epoch": 0.4462723772508555,
-      "grad_norm": 368.3476257324219,
-      "learning_rate": 3.421609450987563e-05,
-      "loss": 43.9047,
-      "step": 110460
-    },
-    {
-      "epoch": 0.44631277851622314,
-      "grad_norm": 507.54205322265625,
-      "learning_rate": 3.421284960603657e-05,
-      "loss": 40.9963,
-      "step": 110470
-    },
-    {
-      "epoch": 0.4463531797815908,
-      "grad_norm": 1140.799560546875,
-      "learning_rate": 3.4209604522588255e-05,
-      "loss": 46.2735,
-      "step": 110480
-    },
-    {
-      "epoch": 0.44639358104695837,
-      "grad_norm": 1338.29541015625,
-      "learning_rate": 3.4206359259593954e-05,
-      "loss": 37.1206,
-      "step": 110490
-    },
-    {
-      "epoch": 0.446433982312326,
-      "grad_norm": 665.2498168945312,
-      "learning_rate": 3.4203113817116957e-05,
-      "loss": 57.0931,
-      "step": 110500
-    },
-    {
-      "epoch": 0.44647438357769365,
-      "grad_norm": 734.5198974609375,
-      "learning_rate": 3.4199868195220505e-05,
-      "loss": 71.1633,
-      "step": 110510
-    },
-    {
-      "epoch": 0.4465147848430613,
-      "grad_norm": 679.3670043945312,
-      "learning_rate": 3.419662239396789e-05,
-      "loss": 52.7156,
-      "step": 110520
-    },
-    {
-      "epoch": 0.4465551861084289,
-      "grad_norm": 538.1387939453125,
-      "learning_rate": 3.419337641342239e-05,
-      "loss": 87.353,
-      "step": 110530
-    },
-    {
-      "epoch": 0.44659558737379657,
-      "grad_norm": 625.4715576171875,
-      "learning_rate": 3.419013025364727e-05,
-      "loss": 52.2098,
-      "step": 110540
-    },
-    {
-      "epoch": 0.4466359886391642,
-      "grad_norm": 1380.083740234375,
-      "learning_rate": 3.4186883914705835e-05,
-      "loss": 110.8045,
-      "step": 110550
-    },
-    {
-      "epoch": 0.4466763899045318,
-      "grad_norm": 664.3757934570312,
-      "learning_rate": 3.418363739666137e-05,
-      "loss": 33.5992,
-      "step": 110560
-    },
-    {
-      "epoch": 0.44671679116989943,
-      "grad_norm": 352.8798522949219,
-      "learning_rate": 3.418039069957717e-05,
-      "loss": 54.5655,
-      "step": 110570
-    },
-    {
-      "epoch": 0.44675719243526707,
-      "grad_norm": 196.41831970214844,
-      "learning_rate": 3.417714382351652e-05,
-      "loss": 46.0049,
-      "step": 110580
-    },
-    {
-      "epoch": 0.4467975937006347,
-      "grad_norm": 1419.017822265625,
-      "learning_rate": 3.417389676854274e-05,
-      "loss": 87.4692,
-      "step": 110590
-    },
-    {
-      "epoch": 0.44683799496600235,
-      "grad_norm": 515.550537109375,
-      "learning_rate": 3.417064953471911e-05,
-      "loss": 46.8774,
-      "step": 110600
-    },
-    {
-      "epoch": 0.44687839623137,
-      "grad_norm": 729.7348022460938,
-      "learning_rate": 3.416740212210894e-05,
-      "loss": 47.5663,
-      "step": 110610
-    },
-    {
-      "epoch": 0.4469187974967376,
-      "grad_norm": 618.8557739257812,
-      "learning_rate": 3.416415453077555e-05,
-      "loss": 56.8461,
-      "step": 110620
-    },
-    {
-      "epoch": 0.4469591987621052,
-      "grad_norm": 1495.1834716796875,
-      "learning_rate": 3.416090676078225e-05,
-      "loss": 71.6594,
-      "step": 110630
-    },
-    {
-      "epoch": 0.44699960002747285,
-      "grad_norm": 537.9067993164062,
-      "learning_rate": 3.415765881219236e-05,
-      "loss": 75.8224,
-      "step": 110640
-    },
-    {
-      "epoch": 0.4470400012928405,
-      "grad_norm": 1021.9192504882812,
-      "learning_rate": 3.4154410685069196e-05,
-      "loss": 71.1588,
-      "step": 110650
-    },
-    {
-      "epoch": 0.44708040255820813,
-      "grad_norm": 1147.8583984375,
-      "learning_rate": 3.4151162379476075e-05,
-      "loss": 99.0359,
-      "step": 110660
-    },
-    {
-      "epoch": 0.4471208038235758,
-      "grad_norm": 951.2720947265625,
-      "learning_rate": 3.414791389547635e-05,
-      "loss": 62.2096,
-      "step": 110670
-    },
-    {
-      "epoch": 0.44716120508894336,
-      "grad_norm": 292.7242736816406,
-      "learning_rate": 3.414466523313332e-05,
-      "loss": 40.3493,
-      "step": 110680
-    },
-    {
-      "epoch": 0.447201606354311,
-      "grad_norm": 1284.24755859375,
-      "learning_rate": 3.414141639251033e-05,
-      "loss": 49.9673,
-      "step": 110690
-    },
-    {
-      "epoch": 0.44724200761967864,
-      "grad_norm": 1174.4100341796875,
-      "learning_rate": 3.413816737367073e-05,
-      "loss": 76.3328,
-      "step": 110700
-    },
-    {
-      "epoch": 0.4472824088850463,
-      "grad_norm": 1334.654052734375,
-      "learning_rate": 3.4134918176677846e-05,
-      "loss": 68.6847,
-      "step": 110710
-    },
-    {
-      "epoch": 0.4473228101504139,
-      "grad_norm": 1329.33203125,
-      "learning_rate": 3.4131668801595027e-05,
-      "loss": 60.1202,
-      "step": 110720
-    },
-    {
-      "epoch": 0.44736321141578156,
-      "grad_norm": 472.8536071777344,
-      "learning_rate": 3.4128419248485635e-05,
-      "loss": 55.691,
-      "step": 110730
-    },
-    {
-      "epoch": 0.4474036126811492,
-      "grad_norm": 1049.84619140625,
-      "learning_rate": 3.4125169517413e-05,
-      "loss": 70.7531,
-      "step": 110740
-    },
-    {
-      "epoch": 0.4474440139465168,
-      "grad_norm": 527.7225952148438,
-      "learning_rate": 3.412191960844049e-05,
-      "loss": 48.9243,
-      "step": 110750
-    },
-    {
-      "epoch": 0.4474844152118844,
-      "grad_norm": 415.2084045410156,
-      "learning_rate": 3.411866952163146e-05,
-      "loss": 52.9608,
-      "step": 110760
-    },
-    {
-      "epoch": 0.44752481647725206,
-      "grad_norm": 1201.3553466796875,
-      "learning_rate": 3.4115419257049286e-05,
-      "loss": 71.9567,
-      "step": 110770
-    },
-    {
-      "epoch": 0.4475652177426197,
-      "grad_norm": 245.35345458984375,
-      "learning_rate": 3.4112168814757307e-05,
-      "loss": 69.7229,
-      "step": 110780
-    },
-    {
-      "epoch": 0.44760561900798734,
-      "grad_norm": 1845.09033203125,
-      "learning_rate": 3.41089181948189e-05,
-      "loss": 68.3285,
-      "step": 110790
-    },
-    {
-      "epoch": 0.447646020273355,
-      "grad_norm": 874.4844360351562,
-      "learning_rate": 3.410566739729746e-05,
-      "loss": 42.1089,
-      "step": 110800
-    },
-    {
-      "epoch": 0.44768642153872257,
-      "grad_norm": 466.1338806152344,
-      "learning_rate": 3.410241642225633e-05,
-      "loss": 53.4753,
-      "step": 110810
-    },
-    {
-      "epoch": 0.4477268228040902,
-      "grad_norm": 391.43475341796875,
-      "learning_rate": 3.409916526975892e-05,
-      "loss": 56.1511,
-      "step": 110820
-    },
-    {
-      "epoch": 0.44776722406945785,
-      "grad_norm": 1110.3125,
-      "learning_rate": 3.409591393986859e-05,
-      "loss": 56.4363,
-      "step": 110830
-    },
-    {
-      "epoch": 0.4478076253348255,
-      "grad_norm": 441.3172302246094,
-      "learning_rate": 3.409266243264874e-05,
-      "loss": 54.6422,
-      "step": 110840
-    },
-    {
-      "epoch": 0.4478480266001931,
-      "grad_norm": 868.8545532226562,
-      "learning_rate": 3.408941074816275e-05,
-      "loss": 56.9401,
-      "step": 110850
-    },
-    {
-      "epoch": 0.44788842786556077,
-      "grad_norm": 988.2617797851562,
-      "learning_rate": 3.408615888647402e-05,
-      "loss": 58.8403,
-      "step": 110860
-    },
-    {
-      "epoch": 0.4479288291309284,
-      "grad_norm": 1164.9776611328125,
-      "learning_rate": 3.408290684764594e-05,
-      "loss": 63.0026,
-      "step": 110870
-    },
-    {
-      "epoch": 0.447969230396296,
-      "grad_norm": 592.5801391601562,
-      "learning_rate": 3.407965463174192e-05,
-      "loss": 52.5244,
-      "step": 110880
-    },
-    {
-      "epoch": 0.44800963166166363,
-      "grad_norm": 647.7515258789062,
-      "learning_rate": 3.407640223882536e-05,
-      "loss": 29.65,
-      "step": 110890
-    },
-    {
-      "epoch": 0.44805003292703127,
-      "grad_norm": 954.7613525390625,
-      "learning_rate": 3.407314966895966e-05,
-      "loss": 56.6861,
-      "step": 110900
-    },
-    {
-      "epoch": 0.4480904341923989,
-      "grad_norm": 997.4801025390625,
-      "learning_rate": 3.406989692220824e-05,
-      "loss": 72.3521,
-      "step": 110910
-    },
-    {
-      "epoch": 0.44813083545776655,
-      "grad_norm": 308.1066589355469,
-      "learning_rate": 3.4066643998634505e-05,
-      "loss": 79.9528,
-      "step": 110920
-    },
-    {
-      "epoch": 0.4481712367231342,
-      "grad_norm": 1413.3211669921875,
-      "learning_rate": 3.406339089830188e-05,
-      "loss": 90.6204,
-      "step": 110930
-    },
-    {
-      "epoch": 0.4482116379885018,
-      "grad_norm": 761.3001098632812,
-      "learning_rate": 3.406013762127379e-05,
-      "loss": 59.6376,
-      "step": 110940
-    },
-    {
-      "epoch": 0.4482520392538694,
-      "grad_norm": 550.8245239257812,
-      "learning_rate": 3.405688416761364e-05,
-      "loss": 61.1043,
-      "step": 110950
-    },
-    {
-      "epoch": 0.44829244051923706,
-      "grad_norm": 1125.8817138671875,
-      "learning_rate": 3.4053630537384885e-05,
-      "loss": 54.3687,
-      "step": 110960
-    },
-    {
-      "epoch": 0.4483328417846047,
-      "grad_norm": 1198.2652587890625,
-      "learning_rate": 3.4050376730650935e-05,
-      "loss": 52.4602,
-      "step": 110970
-    },
-    {
-      "epoch": 0.44837324304997234,
-      "grad_norm": 293.5941162109375,
-      "learning_rate": 3.4047122747475224e-05,
-      "loss": 48.189,
-      "step": 110980
-    },
-    {
-      "epoch": 0.44841364431534,
-      "grad_norm": 1793.2451171875,
-      "learning_rate": 3.40438685879212e-05,
-      "loss": 58.882,
-      "step": 110990
-    },
-    {
-      "epoch": 0.44845404558070756,
-      "grad_norm": 405.7001037597656,
-      "learning_rate": 3.4040614252052305e-05,
-      "loss": 87.0919,
-      "step": 111000
-    },
-    {
-      "epoch": 0.4484944468460752,
-      "grad_norm": 609.4334716796875,
-      "learning_rate": 3.403735973993198e-05,
-      "loss": 88.5648,
-      "step": 111010
-    },
-    {
-      "epoch": 0.44853484811144284,
-      "grad_norm": 476.433837890625,
-      "learning_rate": 3.403410505162369e-05,
-      "loss": 72.3557,
-      "step": 111020
-    },
-    {
-      "epoch": 0.4485752493768105,
-      "grad_norm": 547.7504272460938,
-      "learning_rate": 3.403085018719085e-05,
-      "loss": 71.4732,
-      "step": 111030
-    },
-    {
-      "epoch": 0.4486156506421781,
-      "grad_norm": 618.4702758789062,
-      "learning_rate": 3.402759514669694e-05,
-      "loss": 44.4472,
-      "step": 111040
-    },
-    {
-      "epoch": 0.44865605190754576,
-      "grad_norm": 731.4344482421875,
-      "learning_rate": 3.4024339930205415e-05,
-      "loss": 53.5866,
-      "step": 111050
-    },
-    {
-      "epoch": 0.4486964531729134,
-      "grad_norm": 457.8855895996094,
-      "learning_rate": 3.402108453777974e-05,
-      "loss": 58.6352,
-      "step": 111060
-    },
-    {
-      "epoch": 0.448736854438281,
-      "grad_norm": 612.9066772460938,
-      "learning_rate": 3.401782896948338e-05,
-      "loss": 80.7433,
-      "step": 111070
-    },
-    {
-      "epoch": 0.4487772557036486,
-      "grad_norm": 590.0531616210938,
-      "learning_rate": 3.401457322537979e-05,
-      "loss": 49.937,
-      "step": 111080
-    },
-    {
-      "epoch": 0.44881765696901627,
-      "grad_norm": 579.5253295898438,
-      "learning_rate": 3.401131730553247e-05,
-      "loss": 53.8527,
-      "step": 111090
-    },
-    {
-      "epoch": 0.4488580582343839,
-      "grad_norm": 761.0218505859375,
-      "learning_rate": 3.400806121000487e-05,
-      "loss": 57.2909,
-      "step": 111100
-    },
-    {
-      "epoch": 0.44889845949975155,
-      "grad_norm": 938.6339721679688,
-      "learning_rate": 3.400480493886048e-05,
-      "loss": 54.5442,
-      "step": 111110
-    },
-    {
-      "epoch": 0.4489388607651192,
-      "grad_norm": 800.48486328125,
-      "learning_rate": 3.400154849216278e-05,
-      "loss": 50.155,
-      "step": 111120
-    },
-    {
-      "epoch": 0.44897926203048677,
-      "grad_norm": 1541.491943359375,
-      "learning_rate": 3.3998291869975266e-05,
-      "loss": 47.9149,
-      "step": 111130
-    },
-    {
-      "epoch": 0.4490196632958544,
-      "grad_norm": 631.569580078125,
-      "learning_rate": 3.399503507236141e-05,
-      "loss": 60.8179,
-      "step": 111140
-    },
-    {
-      "epoch": 0.44906006456122205,
-      "grad_norm": 233.70372009277344,
-      "learning_rate": 3.399177809938472e-05,
-      "loss": 33.1188,
-      "step": 111150
-    },
-    {
-      "epoch": 0.4491004658265897,
-      "grad_norm": 492.7761535644531,
-      "learning_rate": 3.398852095110868e-05,
-      "loss": 62.9832,
-      "step": 111160
-    },
-    {
-      "epoch": 0.44914086709195733,
-      "grad_norm": 691.3985595703125,
-      "learning_rate": 3.398526362759681e-05,
-      "loss": 58.8685,
-      "step": 111170
-    },
-    {
-      "epoch": 0.44918126835732497,
-      "grad_norm": 763.817626953125,
-      "learning_rate": 3.3982006128912584e-05,
-      "loss": 34.1555,
-      "step": 111180
-    },
-    {
-      "epoch": 0.4492216696226926,
-      "grad_norm": 794.317626953125,
-      "learning_rate": 3.3978748455119536e-05,
-      "loss": 71.4623,
-      "step": 111190
-    },
-    {
-      "epoch": 0.4492620708880602,
-      "grad_norm": 423.894287109375,
-      "learning_rate": 3.397549060628116e-05,
-      "loss": 56.8908,
-      "step": 111200
-    },
-    {
-      "epoch": 0.44930247215342783,
-      "grad_norm": 2169.064697265625,
-      "learning_rate": 3.3972232582460974e-05,
-      "loss": 52.3652,
-      "step": 111210
-    },
-    {
-      "epoch": 0.4493428734187955,
-      "grad_norm": 363.28289794921875,
-      "learning_rate": 3.3968974383722495e-05,
-      "loss": 40.4274,
-      "step": 111220
-    },
-    {
-      "epoch": 0.4493832746841631,
-      "grad_norm": 973.9365234375,
-      "learning_rate": 3.3965716010129236e-05,
-      "loss": 106.0774,
-      "step": 111230
-    },
-    {
-      "epoch": 0.44942367594953075,
-      "grad_norm": 3008.888671875,
-      "learning_rate": 3.396245746174473e-05,
-      "loss": 44.22,
-      "step": 111240
-    },
-    {
-      "epoch": 0.4494640772148984,
-      "grad_norm": 926.1411743164062,
-      "learning_rate": 3.39591987386325e-05,
-      "loss": 46.0268,
-      "step": 111250
-    },
-    {
-      "epoch": 0.449504478480266,
-      "grad_norm": 177.50221252441406,
-      "learning_rate": 3.3955939840856096e-05,
-      "loss": 44.916,
-      "step": 111260
-    },
-    {
-      "epoch": 0.4495448797456336,
-      "grad_norm": 779.60888671875,
-      "learning_rate": 3.395268076847902e-05,
-      "loss": 56.0308,
-      "step": 111270
-    },
-    {
-      "epoch": 0.44958528101100126,
-      "grad_norm": 323.57818603515625,
-      "learning_rate": 3.394942152156482e-05,
-      "loss": 40.499,
-      "step": 111280
-    },
-    {
-      "epoch": 0.4496256822763689,
-      "grad_norm": 513.7103881835938,
-      "learning_rate": 3.394616210017705e-05,
-      "loss": 44.3521,
-      "step": 111290
-    },
-    {
-      "epoch": 0.44966608354173654,
-      "grad_norm": 657.2388305664062,
-      "learning_rate": 3.3942902504379235e-05,
-      "loss": 72.6372,
-      "step": 111300
-    },
-    {
-      "epoch": 0.4497064848071042,
-      "grad_norm": 513.4599609375,
-      "learning_rate": 3.3939642734234936e-05,
-      "loss": 118.4347,
-      "step": 111310
-    },
-    {
-      "epoch": 0.44974688607247176,
-      "grad_norm": 719.9033203125,
-      "learning_rate": 3.39363827898077e-05,
-      "loss": 85.1415,
-      "step": 111320
-    },
-    {
-      "epoch": 0.4497872873378394,
-      "grad_norm": 1146.169677734375,
-      "learning_rate": 3.393312267116107e-05,
-      "loss": 60.6316,
-      "step": 111330
-    },
-    {
-      "epoch": 0.44982768860320704,
-      "grad_norm": 633.3748168945312,
-      "learning_rate": 3.392986237835863e-05,
-      "loss": 60.0308,
-      "step": 111340
-    },
-    {
-      "epoch": 0.4498680898685747,
-      "grad_norm": 436.7900390625,
-      "learning_rate": 3.3926601911463915e-05,
-      "loss": 54.0903,
-      "step": 111350
-    },
-    {
-      "epoch": 0.4499084911339423,
-      "grad_norm": 425.1293029785156,
-      "learning_rate": 3.392334127054051e-05,
-      "loss": 59.7452,
-      "step": 111360
-    },
-    {
-      "epoch": 0.44994889239930996,
-      "grad_norm": 1201.4161376953125,
-      "learning_rate": 3.392008045565197e-05,
-      "loss": 56.7192,
-      "step": 111370
-    },
-    {
-      "epoch": 0.4499892936646776,
-      "grad_norm": 287.7637939453125,
-      "learning_rate": 3.391681946686186e-05,
-      "loss": 67.3147,
-      "step": 111380
-    },
-    {
-      "epoch": 0.4500296949300452,
-      "grad_norm": 818.6224975585938,
-      "learning_rate": 3.3913558304233776e-05,
-      "loss": 53.3463,
-      "step": 111390
-    },
-    {
-      "epoch": 0.4500700961954128,
-      "grad_norm": 385.80987548828125,
-      "learning_rate": 3.3910296967831266e-05,
-      "loss": 66.1127,
-      "step": 111400
-    },
-    {
-      "epoch": 0.45011049746078047,
-      "grad_norm": 1808.7755126953125,
-      "learning_rate": 3.3907035457717944e-05,
-      "loss": 67.1233,
-      "step": 111410
-    },
-    {
-      "epoch": 0.4501508987261481,
-      "grad_norm": 512.4915771484375,
-      "learning_rate": 3.390377377395738e-05,
-      "loss": 52.8904,
-      "step": 111420
-    },
-    {
-      "epoch": 0.45019129999151575,
-      "grad_norm": 1636.7056884765625,
-      "learning_rate": 3.3900511916613155e-05,
-      "loss": 101.467,
-      "step": 111430
-    },
-    {
-      "epoch": 0.4502317012568834,
-      "grad_norm": 682.1113891601562,
-      "learning_rate": 3.389724988574887e-05,
-      "loss": 47.2666,
-      "step": 111440
-    },
-    {
-      "epoch": 0.45027210252225097,
-      "grad_norm": 401.15771484375,
-      "learning_rate": 3.389398768142812e-05,
-      "loss": 40.7815,
-      "step": 111450
-    },
-    {
-      "epoch": 0.4503125037876186,
-      "grad_norm": 1251.0804443359375,
-      "learning_rate": 3.389072530371451e-05,
-      "loss": 49.2346,
-      "step": 111460
-    },
-    {
-      "epoch": 0.45035290505298625,
-      "grad_norm": 642.7127075195312,
-      "learning_rate": 3.388746275267162e-05,
-      "loss": 63.2051,
-      "step": 111470
-    },
-    {
-      "epoch": 0.4503933063183539,
-      "grad_norm": 605.9111938476562,
-      "learning_rate": 3.388420002836307e-05,
-      "loss": 52.5846,
-      "step": 111480
-    },
-    {
-      "epoch": 0.45043370758372153,
-      "grad_norm": 1575.560302734375,
-      "learning_rate": 3.3880937130852466e-05,
-      "loss": 35.7747,
-      "step": 111490
-    },
-    {
-      "epoch": 0.45047410884908917,
-      "grad_norm": 645.1834716796875,
-      "learning_rate": 3.387767406020343e-05,
-      "loss": 55.3005,
-      "step": 111500
-    },
-    {
-      "epoch": 0.4505145101144568,
-      "grad_norm": 666.5519409179688,
-      "learning_rate": 3.3874410816479564e-05,
-      "loss": 48.5718,
-      "step": 111510
-    },
-    {
-      "epoch": 0.4505549113798244,
-      "grad_norm": 654.6434326171875,
-      "learning_rate": 3.387114739974448e-05,
-      "loss": 81.415,
-      "step": 111520
-    },
-    {
-      "epoch": 0.45059531264519204,
-      "grad_norm": 2661.026123046875,
-      "learning_rate": 3.3867883810061824e-05,
-      "loss": 84.8381,
-      "step": 111530
-    },
-    {
-      "epoch": 0.4506357139105597,
-      "grad_norm": 2447.82470703125,
-      "learning_rate": 3.38646200474952e-05,
-      "loss": 48.1841,
-      "step": 111540
-    },
-    {
-      "epoch": 0.4506761151759273,
-      "grad_norm": 708.4982299804688,
-      "learning_rate": 3.3861356112108247e-05,
-      "loss": 56.9643,
-      "step": 111550
-    },
-    {
-      "epoch": 0.45071651644129496,
-      "grad_norm": 489.48846435546875,
-      "learning_rate": 3.3858092003964594e-05,
-      "loss": 57.1001,
-      "step": 111560
-    },
-    {
-      "epoch": 0.4507569177066626,
-      "grad_norm": 538.01806640625,
-      "learning_rate": 3.385482772312787e-05,
-      "loss": 43.918,
-      "step": 111570
-    },
-    {
-      "epoch": 0.4507973189720302,
-      "grad_norm": 624.0175170898438,
-      "learning_rate": 3.3851563269661726e-05,
-      "loss": 95.1447,
-      "step": 111580
-    },
-    {
-      "epoch": 0.4508377202373978,
-      "grad_norm": 565.3558959960938,
-      "learning_rate": 3.38482986436298e-05,
-      "loss": 44.5088,
-      "step": 111590
-    },
-    {
-      "epoch": 0.45087812150276546,
-      "grad_norm": 746.841796875,
-      "learning_rate": 3.384503384509574e-05,
-      "loss": 44.0926,
-      "step": 111600
-    },
-    {
-      "epoch": 0.4509185227681331,
-      "grad_norm": 969.9041748046875,
-      "learning_rate": 3.384176887412318e-05,
-      "loss": 38.6924,
-      "step": 111610
-    },
-    {
-      "epoch": 0.45095892403350074,
-      "grad_norm": 971.7702026367188,
-      "learning_rate": 3.38385037307758e-05,
-      "loss": 56.2464,
-      "step": 111620
-    },
-    {
-      "epoch": 0.4509993252988684,
-      "grad_norm": 1089.8248291015625,
-      "learning_rate": 3.383523841511723e-05,
-      "loss": 89.415,
-      "step": 111630
-    },
-    {
-      "epoch": 0.45103972656423597,
-      "grad_norm": 535.305419921875,
-      "learning_rate": 3.3831972927211135e-05,
-      "loss": 41.9581,
-      "step": 111640
-    },
-    {
-      "epoch": 0.4510801278296036,
-      "grad_norm": 1719.795166015625,
-      "learning_rate": 3.382870726712119e-05,
-      "loss": 74.1624,
-      "step": 111650
-    },
-    {
-      "epoch": 0.45112052909497125,
-      "grad_norm": 1085.8828125,
-      "learning_rate": 3.382544143491104e-05,
-      "loss": 81.0191,
-      "step": 111660
-    },
-    {
-      "epoch": 0.4511609303603389,
-      "grad_norm": 409.8812561035156,
-      "learning_rate": 3.382217543064438e-05,
-      "loss": 40.052,
-      "step": 111670
-    },
-    {
-      "epoch": 0.4512013316257065,
-      "grad_norm": 1443.2547607421875,
-      "learning_rate": 3.381890925438486e-05,
-      "loss": 37.1698,
-      "step": 111680
-    },
-    {
-      "epoch": 0.45124173289107417,
-      "grad_norm": 1432.521728515625,
-      "learning_rate": 3.3815642906196156e-05,
-      "loss": 75.6335,
-      "step": 111690
-    },
-    {
-      "epoch": 0.4512821341564418,
-      "grad_norm": 1234.38720703125,
-      "learning_rate": 3.381237638614196e-05,
-      "loss": 48.0366,
-      "step": 111700
-    },
-    {
-      "epoch": 0.4513225354218094,
-      "grad_norm": 840.4757690429688,
-      "learning_rate": 3.380910969428596e-05,
-      "loss": 58.3024,
-      "step": 111710
-    },
-    {
-      "epoch": 0.45136293668717703,
-      "grad_norm": 591.4329833984375,
-      "learning_rate": 3.380584283069183e-05,
-      "loss": 42.8441,
-      "step": 111720
-    },
-    {
-      "epoch": 0.45140333795254467,
-      "grad_norm": 579.3867797851562,
-      "learning_rate": 3.380257579542325e-05,
-      "loss": 100.9832,
-      "step": 111730
-    },
-    {
-      "epoch": 0.4514437392179123,
-      "grad_norm": 1018.6928100585938,
-      "learning_rate": 3.379930858854392e-05,
-      "loss": 40.5557,
-      "step": 111740
-    },
-    {
-      "epoch": 0.45148414048327995,
-      "grad_norm": 475.0141906738281,
-      "learning_rate": 3.3796041210117546e-05,
-      "loss": 47.8023,
-      "step": 111750
-    },
-    {
-      "epoch": 0.4515245417486476,
-      "grad_norm": 1127.599853515625,
-      "learning_rate": 3.379277366020782e-05,
-      "loss": 56.3643,
-      "step": 111760
-    },
-    {
-      "epoch": 0.4515649430140152,
-      "grad_norm": 632.768310546875,
-      "learning_rate": 3.3789505938878443e-05,
-      "loss": 42.618,
-      "step": 111770
-    },
-    {
-      "epoch": 0.4516053442793828,
-      "grad_norm": 558.4849853515625,
-      "learning_rate": 3.378623804619313e-05,
-      "loss": 38.2455,
-      "step": 111780
-    },
-    {
-      "epoch": 0.45164574554475045,
-      "grad_norm": 1455.6197509765625,
-      "learning_rate": 3.378296998221557e-05,
-      "loss": 54.4919,
-      "step": 111790
-    },
-    {
-      "epoch": 0.4516861468101181,
-      "grad_norm": 650.8120727539062,
-      "learning_rate": 3.3779701747009504e-05,
-      "loss": 68.0886,
-      "step": 111800
-    },
-    {
-      "epoch": 0.45172654807548573,
-      "grad_norm": 613.3826904296875,
-      "learning_rate": 3.377643334063862e-05,
-      "loss": 47.3737,
-      "step": 111810
-    },
-    {
-      "epoch": 0.4517669493408534,
-      "grad_norm": 814.562744140625,
-      "learning_rate": 3.3773164763166655e-05,
-      "loss": 70.561,
-      "step": 111820
-    },
-    {
-      "epoch": 0.451807350606221,
-      "grad_norm": 785.9385986328125,
-      "learning_rate": 3.376989601465733e-05,
-      "loss": 38.8111,
-      "step": 111830
-    },
-    {
-      "epoch": 0.4518477518715886,
-      "grad_norm": 640.478759765625,
-      "learning_rate": 3.376662709517435e-05,
-      "loss": 55.49,
-      "step": 111840
-    },
-    {
-      "epoch": 0.45188815313695624,
-      "grad_norm": 590.5025024414062,
-      "learning_rate": 3.3763358004781475e-05,
-      "loss": 59.1937,
-      "step": 111850
-    },
-    {
-      "epoch": 0.4519285544023239,
-      "grad_norm": 681.9274291992188,
-      "learning_rate": 3.3760088743542424e-05,
-      "loss": 121.1419,
-      "step": 111860
-    },
-    {
-      "epoch": 0.4519689556676915,
-      "grad_norm": 1086.4017333984375,
-      "learning_rate": 3.375681931152093e-05,
-      "loss": 57.0334,
-      "step": 111870
-    },
-    {
-      "epoch": 0.45200935693305916,
-      "grad_norm": 614.7588500976562,
-      "learning_rate": 3.375354970878073e-05,
-      "loss": 51.712,
-      "step": 111880
-    },
-    {
-      "epoch": 0.4520497581984268,
-      "grad_norm": 1066.3997802734375,
-      "learning_rate": 3.375027993538559e-05,
-      "loss": 62.087,
-      "step": 111890
-    },
-    {
-      "epoch": 0.4520901594637944,
-      "grad_norm": 644.5421142578125,
-      "learning_rate": 3.374700999139923e-05,
-      "loss": 58.5303,
-      "step": 111900
-    },
-    {
-      "epoch": 0.452130560729162,
-      "grad_norm": 840.0023193359375,
-      "learning_rate": 3.37437398768854e-05,
-      "loss": 71.8506,
-      "step": 111910
-    },
-    {
-      "epoch": 0.45217096199452966,
-      "grad_norm": 1441.012451171875,
-      "learning_rate": 3.374046959190786e-05,
-      "loss": 92.4958,
-      "step": 111920
-    },
-    {
-      "epoch": 0.4522113632598973,
-      "grad_norm": 1386.08349609375,
-      "learning_rate": 3.3737199136530364e-05,
-      "loss": 51.7875,
-      "step": 111930
-    },
-    {
-      "epoch": 0.45225176452526494,
-      "grad_norm": 1856.324951171875,
-      "learning_rate": 3.373392851081668e-05,
-      "loss": 74.6996,
-      "step": 111940
-    },
-    {
-      "epoch": 0.4522921657906326,
-      "grad_norm": 371.86566162109375,
-      "learning_rate": 3.373065771483056e-05,
-      "loss": 53.6041,
-      "step": 111950
-    },
-    {
-      "epoch": 0.45233256705600017,
-      "grad_norm": 862.8280639648438,
-      "learning_rate": 3.372738674863577e-05,
-      "loss": 51.361,
-      "step": 111960
-    },
-    {
-      "epoch": 0.4523729683213678,
-      "grad_norm": 923.8464965820312,
-      "learning_rate": 3.372411561229609e-05,
-      "loss": 43.7514,
-      "step": 111970
-    },
-    {
-      "epoch": 0.45241336958673545,
-      "grad_norm": 716.6856079101562,
-      "learning_rate": 3.372084430587528e-05,
-      "loss": 39.6764,
-      "step": 111980
-    },
-    {
-      "epoch": 0.4524537708521031,
-      "grad_norm": 593.4553833007812,
-      "learning_rate": 3.371757282943712e-05,
-      "loss": 39.9165,
-      "step": 111990
-    },
-    {
-      "epoch": 0.4524941721174707,
-      "grad_norm": 496.314208984375,
-      "learning_rate": 3.3714301183045385e-05,
-      "loss": 78.6274,
-      "step": 112000
-    },
-    {
-      "epoch": 0.45253457338283837,
-      "grad_norm": 811.6598510742188,
-      "learning_rate": 3.3711029366763866e-05,
-      "loss": 34.8752,
-      "step": 112010
-    },
-    {
-      "epoch": 0.452574974648206,
-      "grad_norm": 1419.599609375,
-      "learning_rate": 3.370775738065634e-05,
-      "loss": 81.8719,
-      "step": 112020
-    },
-    {
-      "epoch": 0.4526153759135736,
-      "grad_norm": 1418.275146484375,
-      "learning_rate": 3.370448522478661e-05,
-      "loss": 65.6506,
-      "step": 112030
-    },
-    {
-      "epoch": 0.45265577717894123,
-      "grad_norm": 8803.1748046875,
-      "learning_rate": 3.370121289921845e-05,
-      "loss": 112.1481,
-      "step": 112040
-    },
-    {
-      "epoch": 0.45269617844430887,
-      "grad_norm": 328.6112060546875,
-      "learning_rate": 3.369794040401567e-05,
-      "loss": 38.9246,
-      "step": 112050
-    },
-    {
-      "epoch": 0.4527365797096765,
-      "grad_norm": 576.2307739257812,
-      "learning_rate": 3.3694667739242066e-05,
-      "loss": 54.7131,
-      "step": 112060
-    },
-    {
-      "epoch": 0.45277698097504415,
-      "grad_norm": 1210.724365234375,
-      "learning_rate": 3.369139490496144e-05,
-      "loss": 60.4555,
-      "step": 112070
-    },
-    {
-      "epoch": 0.4528173822404118,
-      "grad_norm": 1678.520751953125,
-      "learning_rate": 3.368812190123759e-05,
-      "loss": 53.7549,
-      "step": 112080
-    },
-    {
-      "epoch": 0.4528577835057794,
-      "grad_norm": 754.4735717773438,
-      "learning_rate": 3.3684848728134334e-05,
-      "loss": 55.5762,
-      "step": 112090
-    },
-    {
-      "epoch": 0.452898184771147,
-      "grad_norm": 294.9254455566406,
-      "learning_rate": 3.368157538571548e-05,
-      "loss": 62.4506,
-      "step": 112100
-    },
-    {
-      "epoch": 0.45293858603651466,
-      "grad_norm": 1608.3568115234375,
-      "learning_rate": 3.367830187404484e-05,
-      "loss": 55.6337,
-      "step": 112110
-    },
-    {
-      "epoch": 0.4529789873018823,
-      "grad_norm": 680.6564331054688,
-      "learning_rate": 3.367502819318624e-05,
-      "loss": 49.9141,
-      "step": 112120
-    },
-    {
-      "epoch": 0.45301938856724994,
-      "grad_norm": 1690.9805908203125,
-      "learning_rate": 3.36717543432035e-05,
-      "loss": 58.1772,
-      "step": 112130
-    },
-    {
-      "epoch": 0.4530597898326176,
-      "grad_norm": 564.469482421875,
-      "learning_rate": 3.366848032416045e-05,
-      "loss": 59.8349,
-      "step": 112140
-    },
-    {
-      "epoch": 0.4531001910979852,
-      "grad_norm": 725.934326171875,
-      "learning_rate": 3.3665206136120906e-05,
-      "loss": 72.5532,
-      "step": 112150
-    },
-    {
-      "epoch": 0.4531405923633528,
-      "grad_norm": 949.9281616210938,
-      "learning_rate": 3.3661931779148707e-05,
-      "loss": 65.5223,
-      "step": 112160
-    },
-    {
-      "epoch": 0.45318099362872044,
-      "grad_norm": 1123.207763671875,
-      "learning_rate": 3.365865725330769e-05,
-      "loss": 52.4678,
-      "step": 112170
-    },
-    {
-      "epoch": 0.4532213948940881,
-      "grad_norm": 264.70001220703125,
-      "learning_rate": 3.3655382558661685e-05,
-      "loss": 64.133,
-      "step": 112180
-    },
-    {
-      "epoch": 0.4532617961594557,
-      "grad_norm": 1888.4464111328125,
-      "learning_rate": 3.3652107695274555e-05,
-      "loss": 45.2613,
-      "step": 112190
-    },
-    {
-      "epoch": 0.45330219742482336,
-      "grad_norm": 539.1635131835938,
-      "learning_rate": 3.3648832663210124e-05,
-      "loss": 95.8625,
-      "step": 112200
-    },
-    {
-      "epoch": 0.453342598690191,
-      "grad_norm": 510.55816650390625,
-      "learning_rate": 3.3645557462532245e-05,
-      "loss": 39.8649,
-      "step": 112210
-    },
-    {
-      "epoch": 0.4533829999555586,
-      "grad_norm": 1129.6290283203125,
-      "learning_rate": 3.364228209330477e-05,
-      "loss": 54.8464,
-      "step": 112220
-    },
-    {
-      "epoch": 0.4534234012209262,
-      "grad_norm": 718.903564453125,
-      "learning_rate": 3.363900655559157e-05,
-      "loss": 70.0421,
-      "step": 112230
-    },
-    {
-      "epoch": 0.45346380248629387,
-      "grad_norm": 765.8641967773438,
-      "learning_rate": 3.363573084945648e-05,
-      "loss": 43.2255,
-      "step": 112240
-    },
-    {
-      "epoch": 0.4535042037516615,
-      "grad_norm": 305.63763427734375,
-      "learning_rate": 3.363245497496337e-05,
-      "loss": 55.8746,
-      "step": 112250
-    },
-    {
-      "epoch": 0.45354460501702915,
-      "grad_norm": 359.51715087890625,
-      "learning_rate": 3.362917893217611e-05,
-      "loss": 73.9199,
-      "step": 112260
-    },
-    {
-      "epoch": 0.4535850062823968,
-      "grad_norm": 516.9877319335938,
-      "learning_rate": 3.362590272115855e-05,
-      "loss": 67.4331,
-      "step": 112270
-    },
-    {
-      "epoch": 0.45362540754776437,
-      "grad_norm": 862.2160034179688,
-      "learning_rate": 3.3622626341974594e-05,
-      "loss": 57.4161,
-      "step": 112280
-    },
-    {
-      "epoch": 0.453665808813132,
-      "grad_norm": 1069.54150390625,
-      "learning_rate": 3.361934979468809e-05,
-      "loss": 57.9138,
-      "step": 112290
-    },
-    {
-      "epoch": 0.45370621007849965,
-      "grad_norm": 375.27825927734375,
-      "learning_rate": 3.3616073079362926e-05,
-      "loss": 51.6367,
-      "step": 112300
-    },
-    {
-      "epoch": 0.4537466113438673,
-      "grad_norm": 450.5061950683594,
-      "learning_rate": 3.361279619606299e-05,
-      "loss": 61.73,
-      "step": 112310
-    },
-    {
-      "epoch": 0.45378701260923493,
-      "grad_norm": 1464.5079345703125,
-      "learning_rate": 3.360951914485215e-05,
-      "loss": 52.9956,
-      "step": 112320
-    },
-    {
-      "epoch": 0.45382741387460257,
-      "grad_norm": 324.18194580078125,
-      "learning_rate": 3.3606241925794295e-05,
-      "loss": 47.1093,
-      "step": 112330
-    },
-    {
-      "epoch": 0.4538678151399702,
-      "grad_norm": 836.4578857421875,
-      "learning_rate": 3.360296453895333e-05,
-      "loss": 56.2767,
-      "step": 112340
-    },
-    {
-      "epoch": 0.4539082164053378,
-      "grad_norm": 1074.7415771484375,
-      "learning_rate": 3.3599686984393134e-05,
-      "loss": 50.7534,
-      "step": 112350
-    },
-    {
-      "epoch": 0.45394861767070543,
-      "grad_norm": 225.61444091796875,
-      "learning_rate": 3.359640926217763e-05,
-      "loss": 40.0397,
-      "step": 112360
-    },
-    {
-      "epoch": 0.4539890189360731,
-      "grad_norm": 1213.6234130859375,
-      "learning_rate": 3.359313137237069e-05,
-      "loss": 47.421,
-      "step": 112370
-    },
-    {
-      "epoch": 0.4540294202014407,
-      "grad_norm": 640.165283203125,
-      "learning_rate": 3.3589853315036225e-05,
-      "loss": 60.4271,
-      "step": 112380
-    },
-    {
-      "epoch": 0.45406982146680835,
-      "grad_norm": 354.8425598144531,
-      "learning_rate": 3.358657509023815e-05,
-      "loss": 58.7236,
-      "step": 112390
-    },
-    {
-      "epoch": 0.454110222732176,
-      "grad_norm": 1010.51123046875,
-      "learning_rate": 3.3583296698040384e-05,
-      "loss": 67.7665,
-      "step": 112400
-    },
-    {
-      "epoch": 0.4541506239975436,
-      "grad_norm": 313.4957580566406,
-      "learning_rate": 3.3580018138506824e-05,
-      "loss": 73.5613,
-      "step": 112410
-    },
-    {
-      "epoch": 0.4541910252629112,
-      "grad_norm": 623.6607055664062,
-      "learning_rate": 3.3576739411701394e-05,
-      "loss": 58.9651,
-      "step": 112420
-    },
-    {
-      "epoch": 0.45423142652827886,
-      "grad_norm": 870.5923461914062,
-      "learning_rate": 3.357346051768801e-05,
-      "loss": 64.04,
-      "step": 112430
-    },
-    {
-      "epoch": 0.4542718277936465,
-      "grad_norm": 499.6788330078125,
-      "learning_rate": 3.35701814565306e-05,
-      "loss": 59.2575,
-      "step": 112440
-    },
-    {
-      "epoch": 0.45431222905901414,
-      "grad_norm": 902.0311889648438,
-      "learning_rate": 3.356690222829309e-05,
-      "loss": 37.0818,
-      "step": 112450
-    },
-    {
-      "epoch": 0.4543526303243818,
-      "grad_norm": 713.7149047851562,
-      "learning_rate": 3.356362283303941e-05,
-      "loss": 52.5913,
-      "step": 112460
-    },
-    {
-      "epoch": 0.45439303158974936,
-      "grad_norm": 671.7296142578125,
-      "learning_rate": 3.3560343270833495e-05,
-      "loss": 53.5628,
-      "step": 112470
-    },
-    {
-      "epoch": 0.454433432855117,
-      "grad_norm": 1461.3150634765625,
-      "learning_rate": 3.355706354173928e-05,
-      "loss": 78.7695,
-      "step": 112480
-    },
-    {
-      "epoch": 0.45447383412048464,
-      "grad_norm": 1066.4896240234375,
-      "learning_rate": 3.3553783645820715e-05,
-      "loss": 74.929,
-      "step": 112490
-    },
-    {
-      "epoch": 0.4545142353858523,
-      "grad_norm": 257.38531494140625,
-      "learning_rate": 3.355050358314172e-05,
-      "loss": 39.4283,
-      "step": 112500
-    },
-    {
-      "epoch": 0.4545546366512199,
-      "grad_norm": 685.919677734375,
-      "learning_rate": 3.354722335376626e-05,
-      "loss": 51.0791,
-      "step": 112510
-    },
-    {
-      "epoch": 0.45459503791658756,
-      "grad_norm": 1068.9114990234375,
-      "learning_rate": 3.354394295775829e-05,
-      "loss": 44.3589,
-      "step": 112520
-    },
-    {
-      "epoch": 0.4546354391819552,
-      "grad_norm": 604.07568359375,
-      "learning_rate": 3.354066239518174e-05,
-      "loss": 65.293,
-      "step": 112530
-    },
-    {
-      "epoch": 0.4546758404473228,
-      "grad_norm": 1334.7886962890625,
-      "learning_rate": 3.353738166610058e-05,
-      "loss": 52.1476,
-      "step": 112540
-    },
-    {
-      "epoch": 0.4547162417126904,
-      "grad_norm": 517.1060791015625,
-      "learning_rate": 3.353410077057877e-05,
-      "loss": 80.8657,
-      "step": 112550
-    },
-    {
-      "epoch": 0.45475664297805807,
-      "grad_norm": 479.4462890625,
-      "learning_rate": 3.3530819708680286e-05,
-      "loss": 54.3628,
-      "step": 112560
-    },
-    {
-      "epoch": 0.4547970442434257,
-      "grad_norm": 579.92041015625,
-      "learning_rate": 3.352753848046907e-05,
-      "loss": 41.2108,
-      "step": 112570
-    },
-    {
-      "epoch": 0.45483744550879335,
-      "grad_norm": 557.235107421875,
-      "learning_rate": 3.3524257086009104e-05,
-      "loss": 33.0313,
-      "step": 112580
-    },
-    {
-      "epoch": 0.454877846774161,
-      "grad_norm": 929.7172241210938,
-      "learning_rate": 3.352097552536435e-05,
-      "loss": 40.3618,
-      "step": 112590
-    },
-    {
-      "epoch": 0.45491824803952857,
-      "grad_norm": 1180.1580810546875,
-      "learning_rate": 3.35176937985988e-05,
-      "loss": 73.8001,
-      "step": 112600
-    },
-    {
-      "epoch": 0.4549586493048962,
-      "grad_norm": 931.85009765625,
-      "learning_rate": 3.351441190577642e-05,
-      "loss": 45.0227,
-      "step": 112610
-    },
-    {
-      "epoch": 0.45499905057026385,
-      "grad_norm": 911.3134155273438,
-      "learning_rate": 3.3511129846961184e-05,
-      "loss": 59.2229,
-      "step": 112620
-    },
-    {
-      "epoch": 0.4550394518356315,
-      "grad_norm": 675.170654296875,
-      "learning_rate": 3.35078476222171e-05,
-      "loss": 80.4119,
-      "step": 112630
-    },
-    {
-      "epoch": 0.45507985310099913,
-      "grad_norm": 1176.3988037109375,
-      "learning_rate": 3.350456523160815e-05,
-      "loss": 47.3479,
-      "step": 112640
-    },
-    {
-      "epoch": 0.45512025436636677,
-      "grad_norm": 969.1943359375,
-      "learning_rate": 3.350128267519832e-05,
-      "loss": 81.494,
-      "step": 112650
-    },
-    {
-      "epoch": 0.4551606556317344,
-      "grad_norm": 291.8066711425781,
-      "learning_rate": 3.349799995305162e-05,
-      "loss": 41.662,
-      "step": 112660
-    },
-    {
-      "epoch": 0.455201056897102,
-      "grad_norm": 658.59765625,
-      "learning_rate": 3.3494717065232016e-05,
-      "loss": 61.2986,
-      "step": 112670
-    },
-    {
-      "epoch": 0.45524145816246964,
-      "grad_norm": 917.7549438476562,
-      "learning_rate": 3.349143401180354e-05,
-      "loss": 63.2453,
-      "step": 112680
-    },
-    {
-      "epoch": 0.4552818594278373,
-      "grad_norm": 453.6599426269531,
-      "learning_rate": 3.348815079283018e-05,
-      "loss": 59.7746,
-      "step": 112690
-    },
-    {
-      "epoch": 0.4553222606932049,
-      "grad_norm": 850.1251220703125,
-      "learning_rate": 3.3484867408375954e-05,
-      "loss": 57.4509,
-      "step": 112700
-    },
-    {
-      "epoch": 0.45536266195857256,
-      "grad_norm": 701.5401000976562,
-      "learning_rate": 3.348158385850487e-05,
-      "loss": 41.1296,
-      "step": 112710
-    },
-    {
-      "epoch": 0.4554030632239402,
-      "grad_norm": 819.9852294921875,
-      "learning_rate": 3.347830014328094e-05,
-      "loss": 69.8444,
-      "step": 112720
-    },
-    {
-      "epoch": 0.4554434644893078,
-      "grad_norm": 720.2810668945312,
-      "learning_rate": 3.347501626276819e-05,
-      "loss": 75.9919,
-      "step": 112730
-    },
-    {
-      "epoch": 0.4554838657546754,
-      "grad_norm": 495.089599609375,
-      "learning_rate": 3.3471732217030625e-05,
-      "loss": 40.4414,
-      "step": 112740
-    },
-    {
-      "epoch": 0.45552426702004306,
-      "grad_norm": 573.2144775390625,
-      "learning_rate": 3.346844800613229e-05,
-      "loss": 43.111,
-      "step": 112750
-    },
-    {
-      "epoch": 0.4555646682854107,
-      "grad_norm": 1056.5557861328125,
-      "learning_rate": 3.346516363013719e-05,
-      "loss": 92.2572,
-      "step": 112760
-    },
-    {
-      "epoch": 0.45560506955077834,
-      "grad_norm": 487.2084655761719,
-      "learning_rate": 3.346187908910938e-05,
-      "loss": 36.6314,
-      "step": 112770
-    },
-    {
-      "epoch": 0.455645470816146,
-      "grad_norm": 501.5868225097656,
-      "learning_rate": 3.345859438311287e-05,
-      "loss": 56.4939,
-      "step": 112780
-    },
-    {
-      "epoch": 0.45568587208151357,
-      "grad_norm": 792.0189819335938,
-      "learning_rate": 3.345530951221171e-05,
-      "loss": 94.0912,
-      "step": 112790
-    },
-    {
-      "epoch": 0.4557262733468812,
-      "grad_norm": 778.9989624023438,
-      "learning_rate": 3.3452024476469934e-05,
-      "loss": 40.7863,
-      "step": 112800
-    },
-    {
-      "epoch": 0.45576667461224885,
-      "grad_norm": 1507.0234375,
-      "learning_rate": 3.3448739275951595e-05,
-      "loss": 61.6816,
-      "step": 112810
-    },
-    {
-      "epoch": 0.4558070758776165,
-      "grad_norm": 432.25244140625,
-      "learning_rate": 3.344545391072073e-05,
-      "loss": 44.9511,
-      "step": 112820
-    },
-    {
-      "epoch": 0.4558474771429841,
-      "grad_norm": 655.8712158203125,
-      "learning_rate": 3.34421683808414e-05,
-      "loss": 48.9905,
-      "step": 112830
-    },
-    {
-      "epoch": 0.45588787840835177,
-      "grad_norm": 940.5873413085938,
-      "learning_rate": 3.343888268637765e-05,
-      "loss": 59.8554,
-      "step": 112840
-    },
-    {
-      "epoch": 0.4559282796737194,
-      "grad_norm": 1201.72412109375,
-      "learning_rate": 3.343559682739353e-05,
-      "loss": 85.8798,
-      "step": 112850
-    },
-    {
-      "epoch": 0.455968680939087,
-      "grad_norm": 822.5703735351562,
-      "learning_rate": 3.343231080395312e-05,
-      "loss": 54.6579,
-      "step": 112860
-    },
-    {
-      "epoch": 0.45600908220445463,
-      "grad_norm": 476.291748046875,
-      "learning_rate": 3.342902461612045e-05,
-      "loss": 40.1478,
-      "step": 112870
-    },
-    {
-      "epoch": 0.45604948346982227,
-      "grad_norm": 1140.7523193359375,
-      "learning_rate": 3.3425738263959615e-05,
-      "loss": 36.0184,
-      "step": 112880
-    },
-    {
-      "epoch": 0.4560898847351899,
-      "grad_norm": 627.5087280273438,
-      "learning_rate": 3.3422451747534684e-05,
-      "loss": 67.8078,
-      "step": 112890
-    },
-    {
-      "epoch": 0.45613028600055755,
-      "grad_norm": 525.3790283203125,
-      "learning_rate": 3.3419165066909705e-05,
-      "loss": 47.1049,
-      "step": 112900
-    },
-    {
-      "epoch": 0.4561706872659252,
-      "grad_norm": 361.6174011230469,
-      "learning_rate": 3.3415878222148776e-05,
-      "loss": 66.291,
-      "step": 112910
-    },
-    {
-      "epoch": 0.4562110885312928,
-      "grad_norm": 1267.40966796875,
-      "learning_rate": 3.341259121331597e-05,
-      "loss": 39.7781,
-      "step": 112920
-    },
-    {
-      "epoch": 0.4562514897966604,
-      "grad_norm": 643.3392333984375,
-      "learning_rate": 3.340930404047537e-05,
-      "loss": 37.3639,
-      "step": 112930
-    },
-    {
-      "epoch": 0.45629189106202805,
-      "grad_norm": 481.6955871582031,
-      "learning_rate": 3.3406016703691055e-05,
-      "loss": 55.8988,
-      "step": 112940
-    },
-    {
-      "epoch": 0.4563322923273957,
-      "grad_norm": 650.4725952148438,
-      "learning_rate": 3.340272920302711e-05,
-      "loss": 91.6434,
-      "step": 112950
-    },
-    {
-      "epoch": 0.45637269359276333,
-      "grad_norm": 674.4364013671875,
-      "learning_rate": 3.339944153854764e-05,
-      "loss": 46.4506,
-      "step": 112960
-    },
-    {
-      "epoch": 0.456413094858131,
-      "grad_norm": 542.6135864257812,
-      "learning_rate": 3.3396153710316736e-05,
-      "loss": 54.1574,
-      "step": 112970
-    },
-    {
-      "epoch": 0.4564534961234986,
-      "grad_norm": 1714.6644287109375,
-      "learning_rate": 3.339286571839848e-05,
-      "loss": 96.3534,
-      "step": 112980
-    },
-    {
-      "epoch": 0.4564938973888662,
-      "grad_norm": 525.607177734375,
-      "learning_rate": 3.338957756285699e-05,
-      "loss": 67.7952,
-      "step": 112990
-    },
-    {
-      "epoch": 0.45653429865423384,
-      "grad_norm": 1665.643798828125,
-      "learning_rate": 3.338628924375638e-05,
-      "loss": 69.8561,
-      "step": 113000
-    },
-    {
-      "epoch": 0.4565746999196015,
-      "grad_norm": 1218.552490234375,
-      "learning_rate": 3.338300076116073e-05,
-      "loss": 91.2745,
-      "step": 113010
-    },
-    {
-      "epoch": 0.4566151011849691,
-      "grad_norm": 1521.2366943359375,
-      "learning_rate": 3.337971211513417e-05,
-      "loss": 59.6287,
-      "step": 113020
-    },
-    {
-      "epoch": 0.45665550245033676,
-      "grad_norm": 898.8818969726562,
-      "learning_rate": 3.337642330574081e-05,
-      "loss": 54.526,
-      "step": 113030
-    },
-    {
-      "epoch": 0.4566959037157044,
-      "grad_norm": 390.90435791015625,
-      "learning_rate": 3.3373134333044756e-05,
-      "loss": 35.1512,
-      "step": 113040
-    },
-    {
-      "epoch": 0.456736304981072,
-      "grad_norm": 788.0700073242188,
-      "learning_rate": 3.336984519711015e-05,
-      "loss": 46.5823,
-      "step": 113050
-    },
-    {
-      "epoch": 0.4567767062464396,
-      "grad_norm": 979.811767578125,
-      "learning_rate": 3.336655589800109e-05,
-      "loss": 58.1711,
-      "step": 113060
-    },
-    {
-      "epoch": 0.45681710751180726,
-      "grad_norm": 1041.4178466796875,
-      "learning_rate": 3.336326643578172e-05,
-      "loss": 55.3556,
-      "step": 113070
-    },
-    {
-      "epoch": 0.4568575087771749,
-      "grad_norm": 594.2887573242188,
-      "learning_rate": 3.3359976810516164e-05,
-      "loss": 52.3926,
-      "step": 113080
-    },
-    {
-      "epoch": 0.45689791004254254,
-      "grad_norm": 731.5540771484375,
-      "learning_rate": 3.335668702226856e-05,
-      "loss": 107.6374,
-      "step": 113090
-    },
-    {
-      "epoch": 0.4569383113079102,
-      "grad_norm": 608.2393798828125,
-      "learning_rate": 3.3353397071103046e-05,
-      "loss": 57.6193,
-      "step": 113100
-    },
-    {
-      "epoch": 0.45697871257327777,
-      "grad_norm": 617.5512084960938,
-      "learning_rate": 3.3350106957083744e-05,
-      "loss": 72.52,
-      "step": 113110
-    },
-    {
-      "epoch": 0.4570191138386454,
-      "grad_norm": 658.0664672851562,
-      "learning_rate": 3.334681668027481e-05,
-      "loss": 94.5809,
-      "step": 113120
-    },
-    {
-      "epoch": 0.45705951510401305,
-      "grad_norm": 543.9719848632812,
-      "learning_rate": 3.334352624074039e-05,
-      "loss": 41.8438,
-      "step": 113130
-    },
-    {
-      "epoch": 0.4570999163693807,
-      "grad_norm": 696.774658203125,
-      "learning_rate": 3.334023563854463e-05,
-      "loss": 76.3532,
-      "step": 113140
-    },
-    {
-      "epoch": 0.4571403176347483,
-      "grad_norm": 2398.177734375,
-      "learning_rate": 3.333694487375168e-05,
-      "loss": 38.9571,
-      "step": 113150
-    },
-    {
-      "epoch": 0.45718071890011597,
-      "grad_norm": 455.6088562011719,
-      "learning_rate": 3.33336539464257e-05,
-      "loss": 32.256,
-      "step": 113160
-    },
-    {
-      "epoch": 0.4572211201654836,
-      "grad_norm": 1092.000732421875,
-      "learning_rate": 3.3330362856630845e-05,
-      "loss": 59.1251,
-      "step": 113170
-    },
-    {
-      "epoch": 0.4572615214308512,
-      "grad_norm": 1621.383544921875,
-      "learning_rate": 3.3327071604431275e-05,
-      "loss": 43.2729,
-      "step": 113180
-    },
-    {
-      "epoch": 0.45730192269621883,
-      "grad_norm": 680.7568969726562,
-      "learning_rate": 3.3323780189891166e-05,
-      "loss": 49.1451,
-      "step": 113190
-    },
-    {
-      "epoch": 0.45734232396158647,
-      "grad_norm": 280.8765563964844,
-      "learning_rate": 3.332048861307467e-05,
-      "loss": 54.7998,
-      "step": 113200
-    },
-    {
-      "epoch": 0.4573827252269541,
-      "grad_norm": 457.403564453125,
-      "learning_rate": 3.331719687404597e-05,
-      "loss": 91.7487,
-      "step": 113210
-    },
-    {
-      "epoch": 0.45742312649232175,
-      "grad_norm": 777.8742065429688,
-      "learning_rate": 3.331390497286922e-05,
-      "loss": 44.5807,
-      "step": 113220
-    },
-    {
-      "epoch": 0.4574635277576894,
-      "grad_norm": 1098.077880859375,
-      "learning_rate": 3.331061290960863e-05,
-      "loss": 90.1487,
-      "step": 113230
-    },
-    {
-      "epoch": 0.457503929023057,
-      "grad_norm": 882.6914672851562,
-      "learning_rate": 3.3307320684328354e-05,
-      "loss": 68.7481,
-      "step": 113240
-    },
-    {
-      "epoch": 0.4575443302884246,
-      "grad_norm": 1128.127197265625,
-      "learning_rate": 3.330402829709258e-05,
-      "loss": 65.0538,
-      "step": 113250
-    },
-    {
-      "epoch": 0.45758473155379226,
-      "grad_norm": 937.3806762695312,
-      "learning_rate": 3.3300735747965505e-05,
-      "loss": 57.1227,
-      "step": 113260
-    },
-    {
-      "epoch": 0.4576251328191599,
-      "grad_norm": 911.514404296875,
-      "learning_rate": 3.329744303701132e-05,
-      "loss": 85.5697,
-      "step": 113270
-    },
-    {
-      "epoch": 0.45766553408452754,
-      "grad_norm": 532.0978393554688,
-      "learning_rate": 3.3294150164294204e-05,
-      "loss": 41.7875,
-      "step": 113280
-    },
-    {
-      "epoch": 0.4577059353498952,
-      "grad_norm": 2276.76904296875,
-      "learning_rate": 3.329085712987836e-05,
-      "loss": 58.2938,
-      "step": 113290
-    },
-    {
-      "epoch": 0.4577463366152628,
-      "grad_norm": 633.0645751953125,
-      "learning_rate": 3.3287563933827995e-05,
-      "loss": 45.8421,
-      "step": 113300
-    },
-    {
-      "epoch": 0.4577867378806304,
-      "grad_norm": 884.1083984375,
-      "learning_rate": 3.328427057620729e-05,
-      "loss": 69.57,
-      "step": 113310
-    },
-    {
-      "epoch": 0.45782713914599804,
-      "grad_norm": 832.331298828125,
-      "learning_rate": 3.328097705708047e-05,
-      "loss": 36.6988,
-      "step": 113320
-    },
-    {
-      "epoch": 0.4578675404113657,
-      "grad_norm": 1298.6260986328125,
-      "learning_rate": 3.3277683376511744e-05,
-      "loss": 67.1316,
-      "step": 113330
-    },
-    {
-      "epoch": 0.4579079416767333,
-      "grad_norm": 690.5106811523438,
-      "learning_rate": 3.327438953456532e-05,
-      "loss": 43.1363,
-      "step": 113340
-    },
-    {
-      "epoch": 0.45794834294210096,
-      "grad_norm": 638.0509643554688,
-      "learning_rate": 3.327109553130541e-05,
-      "loss": 64.9828,
-      "step": 113350
-    },
-    {
-      "epoch": 0.4579887442074686,
-      "grad_norm": 712.4091186523438,
-      "learning_rate": 3.326780136679623e-05,
-      "loss": 51.8144,
-      "step": 113360
-    },
-    {
-      "epoch": 0.4580291454728362,
-      "grad_norm": 768.2869873046875,
-      "learning_rate": 3.326450704110201e-05,
-      "loss": 43.4623,
-      "step": 113370
-    },
-    {
-      "epoch": 0.4580695467382038,
-      "grad_norm": 667.932373046875,
-      "learning_rate": 3.3261212554286975e-05,
-      "loss": 60.1954,
-      "step": 113380
-    },
-    {
-      "epoch": 0.45810994800357147,
-      "grad_norm": 797.756591796875,
-      "learning_rate": 3.3257917906415336e-05,
-      "loss": 64.3577,
-      "step": 113390
-    },
-    {
-      "epoch": 0.4581503492689391,
-      "grad_norm": 417.9449462890625,
-      "learning_rate": 3.325462309755134e-05,
-      "loss": 52.2687,
-      "step": 113400
-    },
-    {
-      "epoch": 0.45819075053430675,
-      "grad_norm": 131.72840881347656,
-      "learning_rate": 3.325132812775922e-05,
-      "loss": 53.6158,
-      "step": 113410
-    },
-    {
-      "epoch": 0.4582311517996744,
-      "grad_norm": 1277.050537109375,
-      "learning_rate": 3.324803299710321e-05,
-      "loss": 79.7133,
-      "step": 113420
-    },
-    {
-      "epoch": 0.45827155306504197,
-      "grad_norm": 247.889404296875,
-      "learning_rate": 3.3244737705647554e-05,
-      "loss": 79.97,
-      "step": 113430
-    },
-    {
-      "epoch": 0.4583119543304096,
-      "grad_norm": 1504.002685546875,
-      "learning_rate": 3.324144225345649e-05,
-      "loss": 57.3291,
-      "step": 113440
-    },
-    {
-      "epoch": 0.45835235559577725,
-      "grad_norm": 3024.2744140625,
-      "learning_rate": 3.3238146640594256e-05,
-      "loss": 55.5882,
-      "step": 113450
-    },
-    {
-      "epoch": 0.4583927568611449,
-      "grad_norm": 1827.9388427734375,
-      "learning_rate": 3.323485086712513e-05,
-      "loss": 68.6746,
-      "step": 113460
-    },
-    {
-      "epoch": 0.45843315812651253,
-      "grad_norm": 1997.11279296875,
-      "learning_rate": 3.323155493311334e-05,
-      "loss": 60.3449,
-      "step": 113470
-    },
-    {
-      "epoch": 0.45847355939188017,
-      "grad_norm": 1310.3616943359375,
-      "learning_rate": 3.322825883862314e-05,
-      "loss": 58.0651,
-      "step": 113480
-    },
-    {
-      "epoch": 0.4585139606572478,
-      "grad_norm": 542.6761474609375,
-      "learning_rate": 3.32249625837188e-05,
-      "loss": 45.5163,
-      "step": 113490
-    },
-    {
-      "epoch": 0.4585543619226154,
-      "grad_norm": 109.23764038085938,
-      "learning_rate": 3.322166616846458e-05,
-      "loss": 46.8832,
-      "step": 113500
-    },
-    {
-      "epoch": 0.45859476318798303,
-      "grad_norm": 770.7872314453125,
-      "learning_rate": 3.321836959292475e-05,
-      "loss": 54.6829,
-      "step": 113510
-    },
-    {
-      "epoch": 0.4586351644533507,
-      "grad_norm": 1391.15185546875,
-      "learning_rate": 3.321507285716357e-05,
-      "loss": 76.5766,
-      "step": 113520
-    },
-    {
-      "epoch": 0.4586755657187183,
-      "grad_norm": 1156.2738037109375,
-      "learning_rate": 3.321177596124532e-05,
-      "loss": 44.4677,
-      "step": 113530
-    },
-    {
-      "epoch": 0.45871596698408595,
-      "grad_norm": 882.244140625,
-      "learning_rate": 3.3208478905234274e-05,
-      "loss": 64.2383,
-      "step": 113540
-    },
-    {
-      "epoch": 0.4587563682494536,
-      "grad_norm": 1060.12841796875,
-      "learning_rate": 3.32051816891947e-05,
-      "loss": 49.1175,
-      "step": 113550
-    },
-    {
-      "epoch": 0.4587967695148212,
-      "grad_norm": 1234.741455078125,
-      "learning_rate": 3.320188431319088e-05,
-      "loss": 41.2987,
-      "step": 113560
-    },
-    {
-      "epoch": 0.4588371707801888,
-      "grad_norm": 437.5050048828125,
-      "learning_rate": 3.31985867772871e-05,
-      "loss": 63.9297,
-      "step": 113570
-    },
-    {
-      "epoch": 0.45887757204555646,
-      "grad_norm": 839.0947875976562,
-      "learning_rate": 3.319528908154766e-05,
-      "loss": 55.2399,
-      "step": 113580
-    },
-    {
-      "epoch": 0.4589179733109241,
-      "grad_norm": 1059.20654296875,
-      "learning_rate": 3.319199122603683e-05,
-      "loss": 35.0855,
-      "step": 113590
-    },
-    {
-      "epoch": 0.45895837457629174,
-      "grad_norm": 1006.2760009765625,
-      "learning_rate": 3.318869321081892e-05,
-      "loss": 75.6239,
-      "step": 113600
-    },
-    {
-      "epoch": 0.4589987758416594,
-      "grad_norm": 1427.097412109375,
-      "learning_rate": 3.3185395035958224e-05,
-      "loss": 53.2945,
-      "step": 113610
-    },
-    {
-      "epoch": 0.459039177107027,
-      "grad_norm": 960.8377075195312,
-      "learning_rate": 3.318209670151904e-05,
-      "loss": 74.3849,
-      "step": 113620
-    },
-    {
-      "epoch": 0.4590795783723946,
-      "grad_norm": 780.0336303710938,
-      "learning_rate": 3.317879820756566e-05,
-      "loss": 59.6619,
-      "step": 113630
-    },
-    {
-      "epoch": 0.45911997963776224,
-      "grad_norm": 913.7399291992188,
-      "learning_rate": 3.31754995541624e-05,
-      "loss": 54.3696,
-      "step": 113640
-    },
-    {
-      "epoch": 0.4591603809031299,
-      "grad_norm": 884.738037109375,
-      "learning_rate": 3.3172200741373563e-05,
-      "loss": 87.1222,
-      "step": 113650
-    },
-    {
-      "epoch": 0.4592007821684975,
-      "grad_norm": 495.6034851074219,
-      "learning_rate": 3.3168901769263474e-05,
-      "loss": 52.4804,
-      "step": 113660
-    },
-    {
-      "epoch": 0.45924118343386516,
-      "grad_norm": 1088.6365966796875,
-      "learning_rate": 3.316560263789643e-05,
-      "loss": 46.7347,
-      "step": 113670
-    },
-    {
-      "epoch": 0.4592815846992328,
-      "grad_norm": 724.1251831054688,
-      "learning_rate": 3.3162303347336764e-05,
-      "loss": 53.7025,
-      "step": 113680
-    },
-    {
-      "epoch": 0.4593219859646004,
-      "grad_norm": 982.661376953125,
-      "learning_rate": 3.315900389764879e-05,
-      "loss": 47.1129,
-      "step": 113690
-    },
-    {
-      "epoch": 0.459362387229968,
-      "grad_norm": 662.5863037109375,
-      "learning_rate": 3.315570428889684e-05,
-      "loss": 35.4883,
-      "step": 113700
-    },
-    {
-      "epoch": 0.45940278849533567,
-      "grad_norm": 1490.281005859375,
-      "learning_rate": 3.315240452114523e-05,
-      "loss": 52.5399,
-      "step": 113710
-    },
-    {
-      "epoch": 0.4594431897607033,
-      "grad_norm": 1237.041748046875,
-      "learning_rate": 3.31491045944583e-05,
-      "loss": 59.784,
-      "step": 113720
-    },
-    {
-      "epoch": 0.45948359102607095,
-      "grad_norm": 700.3053588867188,
-      "learning_rate": 3.314580450890038e-05,
-      "loss": 70.0428,
-      "step": 113730
-    },
-    {
-      "epoch": 0.4595239922914386,
-      "grad_norm": 851.7976684570312,
-      "learning_rate": 3.3142504264535804e-05,
-      "loss": 50.4964,
-      "step": 113740
-    },
-    {
-      "epoch": 0.45956439355680617,
-      "grad_norm": 983.1741943359375,
-      "learning_rate": 3.313920386142892e-05,
-      "loss": 59.0461,
-      "step": 113750
-    },
-    {
-      "epoch": 0.4596047948221738,
-      "grad_norm": 1348.2669677734375,
-      "learning_rate": 3.313590329964406e-05,
-      "loss": 59.1032,
-      "step": 113760
-    },
-    {
-      "epoch": 0.45964519608754145,
-      "grad_norm": 487.41680908203125,
-      "learning_rate": 3.313260257924558e-05,
-      "loss": 87.3492,
-      "step": 113770
-    },
-    {
-      "epoch": 0.4596855973529091,
-      "grad_norm": 345.0389099121094,
-      "learning_rate": 3.312930170029783e-05,
-      "loss": 52.9883,
-      "step": 113780
-    },
-    {
-      "epoch": 0.45972599861827673,
-      "grad_norm": 438.7192077636719,
-      "learning_rate": 3.3126000662865156e-05,
-      "loss": 49.1873,
-      "step": 113790
-    },
-    {
-      "epoch": 0.45976639988364437,
-      "grad_norm": 493.15130615234375,
-      "learning_rate": 3.312269946701191e-05,
-      "loss": 50.4528,
-      "step": 113800
-    },
-    {
-      "epoch": 0.459806801149012,
-      "grad_norm": 1520.73974609375,
-      "learning_rate": 3.311939811280246e-05,
-      "loss": 42.8416,
-      "step": 113810
-    },
-    {
-      "epoch": 0.4598472024143796,
-      "grad_norm": 146.88551330566406,
-      "learning_rate": 3.311609660030117e-05,
-      "loss": 52.2549,
-      "step": 113820
-    },
-    {
-      "epoch": 0.45988760367974724,
-      "grad_norm": 610.1900634765625,
-      "learning_rate": 3.311279492957239e-05,
-      "loss": 81.4292,
-      "step": 113830
-    },
-    {
-      "epoch": 0.4599280049451149,
-      "grad_norm": 478.4176330566406,
-      "learning_rate": 3.31094931006805e-05,
-      "loss": 59.0037,
-      "step": 113840
-    },
-    {
-      "epoch": 0.4599684062104825,
-      "grad_norm": 698.8412475585938,
-      "learning_rate": 3.310619111368986e-05,
-      "loss": 60.8137,
-      "step": 113850
-    },
-    {
-      "epoch": 0.46000880747585016,
-      "grad_norm": 719.0433349609375,
-      "learning_rate": 3.310288896866486e-05,
-      "loss": 40.1776,
-      "step": 113860
-    },
-    {
-      "epoch": 0.4600492087412178,
-      "grad_norm": 807.5015258789062,
-      "learning_rate": 3.309958666566986e-05,
-      "loss": 87.1863,
-      "step": 113870
-    },
-    {
-      "epoch": 0.4600896100065854,
-      "grad_norm": 773.2158813476562,
-      "learning_rate": 3.309628420476926e-05,
-      "loss": 65.442,
-      "step": 113880
-    },
-    {
-      "epoch": 0.460130011271953,
-      "grad_norm": 1120.7901611328125,
-      "learning_rate": 3.309298158602742e-05,
-      "loss": 46.1566,
-      "step": 113890
-    },
-    {
-      "epoch": 0.46017041253732066,
-      "grad_norm": 2974.2939453125,
-      "learning_rate": 3.308967880950874e-05,
-      "loss": 86.9134,
-      "step": 113900
-    },
-    {
-      "epoch": 0.4602108138026883,
-      "grad_norm": 1336.4696044921875,
-      "learning_rate": 3.308637587527761e-05,
-      "loss": 79.6506,
-      "step": 113910
-    },
-    {
-      "epoch": 0.46025121506805594,
-      "grad_norm": 1928.9310302734375,
-      "learning_rate": 3.3083072783398416e-05,
-      "loss": 96.6911,
-      "step": 113920
-    },
-    {
-      "epoch": 0.4602916163334236,
-      "grad_norm": 538.5294799804688,
-      "learning_rate": 3.3079769533935556e-05,
-      "loss": 49.8037,
-      "step": 113930
-    },
-    {
-      "epoch": 0.4603320175987912,
-      "grad_norm": 809.7352294921875,
-      "learning_rate": 3.307646612695343e-05,
-      "loss": 61.5456,
-      "step": 113940
-    },
-    {
-      "epoch": 0.4603724188641588,
-      "grad_norm": 550.5420532226562,
-      "learning_rate": 3.307316256251644e-05,
-      "loss": 49.7282,
-      "step": 113950
-    },
-    {
-      "epoch": 0.46041282012952645,
-      "grad_norm": 1422.148193359375,
-      "learning_rate": 3.3069858840688994e-05,
-      "loss": 71.9855,
-      "step": 113960
-    },
-    {
-      "epoch": 0.4604532213948941,
-      "grad_norm": 426.9280090332031,
-      "learning_rate": 3.3066554961535485e-05,
-      "loss": 46.9851,
-      "step": 113970
-    },
-    {
-      "epoch": 0.4604936226602617,
-      "grad_norm": 844.1886596679688,
-      "learning_rate": 3.3063250925120334e-05,
-      "loss": 56.5634,
-      "step": 113980
-    },
-    {
-      "epoch": 0.46053402392562937,
-      "grad_norm": 590.6248779296875,
-      "learning_rate": 3.305994673150797e-05,
-      "loss": 58.149,
-      "step": 113990
-    },
-    {
-      "epoch": 0.460574425190997,
-      "grad_norm": 722.6728515625,
-      "learning_rate": 3.305664238076278e-05,
-      "loss": 48.5719,
-      "step": 114000
-    },
-    {
-      "epoch": 0.4606148264563646,
-      "grad_norm": 761.7833251953125,
-      "learning_rate": 3.30533378729492e-05,
-      "loss": 67.1143,
-      "step": 114010
-    },
-    {
-      "epoch": 0.46065522772173223,
-      "grad_norm": 258.24639892578125,
-      "learning_rate": 3.3050033208131656e-05,
-      "loss": 50.417,
-      "step": 114020
-    },
-    {
-      "epoch": 0.46069562898709987,
-      "grad_norm": 391.4142150878906,
-      "learning_rate": 3.304672838637457e-05,
-      "loss": 65.8437,
-      "step": 114030
-    },
-    {
-      "epoch": 0.4607360302524675,
-      "grad_norm": 432.7169189453125,
-      "learning_rate": 3.3043423407742375e-05,
-      "loss": 60.9631,
-      "step": 114040
-    },
-    {
-      "epoch": 0.46077643151783515,
-      "grad_norm": 893.4138793945312,
-      "learning_rate": 3.3040118272299495e-05,
-      "loss": 131.9281,
-      "step": 114050
-    },
-    {
-      "epoch": 0.4608168327832028,
-      "grad_norm": 1183.3709716796875,
-      "learning_rate": 3.303681298011037e-05,
-      "loss": 74.6201,
-      "step": 114060
-    },
-    {
-      "epoch": 0.4608572340485704,
-      "grad_norm": 327.8617248535156,
-      "learning_rate": 3.303350753123944e-05,
-      "loss": 38.2665,
-      "step": 114070
-    },
-    {
-      "epoch": 0.460897635313938,
-      "grad_norm": 642.1771850585938,
-      "learning_rate": 3.3030201925751145e-05,
-      "loss": 50.6443,
-      "step": 114080
-    },
-    {
-      "epoch": 0.46093803657930565,
-      "grad_norm": 1546.3465576171875,
-      "learning_rate": 3.302689616370993e-05,
-      "loss": 37.8828,
-      "step": 114090
-    },
-    {
-      "epoch": 0.4609784378446733,
-      "grad_norm": 634.73876953125,
-      "learning_rate": 3.302359024518024e-05,
-      "loss": 29.7485,
-      "step": 114100
-    },
-    {
-      "epoch": 0.46101883911004093,
-      "grad_norm": 658.0567626953125,
-      "learning_rate": 3.302028417022653e-05,
-      "loss": 45.2375,
-      "step": 114110
-    },
-    {
-      "epoch": 0.4610592403754086,
-      "grad_norm": 633.6270751953125,
-      "learning_rate": 3.301697793891324e-05,
-      "loss": 59.6125,
-      "step": 114120
-    },
-    {
-      "epoch": 0.4610996416407762,
-      "grad_norm": 1438.4256591796875,
-      "learning_rate": 3.301367155130485e-05,
-      "loss": 89.7295,
-      "step": 114130
-    },
-    {
-      "epoch": 0.4611400429061438,
-      "grad_norm": 1383.4329833984375,
-      "learning_rate": 3.3010365007465805e-05,
-      "loss": 48.6195,
-      "step": 114140
-    },
-    {
-      "epoch": 0.46118044417151144,
-      "grad_norm": 530.7550659179688,
-      "learning_rate": 3.300705830746057e-05,
-      "loss": 46.5861,
-      "step": 114150
-    },
-    {
-      "epoch": 0.4612208454368791,
-      "grad_norm": 633.077880859375,
-      "learning_rate": 3.300375145135361e-05,
-      "loss": 60.9828,
-      "step": 114160
-    },
-    {
-      "epoch": 0.4612612467022467,
-      "grad_norm": 376.91864013671875,
-      "learning_rate": 3.3000444439209396e-05,
-      "loss": 62.8593,
-      "step": 114170
-    },
-    {
-      "epoch": 0.46130164796761436,
-      "grad_norm": 721.732666015625,
-      "learning_rate": 3.299713727109239e-05,
-      "loss": 84.59,
-      "step": 114180
-    },
-    {
-      "epoch": 0.461342049232982,
-      "grad_norm": 473.7373352050781,
-      "learning_rate": 3.299382994706709e-05,
-      "loss": 94.0822,
-      "step": 114190
-    },
-    {
-      "epoch": 0.4613824504983496,
-      "grad_norm": 1528.2044677734375,
-      "learning_rate": 3.299052246719795e-05,
-      "loss": 81.7651,
-      "step": 114200
-    },
-    {
-      "epoch": 0.4614228517637172,
-      "grad_norm": 890.8794555664062,
-      "learning_rate": 3.298721483154946e-05,
-      "loss": 40.4416,
-      "step": 114210
-    },
-    {
-      "epoch": 0.46146325302908486,
-      "grad_norm": 541.0910034179688,
-      "learning_rate": 3.298390704018611e-05,
-      "loss": 43.575,
-      "step": 114220
-    },
-    {
-      "epoch": 0.4615036542944525,
-      "grad_norm": 915.8447875976562,
-      "learning_rate": 3.298059909317239e-05,
-      "loss": 70.035,
-      "step": 114230
-    },
-    {
-      "epoch": 0.46154405555982014,
-      "grad_norm": 1335.07275390625,
-      "learning_rate": 3.297729099057277e-05,
-      "loss": 58.3776,
-      "step": 114240
-    },
-    {
-      "epoch": 0.4615844568251878,
-      "grad_norm": 1571.07373046875,
-      "learning_rate": 3.2973982732451755e-05,
-      "loss": 72.0075,
-      "step": 114250
-    },
-    {
-      "epoch": 0.4616248580905554,
-      "grad_norm": 1263.757080078125,
-      "learning_rate": 3.297067431887384e-05,
-      "loss": 53.4101,
-      "step": 114260
-    },
-    {
-      "epoch": 0.461665259355923,
-      "grad_norm": 916.6146850585938,
-      "learning_rate": 3.296736574990353e-05,
-      "loss": 78.4652,
-      "step": 114270
-    },
-    {
-      "epoch": 0.46170566062129065,
-      "grad_norm": 365.46142578125,
-      "learning_rate": 3.296405702560532e-05,
-      "loss": 80.6376,
-      "step": 114280
-    },
-    {
-      "epoch": 0.4617460618866583,
-      "grad_norm": 944.3591918945312,
-      "learning_rate": 3.2960748146043716e-05,
-      "loss": 90.1059,
-      "step": 114290
-    },
-    {
-      "epoch": 0.46178646315202593,
-      "grad_norm": 679.8099365234375,
-      "learning_rate": 3.295743911128324e-05,
-      "loss": 48.7843,
-      "step": 114300
-    },
-    {
-      "epoch": 0.46182686441739357,
-      "grad_norm": 1261.843017578125,
-      "learning_rate": 3.295412992138838e-05,
-      "loss": 78.7052,
-      "step": 114310
-    },
-    {
-      "epoch": 0.4618672656827612,
-      "grad_norm": 481.5779724121094,
-      "learning_rate": 3.295082057642367e-05,
-      "loss": 41.2433,
-      "step": 114320
-    },
-    {
-      "epoch": 0.4619076669481288,
-      "grad_norm": 903.3895874023438,
-      "learning_rate": 3.294751107645361e-05,
-      "loss": 48.583,
-      "step": 114330
-    },
-    {
-      "epoch": 0.46194806821349643,
-      "grad_norm": 730.4287109375,
-      "learning_rate": 3.294420142154274e-05,
-      "loss": 78.3227,
-      "step": 114340
-    },
-    {
-      "epoch": 0.46198846947886407,
-      "grad_norm": 563.7612915039062,
-      "learning_rate": 3.2940891611755564e-05,
-      "loss": 53.6732,
-      "step": 114350
-    },
-    {
-      "epoch": 0.4620288707442317,
-      "grad_norm": 1336.9444580078125,
-      "learning_rate": 3.293758164715663e-05,
-      "loss": 55.2327,
-      "step": 114360
-    },
-    {
-      "epoch": 0.46206927200959935,
-      "grad_norm": 1316.6781005859375,
-      "learning_rate": 3.293427152781044e-05,
-      "loss": 39.9807,
-      "step": 114370
-    },
-    {
-      "epoch": 0.462109673274967,
-      "grad_norm": 1014.697265625,
-      "learning_rate": 3.2930961253781554e-05,
-      "loss": 81.932,
-      "step": 114380
-    },
-    {
-      "epoch": 0.4621500745403346,
-      "grad_norm": 2144.41064453125,
-      "learning_rate": 3.292765082513449e-05,
-      "loss": 80.4216,
-      "step": 114390
-    },
-    {
-      "epoch": 0.4621904758057022,
-      "grad_norm": 509.5421447753906,
-      "learning_rate": 3.29243402419338e-05,
-      "loss": 40.3778,
-      "step": 114400
-    },
-    {
-      "epoch": 0.46223087707106986,
-      "grad_norm": 432.00933837890625,
-      "learning_rate": 3.2921029504244004e-05,
-      "loss": 76.2071,
-      "step": 114410
-    },
-    {
-      "epoch": 0.4622712783364375,
-      "grad_norm": 722.176025390625,
-      "learning_rate": 3.2917718612129665e-05,
-      "loss": 41.5058,
-      "step": 114420
-    },
-    {
-      "epoch": 0.46231167960180514,
-      "grad_norm": 725.9029541015625,
-      "learning_rate": 3.291440756565533e-05,
-      "loss": 53.8693,
-      "step": 114430
-    },
-    {
-      "epoch": 0.4623520808671728,
-      "grad_norm": 281.3984375,
-      "learning_rate": 3.2911096364885544e-05,
-      "loss": 80.6305,
-      "step": 114440
-    },
-    {
-      "epoch": 0.4623924821325404,
-      "grad_norm": 811.3935546875,
-      "learning_rate": 3.290778500988485e-05,
-      "loss": 55.3654,
-      "step": 114450
-    },
-    {
-      "epoch": 0.462432883397908,
-      "grad_norm": 472.8229064941406,
-      "learning_rate": 3.2904473500717824e-05,
-      "loss": 59.1575,
-      "step": 114460
-    },
-    {
-      "epoch": 0.46247328466327564,
-      "grad_norm": 713.5052490234375,
-      "learning_rate": 3.290116183744902e-05,
-      "loss": 50.5053,
-      "step": 114470
-    },
-    {
-      "epoch": 0.4625136859286433,
-      "grad_norm": 1182.7664794921875,
-      "learning_rate": 3.2897850020143005e-05,
-      "loss": 59.1632,
-      "step": 114480
-    },
-    {
-      "epoch": 0.4625540871940109,
-      "grad_norm": 845.945556640625,
-      "learning_rate": 3.289453804886433e-05,
-      "loss": 85.3159,
-      "step": 114490
-    },
-    {
-      "epoch": 0.46259448845937856,
-      "grad_norm": 406.4460754394531,
-      "learning_rate": 3.289122592367757e-05,
-      "loss": 39.6031,
-      "step": 114500
-    },
-    {
-      "epoch": 0.4626348897247462,
-      "grad_norm": 883.4130249023438,
-      "learning_rate": 3.288791364464729e-05,
-      "loss": 44.7248,
-      "step": 114510
-    },
-    {
-      "epoch": 0.4626752909901138,
-      "grad_norm": 447.2448425292969,
-      "learning_rate": 3.2884601211838085e-05,
-      "loss": 62.0368,
-      "step": 114520
-    },
-    {
-      "epoch": 0.4627156922554814,
-      "grad_norm": 640.0363159179688,
-      "learning_rate": 3.288128862531452e-05,
-      "loss": 56.4858,
-      "step": 114530
-    },
-    {
-      "epoch": 0.46275609352084907,
-      "grad_norm": 891.1563720703125,
-      "learning_rate": 3.287797588514117e-05,
-      "loss": 70.3198,
-      "step": 114540
-    },
-    {
-      "epoch": 0.4627964947862167,
-      "grad_norm": 628.9306030273438,
-      "learning_rate": 3.287466299138262e-05,
-      "loss": 53.9219,
-      "step": 114550
-    },
-    {
-      "epoch": 0.46283689605158435,
-      "grad_norm": 382.36456298828125,
-      "learning_rate": 3.287134994410347e-05,
-      "loss": 42.8895,
-      "step": 114560
-    },
-    {
-      "epoch": 0.462877297316952,
-      "grad_norm": 1126.9713134765625,
-      "learning_rate": 3.28680367433683e-05,
-      "loss": 38.392,
-      "step": 114570
-    },
-    {
-      "epoch": 0.4629176985823196,
-      "grad_norm": 2824.01025390625,
-      "learning_rate": 3.28647233892417e-05,
-      "loss": 77.3813,
-      "step": 114580
-    },
-    {
-      "epoch": 0.4629580998476872,
-      "grad_norm": 413.0010986328125,
-      "learning_rate": 3.286140988178826e-05,
-      "loss": 50.3714,
-      "step": 114590
-    },
-    {
-      "epoch": 0.46299850111305485,
-      "grad_norm": 952.1806030273438,
-      "learning_rate": 3.28580962210726e-05,
-      "loss": 55.7656,
-      "step": 114600
-    },
-    {
-      "epoch": 0.4630389023784225,
-      "grad_norm": 491.023193359375,
-      "learning_rate": 3.2854782407159305e-05,
-      "loss": 47.8107,
-      "step": 114610
-    },
-    {
-      "epoch": 0.46307930364379013,
-      "grad_norm": 413.3840026855469,
-      "learning_rate": 3.285146844011298e-05,
-      "loss": 57.6984,
-      "step": 114620
-    },
-    {
-      "epoch": 0.46311970490915777,
-      "grad_norm": 593.4551391601562,
-      "learning_rate": 3.2848154319998235e-05,
-      "loss": 47.3308,
-      "step": 114630
-    },
-    {
-      "epoch": 0.4631601061745254,
-      "grad_norm": 543.7266845703125,
-      "learning_rate": 3.2844840046879686e-05,
-      "loss": 46.1537,
-      "step": 114640
-    },
-    {
-      "epoch": 0.463200507439893,
-      "grad_norm": 876.9476318359375,
-      "learning_rate": 3.2841525620821945e-05,
-      "loss": 39.367,
-      "step": 114650
-    },
-    {
-      "epoch": 0.46324090870526063,
-      "grad_norm": 470.560791015625,
-      "learning_rate": 3.2838211041889625e-05,
-      "loss": 56.22,
-      "step": 114660
-    },
-    {
-      "epoch": 0.4632813099706283,
-      "grad_norm": 692.4049072265625,
-      "learning_rate": 3.2834896310147336e-05,
-      "loss": 46.9886,
-      "step": 114670
-    },
-    {
-      "epoch": 0.4633217112359959,
-      "grad_norm": 1012.37890625,
-      "learning_rate": 3.283158142565971e-05,
-      "loss": 77.0895,
-      "step": 114680
-    },
-    {
-      "epoch": 0.46336211250136355,
-      "grad_norm": 1028.7650146484375,
-      "learning_rate": 3.282826638849138e-05,
-      "loss": 60.6332,
-      "step": 114690
-    },
-    {
-      "epoch": 0.4634025137667312,
-      "grad_norm": 1001.2539672851562,
-      "learning_rate": 3.2824951198706954e-05,
-      "loss": 55.1329,
-      "step": 114700
-    },
-    {
-      "epoch": 0.4634429150320988,
-      "grad_norm": 707.6488647460938,
-      "learning_rate": 3.2821635856371086e-05,
-      "loss": 68.9933,
-      "step": 114710
-    },
-    {
-      "epoch": 0.4634833162974664,
-      "grad_norm": 640.0159912109375,
-      "learning_rate": 3.28183203615484e-05,
-      "loss": 56.8567,
-      "step": 114720
-    },
-    {
-      "epoch": 0.46352371756283406,
-      "grad_norm": 1112.1865234375,
-      "learning_rate": 3.281500471430353e-05,
-      "loss": 63.7654,
-      "step": 114730
-    },
-    {
-      "epoch": 0.4635641188282017,
-      "grad_norm": 871.5289306640625,
-      "learning_rate": 3.281168891470112e-05,
-      "loss": 64.3242,
-      "step": 114740
-    },
-    {
-      "epoch": 0.46360452009356934,
-      "grad_norm": 232.17013549804688,
-      "learning_rate": 3.2808372962805816e-05,
-      "loss": 51.8217,
-      "step": 114750
-    },
-    {
-      "epoch": 0.463644921358937,
-      "grad_norm": 653.4492797851562,
-      "learning_rate": 3.280505685868226e-05,
-      "loss": 32.7491,
-      "step": 114760
-    },
-    {
-      "epoch": 0.4636853226243046,
-      "grad_norm": 806.7274780273438,
-      "learning_rate": 3.2801740602395105e-05,
-      "loss": 60.581,
-      "step": 114770
-    },
-    {
-      "epoch": 0.4637257238896722,
-      "grad_norm": 698.8681640625,
-      "learning_rate": 3.279842419400899e-05,
-      "loss": 81.199,
-      "step": 114780
-    },
-    {
-      "epoch": 0.46376612515503984,
-      "grad_norm": 1318.1488037109375,
-      "learning_rate": 3.2795107633588586e-05,
-      "loss": 61.5727,
-      "step": 114790
-    },
-    {
-      "epoch": 0.4638065264204075,
-      "grad_norm": 2012.5032958984375,
-      "learning_rate": 3.279179092119855e-05,
-      "loss": 63.0752,
-      "step": 114800
-    },
-    {
-      "epoch": 0.4638469276857751,
-      "grad_norm": 946.7776489257812,
-      "learning_rate": 3.278847405690353e-05,
-      "loss": 104.6616,
-      "step": 114810
-    },
-    {
-      "epoch": 0.46388732895114276,
-      "grad_norm": 332.1543273925781,
-      "learning_rate": 3.278515704076821e-05,
-      "loss": 57.2958,
-      "step": 114820
-    },
-    {
-      "epoch": 0.4639277302165104,
-      "grad_norm": 628.6360473632812,
-      "learning_rate": 3.278183987285724e-05,
-      "loss": 49.5415,
-      "step": 114830
-    },
-    {
-      "epoch": 0.463968131481878,
-      "grad_norm": 407.3933410644531,
-      "learning_rate": 3.277852255323529e-05,
-      "loss": 45.7567,
-      "step": 114840
-    },
-    {
-      "epoch": 0.46400853274724563,
-      "grad_norm": 1091.437255859375,
-      "learning_rate": 3.277520508196705e-05,
-      "loss": 76.9394,
-      "step": 114850
-    },
-    {
-      "epoch": 0.46404893401261327,
-      "grad_norm": 540.4053344726562,
-      "learning_rate": 3.277188745911717e-05,
-      "loss": 42.3653,
-      "step": 114860
-    },
-    {
-      "epoch": 0.4640893352779809,
-      "grad_norm": 1648.4383544921875,
-      "learning_rate": 3.276856968475035e-05,
-      "loss": 78.4631,
-      "step": 114870
-    },
-    {
-      "epoch": 0.46412973654334855,
-      "grad_norm": 469.8823547363281,
-      "learning_rate": 3.276525175893126e-05,
-      "loss": 59.8111,
-      "step": 114880
-    },
-    {
-      "epoch": 0.4641701378087162,
-      "grad_norm": 925.3463745117188,
-      "learning_rate": 3.27619336817246e-05,
-      "loss": 49.3538,
-      "step": 114890
-    },
-    {
-      "epoch": 0.46421053907408383,
-      "grad_norm": 1113.8001708984375,
-      "learning_rate": 3.2758615453195034e-05,
-      "loss": 64.5263,
-      "step": 114900
-    },
-    {
-      "epoch": 0.4642509403394514,
-      "grad_norm": 591.8302612304688,
-      "learning_rate": 3.275529707340728e-05,
-      "loss": 71.7116,
-      "step": 114910
-    },
-    {
-      "epoch": 0.46429134160481905,
-      "grad_norm": 564.6202392578125,
-      "learning_rate": 3.2751978542425995e-05,
-      "loss": 58.8196,
-      "step": 114920
-    },
-    {
-      "epoch": 0.4643317428701867,
-      "grad_norm": 1191.913330078125,
-      "learning_rate": 3.2748659860315916e-05,
-      "loss": 42.0783,
-      "step": 114930
-    },
-    {
-      "epoch": 0.46437214413555433,
-      "grad_norm": 599.6019287109375,
-      "learning_rate": 3.274534102714172e-05,
-      "loss": 50.7316,
-      "step": 114940
-    },
-    {
-      "epoch": 0.46441254540092197,
-      "grad_norm": 527.0614013671875,
-      "learning_rate": 3.2742022042968104e-05,
-      "loss": 51.8205,
-      "step": 114950
-    },
-    {
-      "epoch": 0.4644529466662896,
-      "grad_norm": 665.9529418945312,
-      "learning_rate": 3.273870290785979e-05,
-      "loss": 68.3194,
-      "step": 114960
-    },
-    {
-      "epoch": 0.4644933479316572,
-      "grad_norm": 515.5640258789062,
-      "learning_rate": 3.2735383621881485e-05,
-      "loss": 37.5466,
-      "step": 114970
-    },
-    {
-      "epoch": 0.46453374919702484,
-      "grad_norm": 822.676513671875,
-      "learning_rate": 3.273206418509788e-05,
-      "loss": 69.2502,
-      "step": 114980
-    },
-    {
-      "epoch": 0.4645741504623925,
-      "grad_norm": 1226.153076171875,
-      "learning_rate": 3.272874459757371e-05,
-      "loss": 50.1505,
-      "step": 114990
-    },
-    {
-      "epoch": 0.4646145517277601,
-      "grad_norm": 1813.3848876953125,
-      "learning_rate": 3.272542485937369e-05,
-      "loss": 77.2155,
-      "step": 115000
-    },
-    {
-      "epoch": 0.46465495299312776,
-      "grad_norm": 579.1246337890625,
-      "learning_rate": 3.2722104970562525e-05,
-      "loss": 93.8051,
-      "step": 115010
-    },
-    {
-      "epoch": 0.4646953542584954,
-      "grad_norm": 0.0,
-      "learning_rate": 3.271878493120496e-05,
-      "loss": 37.53,
-      "step": 115020
-    },
-    {
-      "epoch": 0.464735755523863,
-      "grad_norm": 285.748779296875,
-      "learning_rate": 3.27154647413657e-05,
-      "loss": 51.3141,
-      "step": 115030
-    },
-    {
-      "epoch": 0.4647761567892306,
-      "grad_norm": 585.9220581054688,
-      "learning_rate": 3.271214440110948e-05,
-      "loss": 66.3542,
-      "step": 115040
-    },
-    {
-      "epoch": 0.46481655805459826,
-      "grad_norm": 385.7046813964844,
-      "learning_rate": 3.270882391050104e-05,
-      "loss": 43.7319,
-      "step": 115050
-    },
-    {
-      "epoch": 0.4648569593199659,
-      "grad_norm": 799.7272338867188,
-      "learning_rate": 3.270550326960511e-05,
-      "loss": 73.2882,
-      "step": 115060
-    },
-    {
-      "epoch": 0.46489736058533354,
-      "grad_norm": 1642.924560546875,
-      "learning_rate": 3.270218247848642e-05,
-      "loss": 90.7406,
-      "step": 115070
-    },
-    {
-      "epoch": 0.4649377618507012,
-      "grad_norm": 890.0045776367188,
-      "learning_rate": 3.269886153720972e-05,
-      "loss": 65.4913,
-      "step": 115080
-    },
-    {
-      "epoch": 0.4649781631160688,
-      "grad_norm": 1087.22802734375,
-      "learning_rate": 3.2695540445839764e-05,
-      "loss": 66.0933,
-      "step": 115090
-    },
-    {
-      "epoch": 0.4650185643814364,
-      "grad_norm": 436.0012512207031,
-      "learning_rate": 3.269221920444127e-05,
-      "loss": 48.0393,
-      "step": 115100
-    },
-    {
-      "epoch": 0.46505896564680405,
-      "grad_norm": 821.9993286132812,
-      "learning_rate": 3.2688897813079005e-05,
-      "loss": 48.355,
-      "step": 115110
-    },
-    {
-      "epoch": 0.4650993669121717,
-      "grad_norm": 434.6961364746094,
-      "learning_rate": 3.2685576271817716e-05,
-      "loss": 54.5126,
-      "step": 115120
-    },
-    {
-      "epoch": 0.4651397681775393,
-      "grad_norm": 897.4439697265625,
-      "learning_rate": 3.268225458072217e-05,
-      "loss": 62.3177,
-      "step": 115130
-    },
-    {
-      "epoch": 0.46518016944290697,
-      "grad_norm": 2384.05224609375,
-      "learning_rate": 3.267893273985711e-05,
-      "loss": 72.2372,
-      "step": 115140
-    },
-    {
-      "epoch": 0.4652205707082746,
-      "grad_norm": 724.9418334960938,
-      "learning_rate": 3.26756107492873e-05,
-      "loss": 56.3772,
-      "step": 115150
-    },
-    {
-      "epoch": 0.4652609719736422,
-      "grad_norm": 666.811767578125,
-      "learning_rate": 3.267228860907751e-05,
-      "loss": 45.4823,
-      "step": 115160
-    },
-    {
-      "epoch": 0.46530137323900983,
-      "grad_norm": 846.5064086914062,
-      "learning_rate": 3.266896631929251e-05,
-      "loss": 49.0491,
-      "step": 115170
-    },
-    {
-      "epoch": 0.46534177450437747,
-      "grad_norm": 372.02447509765625,
-      "learning_rate": 3.2665643879997056e-05,
-      "loss": 78.3602,
-      "step": 115180
-    },
-    {
-      "epoch": 0.4653821757697451,
-      "grad_norm": 745.8118896484375,
-      "learning_rate": 3.266232129125593e-05,
-      "loss": 39.577,
-      "step": 115190
-    },
-    {
-      "epoch": 0.46542257703511275,
-      "grad_norm": 936.6083984375,
-      "learning_rate": 3.2658998553133895e-05,
-      "loss": 43.7622,
-      "step": 115200
-    },
-    {
-      "epoch": 0.4654629783004804,
-      "grad_norm": 458.98114013671875,
-      "learning_rate": 3.2655675665695754e-05,
-      "loss": 49.8332,
-      "step": 115210
-    },
-    {
-      "epoch": 0.46550337956584803,
-      "grad_norm": 636.2042846679688,
-      "learning_rate": 3.2652352629006275e-05,
-      "loss": 63.9273,
-      "step": 115220
-    },
-    {
-      "epoch": 0.4655437808312156,
-      "grad_norm": 556.1316528320312,
-      "learning_rate": 3.264902944313023e-05,
-      "loss": 46.3902,
-      "step": 115230
-    },
-    {
-      "epoch": 0.46558418209658325,
-      "grad_norm": 724.4549560546875,
-      "learning_rate": 3.2645706108132424e-05,
-      "loss": 54.3436,
-      "step": 115240
-    },
-    {
-      "epoch": 0.4656245833619509,
-      "grad_norm": 546.86767578125,
-      "learning_rate": 3.264238262407764e-05,
-      "loss": 50.2506,
-      "step": 115250
-    },
-    {
-      "epoch": 0.46566498462731853,
-      "grad_norm": 784.9192504882812,
-      "learning_rate": 3.263905899103068e-05,
-      "loss": 40.6821,
-      "step": 115260
-    },
-    {
-      "epoch": 0.4657053858926862,
-      "grad_norm": 686.7406616210938,
-      "learning_rate": 3.263573520905633e-05,
-      "loss": 52.2234,
-      "step": 115270
-    },
-    {
-      "epoch": 0.4657457871580538,
-      "grad_norm": 709.7639770507812,
-      "learning_rate": 3.263241127821938e-05,
-      "loss": 57.2129,
-      "step": 115280
-    },
-    {
-      "epoch": 0.4657861884234214,
-      "grad_norm": 704.1000366210938,
-      "learning_rate": 3.262908719858466e-05,
-      "loss": 44.5228,
-      "step": 115290
-    },
-    {
-      "epoch": 0.46582658968878904,
-      "grad_norm": 444.74029541015625,
-      "learning_rate": 3.262576297021695e-05,
-      "loss": 49.4663,
-      "step": 115300
-    },
-    {
-      "epoch": 0.4658669909541567,
-      "grad_norm": 951.1953125,
-      "learning_rate": 3.262243859318105e-05,
-      "loss": 60.8233,
-      "step": 115310
-    },
-    {
-      "epoch": 0.4659073922195243,
-      "grad_norm": 466.8185119628906,
-      "learning_rate": 3.2619114067541796e-05,
-      "loss": 92.2595,
-      "step": 115320
-    },
-    {
-      "epoch": 0.46594779348489196,
-      "grad_norm": 581.5831298828125,
-      "learning_rate": 3.2615789393363995e-05,
-      "loss": 53.4414,
-      "step": 115330
-    },
-    {
-      "epoch": 0.4659881947502596,
-      "grad_norm": 1562.436767578125,
-      "learning_rate": 3.261246457071245e-05,
-      "loss": 61.0114,
-      "step": 115340
-    },
-    {
-      "epoch": 0.4660285960156272,
-      "grad_norm": 1172.59130859375,
-      "learning_rate": 3.260913959965201e-05,
-      "loss": 71.2536,
-      "step": 115350
-    },
-    {
-      "epoch": 0.4660689972809948,
-      "grad_norm": 751.3859252929688,
-      "learning_rate": 3.260581448024745e-05,
-      "loss": 53.4582,
-      "step": 115360
-    },
-    {
-      "epoch": 0.46610939854636246,
-      "grad_norm": 479.7427673339844,
-      "learning_rate": 3.260248921256364e-05,
-      "loss": 38.3002,
-      "step": 115370
-    },
-    {
-      "epoch": 0.4661497998117301,
-      "grad_norm": 562.0839233398438,
-      "learning_rate": 3.2599163796665376e-05,
-      "loss": 74.0789,
-      "step": 115380
-    },
-    {
-      "epoch": 0.46619020107709774,
-      "grad_norm": 1441.49853515625,
-      "learning_rate": 3.25958382326175e-05,
-      "loss": 46.7784,
-      "step": 115390
-    },
-    {
-      "epoch": 0.4662306023424654,
-      "grad_norm": 586.0812377929688,
-      "learning_rate": 3.2592512520484856e-05,
-      "loss": 54.3181,
-      "step": 115400
-    },
-    {
-      "epoch": 0.466271003607833,
-      "grad_norm": 782.4759521484375,
-      "learning_rate": 3.2589186660332274e-05,
-      "loss": 59.9024,
-      "step": 115410
-    },
-    {
-      "epoch": 0.4663114048732006,
-      "grad_norm": 623.7604370117188,
-      "learning_rate": 3.2585860652224585e-05,
-      "loss": 52.6946,
-      "step": 115420
-    },
-    {
-      "epoch": 0.46635180613856825,
-      "grad_norm": 958.4971313476562,
-      "learning_rate": 3.2582534496226644e-05,
-      "loss": 56.1301,
-      "step": 115430
-    },
-    {
-      "epoch": 0.4663922074039359,
-      "grad_norm": 803.4427490234375,
-      "learning_rate": 3.257920819240328e-05,
-      "loss": 56.7052,
-      "step": 115440
-    },
-    {
-      "epoch": 0.46643260866930353,
-      "grad_norm": 481.8475646972656,
-      "learning_rate": 3.2575881740819355e-05,
-      "loss": 53.755,
-      "step": 115450
-    },
-    {
-      "epoch": 0.46647300993467117,
-      "grad_norm": 542.654052734375,
-      "learning_rate": 3.257255514153971e-05,
-      "loss": 66.2002,
-      "step": 115460
-    },
-    {
-      "epoch": 0.4665134112000388,
-      "grad_norm": 468.3689270019531,
-      "learning_rate": 3.256922839462921e-05,
-      "loss": 60.2367,
-      "step": 115470
-    },
-    {
-      "epoch": 0.4665538124654064,
-      "grad_norm": 1403.767333984375,
-      "learning_rate": 3.25659015001527e-05,
-      "loss": 108.456,
-      "step": 115480
-    },
-    {
-      "epoch": 0.46659421373077403,
-      "grad_norm": 708.3703002929688,
-      "learning_rate": 3.2562574458175044e-05,
-      "loss": 46.7288,
-      "step": 115490
-    },
-    {
-      "epoch": 0.4666346149961417,
-      "grad_norm": 1207.593017578125,
-      "learning_rate": 3.2559247268761115e-05,
-      "loss": 63.8019,
-      "step": 115500
-    },
-    {
-      "epoch": 0.4666750162615093,
-      "grad_norm": 919.4540405273438,
-      "learning_rate": 3.2555919931975766e-05,
-      "loss": 71.2822,
-      "step": 115510
-    },
-    {
-      "epoch": 0.46671541752687695,
-      "grad_norm": 2360.607421875,
-      "learning_rate": 3.2552592447883865e-05,
-      "loss": 63.8303,
-      "step": 115520
-    },
-    {
-      "epoch": 0.4667558187922446,
-      "grad_norm": 1612.3497314453125,
-      "learning_rate": 3.254926481655028e-05,
-      "loss": 45.8642,
-      "step": 115530
-    },
-    {
-      "epoch": 0.4667962200576122,
-      "grad_norm": 1008.3920288085938,
-      "learning_rate": 3.25459370380399e-05,
-      "loss": 49.0776,
-      "step": 115540
-    },
-    {
-      "epoch": 0.4668366213229798,
-      "grad_norm": 2217.866943359375,
-      "learning_rate": 3.254260911241759e-05,
-      "loss": 72.4225,
-      "step": 115550
-    },
-    {
-      "epoch": 0.46687702258834746,
-      "grad_norm": 413.4029235839844,
-      "learning_rate": 3.253928103974823e-05,
-      "loss": 66.9536,
-      "step": 115560
-    },
-    {
-      "epoch": 0.4669174238537151,
-      "grad_norm": 389.1944580078125,
-      "learning_rate": 3.253595282009671e-05,
-      "loss": 32.475,
-      "step": 115570
-    },
-    {
-      "epoch": 0.46695782511908274,
-      "grad_norm": 0.0,
-      "learning_rate": 3.253262445352791e-05,
-      "loss": 60.644,
-      "step": 115580
-    },
-    {
-      "epoch": 0.4669982263844504,
-      "grad_norm": 366.5119934082031,
-      "learning_rate": 3.252929594010671e-05,
-      "loss": 59.2907,
-      "step": 115590
-    },
-    {
-      "epoch": 0.467038627649818,
-      "grad_norm": 882.8370971679688,
-      "learning_rate": 3.2525967279898015e-05,
-      "loss": 70.1194,
-      "step": 115600
-    },
-    {
-      "epoch": 0.4670790289151856,
-      "grad_norm": 955.0043334960938,
-      "learning_rate": 3.252263847296671e-05,
-      "loss": 62.6897,
-      "step": 115610
-    },
-    {
-      "epoch": 0.46711943018055324,
-      "grad_norm": 815.323486328125,
-      "learning_rate": 3.25193095193777e-05,
-      "loss": 30.4579,
-      "step": 115620
-    },
-    {
-      "epoch": 0.4671598314459209,
-      "grad_norm": 721.734130859375,
-      "learning_rate": 3.251598041919587e-05,
-      "loss": 66.6569,
-      "step": 115630
-    },
-    {
-      "epoch": 0.4672002327112885,
-      "grad_norm": 870.7584838867188,
-      "learning_rate": 3.251265117248614e-05,
-      "loss": 52.2032,
-      "step": 115640
-    },
-    {
-      "epoch": 0.46724063397665616,
-      "grad_norm": 4967.96728515625,
-      "learning_rate": 3.25093217793134e-05,
-      "loss": 56.7999,
-      "step": 115650
-    },
-    {
-      "epoch": 0.4672810352420238,
-      "grad_norm": 805.152099609375,
-      "learning_rate": 3.250599223974258e-05,
-      "loss": 52.8073,
-      "step": 115660
-    },
-    {
-      "epoch": 0.4673214365073914,
-      "grad_norm": 614.6864624023438,
-      "learning_rate": 3.250266255383857e-05,
-      "loss": 54.9746,
-      "step": 115670
-    },
-    {
-      "epoch": 0.467361837772759,
-      "grad_norm": 1058.3001708984375,
-      "learning_rate": 3.249933272166629e-05,
-      "loss": 92.6986,
-      "step": 115680
-    },
-    {
-      "epoch": 0.46740223903812667,
-      "grad_norm": 1258.4139404296875,
-      "learning_rate": 3.249600274329066e-05,
-      "loss": 54.6076,
-      "step": 115690
-    },
-    {
-      "epoch": 0.4674426403034943,
-      "grad_norm": 712.8331909179688,
-      "learning_rate": 3.24926726187766e-05,
-      "loss": 65.3554,
-      "step": 115700
-    },
-    {
-      "epoch": 0.46748304156886195,
-      "grad_norm": 826.5805053710938,
-      "learning_rate": 3.248934234818902e-05,
-      "loss": 82.2117,
-      "step": 115710
-    },
-    {
-      "epoch": 0.4675234428342296,
-      "grad_norm": 770.5543212890625,
-      "learning_rate": 3.248601193159287e-05,
-      "loss": 73.6685,
-      "step": 115720
-    },
-    {
-      "epoch": 0.4675638440995972,
-      "grad_norm": 780.5651245117188,
-      "learning_rate": 3.248268136905304e-05,
-      "loss": 44.8846,
-      "step": 115730
-    },
-    {
-      "epoch": 0.4676042453649648,
-      "grad_norm": 1183.1986083984375,
-      "learning_rate": 3.247935066063451e-05,
-      "loss": 57.1673,
-      "step": 115740
-    },
-    {
-      "epoch": 0.46764464663033245,
-      "grad_norm": 461.81573486328125,
-      "learning_rate": 3.247601980640217e-05,
-      "loss": 45.7133,
-      "step": 115750
-    },
-    {
-      "epoch": 0.4676850478957001,
-      "grad_norm": 720.4097290039062,
-      "learning_rate": 3.247268880642098e-05,
-      "loss": 44.3927,
-      "step": 115760
-    },
-    {
-      "epoch": 0.46772544916106773,
-      "grad_norm": 1013.3065795898438,
-      "learning_rate": 3.246935766075588e-05,
-      "loss": 52.5788,
-      "step": 115770
-    },
-    {
-      "epoch": 0.46776585042643537,
-      "grad_norm": 947.147216796875,
-      "learning_rate": 3.24660263694718e-05,
-      "loss": 93.3442,
-      "step": 115780
-    },
-    {
-      "epoch": 0.467806251691803,
-      "grad_norm": 676.0074462890625,
-      "learning_rate": 3.24626949326337e-05,
-      "loss": 48.2604,
-      "step": 115790
-    },
-    {
-      "epoch": 0.4678466529571706,
-      "grad_norm": 433.5167541503906,
-      "learning_rate": 3.245936335030651e-05,
-      "loss": 68.8754,
-      "step": 115800
-    },
-    {
-      "epoch": 0.46788705422253823,
-      "grad_norm": 759.6268920898438,
-      "learning_rate": 3.2456031622555197e-05,
-      "loss": 42.3743,
-      "step": 115810
-    },
-    {
-      "epoch": 0.4679274554879059,
-      "grad_norm": 650.4202880859375,
-      "learning_rate": 3.245269974944471e-05,
-      "loss": 83.6153,
-      "step": 115820
-    },
-    {
-      "epoch": 0.4679678567532735,
-      "grad_norm": 963.0789794921875,
-      "learning_rate": 3.2449367731039996e-05,
-      "loss": 60.6028,
-      "step": 115830
-    },
-    {
-      "epoch": 0.46800825801864115,
-      "grad_norm": 2292.56005859375,
-      "learning_rate": 3.244603556740603e-05,
-      "loss": 74.8116,
-      "step": 115840
-    },
-    {
-      "epoch": 0.4680486592840088,
-      "grad_norm": 457.2271423339844,
-      "learning_rate": 3.2442703258607766e-05,
-      "loss": 37.3418,
-      "step": 115850
-    },
-    {
-      "epoch": 0.4680890605493764,
-      "grad_norm": 474.98126220703125,
-      "learning_rate": 3.243937080471017e-05,
-      "loss": 61.6,
-      "step": 115860
-    },
-    {
-      "epoch": 0.468129461814744,
-      "grad_norm": 528.2426147460938,
-      "learning_rate": 3.243603820577822e-05,
-      "loss": 37.5654,
-      "step": 115870
-    },
-    {
-      "epoch": 0.46816986308011166,
-      "grad_norm": 885.8614501953125,
-      "learning_rate": 3.243270546187687e-05,
-      "loss": 39.1532,
-      "step": 115880
-    },
-    {
-      "epoch": 0.4682102643454793,
-      "grad_norm": 2204.4306640625,
-      "learning_rate": 3.242937257307109e-05,
-      "loss": 67.4763,
-      "step": 115890
-    },
-    {
-      "epoch": 0.46825066561084694,
-      "grad_norm": 1517.2777099609375,
-      "learning_rate": 3.2426039539425876e-05,
-      "loss": 53.94,
-      "step": 115900
-    },
-    {
-      "epoch": 0.4682910668762146,
-      "grad_norm": 2055.982666015625,
-      "learning_rate": 3.2422706361006194e-05,
-      "loss": 58.1359,
-      "step": 115910
-    },
-    {
-      "epoch": 0.4683314681415822,
-      "grad_norm": 607.13330078125,
-      "learning_rate": 3.241937303787703e-05,
-      "loss": 49.6527,
-      "step": 115920
-    },
-    {
-      "epoch": 0.4683718694069498,
-      "grad_norm": 899.364990234375,
-      "learning_rate": 3.2416039570103375e-05,
-      "loss": 57.977,
-      "step": 115930
-    },
-    {
-      "epoch": 0.46841227067231744,
-      "grad_norm": 654.2115478515625,
-      "learning_rate": 3.241270595775021e-05,
-      "loss": 73.5976,
-      "step": 115940
-    },
-    {
-      "epoch": 0.4684526719376851,
-      "grad_norm": 786.1331176757812,
-      "learning_rate": 3.240937220088253e-05,
-      "loss": 35.1766,
-      "step": 115950
-    },
-    {
-      "epoch": 0.4684930732030527,
-      "grad_norm": 1251.7491455078125,
-      "learning_rate": 3.240603829956531e-05,
-      "loss": 62.6786,
-      "step": 115960
-    },
-    {
-      "epoch": 0.46853347446842036,
-      "grad_norm": 1378.3673095703125,
-      "learning_rate": 3.240270425386357e-05,
-      "loss": 75.6629,
-      "step": 115970
-    },
-    {
-      "epoch": 0.468573875733788,
-      "grad_norm": 344.47747802734375,
-      "learning_rate": 3.2399370063842294e-05,
-      "loss": 46.953,
-      "step": 115980
-    },
-    {
-      "epoch": 0.4686142769991556,
-      "grad_norm": 383.6866760253906,
-      "learning_rate": 3.23960357295665e-05,
-      "loss": 47.923,
-      "step": 115990
-    },
-    {
-      "epoch": 0.46865467826452323,
-      "grad_norm": 471.9056091308594,
-      "learning_rate": 3.239270125110117e-05,
-      "loss": 77.3322,
-      "step": 116000
-    },
-    {
-      "epoch": 0.46869507952989087,
-      "grad_norm": 879.043701171875,
-      "learning_rate": 3.238936662851133e-05,
-      "loss": 48.9826,
-      "step": 116010
-    },
-    {
-      "epoch": 0.4687354807952585,
-      "grad_norm": 632.9555053710938,
-      "learning_rate": 3.2386031861861976e-05,
-      "loss": 31.8865,
-      "step": 116020
-    },
-    {
-      "epoch": 0.46877588206062615,
-      "grad_norm": 1467.3824462890625,
-      "learning_rate": 3.2382696951218135e-05,
-      "loss": 95.1815,
-      "step": 116030
-    },
-    {
-      "epoch": 0.4688162833259938,
-      "grad_norm": 336.1209411621094,
-      "learning_rate": 3.2379361896644816e-05,
-      "loss": 54.7575,
-      "step": 116040
-    },
-    {
-      "epoch": 0.46885668459136143,
-      "grad_norm": 873.735595703125,
-      "learning_rate": 3.237602669820704e-05,
-      "loss": 63.7571,
-      "step": 116050
-    },
-    {
-      "epoch": 0.468897085856729,
-      "grad_norm": 537.8552856445312,
-      "learning_rate": 3.2372691355969816e-05,
-      "loss": 55.9651,
-      "step": 116060
-    },
-    {
-      "epoch": 0.46893748712209665,
-      "grad_norm": 353.98614501953125,
-      "learning_rate": 3.2369355869998185e-05,
-      "loss": 39.4995,
-      "step": 116070
-    },
-    {
-      "epoch": 0.4689778883874643,
-      "grad_norm": 195.72950744628906,
-      "learning_rate": 3.236602024035716e-05,
-      "loss": 53.4221,
-      "step": 116080
-    },
-    {
-      "epoch": 0.46901828965283193,
-      "grad_norm": 657.9056396484375,
-      "learning_rate": 3.236268446711179e-05,
-      "loss": 66.8073,
-      "step": 116090
-    },
-    {
-      "epoch": 0.4690586909181996,
-      "grad_norm": 483.935546875,
-      "learning_rate": 3.235934855032709e-05,
-      "loss": 51.066,
-      "step": 116100
-    },
-    {
-      "epoch": 0.4690990921835672,
-      "grad_norm": 1542.11865234375,
-      "learning_rate": 3.23560124900681e-05,
-      "loss": 89.0544,
-      "step": 116110
-    },
-    {
-      "epoch": 0.4691394934489348,
-      "grad_norm": 838.8958740234375,
-      "learning_rate": 3.235267628639987e-05,
-      "loss": 60.8755,
-      "step": 116120
-    },
-    {
-      "epoch": 0.46917989471430244,
-      "grad_norm": 861.8223266601562,
-      "learning_rate": 3.234933993938742e-05,
-      "loss": 47.4594,
-      "step": 116130
-    },
-    {
-      "epoch": 0.4692202959796701,
-      "grad_norm": 550.157958984375,
-      "learning_rate": 3.2346003449095805e-05,
-      "loss": 53.5828,
-      "step": 116140
-    },
-    {
-      "epoch": 0.4692606972450377,
-      "grad_norm": 503.9146728515625,
-      "learning_rate": 3.234266681559007e-05,
-      "loss": 78.4864,
-      "step": 116150
-    },
-    {
-      "epoch": 0.46930109851040536,
-      "grad_norm": 333.4444885253906,
-      "learning_rate": 3.2339330038935265e-05,
-      "loss": 54.1187,
-      "step": 116160
-    },
-    {
-      "epoch": 0.469341499775773,
-      "grad_norm": 666.8385620117188,
-      "learning_rate": 3.233599311919644e-05,
-      "loss": 60.4023,
-      "step": 116170
-    },
-    {
-      "epoch": 0.4693819010411406,
-      "grad_norm": 641.484619140625,
-      "learning_rate": 3.233265605643866e-05,
-      "loss": 63.9357,
-      "step": 116180
-    },
-    {
-      "epoch": 0.4694223023065082,
-      "grad_norm": 1227.126953125,
-      "learning_rate": 3.232931885072697e-05,
-      "loss": 61.4396,
-      "step": 116190
-    },
-    {
-      "epoch": 0.46946270357187586,
-      "grad_norm": 1222.384765625,
-      "learning_rate": 3.2325981502126433e-05,
-      "loss": 61.0376,
-      "step": 116200
-    },
-    {
-      "epoch": 0.4695031048372435,
-      "grad_norm": 706.1589965820312,
-      "learning_rate": 3.232264401070213e-05,
-      "loss": 58.2779,
-      "step": 116210
-    },
-    {
-      "epoch": 0.46954350610261114,
-      "grad_norm": 558.5364990234375,
-      "learning_rate": 3.231930637651909e-05,
-      "loss": 52.5935,
-      "step": 116220
-    },
-    {
-      "epoch": 0.4695839073679788,
-      "grad_norm": 928.905029296875,
-      "learning_rate": 3.231596859964242e-05,
-      "loss": 61.5201,
-      "step": 116230
-    },
-    {
-      "epoch": 0.4696243086333464,
-      "grad_norm": 630.4097290039062,
-      "learning_rate": 3.2312630680137175e-05,
-      "loss": 37.5664,
-      "step": 116240
-    },
-    {
-      "epoch": 0.469664709898714,
-      "grad_norm": 1384.9384765625,
-      "learning_rate": 3.230929261806842e-05,
-      "loss": 53.7496,
-      "step": 116250
-    },
-    {
-      "epoch": 0.46970511116408165,
-      "grad_norm": 147.7491912841797,
-      "learning_rate": 3.230595441350125e-05,
-      "loss": 55.0247,
-      "step": 116260
-    },
-    {
-      "epoch": 0.4697455124294493,
-      "grad_norm": 334.8374938964844,
-      "learning_rate": 3.2302616066500735e-05,
-      "loss": 46.3881,
-      "step": 116270
-    },
-    {
-      "epoch": 0.4697859136948169,
-      "grad_norm": 789.24755859375,
-      "learning_rate": 3.229927757713196e-05,
-      "loss": 70.6806,
-      "step": 116280
-    },
-    {
-      "epoch": 0.46982631496018457,
-      "grad_norm": 1363.5418701171875,
-      "learning_rate": 3.229593894546001e-05,
-      "loss": 58.1901,
-      "step": 116290
-    },
-    {
-      "epoch": 0.4698667162255522,
-      "grad_norm": 609.9053344726562,
-      "learning_rate": 3.229260017154997e-05,
-      "loss": 98.0226,
-      "step": 116300
-    },
-    {
-      "epoch": 0.4699071174909198,
-      "grad_norm": 865.0936279296875,
-      "learning_rate": 3.228926125546695e-05,
-      "loss": 61.3911,
-      "step": 116310
-    },
-    {
-      "epoch": 0.46994751875628743,
-      "grad_norm": 590.865234375,
-      "learning_rate": 3.228592219727602e-05,
-      "loss": 86.893,
-      "step": 116320
-    },
-    {
-      "epoch": 0.46998792002165507,
-      "grad_norm": 789.493408203125,
-      "learning_rate": 3.2282582997042285e-05,
-      "loss": 49.3227,
-      "step": 116330
-    },
-    {
-      "epoch": 0.4700283212870227,
-      "grad_norm": 707.34814453125,
-      "learning_rate": 3.2279243654830836e-05,
-      "loss": 69.9071,
-      "step": 116340
-    },
-    {
-      "epoch": 0.47006872255239035,
-      "grad_norm": 1062.26123046875,
-      "learning_rate": 3.2275904170706797e-05,
-      "loss": 55.3705,
-      "step": 116350
-    },
-    {
-      "epoch": 0.470109123817758,
-      "grad_norm": 967.4932861328125,
-      "learning_rate": 3.227256454473526e-05,
-      "loss": 42.835,
-      "step": 116360
-    },
-    {
-      "epoch": 0.47014952508312563,
-      "grad_norm": 645.4476318359375,
-      "learning_rate": 3.226922477698133e-05,
-      "loss": 57.326,
-      "step": 116370
-    },
-    {
-      "epoch": 0.4701899263484932,
-      "grad_norm": 884.6676025390625,
-      "learning_rate": 3.226588486751012e-05,
-      "loss": 66.296,
-      "step": 116380
-    },
-    {
-      "epoch": 0.47023032761386085,
-      "grad_norm": 3511.986083984375,
-      "learning_rate": 3.2262544816386745e-05,
-      "loss": 49.9027,
-      "step": 116390
-    },
-    {
-      "epoch": 0.4702707288792285,
-      "grad_norm": 476.053955078125,
-      "learning_rate": 3.225920462367632e-05,
-      "loss": 68.0334,
-      "step": 116400
-    },
-    {
-      "epoch": 0.47031113014459613,
-      "grad_norm": 367.8993225097656,
-      "learning_rate": 3.225586428944396e-05,
-      "loss": 41.9805,
-      "step": 116410
-    },
-    {
-      "epoch": 0.4703515314099638,
-      "grad_norm": 500.8946533203125,
-      "learning_rate": 3.225252381375479e-05,
-      "loss": 65.6448,
-      "step": 116420
-    },
-    {
-      "epoch": 0.4703919326753314,
-      "grad_norm": 1217.2655029296875,
-      "learning_rate": 3.224918319667394e-05,
-      "loss": 83.4919,
-      "step": 116430
-    },
-    {
-      "epoch": 0.470432333940699,
-      "grad_norm": 662.4497680664062,
-      "learning_rate": 3.2245842438266526e-05,
-      "loss": 65.4877,
-      "step": 116440
-    },
-    {
-      "epoch": 0.47047273520606664,
-      "grad_norm": 725.7619018554688,
-      "learning_rate": 3.224250153859769e-05,
-      "loss": 55.515,
-      "step": 116450
-    },
-    {
-      "epoch": 0.4705131364714343,
-      "grad_norm": 910.7822265625,
-      "learning_rate": 3.223916049773256e-05,
-      "loss": 43.4893,
-      "step": 116460
-    },
-    {
-      "epoch": 0.4705535377368019,
-      "grad_norm": 2010.1138916015625,
-      "learning_rate": 3.223581931573625e-05,
-      "loss": 58.6502,
-      "step": 116470
-    },
-    {
-      "epoch": 0.47059393900216956,
-      "grad_norm": 715.5496826171875,
-      "learning_rate": 3.223247799267394e-05,
-      "loss": 84.7735,
-      "step": 116480
-    },
-    {
-      "epoch": 0.4706343402675372,
-      "grad_norm": 737.7570190429688,
-      "learning_rate": 3.2229136528610736e-05,
-      "loss": 39.7872,
-      "step": 116490
-    },
-    {
-      "epoch": 0.4706747415329048,
-      "grad_norm": 532.3171997070312,
-      "learning_rate": 3.222579492361179e-05,
-      "loss": 67.2663,
-      "step": 116500
-    },
-    {
-      "epoch": 0.4707151427982724,
-      "grad_norm": 1556.9482421875,
-      "learning_rate": 3.222245317774226e-05,
-      "loss": 57.7589,
-      "step": 116510
-    },
-    {
-      "epoch": 0.47075554406364006,
-      "grad_norm": 1564.6119384765625,
-      "learning_rate": 3.221911129106728e-05,
-      "loss": 41.327,
-      "step": 116520
-    },
-    {
-      "epoch": 0.4707959453290077,
-      "grad_norm": 668.7022094726562,
-      "learning_rate": 3.221576926365202e-05,
-      "loss": 53.1316,
-      "step": 116530
-    },
-    {
-      "epoch": 0.47083634659437534,
-      "grad_norm": 940.4009399414062,
-      "learning_rate": 3.221242709556161e-05,
-      "loss": 52.2775,
-      "step": 116540
-    },
-    {
-      "epoch": 0.470876747859743,
-      "grad_norm": 929.6431884765625,
-      "learning_rate": 3.220908478686123e-05,
-      "loss": 63.8318,
-      "step": 116550
-    },
-    {
-      "epoch": 0.4709171491251106,
-      "grad_norm": 770.2941284179688,
-      "learning_rate": 3.220574233761603e-05,
-      "loss": 50.6914,
-      "step": 116560
-    },
-    {
-      "epoch": 0.4709575503904782,
-      "grad_norm": 2690.2001953125,
-      "learning_rate": 3.220239974789117e-05,
-      "loss": 103.2043,
-      "step": 116570
-    },
-    {
-      "epoch": 0.47099795165584585,
-      "grad_norm": 744.5656127929688,
-      "learning_rate": 3.219905701775182e-05,
-      "loss": 40.3893,
-      "step": 116580
-    },
-    {
-      "epoch": 0.4710383529212135,
-      "grad_norm": 2249.509765625,
-      "learning_rate": 3.219571414726315e-05,
-      "loss": 55.5555,
-      "step": 116590
-    },
-    {
-      "epoch": 0.47107875418658113,
-      "grad_norm": 392.30657958984375,
-      "learning_rate": 3.219237113649032e-05,
-      "loss": 63.3421,
-      "step": 116600
-    },
-    {
-      "epoch": 0.47111915545194877,
-      "grad_norm": 0.0,
-      "learning_rate": 3.2189027985498514e-05,
-      "loss": 82.4716,
-      "step": 116610
-    },
-    {
-      "epoch": 0.4711595567173164,
-      "grad_norm": 626.5792846679688,
-      "learning_rate": 3.2185684694352916e-05,
-      "loss": 38.248,
-      "step": 116620
-    },
-    {
-      "epoch": 0.471199957982684,
-      "grad_norm": 0.0,
-      "learning_rate": 3.218234126311869e-05,
-      "loss": 39.9608,
-      "step": 116630
-    },
-    {
-      "epoch": 0.47124035924805163,
-      "grad_norm": 959.5233154296875,
-      "learning_rate": 3.2178997691861014e-05,
-      "loss": 56.8771,
-      "step": 116640
-    },
-    {
-      "epoch": 0.4712807605134193,
-      "grad_norm": 368.611083984375,
-      "learning_rate": 3.217565398064509e-05,
-      "loss": 30.2148,
-      "step": 116650
-    },
-    {
-      "epoch": 0.4713211617787869,
-      "grad_norm": 2147.029296875,
-      "learning_rate": 3.2172310129536096e-05,
-      "loss": 56.8126,
-      "step": 116660
-    },
-    {
-      "epoch": 0.47136156304415455,
-      "grad_norm": 528.3900146484375,
-      "learning_rate": 3.2168966138599225e-05,
-      "loss": 93.9726,
-      "step": 116670
-    },
-    {
-      "epoch": 0.4714019643095222,
-      "grad_norm": 1705.246337890625,
-      "learning_rate": 3.2165622007899676e-05,
-      "loss": 42.9779,
-      "step": 116680
-    },
-    {
-      "epoch": 0.47144236557488983,
-      "grad_norm": 1022.4556884765625,
-      "learning_rate": 3.216227773750262e-05,
-      "loss": 40.3776,
-      "step": 116690
-    },
-    {
-      "epoch": 0.4714827668402574,
-      "grad_norm": 836.2701416015625,
-      "learning_rate": 3.215893332747328e-05,
-      "loss": 75.0597,
-      "step": 116700
-    },
-    {
-      "epoch": 0.47152316810562506,
-      "grad_norm": 493.19110107421875,
-      "learning_rate": 3.2155588777876856e-05,
-      "loss": 45.9303,
-      "step": 116710
-    },
-    {
-      "epoch": 0.4715635693709927,
-      "grad_norm": 800.6790771484375,
-      "learning_rate": 3.215224408877854e-05,
-      "loss": 52.3173,
-      "step": 116720
-    },
-    {
-      "epoch": 0.47160397063636034,
-      "grad_norm": 1031.7352294921875,
-      "learning_rate": 3.2148899260243545e-05,
-      "loss": 41.5077,
-      "step": 116730
-    },
-    {
-      "epoch": 0.471644371901728,
-      "grad_norm": 524.3075561523438,
-      "learning_rate": 3.214555429233707e-05,
-      "loss": 43.6509,
-      "step": 116740
-    },
-    {
-      "epoch": 0.4716847731670956,
-      "grad_norm": 2053.65576171875,
-      "learning_rate": 3.214220918512434e-05,
-      "loss": 52.429,
-      "step": 116750
-    },
-    {
-      "epoch": 0.4717251744324632,
-      "grad_norm": 660.8265380859375,
-      "learning_rate": 3.213886393867057e-05,
-      "loss": 59.9484,
-      "step": 116760
-    },
-    {
-      "epoch": 0.47176557569783084,
-      "grad_norm": 539.3421630859375,
-      "learning_rate": 3.2135518553040964e-05,
-      "loss": 56.5854,
-      "step": 116770
-    },
-    {
-      "epoch": 0.4718059769631985,
-      "grad_norm": 461.3349304199219,
-      "learning_rate": 3.2132173028300756e-05,
-      "loss": 56.3017,
-      "step": 116780
-    },
-    {
-      "epoch": 0.4718463782285661,
-      "grad_norm": 1058.8426513671875,
-      "learning_rate": 3.212882736451516e-05,
-      "loss": 47.9099,
-      "step": 116790
-    },
-    {
-      "epoch": 0.47188677949393376,
-      "grad_norm": 521.5872802734375,
-      "learning_rate": 3.21254815617494e-05,
-      "loss": 48.9531,
-      "step": 116800
-    },
-    {
-      "epoch": 0.4719271807593014,
-      "grad_norm": 789.1767578125,
-      "learning_rate": 3.212213562006872e-05,
-      "loss": 77.1051,
-      "step": 116810
-    },
-    {
-      "epoch": 0.471967582024669,
-      "grad_norm": 710.3841552734375,
-      "learning_rate": 3.2118789539538335e-05,
-      "loss": 69.6694,
-      "step": 116820
-    },
-    {
-      "epoch": 0.4720079832900366,
-      "grad_norm": 1419.693115234375,
-      "learning_rate": 3.211544332022348e-05,
-      "loss": 65.4239,
-      "step": 116830
-    },
-    {
-      "epoch": 0.47204838455540427,
-      "grad_norm": 518.5504760742188,
-      "learning_rate": 3.21120969621894e-05,
-      "loss": 60.0513,
-      "step": 116840
-    },
-    {
-      "epoch": 0.4720887858207719,
-      "grad_norm": 385.5600280761719,
-      "learning_rate": 3.210875046550132e-05,
-      "loss": 39.933,
-      "step": 116850
-    },
-    {
-      "epoch": 0.47212918708613955,
-      "grad_norm": 850.2272338867188,
-      "learning_rate": 3.210540383022449e-05,
-      "loss": 81.0095,
-      "step": 116860
-    },
-    {
-      "epoch": 0.4721695883515072,
-      "grad_norm": 443.19189453125,
-      "learning_rate": 3.210205705642416e-05,
-      "loss": 81.4084,
-      "step": 116870
-    },
-    {
-      "epoch": 0.4722099896168748,
-      "grad_norm": 490.591796875,
-      "learning_rate": 3.209871014416557e-05,
-      "loss": 44.2058,
-      "step": 116880
-    },
-    {
-      "epoch": 0.4722503908822424,
-      "grad_norm": 1418.0242919921875,
-      "learning_rate": 3.209536309351397e-05,
-      "loss": 80.6199,
-      "step": 116890
-    },
-    {
-      "epoch": 0.47229079214761005,
-      "grad_norm": 1384.6348876953125,
-      "learning_rate": 3.209201590453461e-05,
-      "loss": 90.8569,
-      "step": 116900
-    },
-    {
-      "epoch": 0.4723311934129777,
-      "grad_norm": 595.4918823242188,
-      "learning_rate": 3.208866857729276e-05,
-      "loss": 63.6417,
-      "step": 116910
-    },
-    {
-      "epoch": 0.47237159467834533,
-      "grad_norm": 689.3577880859375,
-      "learning_rate": 3.208532111185365e-05,
-      "loss": 57.5565,
-      "step": 116920
-    },
-    {
-      "epoch": 0.47241199594371297,
-      "grad_norm": 611.20751953125,
-      "learning_rate": 3.208197350828257e-05,
-      "loss": 80.1349,
-      "step": 116930
-    },
-    {
-      "epoch": 0.4724523972090806,
-      "grad_norm": 793.7510375976562,
-      "learning_rate": 3.207862576664477e-05,
-      "loss": 73.2373,
-      "step": 116940
-    },
-    {
-      "epoch": 0.4724927984744482,
-      "grad_norm": 777.3289184570312,
-      "learning_rate": 3.207527788700551e-05,
-      "loss": 74.5804,
-      "step": 116950
-    },
-    {
-      "epoch": 0.47253319973981583,
-      "grad_norm": 1934.6875,
-      "learning_rate": 3.207192986943006e-05,
-      "loss": 78.4664,
-      "step": 116960
-    },
-    {
-      "epoch": 0.4725736010051835,
-      "grad_norm": 573.094482421875,
-      "learning_rate": 3.206858171398371e-05,
-      "loss": 77.9668,
-      "step": 116970
-    },
-    {
-      "epoch": 0.4726140022705511,
-      "grad_norm": 290.15185546875,
-      "learning_rate": 3.206523342073172e-05,
-      "loss": 56.8567,
-      "step": 116980
-    },
-    {
-      "epoch": 0.47265440353591875,
-      "grad_norm": 603.3993530273438,
-      "learning_rate": 3.206188498973935e-05,
-      "loss": 62.5748,
-      "step": 116990
-    },
-    {
-      "epoch": 0.4726948048012864,
-      "grad_norm": 459.7473449707031,
-      "learning_rate": 3.205853642107192e-05,
-      "loss": 55.5894,
-      "step": 117000
-    },
-    {
-      "epoch": 0.47273520606665403,
-      "grad_norm": 2034.221923828125,
-      "learning_rate": 3.2055187714794674e-05,
-      "loss": 94.7735,
-      "step": 117010
-    },
-    {
-      "epoch": 0.4727756073320216,
-      "grad_norm": 556.3843383789062,
-      "learning_rate": 3.205183887097291e-05,
-      "loss": 49.9171,
-      "step": 117020
-    },
-    {
-      "epoch": 0.47281600859738926,
-      "grad_norm": 618.3563232421875,
-      "learning_rate": 3.2048489889671915e-05,
-      "loss": 64.9501,
-      "step": 117030
-    },
-    {
-      "epoch": 0.4728564098627569,
-      "grad_norm": 1050.610107421875,
-      "learning_rate": 3.204514077095699e-05,
-      "loss": 45.0024,
-      "step": 117040
-    },
-    {
-      "epoch": 0.47289681112812454,
-      "grad_norm": 676.1912841796875,
-      "learning_rate": 3.2041791514893416e-05,
-      "loss": 71.348,
-      "step": 117050
-    },
-    {
-      "epoch": 0.4729372123934922,
-      "grad_norm": 338.9596252441406,
-      "learning_rate": 3.2038442121546487e-05,
-      "loss": 60.3322,
-      "step": 117060
-    },
-    {
-      "epoch": 0.4729776136588598,
-      "grad_norm": 1104.4334716796875,
-      "learning_rate": 3.2035092590981514e-05,
-      "loss": 65.3458,
-      "step": 117070
-    },
-    {
-      "epoch": 0.4730180149242274,
-      "grad_norm": 977.1143188476562,
-      "learning_rate": 3.203174292326378e-05,
-      "loss": 65.3839,
-      "step": 117080
-    },
-    {
-      "epoch": 0.47305841618959504,
-      "grad_norm": 552.0875854492188,
-      "learning_rate": 3.20283931184586e-05,
-      "loss": 52.0979,
-      "step": 117090
-    },
-    {
-      "epoch": 0.4730988174549627,
-      "grad_norm": 615.712646484375,
-      "learning_rate": 3.202504317663128e-05,
-      "loss": 32.2021,
-      "step": 117100
-    },
-    {
-      "epoch": 0.4731392187203303,
-      "grad_norm": 550.6535034179688,
-      "learning_rate": 3.2021693097847125e-05,
-      "loss": 70.6825,
-      "step": 117110
-    },
-    {
-      "epoch": 0.47317961998569796,
-      "grad_norm": 610.3817749023438,
-      "learning_rate": 3.2018342882171445e-05,
-      "loss": 36.3148,
-      "step": 117120
-    },
-    {
-      "epoch": 0.4732200212510656,
-      "grad_norm": 755.9873657226562,
-      "learning_rate": 3.2014992529669566e-05,
-      "loss": 44.023,
-      "step": 117130
-    },
-    {
-      "epoch": 0.4732604225164332,
-      "grad_norm": 847.390625,
-      "learning_rate": 3.2011642040406784e-05,
-      "loss": 49.9786,
-      "step": 117140
-    },
-    {
-      "epoch": 0.47330082378180083,
-      "grad_norm": 644.954345703125,
-      "learning_rate": 3.200829141444844e-05,
-      "loss": 93.2963,
-      "step": 117150
-    },
-    {
-      "epoch": 0.47334122504716847,
-      "grad_norm": 646.0951538085938,
-      "learning_rate": 3.2004940651859844e-05,
-      "loss": 55.8603,
-      "step": 117160
-    },
-    {
-      "epoch": 0.4733816263125361,
-      "grad_norm": 1527.83447265625,
-      "learning_rate": 3.200158975270633e-05,
-      "loss": 69.3903,
-      "step": 117170
-    },
-    {
-      "epoch": 0.47342202757790375,
-      "grad_norm": 1161.49658203125,
-      "learning_rate": 3.1998238717053206e-05,
-      "loss": 63.7593,
-      "step": 117180
-    },
-    {
-      "epoch": 0.4734624288432714,
-      "grad_norm": 1293.619140625,
-      "learning_rate": 3.199488754496582e-05,
-      "loss": 63.1005,
-      "step": 117190
-    },
-    {
-      "epoch": 0.47350283010863903,
-      "grad_norm": 505.3819580078125,
-      "learning_rate": 3.19915362365095e-05,
-      "loss": 50.2402,
-      "step": 117200
-    },
-    {
-      "epoch": 0.4735432313740066,
-      "grad_norm": 648.815185546875,
-      "learning_rate": 3.198818479174959e-05,
-      "loss": 53.0908,
-      "step": 117210
-    },
-    {
-      "epoch": 0.47358363263937425,
-      "grad_norm": 601.2451171875,
-      "learning_rate": 3.198483321075141e-05,
-      "loss": 56.0472,
-      "step": 117220
-    },
-    {
-      "epoch": 0.4736240339047419,
-      "grad_norm": 401.19842529296875,
-      "learning_rate": 3.198148149358031e-05,
-      "loss": 58.7673,
-      "step": 117230
-    },
-    {
-      "epoch": 0.47366443517010953,
-      "grad_norm": 942.248046875,
-      "learning_rate": 3.197812964030164e-05,
-      "loss": 42.9004,
-      "step": 117240
-    },
-    {
-      "epoch": 0.4737048364354772,
-      "grad_norm": 1621.0218505859375,
-      "learning_rate": 3.1974777650980735e-05,
-      "loss": 59.7592,
-      "step": 117250
-    },
-    {
-      "epoch": 0.4737452377008448,
-      "grad_norm": 1157.42431640625,
-      "learning_rate": 3.197142552568295e-05,
-      "loss": 53.8315,
-      "step": 117260
-    },
-    {
-      "epoch": 0.4737856389662124,
-      "grad_norm": 2135.51611328125,
-      "learning_rate": 3.196807326447363e-05,
-      "loss": 74.3025,
-      "step": 117270
-    },
-    {
-      "epoch": 0.47382604023158004,
-      "grad_norm": 982.652587890625,
-      "learning_rate": 3.196472086741815e-05,
-      "loss": 56.9442,
-      "step": 117280
-    },
-    {
-      "epoch": 0.4738664414969477,
-      "grad_norm": 303.63623046875,
-      "learning_rate": 3.1961368334581844e-05,
-      "loss": 44.213,
-      "step": 117290
-    },
-    {
-      "epoch": 0.4739068427623153,
-      "grad_norm": 555.4284057617188,
-      "learning_rate": 3.195801566603007e-05,
-      "loss": 53.6509,
-      "step": 117300
-    },
-    {
-      "epoch": 0.47394724402768296,
-      "grad_norm": 946.455078125,
-      "learning_rate": 3.1954662861828204e-05,
-      "loss": 82.5156,
-      "step": 117310
-    },
-    {
-      "epoch": 0.4739876452930506,
-      "grad_norm": 1169.52685546875,
-      "learning_rate": 3.195130992204161e-05,
-      "loss": 57.4077,
-      "step": 117320
-    },
-    {
-      "epoch": 0.47402804655841824,
-      "grad_norm": 595.98291015625,
-      "learning_rate": 3.1947956846735645e-05,
-      "loss": 76.41,
-      "step": 117330
-    },
-    {
-      "epoch": 0.4740684478237858,
-      "grad_norm": 463.3590393066406,
-      "learning_rate": 3.194460363597569e-05,
-      "loss": 44.9645,
-      "step": 117340
-    },
-    {
-      "epoch": 0.47410884908915346,
-      "grad_norm": 443.7630615234375,
-      "learning_rate": 3.1941250289827104e-05,
-      "loss": 49.6183,
-      "step": 117350
-    },
-    {
-      "epoch": 0.4741492503545211,
-      "grad_norm": 2856.91259765625,
-      "learning_rate": 3.193789680835527e-05,
-      "loss": 59.4347,
-      "step": 117360
-    },
-    {
-      "epoch": 0.47418965161988874,
-      "grad_norm": 502.811279296875,
-      "learning_rate": 3.193454319162557e-05,
-      "loss": 52.4119,
-      "step": 117370
-    },
-    {
-      "epoch": 0.4742300528852564,
-      "grad_norm": 867.6959228515625,
-      "learning_rate": 3.193118943970338e-05,
-      "loss": 90.0065,
-      "step": 117380
-    },
-    {
-      "epoch": 0.474270454150624,
-      "grad_norm": 1995.4403076171875,
-      "learning_rate": 3.192783555265408e-05,
-      "loss": 68.9316,
-      "step": 117390
-    },
-    {
-      "epoch": 0.4743108554159916,
-      "grad_norm": 522.066162109375,
-      "learning_rate": 3.192448153054306e-05,
-      "loss": 41.8321,
-      "step": 117400
-    },
-    {
-      "epoch": 0.47435125668135925,
-      "grad_norm": 835.8606567382812,
-      "learning_rate": 3.1921127373435714e-05,
-      "loss": 67.2701,
-      "step": 117410
-    },
-    {
-      "epoch": 0.4743916579467269,
-      "grad_norm": 628.2836303710938,
-      "learning_rate": 3.191777308139742e-05,
-      "loss": 57.3309,
-      "step": 117420
-    },
-    {
-      "epoch": 0.4744320592120945,
-      "grad_norm": 582.8585205078125,
-      "learning_rate": 3.1914418654493586e-05,
-      "loss": 79.4639,
-      "step": 117430
-    },
-    {
-      "epoch": 0.47447246047746217,
-      "grad_norm": 452.00830078125,
-      "learning_rate": 3.191106409278959e-05,
-      "loss": 57.4894,
-      "step": 117440
-    },
-    {
-      "epoch": 0.4745128617428298,
-      "grad_norm": 867.4202270507812,
-      "learning_rate": 3.1907709396350844e-05,
-      "loss": 54.852,
-      "step": 117450
-    },
-    {
-      "epoch": 0.4745532630081974,
-      "grad_norm": 375.7869567871094,
-      "learning_rate": 3.190435456524275e-05,
-      "loss": 42.05,
-      "step": 117460
-    },
-    {
-      "epoch": 0.47459366427356503,
-      "grad_norm": 453.1134033203125,
-      "learning_rate": 3.190099959953071e-05,
-      "loss": 34.9451,
-      "step": 117470
-    },
-    {
-      "epoch": 0.47463406553893267,
-      "grad_norm": 614.6144409179688,
-      "learning_rate": 3.189764449928012e-05,
-      "loss": 86.9492,
-      "step": 117480
-    },
-    {
-      "epoch": 0.4746744668043003,
-      "grad_norm": 673.782470703125,
-      "learning_rate": 3.1894289264556417e-05,
-      "loss": 69.0601,
-      "step": 117490
-    },
-    {
-      "epoch": 0.47471486806966795,
-      "grad_norm": 541.9735107421875,
-      "learning_rate": 3.1890933895424976e-05,
-      "loss": 46.3959,
-      "step": 117500
-    },
-    {
-      "epoch": 0.4747552693350356,
-      "grad_norm": 379.82891845703125,
-      "learning_rate": 3.188757839195125e-05,
-      "loss": 44.2336,
-      "step": 117510
-    },
-    {
-      "epoch": 0.47479567060040323,
-      "grad_norm": 680.2620849609375,
-      "learning_rate": 3.1884222754200625e-05,
-      "loss": 54.9478,
-      "step": 117520
-    },
-    {
-      "epoch": 0.4748360718657708,
-      "grad_norm": 1045.8995361328125,
-      "learning_rate": 3.188086698223853e-05,
-      "loss": 62.2926,
-      "step": 117530
-    },
-    {
-      "epoch": 0.47487647313113845,
-      "grad_norm": 1066.854248046875,
-      "learning_rate": 3.1877511076130404e-05,
-      "loss": 52.8988,
-      "step": 117540
-    },
-    {
-      "epoch": 0.4749168743965061,
-      "grad_norm": 1133.3275146484375,
-      "learning_rate": 3.187415503594166e-05,
-      "loss": 50.2685,
-      "step": 117550
-    },
-    {
-      "epoch": 0.47495727566187373,
-      "grad_norm": 561.2869873046875,
-      "learning_rate": 3.1870798861737705e-05,
-      "loss": 71.2999,
-      "step": 117560
-    },
-    {
-      "epoch": 0.4749976769272414,
-      "grad_norm": 905.6627197265625,
-      "learning_rate": 3.1867442553584e-05,
-      "loss": 35.9215,
-      "step": 117570
-    },
-    {
-      "epoch": 0.475038078192609,
-      "grad_norm": 2765.510009765625,
-      "learning_rate": 3.186408611154597e-05,
-      "loss": 62.86,
-      "step": 117580
-    },
-    {
-      "epoch": 0.4750784794579766,
-      "grad_norm": 3249.31298828125,
-      "learning_rate": 3.186072953568905e-05,
-      "loss": 89.4793,
-      "step": 117590
-    },
-    {
-      "epoch": 0.47511888072334424,
-      "grad_norm": 1029.83642578125,
-      "learning_rate": 3.185737282607867e-05,
-      "loss": 52.9404,
-      "step": 117600
-    },
-    {
-      "epoch": 0.4751592819887119,
-      "grad_norm": 638.6604614257812,
-      "learning_rate": 3.1854015982780275e-05,
-      "loss": 70.1836,
-      "step": 117610
-    },
-    {
-      "epoch": 0.4751996832540795,
-      "grad_norm": 866.2382202148438,
-      "learning_rate": 3.185065900585931e-05,
-      "loss": 60.6042,
-      "step": 117620
-    },
-    {
-      "epoch": 0.47524008451944716,
-      "grad_norm": 967.8104858398438,
-      "learning_rate": 3.184730189538122e-05,
-      "loss": 67.1052,
-      "step": 117630
-    },
-    {
-      "epoch": 0.4752804857848148,
-      "grad_norm": 990.4212646484375,
-      "learning_rate": 3.1843944651411456e-05,
-      "loss": 44.4866,
-      "step": 117640
-    },
-    {
-      "epoch": 0.47532088705018244,
-      "grad_norm": 528.0682983398438,
-      "learning_rate": 3.184058727401546e-05,
-      "loss": 51.6501,
-      "step": 117650
-    },
-    {
-      "epoch": 0.47536128831555,
-      "grad_norm": 936.557373046875,
-      "learning_rate": 3.1837229763258705e-05,
-      "loss": 97.2483,
-      "step": 117660
-    },
-    {
-      "epoch": 0.47540168958091766,
-      "grad_norm": 870.8411865234375,
-      "learning_rate": 3.183387211920663e-05,
-      "loss": 81.8832,
-      "step": 117670
-    },
-    {
-      "epoch": 0.4754420908462853,
-      "grad_norm": 628.0894165039062,
-      "learning_rate": 3.183051434192471e-05,
-      "loss": 44.0012,
-      "step": 117680
-    },
-    {
-      "epoch": 0.47548249211165294,
-      "grad_norm": 423.3321228027344,
-      "learning_rate": 3.1827156431478386e-05,
-      "loss": 77.5057,
-      "step": 117690
-    },
-    {
-      "epoch": 0.4755228933770206,
-      "grad_norm": 483.94091796875,
-      "learning_rate": 3.1823798387933134e-05,
-      "loss": 39.6376,
-      "step": 117700
-    },
-    {
-      "epoch": 0.4755632946423882,
-      "grad_norm": 791.345703125,
-      "learning_rate": 3.182044021135442e-05,
-      "loss": 55.5527,
-      "step": 117710
-    },
-    {
-      "epoch": 0.4756036959077558,
-      "grad_norm": 498.8603210449219,
-      "learning_rate": 3.181708190180771e-05,
-      "loss": 41.6617,
-      "step": 117720
-    },
-    {
-      "epoch": 0.47564409717312345,
-      "grad_norm": 771.0953979492188,
-      "learning_rate": 3.181372345935848e-05,
-      "loss": 48.7883,
-      "step": 117730
-    },
-    {
-      "epoch": 0.4756844984384911,
-      "grad_norm": 661.4583740234375,
-      "learning_rate": 3.1810364884072205e-05,
-      "loss": 72.9009,
-      "step": 117740
-    },
-    {
-      "epoch": 0.47572489970385873,
-      "grad_norm": 632.1604614257812,
-      "learning_rate": 3.180700617601436e-05,
-      "loss": 59.3145,
-      "step": 117750
-    },
-    {
-      "epoch": 0.47576530096922637,
-      "grad_norm": 806.9335327148438,
-      "learning_rate": 3.180364733525043e-05,
-      "loss": 41.7695,
-      "step": 117760
-    },
-    {
-      "epoch": 0.475805702234594,
-      "grad_norm": 776.6650390625,
-      "learning_rate": 3.1800288361845883e-05,
-      "loss": 62.7163,
-      "step": 117770
-    },
-    {
-      "epoch": 0.4758461034999616,
-      "grad_norm": 601.4182739257812,
-      "learning_rate": 3.179692925586622e-05,
-      "loss": 54.4891,
-      "step": 117780
-    },
-    {
-      "epoch": 0.47588650476532923,
-      "grad_norm": 907.4159545898438,
-      "learning_rate": 3.179357001737692e-05,
-      "loss": 54.7433,
-      "step": 117790
-    },
-    {
-      "epoch": 0.4759269060306969,
-      "grad_norm": 792.1842651367188,
-      "learning_rate": 3.179021064644347e-05,
-      "loss": 73.5924,
-      "step": 117800
-    },
-    {
-      "epoch": 0.4759673072960645,
-      "grad_norm": 641.330810546875,
-      "learning_rate": 3.178685114313137e-05,
-      "loss": 35.0172,
-      "step": 117810
-    },
-    {
-      "epoch": 0.47600770856143215,
-      "grad_norm": 272.5650939941406,
-      "learning_rate": 3.178349150750612e-05,
-      "loss": 77.2498,
-      "step": 117820
-    },
-    {
-      "epoch": 0.4760481098267998,
-      "grad_norm": 917.3323974609375,
-      "learning_rate": 3.1780131739633204e-05,
-      "loss": 66.6369,
-      "step": 117830
-    },
-    {
-      "epoch": 0.47608851109216743,
-      "grad_norm": 924.8410034179688,
-      "learning_rate": 3.177677183957813e-05,
-      "loss": 50.9081,
-      "step": 117840
-    },
-    {
-      "epoch": 0.476128912357535,
-      "grad_norm": 539.0849609375,
-      "learning_rate": 3.17734118074064e-05,
-      "loss": 43.3925,
-      "step": 117850
-    },
-    {
-      "epoch": 0.47616931362290266,
-      "grad_norm": 1520.1893310546875,
-      "learning_rate": 3.177005164318353e-05,
-      "loss": 52.2887,
-      "step": 117860
-    },
-    {
-      "epoch": 0.4762097148882703,
-      "grad_norm": 0.0,
-      "learning_rate": 3.1766691346974996e-05,
-      "loss": 31.6033,
-      "step": 117870
-    },
-    {
-      "epoch": 0.47625011615363794,
-      "grad_norm": 0.0,
-      "learning_rate": 3.176333091884635e-05,
-      "loss": 59.7049,
-      "step": 117880
-    },
-    {
-      "epoch": 0.4762905174190056,
-      "grad_norm": 1278.6094970703125,
-      "learning_rate": 3.175997035886307e-05,
-      "loss": 56.4383,
-      "step": 117890
-    },
-    {
-      "epoch": 0.4763309186843732,
-      "grad_norm": 1077.4521484375,
-      "learning_rate": 3.1756609667090696e-05,
-      "loss": 70.2687,
-      "step": 117900
-    },
-    {
-      "epoch": 0.4763713199497408,
-      "grad_norm": 842.562255859375,
-      "learning_rate": 3.175324884359474e-05,
-      "loss": 112.2075,
-      "step": 117910
-    },
-    {
-      "epoch": 0.47641172121510844,
-      "grad_norm": 310.8381042480469,
-      "learning_rate": 3.174988788844072e-05,
-      "loss": 48.3912,
-      "step": 117920
-    },
-    {
-      "epoch": 0.4764521224804761,
-      "grad_norm": 1312.25244140625,
-      "learning_rate": 3.1746526801694156e-05,
-      "loss": 52.8015,
-      "step": 117930
-    },
-    {
-      "epoch": 0.4764925237458437,
-      "grad_norm": 885.5872802734375,
-      "learning_rate": 3.174316558342059e-05,
-      "loss": 78.4678,
-      "step": 117940
-    },
-    {
-      "epoch": 0.47653292501121136,
-      "grad_norm": 629.9498291015625,
-      "learning_rate": 3.173980423368553e-05,
-      "loss": 57.0286,
-      "step": 117950
-    },
-    {
-      "epoch": 0.476573326276579,
-      "grad_norm": 622.9977416992188,
-      "learning_rate": 3.173644275255451e-05,
-      "loss": 61.0304,
-      "step": 117960
-    },
-    {
-      "epoch": 0.47661372754194664,
-      "grad_norm": 772.329833984375,
-      "learning_rate": 3.173308114009308e-05,
-      "loss": 34.8157,
-      "step": 117970
-    },
-    {
-      "epoch": 0.4766541288073142,
-      "grad_norm": 884.4105834960938,
-      "learning_rate": 3.1729719396366765e-05,
-      "loss": 44.0258,
-      "step": 117980
-    },
-    {
-      "epoch": 0.47669453007268187,
-      "grad_norm": 1267.8201904296875,
-      "learning_rate": 3.172635752144111e-05,
-      "loss": 53.9944,
-      "step": 117990
-    },
-    {
-      "epoch": 0.4767349313380495,
-      "grad_norm": 1093.1011962890625,
-      "learning_rate": 3.172299551538164e-05,
-      "loss": 49.2085,
-      "step": 118000
-    },
-    {
-      "epoch": 0.47677533260341715,
-      "grad_norm": 345.0997009277344,
-      "learning_rate": 3.1719633378253924e-05,
-      "loss": 48.0018,
-      "step": 118010
-    },
-    {
-      "epoch": 0.4768157338687848,
-      "grad_norm": 610.406494140625,
-      "learning_rate": 3.171627111012349e-05,
-      "loss": 52.8992,
-      "step": 118020
-    },
-    {
-      "epoch": 0.4768561351341524,
-      "grad_norm": 1354.5128173828125,
-      "learning_rate": 3.1712908711055897e-05,
-      "loss": 72.9081,
-      "step": 118030
-    },
-    {
-      "epoch": 0.47689653639952,
-      "grad_norm": 446.1787109375,
-      "learning_rate": 3.170954618111669e-05,
-      "loss": 63.6848,
-      "step": 118040
-    },
-    {
-      "epoch": 0.47693693766488765,
-      "grad_norm": 704.4627075195312,
-      "learning_rate": 3.170618352037142e-05,
-      "loss": 58.9005,
-      "step": 118050
-    },
-    {
-      "epoch": 0.4769773389302553,
-      "grad_norm": 481.2560119628906,
-      "learning_rate": 3.170282072888566e-05,
-      "loss": 46.0265,
-      "step": 118060
-    },
-    {
-      "epoch": 0.47701774019562293,
-      "grad_norm": 560.6388549804688,
-      "learning_rate": 3.169945780672495e-05,
-      "loss": 42.9609,
-      "step": 118070
-    },
-    {
-      "epoch": 0.47705814146099057,
-      "grad_norm": 2451.45556640625,
-      "learning_rate": 3.169609475395486e-05,
-      "loss": 84.6672,
-      "step": 118080
-    },
-    {
-      "epoch": 0.4770985427263582,
-      "grad_norm": 458.1382141113281,
-      "learning_rate": 3.169273157064097e-05,
-      "loss": 59.9135,
-      "step": 118090
-    },
-    {
-      "epoch": 0.4771389439917258,
-      "grad_norm": 736.26416015625,
-      "learning_rate": 3.168936825684882e-05,
-      "loss": 42.6383,
-      "step": 118100
-    },
-    {
-      "epoch": 0.47717934525709343,
-      "grad_norm": 3540.630126953125,
-      "learning_rate": 3.1686004812644e-05,
-      "loss": 58.3743,
-      "step": 118110
-    },
-    {
-      "epoch": 0.4772197465224611,
-      "grad_norm": 609.9805908203125,
-      "learning_rate": 3.1682641238092064e-05,
-      "loss": 44.9807,
-      "step": 118120
-    },
-    {
-      "epoch": 0.4772601477878287,
-      "grad_norm": 864.6240234375,
-      "learning_rate": 3.16792775332586e-05,
-      "loss": 91.4741,
-      "step": 118130
-    },
-    {
-      "epoch": 0.47730054905319635,
-      "grad_norm": 804.3956298828125,
-      "learning_rate": 3.167591369820918e-05,
-      "loss": 64.7296,
-      "step": 118140
-    },
-    {
-      "epoch": 0.477340950318564,
-      "grad_norm": 365.2998962402344,
-      "learning_rate": 3.1672549733009396e-05,
-      "loss": 50.6288,
-      "step": 118150
-    },
-    {
-      "epoch": 0.47738135158393163,
-      "grad_norm": 402.1658630371094,
-      "learning_rate": 3.166918563772481e-05,
-      "loss": 53.5217,
-      "step": 118160
-    },
-    {
-      "epoch": 0.4774217528492992,
-      "grad_norm": 668.7144165039062,
-      "learning_rate": 3.1665821412421015e-05,
-      "loss": 54.7511,
-      "step": 118170
-    },
-    {
-      "epoch": 0.47746215411466686,
-      "grad_norm": 919.4798583984375,
-      "learning_rate": 3.1662457057163604e-05,
-      "loss": 47.6506,
-      "step": 118180
-    },
-    {
-      "epoch": 0.4775025553800345,
-      "grad_norm": 1031.98193359375,
-      "learning_rate": 3.165909257201816e-05,
-      "loss": 71.545,
-      "step": 118190
-    },
-    {
-      "epoch": 0.47754295664540214,
-      "grad_norm": 390.0589294433594,
-      "learning_rate": 3.1655727957050285e-05,
-      "loss": 49.304,
-      "step": 118200
-    },
-    {
-      "epoch": 0.4775833579107698,
-      "grad_norm": 1422.1484375,
-      "learning_rate": 3.165236321232557e-05,
-      "loss": 52.9677,
-      "step": 118210
-    },
-    {
-      "epoch": 0.4776237591761374,
-      "grad_norm": 846.0404052734375,
-      "learning_rate": 3.1648998337909594e-05,
-      "loss": 62.2265,
-      "step": 118220
-    },
-    {
-      "epoch": 0.477664160441505,
-      "grad_norm": 408.1695556640625,
-      "learning_rate": 3.164563333386798e-05,
-      "loss": 67.1796,
-      "step": 118230
-    },
-    {
-      "epoch": 0.47770456170687264,
-      "grad_norm": 256.5215759277344,
-      "learning_rate": 3.1642268200266317e-05,
-      "loss": 40.5587,
-      "step": 118240
-    },
-    {
-      "epoch": 0.4777449629722403,
-      "grad_norm": 1751.73486328125,
-      "learning_rate": 3.163890293717022e-05,
-      "loss": 52.8397,
-      "step": 118250
-    },
-    {
-      "epoch": 0.4777853642376079,
-      "grad_norm": 564.2115478515625,
-      "learning_rate": 3.1635537544645296e-05,
-      "loss": 67.6077,
-      "step": 118260
-    },
-    {
-      "epoch": 0.47782576550297556,
-      "grad_norm": 1232.5323486328125,
-      "learning_rate": 3.163217202275715e-05,
-      "loss": 49.1063,
-      "step": 118270
-    },
-    {
-      "epoch": 0.4778661667683432,
-      "grad_norm": 3210.296142578125,
-      "learning_rate": 3.162880637157139e-05,
-      "loss": 69.347,
-      "step": 118280
-    },
-    {
-      "epoch": 0.47790656803371084,
-      "grad_norm": 674.1715087890625,
-      "learning_rate": 3.1625440591153645e-05,
-      "loss": 63.0254,
-      "step": 118290
-    },
-    {
-      "epoch": 0.47794696929907843,
-      "grad_norm": 369.24017333984375,
-      "learning_rate": 3.162207468156952e-05,
-      "loss": 44.4211,
-      "step": 118300
-    },
-    {
-      "epoch": 0.47798737056444607,
-      "grad_norm": 634.0050659179688,
-      "learning_rate": 3.161870864288464e-05,
-      "loss": 58.934,
-      "step": 118310
-    },
-    {
-      "epoch": 0.4780277718298137,
-      "grad_norm": 1877.2578125,
-      "learning_rate": 3.1615342475164636e-05,
-      "loss": 62.1043,
-      "step": 118320
-    },
-    {
-      "epoch": 0.47806817309518135,
-      "grad_norm": 550.0879516601562,
-      "learning_rate": 3.161197617847511e-05,
-      "loss": 41.5093,
-      "step": 118330
-    },
-    {
-      "epoch": 0.478108574360549,
-      "grad_norm": 544.1702270507812,
-      "learning_rate": 3.160860975288171e-05,
-      "loss": 59.2241,
-      "step": 118340
-    },
-    {
-      "epoch": 0.47814897562591663,
-      "grad_norm": 316.4408264160156,
-      "learning_rate": 3.1605243198450066e-05,
-      "loss": 60.8573,
-      "step": 118350
-    },
-    {
-      "epoch": 0.4781893768912842,
-      "grad_norm": 412.7470703125,
-      "learning_rate": 3.16018765152458e-05,
-      "loss": 47.1499,
-      "step": 118360
-    },
-    {
-      "epoch": 0.47822977815665185,
-      "grad_norm": 998.5579833984375,
-      "learning_rate": 3.159850970333456e-05,
-      "loss": 62.5579,
-      "step": 118370
-    },
-    {
-      "epoch": 0.4782701794220195,
-      "grad_norm": 425.9685974121094,
-      "learning_rate": 3.159514276278197e-05,
-      "loss": 55.9611,
-      "step": 118380
-    },
-    {
-      "epoch": 0.47831058068738713,
-      "grad_norm": 393.94635009765625,
-      "learning_rate": 3.1591775693653674e-05,
-      "loss": 50.0992,
-      "step": 118390
-    },
-    {
-      "epoch": 0.4783509819527548,
-      "grad_norm": 1880.3455810546875,
-      "learning_rate": 3.158840849601532e-05,
-      "loss": 65.1443,
-      "step": 118400
-    },
-    {
-      "epoch": 0.4783913832181224,
-      "grad_norm": 503.0107421875,
-      "learning_rate": 3.1585041169932545e-05,
-      "loss": 44.3258,
-      "step": 118410
-    },
-    {
-      "epoch": 0.47843178448349,
-      "grad_norm": 0.0,
-      "learning_rate": 3.1581673715471006e-05,
-      "loss": 53.8743,
-      "step": 118420
-    },
-    {
-      "epoch": 0.47847218574885764,
-      "grad_norm": 608.0613403320312,
-      "learning_rate": 3.157830613269635e-05,
-      "loss": 54.61,
-      "step": 118430
-    },
-    {
-      "epoch": 0.4785125870142253,
-      "grad_norm": 427.8965148925781,
-      "learning_rate": 3.157493842167423e-05,
-      "loss": 63.0965,
-      "step": 118440
-    },
-    {
-      "epoch": 0.4785529882795929,
-      "grad_norm": 507.14190673828125,
-      "learning_rate": 3.15715705824703e-05,
-      "loss": 50.7432,
-      "step": 118450
-    },
-    {
-      "epoch": 0.47859338954496056,
-      "grad_norm": 983.0687255859375,
-      "learning_rate": 3.156820261515022e-05,
-      "loss": 87.0616,
-      "step": 118460
-    },
-    {
-      "epoch": 0.4786337908103282,
-      "grad_norm": 1234.0418701171875,
-      "learning_rate": 3.1564834519779647e-05,
-      "loss": 48.0925,
-      "step": 118470
-    },
-    {
-      "epoch": 0.47867419207569584,
-      "grad_norm": 935.4703369140625,
-      "learning_rate": 3.156146629642425e-05,
-      "loss": 53.1847,
-      "step": 118480
-    },
-    {
-      "epoch": 0.4787145933410634,
-      "grad_norm": 820.0416259765625,
-      "learning_rate": 3.155809794514968e-05,
-      "loss": 77.3875,
-      "step": 118490
-    },
-    {
-      "epoch": 0.47875499460643106,
-      "grad_norm": 877.6055297851562,
-      "learning_rate": 3.155472946602162e-05,
-      "loss": 59.0853,
-      "step": 118500
-    },
-    {
-      "epoch": 0.4787953958717987,
-      "grad_norm": 979.28662109375,
-      "learning_rate": 3.155136085910573e-05,
-      "loss": 60.1543,
-      "step": 118510
-    },
-    {
-      "epoch": 0.47883579713716634,
-      "grad_norm": 702.0006713867188,
-      "learning_rate": 3.15479921244677e-05,
-      "loss": 44.0273,
-      "step": 118520
-    },
-    {
-      "epoch": 0.478876198402534,
-      "grad_norm": 1405.5360107421875,
-      "learning_rate": 3.1544623262173176e-05,
-      "loss": 63.5543,
-      "step": 118530
-    },
-    {
-      "epoch": 0.4789165996679016,
-      "grad_norm": 971.0574340820312,
-      "learning_rate": 3.1541254272287865e-05,
-      "loss": 55.8479,
-      "step": 118540
-    },
-    {
-      "epoch": 0.4789570009332692,
-      "grad_norm": 1137.547119140625,
-      "learning_rate": 3.153788515487742e-05,
-      "loss": 66.1714,
-      "step": 118550
-    },
-    {
-      "epoch": 0.47899740219863685,
-      "grad_norm": 813.9231567382812,
-      "learning_rate": 3.153451591000756e-05,
-      "loss": 102.5478,
-      "step": 118560
-    },
-    {
-      "epoch": 0.4790378034640045,
-      "grad_norm": 870.7437133789062,
-      "learning_rate": 3.153114653774393e-05,
-      "loss": 81.2061,
-      "step": 118570
-    },
-    {
-      "epoch": 0.4790782047293721,
-      "grad_norm": 984.029296875,
-      "learning_rate": 3.152777703815223e-05,
-      "loss": 64.9325,
-      "step": 118580
-    },
-    {
-      "epoch": 0.47911860599473977,
-      "grad_norm": 853.7675170898438,
-      "learning_rate": 3.152440741129817e-05,
-      "loss": 69.4327,
-      "step": 118590
-    },
-    {
-      "epoch": 0.4791590072601074,
-      "grad_norm": 1823.9547119140625,
-      "learning_rate": 3.152103765724743e-05,
-      "loss": 51.6635,
-      "step": 118600
-    },
-    {
-      "epoch": 0.479199408525475,
-      "grad_norm": 1002.6013793945312,
-      "learning_rate": 3.1517667776065696e-05,
-      "loss": 53.4132,
-      "step": 118610
-    },
-    {
-      "epoch": 0.47923980979084263,
-      "grad_norm": 1199.445556640625,
-      "learning_rate": 3.151429776781868e-05,
-      "loss": 57.5089,
-      "step": 118620
-    },
-    {
-      "epoch": 0.47928021105621027,
-      "grad_norm": 545.897705078125,
-      "learning_rate": 3.151092763257206e-05,
-      "loss": 66.9297,
-      "step": 118630
-    },
-    {
-      "epoch": 0.4793206123215779,
-      "grad_norm": 1726.234375,
-      "learning_rate": 3.150755737039157e-05,
-      "loss": 53.6566,
-      "step": 118640
-    },
-    {
-      "epoch": 0.47936101358694555,
-      "grad_norm": 719.9454956054688,
-      "learning_rate": 3.150418698134289e-05,
-      "loss": 48.0167,
-      "step": 118650
-    },
-    {
-      "epoch": 0.4794014148523132,
-      "grad_norm": 864.4921264648438,
-      "learning_rate": 3.150081646549174e-05,
-      "loss": 84.6051,
-      "step": 118660
-    },
-    {
-      "epoch": 0.47944181611768083,
-      "grad_norm": 824.414306640625,
-      "learning_rate": 3.149744582290383e-05,
-      "loss": 53.0983,
-      "step": 118670
-    },
-    {
-      "epoch": 0.4794822173830484,
-      "grad_norm": 471.1961975097656,
-      "learning_rate": 3.149407505364486e-05,
-      "loss": 69.1806,
-      "step": 118680
-    },
-    {
-      "epoch": 0.47952261864841605,
-      "grad_norm": 684.9292602539062,
-      "learning_rate": 3.149070415778056e-05,
-      "loss": 43.8025,
-      "step": 118690
-    },
-    {
-      "epoch": 0.4795630199137837,
-      "grad_norm": 507.1294860839844,
-      "learning_rate": 3.148733313537664e-05,
-      "loss": 53.6068,
-      "step": 118700
-    },
-    {
-      "epoch": 0.47960342117915133,
-      "grad_norm": 1452.51953125,
-      "learning_rate": 3.148396198649882e-05,
-      "loss": 62.2304,
-      "step": 118710
-    },
-    {
-      "epoch": 0.479643822444519,
-      "grad_norm": 568.4696655273438,
-      "learning_rate": 3.148059071121282e-05,
-      "loss": 69.3875,
-      "step": 118720
-    },
-    {
-      "epoch": 0.4796842237098866,
-      "grad_norm": 527.501953125,
-      "learning_rate": 3.147721930958437e-05,
-      "loss": 53.4141,
-      "step": 118730
-    },
-    {
-      "epoch": 0.4797246249752542,
-      "grad_norm": 1219.31640625,
-      "learning_rate": 3.14738477816792e-05,
-      "loss": 58.4269,
-      "step": 118740
-    },
-    {
-      "epoch": 0.47976502624062184,
-      "grad_norm": 1039.8255615234375,
-      "learning_rate": 3.147047612756302e-05,
-      "loss": 39.0244,
-      "step": 118750
-    },
-    {
-      "epoch": 0.4798054275059895,
-      "grad_norm": 1056.8941650390625,
-      "learning_rate": 3.146710434730159e-05,
-      "loss": 54.2339,
-      "step": 118760
-    },
-    {
-      "epoch": 0.4798458287713571,
-      "grad_norm": 756.7150268554688,
-      "learning_rate": 3.1463732440960625e-05,
-      "loss": 53.7804,
-      "step": 118770
-    },
-    {
-      "epoch": 0.47988623003672476,
-      "grad_norm": 990.2318725585938,
-      "learning_rate": 3.1460360408605866e-05,
-      "loss": 45.0402,
-      "step": 118780
-    },
-    {
-      "epoch": 0.4799266313020924,
-      "grad_norm": 782.805908203125,
-      "learning_rate": 3.145698825030307e-05,
-      "loss": 52.3285,
-      "step": 118790
-    },
-    {
-      "epoch": 0.47996703256746004,
-      "grad_norm": 781.7633666992188,
-      "learning_rate": 3.145361596611795e-05,
-      "loss": 53.3584,
-      "step": 118800
-    },
-    {
-      "epoch": 0.4800074338328276,
-      "grad_norm": 429.9420166015625,
-      "learning_rate": 3.1450243556116266e-05,
-      "loss": 61.5989,
-      "step": 118810
-    },
-    {
-      "epoch": 0.48004783509819526,
-      "grad_norm": 1086.17822265625,
-      "learning_rate": 3.144687102036376e-05,
-      "loss": 44.8307,
-      "step": 118820
-    },
-    {
-      "epoch": 0.4800882363635629,
-      "grad_norm": 858.8822631835938,
-      "learning_rate": 3.1443498358926186e-05,
-      "loss": 63.265,
-      "step": 118830
-    },
-    {
-      "epoch": 0.48012863762893054,
-      "grad_norm": 530.9268188476562,
-      "learning_rate": 3.1440125571869306e-05,
-      "loss": 45.8561,
-      "step": 118840
-    },
-    {
-      "epoch": 0.4801690388942982,
-      "grad_norm": 426.1288146972656,
-      "learning_rate": 3.143675265925885e-05,
-      "loss": 52.3246,
-      "step": 118850
-    },
-    {
-      "epoch": 0.4802094401596658,
-      "grad_norm": 374.76849365234375,
-      "learning_rate": 3.1433379621160586e-05,
-      "loss": 44.3146,
-      "step": 118860
-    },
-    {
-      "epoch": 0.4802498414250334,
-      "grad_norm": 727.5894165039062,
-      "learning_rate": 3.143000645764028e-05,
-      "loss": 45.2602,
-      "step": 118870
-    },
-    {
-      "epoch": 0.48029024269040105,
-      "grad_norm": 914.193603515625,
-      "learning_rate": 3.142663316876368e-05,
-      "loss": 67.587,
-      "step": 118880
-    },
-    {
-      "epoch": 0.4803306439557687,
-      "grad_norm": 119.14258575439453,
-      "learning_rate": 3.1423259754596576e-05,
-      "loss": 48.5501,
-      "step": 118890
-    },
-    {
-      "epoch": 0.48037104522113633,
-      "grad_norm": 947.4161376953125,
-      "learning_rate": 3.1419886215204694e-05,
-      "loss": 56.8426,
-      "step": 118900
-    },
-    {
-      "epoch": 0.48041144648650397,
-      "grad_norm": 593.0452270507812,
-      "learning_rate": 3.1416512550653835e-05,
-      "loss": 70.3968,
-      "step": 118910
-    },
-    {
-      "epoch": 0.4804518477518716,
-      "grad_norm": 757.6504516601562,
-      "learning_rate": 3.141313876100976e-05,
-      "loss": 59.1005,
-      "step": 118920
-    },
-    {
-      "epoch": 0.4804922490172392,
-      "grad_norm": 3180.365966796875,
-      "learning_rate": 3.1409764846338245e-05,
-      "loss": 64.8498,
-      "step": 118930
-    },
-    {
-      "epoch": 0.48053265028260683,
-      "grad_norm": 566.63671875,
-      "learning_rate": 3.140639080670507e-05,
-      "loss": 64.9276,
-      "step": 118940
-    },
-    {
-      "epoch": 0.4805730515479745,
-      "grad_norm": 714.4104614257812,
-      "learning_rate": 3.140301664217599e-05,
-      "loss": 59.2069,
-      "step": 118950
-    },
-    {
-      "epoch": 0.4806134528133421,
-      "grad_norm": 575.2021484375,
-      "learning_rate": 3.139964235281682e-05,
-      "loss": 43.4193,
-      "step": 118960
-    },
-    {
-      "epoch": 0.48065385407870975,
-      "grad_norm": 1377.7861328125,
-      "learning_rate": 3.1396267938693316e-05,
-      "loss": 63.2737,
-      "step": 118970
-    },
-    {
-      "epoch": 0.4806942553440774,
-      "grad_norm": 661.6658935546875,
-      "learning_rate": 3.1392893399871295e-05,
-      "loss": 60.5852,
-      "step": 118980
-    },
-    {
-      "epoch": 0.48073465660944503,
-      "grad_norm": 513.0794067382812,
-      "learning_rate": 3.1389518736416507e-05,
-      "loss": 50.5535,
-      "step": 118990
-    },
-    {
-      "epoch": 0.4807750578748126,
-      "grad_norm": 1068.8321533203125,
-      "learning_rate": 3.138614394839476e-05,
-      "loss": 56.0899,
-      "step": 119000
-    },
-    {
-      "epoch": 0.48081545914018026,
-      "grad_norm": 346.4224548339844,
-      "learning_rate": 3.138276903587186e-05,
-      "loss": 62.7875,
-      "step": 119010
-    },
-    {
-      "epoch": 0.4808558604055479,
-      "grad_norm": 591.9736328125,
-      "learning_rate": 3.137939399891359e-05,
-      "loss": 50.5829,
-      "step": 119020
-    },
-    {
-      "epoch": 0.48089626167091554,
-      "grad_norm": 762.9810791015625,
-      "learning_rate": 3.1376018837585747e-05,
-      "loss": 66.904,
-      "step": 119030
-    },
-    {
-      "epoch": 0.4809366629362832,
-      "grad_norm": 700.6834106445312,
-      "learning_rate": 3.137264355195413e-05,
-      "loss": 66.1691,
-      "step": 119040
-    },
-    {
-      "epoch": 0.4809770642016508,
-      "grad_norm": 315.9108581542969,
-      "learning_rate": 3.1369268142084556e-05,
-      "loss": 49.7968,
-      "step": 119050
-    },
-    {
-      "epoch": 0.4810174654670184,
-      "grad_norm": 0.0,
-      "learning_rate": 3.136589260804282e-05,
-      "loss": 47.6848,
-      "step": 119060
-    },
-    {
-      "epoch": 0.48105786673238604,
-      "grad_norm": 395.9146423339844,
-      "learning_rate": 3.1362516949894725e-05,
-      "loss": 57.0576,
-      "step": 119070
-    },
-    {
-      "epoch": 0.4810982679977537,
-      "grad_norm": 476.3524475097656,
-      "learning_rate": 3.135914116770609e-05,
-      "loss": 58.1471,
-      "step": 119080
-    },
-    {
-      "epoch": 0.4811386692631213,
-      "grad_norm": 626.7835083007812,
-      "learning_rate": 3.135576526154272e-05,
-      "loss": 39.7096,
-      "step": 119090
-    },
-    {
-      "epoch": 0.48117907052848896,
-      "grad_norm": 1119.859619140625,
-      "learning_rate": 3.135238923147043e-05,
-      "loss": 87.1097,
-      "step": 119100
-    },
-    {
-      "epoch": 0.4812194717938566,
-      "grad_norm": 804.6915283203125,
-      "learning_rate": 3.1349013077555045e-05,
-      "loss": 36.7657,
-      "step": 119110
-    },
-    {
-      "epoch": 0.48125987305922424,
-      "grad_norm": 769.036865234375,
-      "learning_rate": 3.134563679986238e-05,
-      "loss": 40.638,
-      "step": 119120
-    },
-    {
-      "epoch": 0.4813002743245918,
-      "grad_norm": 191.26966857910156,
-      "learning_rate": 3.134226039845827e-05,
-      "loss": 50.2752,
-      "step": 119130
-    },
-    {
-      "epoch": 0.48134067558995947,
-      "grad_norm": 778.8662109375,
-      "learning_rate": 3.1338883873408516e-05,
-      "loss": 58.7929,
-      "step": 119140
-    },
-    {
-      "epoch": 0.4813810768553271,
-      "grad_norm": 514.7074584960938,
-      "learning_rate": 3.133550722477896e-05,
-      "loss": 66.8902,
-      "step": 119150
-    },
-    {
-      "epoch": 0.48142147812069475,
-      "grad_norm": 355.43096923828125,
-      "learning_rate": 3.133213045263543e-05,
-      "loss": 58.968,
-      "step": 119160
-    },
-    {
-      "epoch": 0.4814618793860624,
-      "grad_norm": 3042.458984375,
-      "learning_rate": 3.132875355704376e-05,
-      "loss": 88.0314,
-      "step": 119170
-    },
-    {
-      "epoch": 0.48150228065143,
-      "grad_norm": 656.7515869140625,
-      "learning_rate": 3.1325376538069776e-05,
-      "loss": 68.5067,
-      "step": 119180
-    },
-    {
-      "epoch": 0.4815426819167976,
-      "grad_norm": 495.6517028808594,
-      "learning_rate": 3.132199939577932e-05,
-      "loss": 73.5789,
-      "step": 119190
-    },
-    {
-      "epoch": 0.48158308318216525,
-      "grad_norm": 1507.7938232421875,
-      "learning_rate": 3.1318622130238236e-05,
-      "loss": 55.6776,
-      "step": 119200
-    },
-    {
-      "epoch": 0.4816234844475329,
-      "grad_norm": 802.6456298828125,
-      "learning_rate": 3.1315244741512356e-05,
-      "loss": 49.8764,
-      "step": 119210
-    },
-    {
-      "epoch": 0.48166388571290053,
-      "grad_norm": 814.6461181640625,
-      "learning_rate": 3.131186722966753e-05,
-      "loss": 51.0948,
-      "step": 119220
-    },
-    {
-      "epoch": 0.48170428697826817,
-      "grad_norm": 826.6341552734375,
-      "learning_rate": 3.1308489594769605e-05,
-      "loss": 54.8559,
-      "step": 119230
-    },
-    {
-      "epoch": 0.4817446882436358,
-      "grad_norm": 536.5158081054688,
-      "learning_rate": 3.1305111836884425e-05,
-      "loss": 49.7287,
-      "step": 119240
-    },
-    {
-      "epoch": 0.4817850895090034,
-      "grad_norm": 1080.1739501953125,
-      "learning_rate": 3.130173395607785e-05,
-      "loss": 96.4672,
-      "step": 119250
-    },
-    {
-      "epoch": 0.48182549077437103,
-      "grad_norm": 875.5479125976562,
-      "learning_rate": 3.129835595241571e-05,
-      "loss": 41.3818,
-      "step": 119260
-    },
-    {
-      "epoch": 0.4818658920397387,
-      "grad_norm": 977.9671020507812,
-      "learning_rate": 3.129497782596389e-05,
-      "loss": 44.1907,
-      "step": 119270
-    },
-    {
-      "epoch": 0.4819062933051063,
-      "grad_norm": 469.0617370605469,
-      "learning_rate": 3.129159957678824e-05,
-      "loss": 50.5765,
-      "step": 119280
-    },
-    {
-      "epoch": 0.48194669457047395,
-      "grad_norm": 851.140869140625,
-      "learning_rate": 3.128822120495462e-05,
-      "loss": 50.6099,
-      "step": 119290
-    },
-    {
-      "epoch": 0.4819870958358416,
-      "grad_norm": 608.3344116210938,
-      "learning_rate": 3.1284842710528876e-05,
-      "loss": 72.0118,
-      "step": 119300
-    },
-    {
-      "epoch": 0.48202749710120923,
-      "grad_norm": 380.56195068359375,
-      "learning_rate": 3.128146409357689e-05,
-      "loss": 83.0865,
-      "step": 119310
-    },
-    {
-      "epoch": 0.4820678983665768,
-      "grad_norm": 601.8255615234375,
-      "learning_rate": 3.127808535416454e-05,
-      "loss": 48.3086,
-      "step": 119320
-    },
-    {
-      "epoch": 0.48210829963194446,
-      "grad_norm": 949.0869140625,
-      "learning_rate": 3.127470649235768e-05,
-      "loss": 94.7472,
-      "step": 119330
-    },
-    {
-      "epoch": 0.4821487008973121,
-      "grad_norm": 398.5122985839844,
-      "learning_rate": 3.1271327508222174e-05,
-      "loss": 73.9752,
-      "step": 119340
-    },
-    {
-      "epoch": 0.48218910216267974,
-      "grad_norm": 683.5873413085938,
-      "learning_rate": 3.126794840182392e-05,
-      "loss": 65.0503,
-      "step": 119350
-    },
-    {
-      "epoch": 0.4822295034280474,
-      "grad_norm": 574.1619873046875,
-      "learning_rate": 3.126456917322878e-05,
-      "loss": 55.5617,
-      "step": 119360
-    },
-    {
-      "epoch": 0.482269904693415,
-      "grad_norm": 918.1862182617188,
-      "learning_rate": 3.1261189822502644e-05,
-      "loss": 48.1121,
-      "step": 119370
-    },
-    {
-      "epoch": 0.4823103059587826,
-      "grad_norm": 1102.67919921875,
-      "learning_rate": 3.125781034971139e-05,
-      "loss": 79.4511,
-      "step": 119380
-    },
-    {
-      "epoch": 0.48235070722415024,
-      "grad_norm": 865.8079223632812,
-      "learning_rate": 3.125443075492089e-05,
-      "loss": 58.5068,
-      "step": 119390
-    },
-    {
-      "epoch": 0.4823911084895179,
-      "grad_norm": 644.5775756835938,
-      "learning_rate": 3.1251051038197055e-05,
-      "loss": 53.9188,
-      "step": 119400
-    },
-    {
-      "epoch": 0.4824315097548855,
-      "grad_norm": 472.7004699707031,
-      "learning_rate": 3.124767119960576e-05,
-      "loss": 46.8185,
-      "step": 119410
-    },
-    {
-      "epoch": 0.48247191102025316,
-      "grad_norm": 988.4273681640625,
-      "learning_rate": 3.1244291239212896e-05,
-      "loss": 64.0862,
-      "step": 119420
-    },
-    {
-      "epoch": 0.4825123122856208,
-      "grad_norm": 320.8127746582031,
-      "learning_rate": 3.124091115708436e-05,
-      "loss": 66.8379,
-      "step": 119430
-    },
-    {
-      "epoch": 0.48255271355098844,
-      "grad_norm": 703.6538696289062,
-      "learning_rate": 3.123753095328604e-05,
-      "loss": 74.7884,
-      "step": 119440
-    },
-    {
-      "epoch": 0.48259311481635603,
-      "grad_norm": 741.8955078125,
-      "learning_rate": 3.123415062788385e-05,
-      "loss": 47.4477,
-      "step": 119450
-    },
-    {
-      "epoch": 0.48263351608172367,
-      "grad_norm": 764.6326904296875,
-      "learning_rate": 3.123077018094369e-05,
-      "loss": 48.5175,
-      "step": 119460
-    },
-    {
-      "epoch": 0.4826739173470913,
-      "grad_norm": 436.60516357421875,
-      "learning_rate": 3.122738961253145e-05,
-      "loss": 66.414,
-      "step": 119470
-    },
-    {
-      "epoch": 0.48271431861245895,
-      "grad_norm": 369.33343505859375,
-      "learning_rate": 3.1224008922713044e-05,
-      "loss": 71.549,
-      "step": 119480
-    },
-    {
-      "epoch": 0.4827547198778266,
-      "grad_norm": 921.5698852539062,
-      "learning_rate": 3.122062811155438e-05,
-      "loss": 50.5309,
-      "step": 119490
-    },
-    {
-      "epoch": 0.48279512114319423,
-      "grad_norm": 597.4305419921875,
-      "learning_rate": 3.121724717912138e-05,
-      "loss": 50.5436,
-      "step": 119500
-    },
-    {
-      "epoch": 0.4828355224085618,
-      "grad_norm": 1035.259765625,
-      "learning_rate": 3.121386612547993e-05,
-      "loss": 31.3353,
-      "step": 119510
-    },
-    {
-      "epoch": 0.48287592367392945,
-      "grad_norm": 1129.0030517578125,
-      "learning_rate": 3.121048495069596e-05,
-      "loss": 74.6693,
-      "step": 119520
-    },
-    {
-      "epoch": 0.4829163249392971,
-      "grad_norm": 1025.2454833984375,
-      "learning_rate": 3.1207103654835394e-05,
-      "loss": 102.4706,
-      "step": 119530
-    },
-    {
-      "epoch": 0.48295672620466473,
-      "grad_norm": 379.8052062988281,
-      "learning_rate": 3.120372223796415e-05,
-      "loss": 49.5203,
-      "step": 119540
-    },
-    {
-      "epoch": 0.4829971274700324,
-      "grad_norm": 607.2550048828125,
-      "learning_rate": 3.120034070014814e-05,
-      "loss": 70.4186,
-      "step": 119550
-    },
-    {
-      "epoch": 0.4830375287354,
-      "grad_norm": 690.1021728515625,
-      "learning_rate": 3.11969590414533e-05,
-      "loss": 43.4509,
-      "step": 119560
-    },
-    {
-      "epoch": 0.4830779300007676,
-      "grad_norm": 1215.8643798828125,
-      "learning_rate": 3.119357726194556e-05,
-      "loss": 42.1141,
-      "step": 119570
-    },
-    {
-      "epoch": 0.48311833126613524,
-      "grad_norm": 259.37872314453125,
-      "learning_rate": 3.119019536169083e-05,
-      "loss": 58.9376,
-      "step": 119580
-    },
-    {
-      "epoch": 0.4831587325315029,
-      "grad_norm": 755.9671020507812,
-      "learning_rate": 3.118681334075506e-05,
-      "loss": 45.3444,
-      "step": 119590
-    },
-    {
-      "epoch": 0.4831991337968705,
-      "grad_norm": 1142.39892578125,
-      "learning_rate": 3.118343119920418e-05,
-      "loss": 58.759,
-      "step": 119600
-    },
-    {
-      "epoch": 0.48323953506223816,
-      "grad_norm": 651.2678833007812,
-      "learning_rate": 3.1180048937104114e-05,
-      "loss": 51.0291,
-      "step": 119610
-    },
-    {
-      "epoch": 0.4832799363276058,
-      "grad_norm": 555.2761840820312,
-      "learning_rate": 3.117666655452083e-05,
-      "loss": 48.9106,
-      "step": 119620
-    },
-    {
-      "epoch": 0.48332033759297344,
-      "grad_norm": 1292.247802734375,
-      "learning_rate": 3.117328405152024e-05,
-      "loss": 67.4017,
-      "step": 119630
-    },
-    {
-      "epoch": 0.483360738858341,
-      "grad_norm": 579.1093139648438,
-      "learning_rate": 3.11699014281683e-05,
-      "loss": 56.0868,
-      "step": 119640
-    },
-    {
-      "epoch": 0.48340114012370866,
-      "grad_norm": 571.940673828125,
-      "learning_rate": 3.116651868453097e-05,
-      "loss": 57.993,
-      "step": 119650
-    },
-    {
-      "epoch": 0.4834415413890763,
-      "grad_norm": 592.5040893554688,
-      "learning_rate": 3.116313582067416e-05,
-      "loss": 51.3045,
-      "step": 119660
-    },
-    {
-      "epoch": 0.48348194265444394,
-      "grad_norm": 820.21337890625,
-      "learning_rate": 3.115975283666386e-05,
-      "loss": 50.0175,
-      "step": 119670
-    },
-    {
-      "epoch": 0.4835223439198116,
-      "grad_norm": 1067.8109130859375,
-      "learning_rate": 3.1156369732566006e-05,
-      "loss": 31.4268,
-      "step": 119680
-    },
-    {
-      "epoch": 0.4835627451851792,
-      "grad_norm": 589.9420776367188,
-      "learning_rate": 3.115298650844655e-05,
-      "loss": 85.4477,
-      "step": 119690
-    },
-    {
-      "epoch": 0.4836031464505468,
-      "grad_norm": 485.7039489746094,
-      "learning_rate": 3.114960316437145e-05,
-      "loss": 53.188,
-      "step": 119700
-    },
-    {
-      "epoch": 0.48364354771591445,
-      "grad_norm": 809.940673828125,
-      "learning_rate": 3.1146219700406674e-05,
-      "loss": 73.8831,
-      "step": 119710
-    },
-    {
-      "epoch": 0.4836839489812821,
-      "grad_norm": 1266.80419921875,
-      "learning_rate": 3.114283611661818e-05,
-      "loss": 65.2104,
-      "step": 119720
-    },
-    {
-      "epoch": 0.4837243502466497,
-      "grad_norm": 1203.2113037109375,
-      "learning_rate": 3.113945241307194e-05,
-      "loss": 64.743,
-      "step": 119730
-    },
-    {
-      "epoch": 0.48376475151201737,
-      "grad_norm": 655.8514404296875,
-      "learning_rate": 3.1136068589833914e-05,
-      "loss": 70.6579,
-      "step": 119740
-    },
-    {
-      "epoch": 0.483805152777385,
-      "grad_norm": 842.0756225585938,
-      "learning_rate": 3.1132684646970064e-05,
-      "loss": 41.7421,
-      "step": 119750
-    },
-    {
-      "epoch": 0.48384555404275265,
-      "grad_norm": 859.2930297851562,
-      "learning_rate": 3.1129300584546375e-05,
-      "loss": 61.2274,
-      "step": 119760
-    },
-    {
-      "epoch": 0.48388595530812023,
-      "grad_norm": 972.3688354492188,
-      "learning_rate": 3.1125916402628814e-05,
-      "loss": 58.9962,
-      "step": 119770
-    },
-    {
-      "epoch": 0.48392635657348787,
-      "grad_norm": 983.5337524414062,
-      "learning_rate": 3.112253210128336e-05,
-      "loss": 57.9547,
-      "step": 119780
-    },
-    {
-      "epoch": 0.4839667578388555,
-      "grad_norm": 997.2120361328125,
-      "learning_rate": 3.111914768057599e-05,
-      "loss": 47.62,
-      "step": 119790
-    },
-    {
-      "epoch": 0.48400715910422315,
-      "grad_norm": 1432.3392333984375,
-      "learning_rate": 3.111576314057268e-05,
-      "loss": 48.3912,
-      "step": 119800
-    },
-    {
-      "epoch": 0.4840475603695908,
-      "grad_norm": 502.978759765625,
-      "learning_rate": 3.1112378481339425e-05,
-      "loss": 60.9147,
-      "step": 119810
-    },
-    {
-      "epoch": 0.48408796163495843,
-      "grad_norm": 881.69580078125,
-      "learning_rate": 3.1108993702942205e-05,
-      "loss": 63.0135,
-      "step": 119820
-    },
-    {
-      "epoch": 0.484128362900326,
-      "grad_norm": 619.3832397460938,
-      "learning_rate": 3.110560880544701e-05,
-      "loss": 55.6975,
-      "step": 119830
-    },
-    {
-      "epoch": 0.48416876416569365,
-      "grad_norm": 933.4210815429688,
-      "learning_rate": 3.1102223788919824e-05,
-      "loss": 45.4596,
-      "step": 119840
-    },
-    {
-      "epoch": 0.4842091654310613,
-      "grad_norm": 1533.591552734375,
-      "learning_rate": 3.1098838653426645e-05,
-      "loss": 62.3054,
-      "step": 119850
-    },
-    {
-      "epoch": 0.48424956669642893,
-      "grad_norm": 815.6768798828125,
-      "learning_rate": 3.1095453399033466e-05,
-      "loss": 53.5172,
-      "step": 119860
-    },
-    {
-      "epoch": 0.4842899679617966,
-      "grad_norm": 1024.9600830078125,
-      "learning_rate": 3.109206802580629e-05,
-      "loss": 63.8872,
-      "step": 119870
-    },
-    {
-      "epoch": 0.4843303692271642,
-      "grad_norm": 0.0,
-      "learning_rate": 3.10886825338111e-05,
-      "loss": 48.7491,
-      "step": 119880
-    },
-    {
-      "epoch": 0.4843707704925318,
-      "grad_norm": 610.7237548828125,
-      "learning_rate": 3.108529692311391e-05,
-      "loss": 59.5073,
-      "step": 119890
-    },
-    {
-      "epoch": 0.48441117175789944,
-      "grad_norm": 999.9380493164062,
-      "learning_rate": 3.108191119378073e-05,
-      "loss": 54.3617,
-      "step": 119900
-    },
-    {
-      "epoch": 0.4844515730232671,
-      "grad_norm": 589.252685546875,
-      "learning_rate": 3.107852534587756e-05,
-      "loss": 70.6535,
-      "step": 119910
-    },
-    {
-      "epoch": 0.4844919742886347,
-      "grad_norm": 1168.3255615234375,
-      "learning_rate": 3.107513937947041e-05,
-      "loss": 65.7924,
-      "step": 119920
-    },
-    {
-      "epoch": 0.48453237555400236,
-      "grad_norm": 452.32415771484375,
-      "learning_rate": 3.107175329462529e-05,
-      "loss": 33.2393,
-      "step": 119930
-    },
-    {
-      "epoch": 0.48457277681937,
-      "grad_norm": 365.3132629394531,
-      "learning_rate": 3.106836709140821e-05,
-      "loss": 61.387,
-      "step": 119940
-    },
-    {
-      "epoch": 0.48461317808473764,
-      "grad_norm": 431.0849914550781,
-      "learning_rate": 3.1064980769885187e-05,
-      "loss": 49.9083,
-      "step": 119950
-    },
-    {
-      "epoch": 0.4846535793501052,
-      "grad_norm": 706.0770874023438,
-      "learning_rate": 3.1061594330122246e-05,
-      "loss": 41.8204,
-      "step": 119960
-    },
-    {
-      "epoch": 0.48469398061547286,
-      "grad_norm": 519.536865234375,
-      "learning_rate": 3.10582077721854e-05,
-      "loss": 39.0427,
-      "step": 119970
-    },
-    {
-      "epoch": 0.4847343818808405,
-      "grad_norm": 1396.991455078125,
-      "learning_rate": 3.1054821096140676e-05,
-      "loss": 70.4659,
-      "step": 119980
-    },
-    {
-      "epoch": 0.48477478314620814,
-      "grad_norm": 1468.1710205078125,
-      "learning_rate": 3.10514343020541e-05,
-      "loss": 91.4469,
-      "step": 119990
-    },
-    {
-      "epoch": 0.4848151844115758,
-      "grad_norm": 483.8679504394531,
-      "learning_rate": 3.104804738999169e-05,
-      "loss": 31.8263,
-      "step": 120000
-    },
-    {
-      "epoch": 0.4848555856769434,
-      "grad_norm": 1103.7783203125,
-      "learning_rate": 3.10446603600195e-05,
-      "loss": 48.717,
-      "step": 120010
-    },
-    {
-      "epoch": 0.484895986942311,
-      "grad_norm": 0.0,
-      "learning_rate": 3.104127321220353e-05,
-      "loss": 44.1308,
-      "step": 120020
-    },
-    {
-      "epoch": 0.48493638820767865,
-      "grad_norm": 2047.26025390625,
-      "learning_rate": 3.1037885946609824e-05,
-      "loss": 61.4208,
-      "step": 120030
-    },
-    {
-      "epoch": 0.4849767894730463,
-      "grad_norm": 2711.773681640625,
-      "learning_rate": 3.103449856330443e-05,
-      "loss": 70.1431,
-      "step": 120040
-    },
-    {
-      "epoch": 0.48501719073841393,
-      "grad_norm": 1064.7301025390625,
-      "learning_rate": 3.1031111062353373e-05,
-      "loss": 59.401,
-      "step": 120050
-    },
-    {
-      "epoch": 0.48505759200378157,
-      "grad_norm": 1243.82763671875,
-      "learning_rate": 3.102772344382271e-05,
-      "loss": 78.9579,
-      "step": 120060
-    },
-    {
-      "epoch": 0.4850979932691492,
-      "grad_norm": 935.9784545898438,
-      "learning_rate": 3.102433570777847e-05,
-      "loss": 56.5204,
-      "step": 120070
-    },
-    {
-      "epoch": 0.48513839453451685,
-      "grad_norm": 583.158935546875,
-      "learning_rate": 3.102094785428671e-05,
-      "loss": 63.3833,
-      "step": 120080
-    },
-    {
-      "epoch": 0.48517879579988443,
-      "grad_norm": 749.0221557617188,
-      "learning_rate": 3.101755988341347e-05,
-      "loss": 48.2709,
-      "step": 120090
-    },
-    {
-      "epoch": 0.4852191970652521,
-      "grad_norm": 876.2115478515625,
-      "learning_rate": 3.101417179522479e-05,
-      "loss": 53.4147,
-      "step": 120100
-    },
-    {
-      "epoch": 0.4852595983306197,
-      "grad_norm": 1196.973388671875,
-      "learning_rate": 3.101078358978675e-05,
-      "loss": 56.0996,
-      "step": 120110
-    },
-    {
-      "epoch": 0.48529999959598735,
-      "grad_norm": 580.7915649414062,
-      "learning_rate": 3.100739526716538e-05,
-      "loss": 58.37,
-      "step": 120120
-    },
-    {
-      "epoch": 0.485340400861355,
-      "grad_norm": 942.7879028320312,
-      "learning_rate": 3.100400682742675e-05,
-      "loss": 64.2288,
-      "step": 120130
-    },
-    {
-      "epoch": 0.48538080212672263,
-      "grad_norm": 627.4921264648438,
-      "learning_rate": 3.100061827063692e-05,
-      "loss": 58.1287,
-      "step": 120140
-    },
-    {
-      "epoch": 0.4854212033920902,
-      "grad_norm": 1478.2410888671875,
-      "learning_rate": 3.0997229596861944e-05,
-      "loss": 91.4508,
-      "step": 120150
-    },
-    {
-      "epoch": 0.48546160465745786,
-      "grad_norm": 1782.7314453125,
-      "learning_rate": 3.099384080616789e-05,
-      "loss": 56.9644,
-      "step": 120160
-    },
-    {
-      "epoch": 0.4855020059228255,
-      "grad_norm": 359.5166320800781,
-      "learning_rate": 3.099045189862081e-05,
-      "loss": 43.2427,
-      "step": 120170
-    },
-    {
-      "epoch": 0.48554240718819314,
-      "grad_norm": 604.357177734375,
-      "learning_rate": 3.0987062874286804e-05,
-      "loss": 43.9985,
-      "step": 120180
-    },
-    {
-      "epoch": 0.4855828084535608,
-      "grad_norm": 536.3892211914062,
-      "learning_rate": 3.098367373323192e-05,
-      "loss": 60.2531,
-      "step": 120190
-    },
-    {
-      "epoch": 0.4856232097189284,
-      "grad_norm": 762.7628784179688,
-      "learning_rate": 3.098028447552224e-05,
-      "loss": 45.4865,
-      "step": 120200
-    },
-    {
-      "epoch": 0.485663610984296,
-      "grad_norm": 621.0496215820312,
-      "learning_rate": 3.097689510122382e-05,
-      "loss": 49.9217,
-      "step": 120210
-    },
-    {
-      "epoch": 0.48570401224966364,
-      "grad_norm": 1028.444580078125,
-      "learning_rate": 3.0973505610402765e-05,
-      "loss": 54.5276,
-      "step": 120220
-    },
-    {
-      "epoch": 0.4857444135150313,
-      "grad_norm": 639.4829711914062,
-      "learning_rate": 3.0970116003125146e-05,
-      "loss": 46.6422,
-      "step": 120230
-    },
-    {
-      "epoch": 0.4857848147803989,
-      "grad_norm": 955.115234375,
-      "learning_rate": 3.0966726279457034e-05,
-      "loss": 59.0818,
-      "step": 120240
-    },
-    {
-      "epoch": 0.48582521604576656,
-      "grad_norm": 1233.369873046875,
-      "learning_rate": 3.0963336439464526e-05,
-      "loss": 56.1891,
-      "step": 120250
-    },
-    {
-      "epoch": 0.4858656173111342,
-      "grad_norm": 3286.7587890625,
-      "learning_rate": 3.09599464832137e-05,
-      "loss": 70.8362,
-      "step": 120260
-    },
-    {
-      "epoch": 0.48590601857650184,
-      "grad_norm": 749.9498291015625,
-      "learning_rate": 3.0956556410770655e-05,
-      "loss": 71.6872,
-      "step": 120270
-    },
-    {
-      "epoch": 0.4859464198418694,
-      "grad_norm": 337.7810974121094,
-      "learning_rate": 3.0953166222201476e-05,
-      "loss": 34.0317,
-      "step": 120280
-    },
-    {
-      "epoch": 0.48598682110723707,
-      "grad_norm": 969.7774047851562,
-      "learning_rate": 3.094977591757224e-05,
-      "loss": 39.0394,
-      "step": 120290
-    },
-    {
-      "epoch": 0.4860272223726047,
-      "grad_norm": 600.6663208007812,
-      "learning_rate": 3.094638549694908e-05,
-      "loss": 72.5898,
-      "step": 120300
-    },
-    {
-      "epoch": 0.48606762363797235,
-      "grad_norm": 388.6318664550781,
-      "learning_rate": 3.0942994960398064e-05,
-      "loss": 58.7893,
-      "step": 120310
-    },
-    {
-      "epoch": 0.48610802490334,
-      "grad_norm": 1114.3638916015625,
-      "learning_rate": 3.09396043079853e-05,
-      "loss": 67.1577,
-      "step": 120320
-    },
-    {
-      "epoch": 0.4861484261687076,
-      "grad_norm": 723.9693603515625,
-      "learning_rate": 3.0936213539776895e-05,
-      "loss": 66.5562,
-      "step": 120330
-    },
-    {
-      "epoch": 0.4861888274340752,
-      "grad_norm": 608.6270141601562,
-      "learning_rate": 3.093282265583895e-05,
-      "loss": 61.2591,
-      "step": 120340
-    },
-    {
-      "epoch": 0.48622922869944285,
-      "grad_norm": 455.3746643066406,
-      "learning_rate": 3.092943165623758e-05,
-      "loss": 57.6604,
-      "step": 120350
-    },
-    {
-      "epoch": 0.4862696299648105,
-      "grad_norm": 494.3013610839844,
-      "learning_rate": 3.092604054103888e-05,
-      "loss": 42.1755,
-      "step": 120360
-    },
-    {
-      "epoch": 0.48631003123017813,
-      "grad_norm": 2547.077392578125,
-      "learning_rate": 3.092264931030897e-05,
-      "loss": 86.2226,
-      "step": 120370
-    },
-    {
-      "epoch": 0.48635043249554577,
-      "grad_norm": 671.0562133789062,
-      "learning_rate": 3.0919257964113964e-05,
-      "loss": 60.703,
-      "step": 120380
-    },
-    {
-      "epoch": 0.4863908337609134,
-      "grad_norm": 471.1807556152344,
-      "learning_rate": 3.0915866502519975e-05,
-      "loss": 55.5357,
-      "step": 120390
-    },
-    {
-      "epoch": 0.48643123502628105,
-      "grad_norm": 656.9327392578125,
-      "learning_rate": 3.091247492559312e-05,
-      "loss": 66.8196,
-      "step": 120400
-    },
-    {
-      "epoch": 0.48647163629164863,
-      "grad_norm": 2174.91064453125,
-      "learning_rate": 3.090908323339952e-05,
-      "loss": 62.3699,
-      "step": 120410
-    },
-    {
-      "epoch": 0.4865120375570163,
-      "grad_norm": 663.234619140625,
-      "learning_rate": 3.090569142600531e-05,
-      "loss": 58.3517,
-      "step": 120420
-    },
-    {
-      "epoch": 0.4865524388223839,
-      "grad_norm": 1044.9910888671875,
-      "learning_rate": 3.09022995034766e-05,
-      "loss": 45.2753,
-      "step": 120430
-    },
-    {
-      "epoch": 0.48659284008775155,
-      "grad_norm": 1259.57666015625,
-      "learning_rate": 3.089890746587953e-05,
-      "loss": 37.8244,
-      "step": 120440
-    },
-    {
-      "epoch": 0.4866332413531192,
-      "grad_norm": 619.0317993164062,
-      "learning_rate": 3.089551531328021e-05,
-      "loss": 52.6852,
-      "step": 120450
-    },
-    {
-      "epoch": 0.48667364261848683,
-      "grad_norm": 547.9066772460938,
-      "learning_rate": 3.0892123045744785e-05,
-      "loss": 35.2198,
-      "step": 120460
-    },
-    {
-      "epoch": 0.4867140438838544,
-      "grad_norm": 538.3446044921875,
-      "learning_rate": 3.08887306633394e-05,
-      "loss": 49.3895,
-      "step": 120470
-    },
-    {
-      "epoch": 0.48675444514922206,
-      "grad_norm": 1102.380859375,
-      "learning_rate": 3.088533816613017e-05,
-      "loss": 78.057,
-      "step": 120480
-    },
-    {
-      "epoch": 0.4867948464145897,
-      "grad_norm": 836.9898071289062,
-      "learning_rate": 3.0881945554183235e-05,
-      "loss": 73.6896,
-      "step": 120490
-    },
-    {
-      "epoch": 0.48683524767995734,
-      "grad_norm": 657.3259887695312,
-      "learning_rate": 3.087855282756475e-05,
-      "loss": 45.1531,
-      "step": 120500
-    },
-    {
-      "epoch": 0.486875648945325,
-      "grad_norm": 440.7289733886719,
-      "learning_rate": 3.087515998634085e-05,
-      "loss": 57.9621,
-      "step": 120510
-    },
-    {
-      "epoch": 0.4869160502106926,
-      "grad_norm": 955.15966796875,
-      "learning_rate": 3.087176703057769e-05,
-      "loss": 56.2972,
-      "step": 120520
-    },
-    {
-      "epoch": 0.4869564514760602,
-      "grad_norm": 293.07012939453125,
-      "learning_rate": 3.08683739603414e-05,
-      "loss": 42.8358,
-      "step": 120530
-    },
-    {
-      "epoch": 0.48699685274142784,
-      "grad_norm": 516.2140502929688,
-      "learning_rate": 3.0864980775698145e-05,
-      "loss": 92.3896,
-      "step": 120540
-    },
-    {
-      "epoch": 0.4870372540067955,
-      "grad_norm": 425.08343505859375,
-      "learning_rate": 3.086158747671406e-05,
-      "loss": 90.2946,
-      "step": 120550
-    },
-    {
-      "epoch": 0.4870776552721631,
-      "grad_norm": 302.85626220703125,
-      "learning_rate": 3.085819406345532e-05,
-      "loss": 73.1443,
-      "step": 120560
-    },
-    {
-      "epoch": 0.48711805653753076,
-      "grad_norm": 910.6437377929688,
-      "learning_rate": 3.0854800535988064e-05,
-      "loss": 54.2145,
-      "step": 120570
-    },
-    {
-      "epoch": 0.4871584578028984,
-      "grad_norm": 1036.3970947265625,
-      "learning_rate": 3.085140689437846e-05,
-      "loss": 49.4283,
-      "step": 120580
-    },
-    {
-      "epoch": 0.48719885906826604,
-      "grad_norm": 1020.8902587890625,
-      "learning_rate": 3.084801313869266e-05,
-      "loss": 50.4306,
-      "step": 120590
-    },
-    {
-      "epoch": 0.48723926033363363,
-      "grad_norm": 683.2181396484375,
-      "learning_rate": 3.0844619268996845e-05,
-      "loss": 52.7355,
-      "step": 120600
-    },
-    {
-      "epoch": 0.48727966159900127,
-      "grad_norm": 712.2384033203125,
-      "learning_rate": 3.084122528535717e-05,
-      "loss": 76.377,
-      "step": 120610
-    },
-    {
-      "epoch": 0.4873200628643689,
-      "grad_norm": 1040.790771484375,
-      "learning_rate": 3.0837831187839784e-05,
-      "loss": 66.5839,
-      "step": 120620
-    },
-    {
-      "epoch": 0.48736046412973655,
-      "grad_norm": 826.3641967773438,
-      "learning_rate": 3.083443697651088e-05,
-      "loss": 81.0511,
-      "step": 120630
-    },
-    {
-      "epoch": 0.4874008653951042,
-      "grad_norm": 1687.8861083984375,
-      "learning_rate": 3.083104265143663e-05,
-      "loss": 73.7149,
-      "step": 120640
-    },
-    {
-      "epoch": 0.48744126666047183,
-      "grad_norm": 700.1195678710938,
-      "learning_rate": 3.08276482126832e-05,
-      "loss": 49.3427,
-      "step": 120650
-    },
-    {
-      "epoch": 0.4874816679258394,
-      "grad_norm": 661.125244140625,
-      "learning_rate": 3.082425366031676e-05,
-      "loss": 37.9126,
-      "step": 120660
-    },
-    {
-      "epoch": 0.48752206919120705,
-      "grad_norm": 773.9281005859375,
-      "learning_rate": 3.08208589944035e-05,
-      "loss": 47.0344,
-      "step": 120670
-    },
-    {
-      "epoch": 0.4875624704565747,
-      "grad_norm": 1276.460205078125,
-      "learning_rate": 3.08174642150096e-05,
-      "loss": 53.2406,
-      "step": 120680
-    },
-    {
-      "epoch": 0.48760287172194233,
-      "grad_norm": 669.7227172851562,
-      "learning_rate": 3.081406932220123e-05,
-      "loss": 39.3701,
-      "step": 120690
-    },
-    {
-      "epoch": 0.48764327298731,
-      "grad_norm": 2131.654541015625,
-      "learning_rate": 3.08106743160446e-05,
-      "loss": 64.2498,
-      "step": 120700
-    },
-    {
-      "epoch": 0.4876836742526776,
-      "grad_norm": 621.3330078125,
-      "learning_rate": 3.0807279196605876e-05,
-      "loss": 64.5872,
-      "step": 120710
-    },
-    {
-      "epoch": 0.48772407551804525,
-      "grad_norm": 565.8031616210938,
-      "learning_rate": 3.0803883963951255e-05,
-      "loss": 59.7642,
-      "step": 120720
-    },
-    {
-      "epoch": 0.48776447678341284,
-      "grad_norm": 936.9571533203125,
-      "learning_rate": 3.080048861814693e-05,
-      "loss": 34.2556,
-      "step": 120730
-    },
-    {
-      "epoch": 0.4878048780487805,
-      "grad_norm": 361.34063720703125,
-      "learning_rate": 3.0797093159259085e-05,
-      "loss": 52.7916,
-      "step": 120740
-    },
-    {
-      "epoch": 0.4878452793141481,
-      "grad_norm": 325.8556213378906,
-      "learning_rate": 3.079369758735393e-05,
-      "loss": 41.6967,
-      "step": 120750
-    },
-    {
-      "epoch": 0.48788568057951576,
-      "grad_norm": 522.2239990234375,
-      "learning_rate": 3.0790301902497666e-05,
-      "loss": 44.1217,
-      "step": 120760
-    },
-    {
-      "epoch": 0.4879260818448834,
-      "grad_norm": 1932.948486328125,
-      "learning_rate": 3.078690610475647e-05,
-      "loss": 70.4253,
-      "step": 120770
-    },
-    {
-      "epoch": 0.48796648311025104,
-      "grad_norm": 1131.1263427734375,
-      "learning_rate": 3.0783510194196576e-05,
-      "loss": 101.964,
-      "step": 120780
-    },
-    {
-      "epoch": 0.4880068843756186,
-      "grad_norm": 1616.0198974609375,
-      "learning_rate": 3.078011417088416e-05,
-      "loss": 55.159,
-      "step": 120790
-    },
-    {
-      "epoch": 0.48804728564098626,
-      "grad_norm": 942.608642578125,
-      "learning_rate": 3.0776718034885454e-05,
-      "loss": 47.1413,
-      "step": 120800
-    },
-    {
-      "epoch": 0.4880876869063539,
-      "grad_norm": 503.5368347167969,
-      "learning_rate": 3.0773321786266644e-05,
-      "loss": 36.5475,
-      "step": 120810
-    },
-    {
-      "epoch": 0.48812808817172154,
-      "grad_norm": 844.5652465820312,
-      "learning_rate": 3.076992542509396e-05,
-      "loss": 52.4938,
-      "step": 120820
-    },
-    {
-      "epoch": 0.4881684894370892,
-      "grad_norm": 856.3523559570312,
-      "learning_rate": 3.07665289514336e-05,
-      "loss": 83.7883,
-      "step": 120830
-    },
-    {
-      "epoch": 0.4882088907024568,
-      "grad_norm": 400.933349609375,
-      "learning_rate": 3.07631323653518e-05,
-      "loss": 55.8697,
-      "step": 120840
-    },
-    {
-      "epoch": 0.4882492919678244,
-      "grad_norm": 604.0474853515625,
-      "learning_rate": 3.075973566691477e-05,
-      "loss": 72.4179,
-      "step": 120850
-    },
-    {
-      "epoch": 0.48828969323319205,
-      "grad_norm": 584.9152221679688,
-      "learning_rate": 3.0756338856188716e-05,
-      "loss": 59.9413,
-      "step": 120860
-    },
-    {
-      "epoch": 0.4883300944985597,
-      "grad_norm": 734.94287109375,
-      "learning_rate": 3.075294193323988e-05,
-      "loss": 81.9922,
-      "step": 120870
-    },
-    {
-      "epoch": 0.4883704957639273,
-      "grad_norm": 775.8441162109375,
-      "learning_rate": 3.074954489813449e-05,
-      "loss": 59.0604,
-      "step": 120880
-    },
-    {
-      "epoch": 0.48841089702929497,
-      "grad_norm": 464.788330078125,
-      "learning_rate": 3.074614775093874e-05,
-      "loss": 77.2058,
-      "step": 120890
-    },
-    {
-      "epoch": 0.4884512982946626,
-      "grad_norm": 250.41958618164062,
-      "learning_rate": 3.074275049171889e-05,
-      "loss": 40.2773,
-      "step": 120900
-    },
-    {
-      "epoch": 0.48849169956003025,
-      "grad_norm": 608.3348388671875,
-      "learning_rate": 3.073935312054117e-05,
-      "loss": 34.6437,
-      "step": 120910
-    },
-    {
-      "epoch": 0.48853210082539783,
-      "grad_norm": 1844.59033203125,
-      "learning_rate": 3.0735955637471794e-05,
-      "loss": 71.1501,
-      "step": 120920
-    },
-    {
-      "epoch": 0.48857250209076547,
-      "grad_norm": 385.8973388671875,
-      "learning_rate": 3.073255804257702e-05,
-      "loss": 56.5991,
-      "step": 120930
-    },
-    {
-      "epoch": 0.4886129033561331,
-      "grad_norm": 503.91558837890625,
-      "learning_rate": 3.072916033592307e-05,
-      "loss": 62.6247,
-      "step": 120940
-    },
-    {
-      "epoch": 0.48865330462150075,
-      "grad_norm": 590.6072387695312,
-      "learning_rate": 3.0725762517576195e-05,
-      "loss": 39.7041,
-      "step": 120950
-    },
-    {
-      "epoch": 0.4886937058868684,
-      "grad_norm": 369.4613952636719,
-      "learning_rate": 3.072236458760262e-05,
-      "loss": 37.5118,
-      "step": 120960
-    },
-    {
-      "epoch": 0.48873410715223603,
-      "grad_norm": 743.9142456054688,
-      "learning_rate": 3.071896654606862e-05,
-      "loss": 48.3398,
-      "step": 120970
-    },
-    {
-      "epoch": 0.4887745084176036,
-      "grad_norm": 611.325439453125,
-      "learning_rate": 3.0715568393040405e-05,
-      "loss": 82.7277,
-      "step": 120980
-    },
-    {
-      "epoch": 0.48881490968297125,
-      "grad_norm": 1614.0037841796875,
-      "learning_rate": 3.071217012858425e-05,
-      "loss": 102.5452,
-      "step": 120990
-    },
-    {
-      "epoch": 0.4888553109483389,
-      "grad_norm": 984.4487915039062,
-      "learning_rate": 3.0708771752766394e-05,
-      "loss": 56.9584,
-      "step": 121000
-    },
-    {
-      "epoch": 0.48889571221370653,
-      "grad_norm": 1377.5753173828125,
-      "learning_rate": 3.07053732656531e-05,
-      "loss": 75.7293,
-      "step": 121010
-    },
-    {
-      "epoch": 0.4889361134790742,
-      "grad_norm": 438.0140686035156,
-      "learning_rate": 3.070197466731061e-05,
-      "loss": 43.8571,
-      "step": 121020
-    },
-    {
-      "epoch": 0.4889765147444418,
-      "grad_norm": 710.8450317382812,
-      "learning_rate": 3.069857595780519e-05,
-      "loss": 66.9329,
-      "step": 121030
-    },
-    {
-      "epoch": 0.48901691600980945,
-      "grad_norm": 762.4956665039062,
-      "learning_rate": 3.06951771372031e-05,
-      "loss": 52.6469,
-      "step": 121040
-    },
-    {
-      "epoch": 0.48905731727517704,
-      "grad_norm": 732.955322265625,
-      "learning_rate": 3.06917782055706e-05,
-      "loss": 84.5691,
-      "step": 121050
-    },
-    {
-      "epoch": 0.4890977185405447,
-      "grad_norm": 899.1510009765625,
-      "learning_rate": 3.0688379162973955e-05,
-      "loss": 78.2393,
-      "step": 121060
-    },
-    {
-      "epoch": 0.4891381198059123,
-      "grad_norm": 798.7470092773438,
-      "learning_rate": 3.0684980009479424e-05,
-      "loss": 58.7748,
-      "step": 121070
-    },
-    {
-      "epoch": 0.48917852107127996,
-      "grad_norm": 897.6027221679688,
-      "learning_rate": 3.068158074515328e-05,
-      "loss": 52.4273,
-      "step": 121080
-    },
-    {
-      "epoch": 0.4892189223366476,
-      "grad_norm": 503.83392333984375,
-      "learning_rate": 3.0678181370061805e-05,
-      "loss": 59.4657,
-      "step": 121090
-    },
-    {
-      "epoch": 0.48925932360201524,
-      "grad_norm": 1317.728759765625,
-      "learning_rate": 3.0674781884271254e-05,
-      "loss": 93.741,
-      "step": 121100
-    },
-    {
-      "epoch": 0.4892997248673828,
-      "grad_norm": 1355.2291259765625,
-      "learning_rate": 3.067138228784791e-05,
-      "loss": 84.8171,
-      "step": 121110
-    },
-    {
-      "epoch": 0.48934012613275046,
-      "grad_norm": 1263.6253662109375,
-      "learning_rate": 3.0667982580858044e-05,
-      "loss": 65.6653,
-      "step": 121120
-    },
-    {
-      "epoch": 0.4893805273981181,
-      "grad_norm": 630.547607421875,
-      "learning_rate": 3.066458276336794e-05,
-      "loss": 62.2357,
-      "step": 121130
-    },
-    {
-      "epoch": 0.48942092866348574,
-      "grad_norm": 1021.1478271484375,
-      "learning_rate": 3.0661182835443884e-05,
-      "loss": 57.1426,
-      "step": 121140
-    },
-    {
-      "epoch": 0.4894613299288534,
-      "grad_norm": 1333.4888916015625,
-      "learning_rate": 3.065778279715215e-05,
-      "loss": 49.1796,
-      "step": 121150
-    },
-    {
-      "epoch": 0.489501731194221,
-      "grad_norm": 481.13165283203125,
-      "learning_rate": 3.0654382648559026e-05,
-      "loss": 65.8724,
-      "step": 121160
-    },
-    {
-      "epoch": 0.4895421324595886,
-      "grad_norm": 359.9427795410156,
-      "learning_rate": 3.065098238973081e-05,
-      "loss": 77.4488,
-      "step": 121170
-    },
-    {
-      "epoch": 0.48958253372495625,
-      "grad_norm": 274.55328369140625,
-      "learning_rate": 3.064758202073377e-05,
-      "loss": 54.833,
-      "step": 121180
-    },
-    {
-      "epoch": 0.4896229349903239,
-      "grad_norm": 489.7810974121094,
-      "learning_rate": 3.064418154163422e-05,
-      "loss": 54.8473,
-      "step": 121190
-    },
-    {
-      "epoch": 0.48966333625569153,
-      "grad_norm": 674.3745727539062,
-      "learning_rate": 3.064078095249844e-05,
-      "loss": 51.562,
-      "step": 121200
-    },
-    {
-      "epoch": 0.48970373752105917,
-      "grad_norm": 443.5577087402344,
-      "learning_rate": 3.0637380253392736e-05,
-      "loss": 68.5722,
-      "step": 121210
-    },
-    {
-      "epoch": 0.4897441387864268,
-      "grad_norm": 824.37158203125,
-      "learning_rate": 3.06339794443834e-05,
-      "loss": 46.6327,
-      "step": 121220
-    },
-    {
-      "epoch": 0.48978454005179445,
-      "grad_norm": 275.5827331542969,
-      "learning_rate": 3.063057852553674e-05,
-      "loss": 32.1468,
-      "step": 121230
-    },
-    {
-      "epoch": 0.48982494131716203,
-      "grad_norm": 573.8633422851562,
-      "learning_rate": 3.062717749691904e-05,
-      "loss": 58.9486,
-      "step": 121240
-    },
-    {
-      "epoch": 0.4898653425825297,
-      "grad_norm": 383.3703918457031,
-      "learning_rate": 3.062377635859663e-05,
-      "loss": 39.3934,
-      "step": 121250
-    },
-    {
-      "epoch": 0.4899057438478973,
-      "grad_norm": 942.3863525390625,
-      "learning_rate": 3.06203751106358e-05,
-      "loss": 49.6613,
-      "step": 121260
-    },
-    {
-      "epoch": 0.48994614511326495,
-      "grad_norm": 0.0,
-      "learning_rate": 3.0616973753102856e-05,
-      "loss": 37.8872,
-      "step": 121270
-    },
-    {
-      "epoch": 0.4899865463786326,
-      "grad_norm": 1657.1973876953125,
-      "learning_rate": 3.0613572286064125e-05,
-      "loss": 68.007,
-      "step": 121280
-    },
-    {
-      "epoch": 0.49002694764400023,
-      "grad_norm": 647.4043579101562,
-      "learning_rate": 3.061017070958591e-05,
-      "loss": 92.0626,
-      "step": 121290
-    },
-    {
-      "epoch": 0.4900673489093678,
-      "grad_norm": 885.7959594726562,
-      "learning_rate": 3.0606769023734536e-05,
-      "loss": 30.4313,
-      "step": 121300
-    },
-    {
-      "epoch": 0.49010775017473546,
-      "grad_norm": 1316.2650146484375,
-      "learning_rate": 3.060336722857631e-05,
-      "loss": 83.0374,
-      "step": 121310
-    },
-    {
-      "epoch": 0.4901481514401031,
-      "grad_norm": 1012.3235473632812,
-      "learning_rate": 3.059996532417754e-05,
-      "loss": 41.8788,
-      "step": 121320
-    },
-    {
-      "epoch": 0.49018855270547074,
-      "grad_norm": 449.5831604003906,
-      "learning_rate": 3.059656331060458e-05,
-      "loss": 55.4948,
-      "step": 121330
-    },
-    {
-      "epoch": 0.4902289539708384,
-      "grad_norm": 723.0786743164062,
-      "learning_rate": 3.0593161187923736e-05,
-      "loss": 68.319,
-      "step": 121340
-    },
-    {
-      "epoch": 0.490269355236206,
-      "grad_norm": 569.5224609375,
-      "learning_rate": 3.0589758956201327e-05,
-      "loss": 62.8122,
-      "step": 121350
-    },
-    {
-      "epoch": 0.49030975650157366,
-      "grad_norm": 572.9276733398438,
-      "learning_rate": 3.058635661550369e-05,
-      "loss": 60.2772,
-      "step": 121360
-    },
-    {
-      "epoch": 0.49035015776694124,
-      "grad_norm": 379.9477844238281,
-      "learning_rate": 3.058295416589716e-05,
-      "loss": 69.8202,
-      "step": 121370
-    },
-    {
-      "epoch": 0.4903905590323089,
-      "grad_norm": 534.4418334960938,
-      "learning_rate": 3.0579551607448066e-05,
-      "loss": 51.7835,
-      "step": 121380
-    },
-    {
-      "epoch": 0.4904309602976765,
-      "grad_norm": 803.1372680664062,
-      "learning_rate": 3.057614894022274e-05,
-      "loss": 43.7992,
-      "step": 121390
-    },
-    {
-      "epoch": 0.49047136156304416,
-      "grad_norm": 2386.29736328125,
-      "learning_rate": 3.0572746164287514e-05,
-      "loss": 54.0955,
-      "step": 121400
-    },
-    {
-      "epoch": 0.4905117628284118,
-      "grad_norm": 997.156005859375,
-      "learning_rate": 3.0569343279708734e-05,
-      "loss": 65.676,
-      "step": 121410
-    },
-    {
-      "epoch": 0.49055216409377944,
-      "grad_norm": 525.251953125,
-      "learning_rate": 3.056594028655274e-05,
-      "loss": 52.9306,
-      "step": 121420
-    },
-    {
-      "epoch": 0.490592565359147,
-      "grad_norm": 1752.92236328125,
-      "learning_rate": 3.056253718488588e-05,
-      "loss": 56.5857,
-      "step": 121430
-    },
-    {
-      "epoch": 0.49063296662451467,
-      "grad_norm": 704.6538696289062,
-      "learning_rate": 3.055913397477448e-05,
-      "loss": 68.621,
-      "step": 121440
-    },
-    {
-      "epoch": 0.4906733678898823,
-      "grad_norm": 615.273681640625,
-      "learning_rate": 3.0555730656284914e-05,
-      "loss": 54.718,
-      "step": 121450
-    },
-    {
-      "epoch": 0.49071376915524995,
-      "grad_norm": 676.6206665039062,
-      "learning_rate": 3.0552327229483515e-05,
-      "loss": 60.8578,
-      "step": 121460
-    },
-    {
-      "epoch": 0.4907541704206176,
-      "grad_norm": 512.4281005859375,
-      "learning_rate": 3.054892369443663e-05,
-      "loss": 48.8229,
-      "step": 121470
-    },
-    {
-      "epoch": 0.4907945716859852,
-      "grad_norm": 885.903076171875,
-      "learning_rate": 3.054552005121064e-05,
-      "loss": 83.9689,
-      "step": 121480
-    },
-    {
-      "epoch": 0.4908349729513528,
-      "grad_norm": 405.67919921875,
-      "learning_rate": 3.054211629987187e-05,
-      "loss": 50.5087,
-      "step": 121490
-    },
-    {
-      "epoch": 0.49087537421672045,
-      "grad_norm": 430.7709045410156,
-      "learning_rate": 3.053871244048669e-05,
-      "loss": 54.2436,
-      "step": 121500
-    },
-    {
-      "epoch": 0.4909157754820881,
-      "grad_norm": 394.2481689453125,
-      "learning_rate": 3.0535308473121455e-05,
-      "loss": 37.6645,
-      "step": 121510
-    },
-    {
-      "epoch": 0.49095617674745573,
-      "grad_norm": 2119.940673828125,
-      "learning_rate": 3.053190439784253e-05,
-      "loss": 85.2859,
-      "step": 121520
-    },
-    {
-      "epoch": 0.49099657801282337,
-      "grad_norm": 509.3744201660156,
-      "learning_rate": 3.052850021471629e-05,
-      "loss": 74.4669,
-      "step": 121530
-    },
-    {
-      "epoch": 0.491036979278191,
-      "grad_norm": 0.0,
-      "learning_rate": 3.052509592380909e-05,
-      "loss": 51.207,
-      "step": 121540
-    },
-    {
-      "epoch": 0.49107738054355865,
-      "grad_norm": 615.7112426757812,
-      "learning_rate": 3.052169152518729e-05,
-      "loss": 37.4327,
-      "step": 121550
-    },
-    {
-      "epoch": 0.49111778180892623,
-      "grad_norm": 624.2453002929688,
-      "learning_rate": 3.051828701891729e-05,
-      "loss": 52.5991,
-      "step": 121560
-    },
-    {
-      "epoch": 0.4911581830742939,
-      "grad_norm": 558.4168090820312,
-      "learning_rate": 3.0514882405065432e-05,
-      "loss": 63.3106,
-      "step": 121570
-    },
-    {
-      "epoch": 0.4911985843396615,
-      "grad_norm": 509.6629333496094,
-      "learning_rate": 3.0511477683698108e-05,
-      "loss": 63.4842,
-      "step": 121580
-    },
-    {
-      "epoch": 0.49123898560502915,
-      "grad_norm": 806.631103515625,
-      "learning_rate": 3.050807285488168e-05,
-      "loss": 62.4659,
-      "step": 121590
-    },
-    {
-      "epoch": 0.4912793868703968,
-      "grad_norm": 814.9642944335938,
-      "learning_rate": 3.050466791868254e-05,
-      "loss": 66.1909,
-      "step": 121600
-    },
-    {
-      "epoch": 0.49131978813576443,
-      "grad_norm": 712.5091552734375,
-      "learning_rate": 3.0501262875167063e-05,
-      "loss": 74.5207,
-      "step": 121610
-    },
-    {
-      "epoch": 0.491360189401132,
-      "grad_norm": 615.8709106445312,
-      "learning_rate": 3.0497857724401642e-05,
-      "loss": 79.785,
-      "step": 121620
-    },
-    {
-      "epoch": 0.49140059066649966,
-      "grad_norm": 729.620361328125,
-      "learning_rate": 3.0494452466452644e-05,
-      "loss": 75.2466,
-      "step": 121630
-    },
-    {
-      "epoch": 0.4914409919318673,
-      "grad_norm": 232.82257080078125,
-      "learning_rate": 3.049104710138647e-05,
-      "loss": 40.4371,
-      "step": 121640
-    },
-    {
-      "epoch": 0.49148139319723494,
-      "grad_norm": 528.4829711914062,
-      "learning_rate": 3.0487641629269516e-05,
-      "loss": 49.6708,
-      "step": 121650
-    },
-    {
-      "epoch": 0.4915217944626026,
-      "grad_norm": 2002.76416015625,
-      "learning_rate": 3.0484236050168153e-05,
-      "loss": 52.0956,
-      "step": 121660
-    },
-    {
-      "epoch": 0.4915621957279702,
-      "grad_norm": 575.5763549804688,
-      "learning_rate": 3.048083036414878e-05,
-      "loss": 39.8568,
-      "step": 121670
-    },
-    {
-      "epoch": 0.4916025969933378,
-      "grad_norm": 0.0,
-      "learning_rate": 3.0477424571277807e-05,
-      "loss": 39.302,
-      "step": 121680
-    },
-    {
-      "epoch": 0.49164299825870544,
-      "grad_norm": 412.2666015625,
-      "learning_rate": 3.047401867162162e-05,
-      "loss": 46.3011,
-      "step": 121690
-    },
-    {
-      "epoch": 0.4916833995240731,
-      "grad_norm": 749.7333374023438,
-      "learning_rate": 3.0470612665246618e-05,
-      "loss": 63.5483,
-      "step": 121700
-    },
-    {
-      "epoch": 0.4917238007894407,
-      "grad_norm": 686.0115966796875,
-      "learning_rate": 3.0467206552219208e-05,
-      "loss": 37.7942,
-      "step": 121710
-    },
-    {
-      "epoch": 0.49176420205480836,
-      "grad_norm": 1040.0323486328125,
-      "learning_rate": 3.0463800332605784e-05,
-      "loss": 69.1257,
-      "step": 121720
-    },
-    {
-      "epoch": 0.491804603320176,
-      "grad_norm": 547.2194213867188,
-      "learning_rate": 3.046039400647277e-05,
-      "loss": 28.7517,
-      "step": 121730
-    },
-    {
-      "epoch": 0.49184500458554364,
-      "grad_norm": 1607.9515380859375,
-      "learning_rate": 3.0456987573886564e-05,
-      "loss": 67.1605,
-      "step": 121740
-    },
-    {
-      "epoch": 0.49188540585091123,
-      "grad_norm": 1343.8623046875,
-      "learning_rate": 3.045358103491357e-05,
-      "loss": 61.9395,
-      "step": 121750
-    },
-    {
-      "epoch": 0.49192580711627887,
-      "grad_norm": 360.9034729003906,
-      "learning_rate": 3.0450174389620205e-05,
-      "loss": 60.9107,
-      "step": 121760
-    },
-    {
-      "epoch": 0.4919662083816465,
-      "grad_norm": 658.2755126953125,
-      "learning_rate": 3.044676763807288e-05,
-      "loss": 44.2099,
-      "step": 121770
-    },
-    {
-      "epoch": 0.49200660964701415,
-      "grad_norm": 463.0184326171875,
-      "learning_rate": 3.044336078033803e-05,
-      "loss": 50.352,
-      "step": 121780
-    },
-    {
-      "epoch": 0.4920470109123818,
-      "grad_norm": 1313.499755859375,
-      "learning_rate": 3.043995381648205e-05,
-      "loss": 65.2679,
-      "step": 121790
-    },
-    {
-      "epoch": 0.49208741217774943,
-      "grad_norm": 662.1671142578125,
-      "learning_rate": 3.0436546746571372e-05,
-      "loss": 47.1219,
-      "step": 121800
-    },
-    {
-      "epoch": 0.492127813443117,
-      "grad_norm": 3354.996826171875,
-      "learning_rate": 3.0433139570672407e-05,
-      "loss": 87.668,
-      "step": 121810
-    },
-    {
-      "epoch": 0.49216821470848465,
-      "grad_norm": 931.5249633789062,
-      "learning_rate": 3.0429732288851603e-05,
-      "loss": 44.4428,
-      "step": 121820
-    },
-    {
-      "epoch": 0.4922086159738523,
-      "grad_norm": 798.1958618164062,
-      "learning_rate": 3.0426324901175374e-05,
-      "loss": 51.0042,
-      "step": 121830
-    },
-    {
-      "epoch": 0.49224901723921993,
-      "grad_norm": 674.7509765625,
-      "learning_rate": 3.0422917407710137e-05,
-      "loss": 33.4516,
-      "step": 121840
-    },
-    {
-      "epoch": 0.4922894185045876,
-      "grad_norm": 734.49169921875,
-      "learning_rate": 3.0419509808522334e-05,
-      "loss": 77.9193,
-      "step": 121850
-    },
-    {
-      "epoch": 0.4923298197699552,
-      "grad_norm": 561.5974731445312,
-      "learning_rate": 3.0416102103678402e-05,
-      "loss": 52.3013,
-      "step": 121860
-    },
-    {
-      "epoch": 0.49237022103532285,
-      "grad_norm": 292.3045959472656,
-      "learning_rate": 3.041269429324477e-05,
-      "loss": 31.1926,
-      "step": 121870
-    },
-    {
-      "epoch": 0.49241062230069044,
-      "grad_norm": 613.70361328125,
-      "learning_rate": 3.040928637728787e-05,
-      "loss": 56.5007,
-      "step": 121880
-    },
-    {
-      "epoch": 0.4924510235660581,
-      "grad_norm": 1232.8988037109375,
-      "learning_rate": 3.040587835587415e-05,
-      "loss": 73.1134,
-      "step": 121890
-    },
-    {
-      "epoch": 0.4924914248314257,
-      "grad_norm": 831.9037475585938,
-      "learning_rate": 3.0402470229070056e-05,
-      "loss": 46.9302,
-      "step": 121900
-    },
-    {
-      "epoch": 0.49253182609679336,
-      "grad_norm": 409.0111389160156,
-      "learning_rate": 3.039906199694202e-05,
-      "loss": 39.5951,
-      "step": 121910
-    },
-    {
-      "epoch": 0.492572227362161,
-      "grad_norm": 646.8606567382812,
-      "learning_rate": 3.0395653659556488e-05,
-      "loss": 47.5818,
-      "step": 121920
-    },
-    {
-      "epoch": 0.49261262862752864,
-      "grad_norm": 363.55609130859375,
-      "learning_rate": 3.039224521697991e-05,
-      "loss": 72.8517,
-      "step": 121930
-    },
-    {
-      "epoch": 0.4926530298928962,
-      "grad_norm": 498.0539855957031,
-      "learning_rate": 3.0388836669278738e-05,
-      "loss": 53.5023,
-      "step": 121940
-    },
-    {
-      "epoch": 0.49269343115826386,
-      "grad_norm": 943.4515991210938,
-      "learning_rate": 3.038542801651941e-05,
-      "loss": 59.0278,
-      "step": 121950
-    },
-    {
-      "epoch": 0.4927338324236315,
-      "grad_norm": 3113.823486328125,
-      "learning_rate": 3.0382019258768403e-05,
-      "loss": 72.7263,
-      "step": 121960
-    },
-    {
-      "epoch": 0.49277423368899914,
-      "grad_norm": 509.8355407714844,
-      "learning_rate": 3.0378610396092154e-05,
-      "loss": 51.3322,
-      "step": 121970
-    },
-    {
-      "epoch": 0.4928146349543668,
-      "grad_norm": 791.736572265625,
-      "learning_rate": 3.0375201428557132e-05,
-      "loss": 48.9846,
-      "step": 121980
-    },
-    {
-      "epoch": 0.4928550362197344,
-      "grad_norm": 532.8504638671875,
-      "learning_rate": 3.0371792356229783e-05,
-      "loss": 35.4836,
-      "step": 121990
-    },
-    {
-      "epoch": 0.492895437485102,
-      "grad_norm": 419.5154113769531,
-      "learning_rate": 3.0368383179176585e-05,
-      "loss": 69.5128,
-      "step": 122000
-    },
-    {
-      "epoch": 0.49293583875046965,
-      "grad_norm": 3489.129150390625,
-      "learning_rate": 3.036497389746399e-05,
-      "loss": 68.8055,
-      "step": 122010
-    },
-    {
-      "epoch": 0.4929762400158373,
-      "grad_norm": 1044.6541748046875,
-      "learning_rate": 3.0361564511158457e-05,
-      "loss": 62.8193,
-      "step": 122020
-    },
-    {
-      "epoch": 0.4930166412812049,
-      "grad_norm": 1858.55615234375,
-      "learning_rate": 3.0358155020326477e-05,
-      "loss": 77.7083,
-      "step": 122030
-    },
-    {
-      "epoch": 0.49305704254657257,
-      "grad_norm": 682.7332153320312,
-      "learning_rate": 3.0354745425034498e-05,
-      "loss": 64.1814,
-      "step": 122040
-    },
-    {
-      "epoch": 0.4930974438119402,
-      "grad_norm": 653.3331298828125,
-      "learning_rate": 3.0351335725349e-05,
-      "loss": 49.824,
-      "step": 122050
-    },
-    {
-      "epoch": 0.49313784507730785,
-      "grad_norm": 859.71337890625,
-      "learning_rate": 3.0347925921336463e-05,
-      "loss": 54.182,
-      "step": 122060
-    },
-    {
-      "epoch": 0.49317824634267543,
-      "grad_norm": 1686.3477783203125,
-      "learning_rate": 3.0344516013063357e-05,
-      "loss": 43.8826,
-      "step": 122070
-    },
-    {
-      "epoch": 0.49321864760804307,
-      "grad_norm": 1858.2529296875,
-      "learning_rate": 3.034110600059616e-05,
-      "loss": 42.0115,
-      "step": 122080
-    },
-    {
-      "epoch": 0.4932590488734107,
-      "grad_norm": 857.298828125,
-      "learning_rate": 3.0337695884001343e-05,
-      "loss": 44.892,
-      "step": 122090
-    },
-    {
-      "epoch": 0.49329945013877835,
-      "grad_norm": 3229.34765625,
-      "learning_rate": 3.0334285663345404e-05,
-      "loss": 49.8066,
-      "step": 122100
-    },
-    {
-      "epoch": 0.493339851404146,
-      "grad_norm": 782.3743896484375,
-      "learning_rate": 3.033087533869482e-05,
-      "loss": 46.6393,
-      "step": 122110
-    },
-    {
-      "epoch": 0.49338025266951363,
-      "grad_norm": 1177.9285888671875,
-      "learning_rate": 3.032746491011607e-05,
-      "loss": 54.9906,
-      "step": 122120
-    },
-    {
-      "epoch": 0.4934206539348812,
-      "grad_norm": 997.4578247070312,
-      "learning_rate": 3.0324054377675654e-05,
-      "loss": 47.8525,
-      "step": 122130
-    },
-    {
-      "epoch": 0.49346105520024885,
-      "grad_norm": 1065.852294921875,
-      "learning_rate": 3.032064374144005e-05,
-      "loss": 58.0819,
-      "step": 122140
-    },
-    {
-      "epoch": 0.4935014564656165,
-      "grad_norm": 2079.648681640625,
-      "learning_rate": 3.031723300147577e-05,
-      "loss": 95.1813,
-      "step": 122150
-    },
-    {
-      "epoch": 0.49354185773098413,
-      "grad_norm": 759.1595458984375,
-      "learning_rate": 3.0313822157849287e-05,
-      "loss": 54.4963,
-      "step": 122160
-    },
-    {
-      "epoch": 0.4935822589963518,
-      "grad_norm": 2073.466064453125,
-      "learning_rate": 3.031041121062711e-05,
-      "loss": 70.8082,
-      "step": 122170
-    },
-    {
-      "epoch": 0.4936226602617194,
-      "grad_norm": 1743.0040283203125,
-      "learning_rate": 3.030700015987573e-05,
-      "loss": 57.2004,
-      "step": 122180
-    },
-    {
-      "epoch": 0.49366306152708705,
-      "grad_norm": 714.07275390625,
-      "learning_rate": 3.030358900566165e-05,
-      "loss": 60.7911,
-      "step": 122190
-    },
-    {
-      "epoch": 0.49370346279245464,
-      "grad_norm": 1210.780029296875,
-      "learning_rate": 3.0300177748051373e-05,
-      "loss": 57.1338,
-      "step": 122200
-    },
-    {
-      "epoch": 0.4937438640578223,
-      "grad_norm": 815.665771484375,
-      "learning_rate": 3.02967663871114e-05,
-      "loss": 74.5411,
-      "step": 122210
-    },
-    {
-      "epoch": 0.4937842653231899,
-      "grad_norm": 1104.837646484375,
-      "learning_rate": 3.0293354922908235e-05,
-      "loss": 69.4025,
-      "step": 122220
-    },
-    {
-      "epoch": 0.49382466658855756,
-      "grad_norm": 693.127197265625,
-      "learning_rate": 3.0289943355508392e-05,
-      "loss": 63.0513,
-      "step": 122230
-    },
-    {
-      "epoch": 0.4938650678539252,
-      "grad_norm": 683.1317749023438,
-      "learning_rate": 3.028653168497838e-05,
-      "loss": 46.7121,
-      "step": 122240
-    },
-    {
-      "epoch": 0.49390546911929284,
-      "grad_norm": 317.39532470703125,
-      "learning_rate": 3.028311991138472e-05,
-      "loss": 63.5555,
-      "step": 122250
-    },
-    {
-      "epoch": 0.4939458703846604,
-      "grad_norm": 466.8677062988281,
-      "learning_rate": 3.0279708034793907e-05,
-      "loss": 45.8454,
-      "step": 122260
-    },
-    {
-      "epoch": 0.49398627165002806,
-      "grad_norm": 1304.4661865234375,
-      "learning_rate": 3.027629605527248e-05,
-      "loss": 67.0937,
-      "step": 122270
-    },
-    {
-      "epoch": 0.4940266729153957,
-      "grad_norm": 341.1806335449219,
-      "learning_rate": 3.0272883972886935e-05,
-      "loss": 32.4989,
-      "step": 122280
-    },
-    {
-      "epoch": 0.49406707418076334,
-      "grad_norm": 668.6641235351562,
-      "learning_rate": 3.02694717877038e-05,
-      "loss": 58.9114,
-      "step": 122290
-    },
-    {
-      "epoch": 0.494107475446131,
-      "grad_norm": 1312.1033935546875,
-      "learning_rate": 3.02660594997896e-05,
-      "loss": 52.9595,
-      "step": 122300
-    },
-    {
-      "epoch": 0.4941478767114986,
-      "grad_norm": 206.2220916748047,
-      "learning_rate": 3.0262647109210867e-05,
-      "loss": 42.2671,
-      "step": 122310
-    },
-    {
-      "epoch": 0.4941882779768662,
-      "grad_norm": 533.5427856445312,
-      "learning_rate": 3.0259234616034116e-05,
-      "loss": 54.6035,
-      "step": 122320
-    },
-    {
-      "epoch": 0.49422867924223385,
-      "grad_norm": 389.654296875,
-      "learning_rate": 3.0255822020325873e-05,
-      "loss": 46.6373,
-      "step": 122330
-    },
-    {
-      "epoch": 0.4942690805076015,
-      "grad_norm": 509.89276123046875,
-      "learning_rate": 3.025240932215268e-05,
-      "loss": 65.848,
-      "step": 122340
-    },
-    {
-      "epoch": 0.49430948177296913,
-      "grad_norm": 738.1375732421875,
-      "learning_rate": 3.024899652158107e-05,
-      "loss": 46.7456,
-      "step": 122350
-    },
-    {
-      "epoch": 0.49434988303833677,
-      "grad_norm": 272.38751220703125,
-      "learning_rate": 3.0245583618677558e-05,
-      "loss": 58.7914,
-      "step": 122360
-    },
-    {
-      "epoch": 0.4943902843037044,
-      "grad_norm": 599.438232421875,
-      "learning_rate": 3.0242170613508692e-05,
-      "loss": 57.4107,
-      "step": 122370
-    },
-    {
-      "epoch": 0.49443068556907205,
-      "grad_norm": 323.33062744140625,
-      "learning_rate": 3.0238757506141012e-05,
-      "loss": 48.1619,
-      "step": 122380
-    },
-    {
-      "epoch": 0.49447108683443963,
-      "grad_norm": 1464.4964599609375,
-      "learning_rate": 3.0235344296641067e-05,
-      "loss": 58.9648,
-      "step": 122390
-    },
-    {
-      "epoch": 0.4945114880998073,
-      "grad_norm": 2316.94873046875,
-      "learning_rate": 3.023193098507538e-05,
-      "loss": 80.6663,
-      "step": 122400
-    },
-    {
-      "epoch": 0.4945518893651749,
-      "grad_norm": 777.03173828125,
-      "learning_rate": 3.0228517571510507e-05,
-      "loss": 74.6564,
-      "step": 122410
-    },
-    {
-      "epoch": 0.49459229063054255,
-      "grad_norm": 832.2376098632812,
-      "learning_rate": 3.0225104056013e-05,
-      "loss": 50.8063,
-      "step": 122420
-    },
-    {
-      "epoch": 0.4946326918959102,
-      "grad_norm": 1374.7584228515625,
-      "learning_rate": 3.0221690438649386e-05,
-      "loss": 84.333,
-      "step": 122430
-    },
-    {
-      "epoch": 0.49467309316127783,
-      "grad_norm": 586.2657470703125,
-      "learning_rate": 3.0218276719486244e-05,
-      "loss": 54.8292,
-      "step": 122440
-    },
-    {
-      "epoch": 0.4947134944266454,
-      "grad_norm": 362.1689147949219,
-      "learning_rate": 3.0214862898590095e-05,
-      "loss": 45.3425,
-      "step": 122450
-    },
-    {
-      "epoch": 0.49475389569201306,
-      "grad_norm": 740.5028076171875,
-      "learning_rate": 3.021144897602752e-05,
-      "loss": 65.865,
-      "step": 122460
-    },
-    {
-      "epoch": 0.4947942969573807,
-      "grad_norm": 776.755615234375,
-      "learning_rate": 3.020803495186506e-05,
-      "loss": 45.3244,
-      "step": 122470
-    },
-    {
-      "epoch": 0.49483469822274834,
-      "grad_norm": 1503.520263671875,
-      "learning_rate": 3.020462082616928e-05,
-      "loss": 44.012,
-      "step": 122480
-    },
-    {
-      "epoch": 0.494875099488116,
-      "grad_norm": 772.3239135742188,
-      "learning_rate": 3.0201206599006733e-05,
-      "loss": 50.5993,
-      "step": 122490
-    },
-    {
-      "epoch": 0.4949155007534836,
-      "grad_norm": 343.9445495605469,
-      "learning_rate": 3.0197792270443982e-05,
-      "loss": 43.6323,
-      "step": 122500
-    },
-    {
-      "epoch": 0.49495590201885126,
-      "grad_norm": 765.3672485351562,
-      "learning_rate": 3.0194377840547606e-05,
-      "loss": 48.1465,
-      "step": 122510
-    },
-    {
-      "epoch": 0.49499630328421884,
-      "grad_norm": 936.2794799804688,
-      "learning_rate": 3.0190963309384156e-05,
-      "loss": 58.2882,
-      "step": 122520
-    },
-    {
-      "epoch": 0.4950367045495865,
-      "grad_norm": 770.3196411132812,
-      "learning_rate": 3.01875486770202e-05,
-      "loss": 43.9133,
-      "step": 122530
-    },
-    {
-      "epoch": 0.4950771058149541,
-      "grad_norm": 411.43255615234375,
-      "learning_rate": 3.0184133943522314e-05,
-      "loss": 50.147,
-      "step": 122540
-    },
-    {
-      "epoch": 0.49511750708032176,
-      "grad_norm": 589.0484619140625,
-      "learning_rate": 3.0180719108957063e-05,
-      "loss": 48.991,
-      "step": 122550
-    },
-    {
-      "epoch": 0.4951579083456894,
-      "grad_norm": 317.48046875,
-      "learning_rate": 3.0177304173391037e-05,
-      "loss": 47.1381,
-      "step": 122560
-    },
-    {
-      "epoch": 0.49519830961105704,
-      "grad_norm": 2199.10888671875,
-      "learning_rate": 3.0173889136890786e-05,
-      "loss": 60.0664,
-      "step": 122570
-    },
-    {
-      "epoch": 0.4952387108764246,
-      "grad_norm": 2371.26025390625,
-      "learning_rate": 3.0170473999522915e-05,
-      "loss": 53.2513,
-      "step": 122580
-    },
-    {
-      "epoch": 0.49527911214179227,
-      "grad_norm": 2204.5810546875,
-      "learning_rate": 3.016705876135399e-05,
-      "loss": 75.0456,
-      "step": 122590
-    },
-    {
-      "epoch": 0.4953195134071599,
-      "grad_norm": 1063.3651123046875,
-      "learning_rate": 3.016364342245059e-05,
-      "loss": 85.0364,
-      "step": 122600
-    },
-    {
-      "epoch": 0.49535991467252755,
-      "grad_norm": 1718.4310302734375,
-      "learning_rate": 3.016022798287931e-05,
-      "loss": 66.0194,
-      "step": 122610
-    },
-    {
-      "epoch": 0.4954003159378952,
-      "grad_norm": 953.0413208007812,
-      "learning_rate": 3.0156812442706715e-05,
-      "loss": 54.6552,
-      "step": 122620
-    },
-    {
-      "epoch": 0.4954407172032628,
-      "grad_norm": 702.347900390625,
-      "learning_rate": 3.015339680199941e-05,
-      "loss": 39.405,
-      "step": 122630
-    },
-    {
-      "epoch": 0.4954811184686304,
-      "grad_norm": 477.78302001953125,
-      "learning_rate": 3.0149981060823995e-05,
-      "loss": 39.2307,
-      "step": 122640
-    },
-    {
-      "epoch": 0.49552151973399805,
-      "grad_norm": 893.2791137695312,
-      "learning_rate": 3.0146565219247036e-05,
-      "loss": 70.0523,
-      "step": 122650
-    },
-    {
-      "epoch": 0.4955619209993657,
-      "grad_norm": 493.2530517578125,
-      "learning_rate": 3.0143149277335138e-05,
-      "loss": 59.8175,
-      "step": 122660
-    },
-    {
-      "epoch": 0.49560232226473333,
-      "grad_norm": 2252.927734375,
-      "learning_rate": 3.01397332351549e-05,
-      "loss": 79.7985,
-      "step": 122670
-    },
-    {
-      "epoch": 0.49564272353010097,
-      "grad_norm": 639.860595703125,
-      "learning_rate": 3.013631709277292e-05,
-      "loss": 80.3618,
-      "step": 122680
-    },
-    {
-      "epoch": 0.4956831247954686,
-      "grad_norm": 542.6490478515625,
-      "learning_rate": 3.013290085025579e-05,
-      "loss": 64.4751,
-      "step": 122690
-    },
-    {
-      "epoch": 0.49572352606083625,
-      "grad_norm": 933.4337158203125,
-      "learning_rate": 3.0129484507670115e-05,
-      "loss": 57.9422,
-      "step": 122700
-    },
-    {
-      "epoch": 0.49576392732620383,
-      "grad_norm": 1414.308837890625,
-      "learning_rate": 3.0126068065082504e-05,
-      "loss": 50.3865,
-      "step": 122710
-    },
-    {
-      "epoch": 0.4958043285915715,
-      "grad_norm": 900.0178833007812,
-      "learning_rate": 3.0122651522559553e-05,
-      "loss": 54.273,
-      "step": 122720
-    },
-    {
-      "epoch": 0.4958447298569391,
-      "grad_norm": 1027.331787109375,
-      "learning_rate": 3.0119234880167867e-05,
-      "loss": 46.3332,
-      "step": 122730
-    },
-    {
-      "epoch": 0.49588513112230675,
-      "grad_norm": 527.3615112304688,
-      "learning_rate": 3.0115818137974067e-05,
-      "loss": 98.7442,
-      "step": 122740
-    },
-    {
-      "epoch": 0.4959255323876744,
-      "grad_norm": 546.8230590820312,
-      "learning_rate": 3.0112401296044757e-05,
-      "loss": 68.4313,
-      "step": 122750
-    },
-    {
-      "epoch": 0.49596593365304203,
-      "grad_norm": 835.651611328125,
-      "learning_rate": 3.0108984354446556e-05,
-      "loss": 45.3258,
-      "step": 122760
-    },
-    {
-      "epoch": 0.4960063349184096,
-      "grad_norm": 459.29144287109375,
-      "learning_rate": 3.0105567313246074e-05,
-      "loss": 70.6443,
-      "step": 122770
-    },
-    {
-      "epoch": 0.49604673618377726,
-      "grad_norm": 1120.5982666015625,
-      "learning_rate": 3.010215017250993e-05,
-      "loss": 73.4087,
-      "step": 122780
-    },
-    {
-      "epoch": 0.4960871374491449,
-      "grad_norm": 815.575927734375,
-      "learning_rate": 3.0098732932304734e-05,
-      "loss": 51.7506,
-      "step": 122790
-    },
-    {
-      "epoch": 0.49612753871451254,
-      "grad_norm": 529.2800903320312,
-      "learning_rate": 3.0095315592697126e-05,
-      "loss": 43.3278,
-      "step": 122800
-    },
-    {
-      "epoch": 0.4961679399798802,
-      "grad_norm": 352.13128662109375,
-      "learning_rate": 3.0091898153753705e-05,
-      "loss": 52.9101,
-      "step": 122810
-    },
-    {
-      "epoch": 0.4962083412452478,
-      "grad_norm": 2164.66064453125,
-      "learning_rate": 3.0088480615541113e-05,
-      "loss": 69.647,
-      "step": 122820
-    },
-    {
-      "epoch": 0.49624874251061546,
-      "grad_norm": 468.87158203125,
-      "learning_rate": 3.0085062978125967e-05,
-      "loss": 65.0103,
-      "step": 122830
-    },
-    {
-      "epoch": 0.49628914377598304,
-      "grad_norm": 771.53857421875,
-      "learning_rate": 3.008164524157491e-05,
-      "loss": 47.0131,
-      "step": 122840
-    },
-    {
-      "epoch": 0.4963295450413507,
-      "grad_norm": 630.9580078125,
-      "learning_rate": 3.0078227405954557e-05,
-      "loss": 36.1967,
-      "step": 122850
-    },
-    {
-      "epoch": 0.4963699463067183,
-      "grad_norm": 632.0338745117188,
-      "learning_rate": 3.007480947133155e-05,
-      "loss": 78.9644,
-      "step": 122860
-    },
-    {
-      "epoch": 0.49641034757208596,
-      "grad_norm": 748.582763671875,
-      "learning_rate": 3.0071391437772516e-05,
-      "loss": 52.5742,
-      "step": 122870
-    },
-    {
-      "epoch": 0.4964507488374536,
-      "grad_norm": 402.97882080078125,
-      "learning_rate": 3.00679733053441e-05,
-      "loss": 49.2446,
-      "step": 122880
-    },
-    {
-      "epoch": 0.49649115010282124,
-      "grad_norm": 814.5822143554688,
-      "learning_rate": 3.0064555074112927e-05,
-      "loss": 49.1269,
-      "step": 122890
-    },
-    {
-      "epoch": 0.49653155136818883,
-      "grad_norm": 484.45751953125,
-      "learning_rate": 3.0061136744145652e-05,
-      "loss": 46.5377,
-      "step": 122900
-    },
-    {
-      "epoch": 0.49657195263355647,
-      "grad_norm": 1381.5684814453125,
-      "learning_rate": 3.0057718315508905e-05,
-      "loss": 56.9945,
-      "step": 122910
-    },
-    {
-      "epoch": 0.4966123538989241,
-      "grad_norm": 1052.2681884765625,
-      "learning_rate": 3.005429978826934e-05,
-      "loss": 54.9941,
-      "step": 122920
-    },
-    {
-      "epoch": 0.49665275516429175,
-      "grad_norm": 1034.386474609375,
-      "learning_rate": 3.0050881162493593e-05,
-      "loss": 46.4848,
-      "step": 122930
-    },
-    {
-      "epoch": 0.4966931564296594,
-      "grad_norm": 919.5922241210938,
-      "learning_rate": 3.004746243824833e-05,
-      "loss": 56.9942,
-      "step": 122940
-    },
-    {
-      "epoch": 0.49673355769502703,
-      "grad_norm": 787.20947265625,
-      "learning_rate": 3.0044043615600175e-05,
-      "loss": 65.2977,
-      "step": 122950
-    },
-    {
-      "epoch": 0.4967739589603946,
-      "grad_norm": 0.0,
-      "learning_rate": 3.0040624694615803e-05,
-      "loss": 41.2134,
-      "step": 122960
-    },
-    {
-      "epoch": 0.49681436022576225,
-      "grad_norm": 683.5494995117188,
-      "learning_rate": 3.003720567536185e-05,
-      "loss": 72.0073,
-      "step": 122970
-    },
-    {
-      "epoch": 0.4968547614911299,
-      "grad_norm": 575.966796875,
-      "learning_rate": 3.003378655790498e-05,
-      "loss": 65.4787,
-      "step": 122980
-    },
-    {
-      "epoch": 0.49689516275649753,
-      "grad_norm": 1353.0836181640625,
-      "learning_rate": 3.0030367342311848e-05,
-      "loss": 67.0195,
-      "step": 122990
-    },
-    {
-      "epoch": 0.4969355640218652,
-      "grad_norm": 590.2861328125,
-      "learning_rate": 3.002694802864912e-05,
-      "loss": 53.1483,
-      "step": 123000
-    },
-    {
-      "epoch": 0.4969759652872328,
-      "grad_norm": 1127.41259765625,
-      "learning_rate": 3.002352861698345e-05,
-      "loss": 88.6338,
-      "step": 123010
-    },
-    {
-      "epoch": 0.49701636655260045,
-      "grad_norm": 902.4027099609375,
-      "learning_rate": 3.00201091073815e-05,
-      "loss": 39.8632,
-      "step": 123020
-    },
-    {
-      "epoch": 0.49705676781796804,
-      "grad_norm": 760.8033447265625,
-      "learning_rate": 3.0016689499909945e-05,
-      "loss": 37.9242,
-      "step": 123030
-    },
-    {
-      "epoch": 0.4970971690833357,
-      "grad_norm": 612.1544189453125,
-      "learning_rate": 3.0013269794635446e-05,
-      "loss": 56.6241,
-      "step": 123040
-    },
-    {
-      "epoch": 0.4971375703487033,
-      "grad_norm": 777.5595092773438,
-      "learning_rate": 3.0009849991624662e-05,
-      "loss": 82.5385,
-      "step": 123050
-    },
-    {
-      "epoch": 0.49717797161407096,
-      "grad_norm": 1702.484619140625,
-      "learning_rate": 3.0006430090944277e-05,
-      "loss": 59.1968,
-      "step": 123060
-    },
-    {
-      "epoch": 0.4972183728794386,
-      "grad_norm": 537.38134765625,
-      "learning_rate": 3.000301009266096e-05,
-      "loss": 46.0222,
-      "step": 123070
-    },
-    {
-      "epoch": 0.49725877414480624,
-      "grad_norm": 305.5016784667969,
-      "learning_rate": 2.9999589996841386e-05,
-      "loss": 57.2696,
-      "step": 123080
-    },
-    {
-      "epoch": 0.4972991754101738,
-      "grad_norm": 844.5637817382812,
-      "learning_rate": 2.9996169803552233e-05,
-      "loss": 56.0423,
-      "step": 123090
-    },
-    {
-      "epoch": 0.49733957667554146,
-      "grad_norm": 883.1146850585938,
-      "learning_rate": 2.9992749512860173e-05,
-      "loss": 32.5165,
-      "step": 123100
-    },
-    {
-      "epoch": 0.4973799779409091,
-      "grad_norm": 1908.025634765625,
-      "learning_rate": 2.99893291248319e-05,
-      "loss": 110.2845,
-      "step": 123110
-    },
-    {
-      "epoch": 0.49742037920627674,
-      "grad_norm": 1058.1068115234375,
-      "learning_rate": 2.9985908639534075e-05,
-      "loss": 67.6201,
-      "step": 123120
-    },
-    {
-      "epoch": 0.4974607804716444,
-      "grad_norm": 1846.614013671875,
-      "learning_rate": 2.998248805703341e-05,
-      "loss": 49.1556,
-      "step": 123130
-    },
-    {
-      "epoch": 0.497501181737012,
-      "grad_norm": 1011.847412109375,
-      "learning_rate": 2.9979067377396565e-05,
-      "loss": 25.5965,
-      "step": 123140
-    },
-    {
-      "epoch": 0.49754158300237966,
-      "grad_norm": 627.6660766601562,
-      "learning_rate": 2.9975646600690234e-05,
-      "loss": 34.3208,
-      "step": 123150
-    },
-    {
-      "epoch": 0.49758198426774725,
-      "grad_norm": 770.2060546875,
-      "learning_rate": 2.9972225726981113e-05,
-      "loss": 52.1218,
-      "step": 123160
-    },
-    {
-      "epoch": 0.4976223855331149,
-      "grad_norm": 398.234130859375,
-      "learning_rate": 2.99688047563359e-05,
-      "loss": 60.0683,
-      "step": 123170
-    },
-    {
-      "epoch": 0.4976627867984825,
-      "grad_norm": 1440.9393310546875,
-      "learning_rate": 2.996538368882127e-05,
-      "loss": 94.9321,
-      "step": 123180
-    },
-    {
-      "epoch": 0.49770318806385017,
-      "grad_norm": 580.39892578125,
-      "learning_rate": 2.9961962524503927e-05,
-      "loss": 46.6695,
-      "step": 123190
-    },
-    {
-      "epoch": 0.4977435893292178,
-      "grad_norm": 402.80419921875,
-      "learning_rate": 2.9958541263450584e-05,
-      "loss": 58.4956,
-      "step": 123200
-    },
-    {
-      "epoch": 0.49778399059458545,
-      "grad_norm": 709.5952758789062,
-      "learning_rate": 2.9955119905727925e-05,
-      "loss": 54.5307,
-      "step": 123210
-    },
-    {
-      "epoch": 0.49782439185995303,
-      "grad_norm": 418.1485290527344,
-      "learning_rate": 2.995169845140264e-05,
-      "loss": 33.9692,
-      "step": 123220
-    },
-    {
-      "epoch": 0.49786479312532067,
-      "grad_norm": 555.658203125,
-      "learning_rate": 2.994827690054145e-05,
-      "loss": 57.7686,
-      "step": 123230
-    },
-    {
-      "epoch": 0.4979051943906883,
-      "grad_norm": 1826.1614990234375,
-      "learning_rate": 2.9944855253211052e-05,
-      "loss": 59.2031,
-      "step": 123240
-    },
-    {
-      "epoch": 0.49794559565605595,
-      "grad_norm": 728.795166015625,
-      "learning_rate": 2.9941433509478156e-05,
-      "loss": 64.7671,
-      "step": 123250
-    },
-    {
-      "epoch": 0.4979859969214236,
-      "grad_norm": 562.8650512695312,
-      "learning_rate": 2.993801166940947e-05,
-      "loss": 30.4642,
-      "step": 123260
-    },
-    {
-      "epoch": 0.49802639818679123,
-      "grad_norm": 718.48681640625,
-      "learning_rate": 2.9934589733071704e-05,
-      "loss": 43.2636,
-      "step": 123270
-    },
-    {
-      "epoch": 0.4980667994521588,
-      "grad_norm": 775.666748046875,
-      "learning_rate": 2.9931167700531578e-05,
-      "loss": 64.5946,
-      "step": 123280
-    },
-    {
-      "epoch": 0.49810720071752645,
-      "grad_norm": 941.3685302734375,
-      "learning_rate": 2.9927745571855786e-05,
-      "loss": 68.1941,
-      "step": 123290
-    },
-    {
-      "epoch": 0.4981476019828941,
-      "grad_norm": 806.8750610351562,
-      "learning_rate": 2.9924323347111073e-05,
-      "loss": 75.4404,
-      "step": 123300
-    },
-    {
-      "epoch": 0.49818800324826173,
-      "grad_norm": 1879.86279296875,
-      "learning_rate": 2.992090102636413e-05,
-      "loss": 82.7253,
-      "step": 123310
-    },
-    {
-      "epoch": 0.4982284045136294,
-      "grad_norm": 500.6525573730469,
-      "learning_rate": 2.991747860968168e-05,
-      "loss": 59.617,
-      "step": 123320
-    },
-    {
-      "epoch": 0.498268805778997,
-      "grad_norm": 576.9788818359375,
-      "learning_rate": 2.9914056097130473e-05,
-      "loss": 37.7997,
-      "step": 123330
-    },
-    {
-      "epoch": 0.49830920704436465,
-      "grad_norm": 1132.8011474609375,
-      "learning_rate": 2.9910633488777196e-05,
-      "loss": 60.6151,
-      "step": 123340
-    },
-    {
-      "epoch": 0.49834960830973224,
-      "grad_norm": 1793.87109375,
-      "learning_rate": 2.99072107846886e-05,
-      "loss": 49.606,
-      "step": 123350
-    },
-    {
-      "epoch": 0.4983900095750999,
-      "grad_norm": 226.36434936523438,
-      "learning_rate": 2.9903787984931396e-05,
-      "loss": 29.6053,
-      "step": 123360
-    },
-    {
-      "epoch": 0.4984304108404675,
-      "grad_norm": 257.8052673339844,
-      "learning_rate": 2.9900365089572328e-05,
-      "loss": 42.242,
-      "step": 123370
-    },
-    {
-      "epoch": 0.49847081210583516,
-      "grad_norm": 719.3456420898438,
-      "learning_rate": 2.9896942098678122e-05,
-      "loss": 46.6997,
-      "step": 123380
-    },
-    {
-      "epoch": 0.4985112133712028,
-      "grad_norm": 641.3980102539062,
-      "learning_rate": 2.9893519012315503e-05,
-      "loss": 67.5891,
-      "step": 123390
-    },
-    {
-      "epoch": 0.49855161463657044,
-      "grad_norm": 685.2548828125,
-      "learning_rate": 2.9890095830551207e-05,
-      "loss": 60.4898,
-      "step": 123400
-    },
-    {
-      "epoch": 0.498592015901938,
-      "grad_norm": 632.763671875,
-      "learning_rate": 2.9886672553451985e-05,
-      "loss": 61.0664,
-      "step": 123410
-    },
-    {
-      "epoch": 0.49863241716730566,
-      "grad_norm": 0.0,
-      "learning_rate": 2.988324918108456e-05,
-      "loss": 56.5434,
-      "step": 123420
-    },
-    {
-      "epoch": 0.4986728184326733,
-      "grad_norm": 1752.1431884765625,
-      "learning_rate": 2.9879825713515676e-05,
-      "loss": 105.4006,
-      "step": 123430
-    },
-    {
-      "epoch": 0.49871321969804094,
-      "grad_norm": 575.599853515625,
-      "learning_rate": 2.9876402150812078e-05,
-      "loss": 52.491,
-      "step": 123440
-    },
-    {
-      "epoch": 0.4987536209634086,
-      "grad_norm": 620.3259887695312,
-      "learning_rate": 2.9872978493040514e-05,
-      "loss": 48.9634,
-      "step": 123450
-    },
-    {
-      "epoch": 0.4987940222287762,
-      "grad_norm": 837.0609741210938,
-      "learning_rate": 2.9869554740267724e-05,
-      "loss": 46.9197,
-      "step": 123460
-    },
-    {
-      "epoch": 0.49883442349414386,
-      "grad_norm": 231.35792541503906,
-      "learning_rate": 2.986613089256046e-05,
-      "loss": 62.4719,
-      "step": 123470
-    },
-    {
-      "epoch": 0.49887482475951145,
-      "grad_norm": 848.2174072265625,
-      "learning_rate": 2.9862706949985463e-05,
-      "loss": 63.5692,
-      "step": 123480
-    },
-    {
-      "epoch": 0.4989152260248791,
-      "grad_norm": 482.8079833984375,
-      "learning_rate": 2.9859282912609497e-05,
-      "loss": 48.4904,
-      "step": 123490
-    },
-    {
-      "epoch": 0.49895562729024673,
-      "grad_norm": 786.4293212890625,
-      "learning_rate": 2.98558587804993e-05,
-      "loss": 51.905,
-      "step": 123500
-    },
-    {
-      "epoch": 0.49899602855561437,
-      "grad_norm": 967.361083984375,
-      "learning_rate": 2.9852434553721642e-05,
-      "loss": 55.1196,
-      "step": 123510
-    },
-    {
-      "epoch": 0.499036429820982,
-      "grad_norm": 691.3804321289062,
-      "learning_rate": 2.984901023234327e-05,
-      "loss": 63.2347,
-      "step": 123520
-    },
-    {
-      "epoch": 0.49907683108634965,
-      "grad_norm": 2461.232421875,
-      "learning_rate": 2.9845585816430955e-05,
-      "loss": 86.1027,
-      "step": 123530
-    },
-    {
-      "epoch": 0.49911723235171723,
-      "grad_norm": 625.51220703125,
-      "learning_rate": 2.9842161306051446e-05,
-      "loss": 59.4591,
-      "step": 123540
-    },
-    {
-      "epoch": 0.4991576336170849,
-      "grad_norm": 937.5235595703125,
-      "learning_rate": 2.9838736701271514e-05,
-      "loss": 86.4743,
-      "step": 123550
-    },
-    {
-      "epoch": 0.4991980348824525,
-      "grad_norm": 719.2523803710938,
-      "learning_rate": 2.9835312002157913e-05,
-      "loss": 57.3074,
-      "step": 123560
-    },
-    {
-      "epoch": 0.49923843614782015,
-      "grad_norm": 1146.9417724609375,
-      "learning_rate": 2.983188720877741e-05,
-      "loss": 59.0976,
-      "step": 123570
-    },
-    {
-      "epoch": 0.4992788374131878,
-      "grad_norm": 643.1884155273438,
-      "learning_rate": 2.9828462321196788e-05,
-      "loss": 59.2565,
-      "step": 123580
-    },
-    {
-      "epoch": 0.49931923867855543,
-      "grad_norm": 717.5197143554688,
-      "learning_rate": 2.9825037339482804e-05,
-      "loss": 62.1937,
-      "step": 123590
-    },
-    {
-      "epoch": 0.499359639943923,
-      "grad_norm": 911.0812377929688,
-      "learning_rate": 2.9821612263702226e-05,
-      "loss": 53.7806,
-      "step": 123600
-    },
-    {
-      "epoch": 0.49940004120929066,
-      "grad_norm": 655.2685546875,
-      "learning_rate": 2.981818709392184e-05,
-      "loss": 76.8916,
-      "step": 123610
-    },
-    {
-      "epoch": 0.4994404424746583,
-      "grad_norm": 694.6813354492188,
-      "learning_rate": 2.981476183020842e-05,
-      "loss": 54.2385,
-      "step": 123620
-    },
-    {
-      "epoch": 0.49948084374002594,
-      "grad_norm": 277.7393798828125,
-      "learning_rate": 2.9811336472628737e-05,
-      "loss": 56.6968,
-      "step": 123630
-    },
-    {
-      "epoch": 0.4995212450053936,
-      "grad_norm": 2142.0654296875,
-      "learning_rate": 2.9807911021249573e-05,
-      "loss": 71.4143,
-      "step": 123640
-    },
-    {
-      "epoch": 0.4995616462707612,
-      "grad_norm": 685.448974609375,
-      "learning_rate": 2.9804485476137706e-05,
-      "loss": 41.1674,
-      "step": 123650
-    },
-    {
-      "epoch": 0.49960204753612886,
-      "grad_norm": 0.0,
-      "learning_rate": 2.9801059837359925e-05,
-      "loss": 67.2189,
-      "step": 123660
-    },
-    {
-      "epoch": 0.49964244880149644,
-      "grad_norm": 563.3724975585938,
-      "learning_rate": 2.979763410498301e-05,
-      "loss": 42.097,
-      "step": 123670
-    },
-    {
-      "epoch": 0.4996828500668641,
-      "grad_norm": 823.16259765625,
-      "learning_rate": 2.9794208279073743e-05,
-      "loss": 46.5123,
-      "step": 123680
-    },
-    {
-      "epoch": 0.4997232513322317,
-      "grad_norm": 396.4298095703125,
-      "learning_rate": 2.9790782359698914e-05,
-      "loss": 38.3301,
-      "step": 123690
-    },
-    {
-      "epoch": 0.49976365259759936,
-      "grad_norm": 487.17205810546875,
-      "learning_rate": 2.9787356346925327e-05,
-      "loss": 54.4319,
-      "step": 123700
-    },
-    {
-      "epoch": 0.499804053862967,
-      "grad_norm": 1401.6353759765625,
-      "learning_rate": 2.9783930240819758e-05,
-      "loss": 89.3769,
-      "step": 123710
-    },
-    {
-      "epoch": 0.49984445512833464,
-      "grad_norm": 401.6158447265625,
-      "learning_rate": 2.978050404144901e-05,
-      "loss": 39.3288,
-      "step": 123720
-    },
-    {
-      "epoch": 0.4998848563937022,
-      "grad_norm": 1638.397216796875,
-      "learning_rate": 2.977707774887987e-05,
-      "loss": 85.3148,
-      "step": 123730
-    },
-    {
-      "epoch": 0.49992525765906987,
-      "grad_norm": 525.0819702148438,
-      "learning_rate": 2.9773651363179144e-05,
-      "loss": 62.3121,
-      "step": 123740
-    },
-    {
-      "epoch": 0.4999656589244375,
-      "grad_norm": 576.019775390625,
-      "learning_rate": 2.9770224884413623e-05,
-      "loss": 34.6271,
-      "step": 123750
-    },
-    {
-      "epoch": 0.5000060601898051,
-      "grad_norm": 596.5224609375,
-      "learning_rate": 2.9766798312650112e-05,
-      "loss": 45.5462,
-      "step": 123760
-    },
-    {
-      "epoch": 0.5000464614551727,
-      "grad_norm": 1210.31005859375,
-      "learning_rate": 2.976337164795541e-05,
-      "loss": 43.6167,
-      "step": 123770
-    },
-    {
-      "epoch": 0.5000868627205404,
-      "grad_norm": 1921.7060546875,
-      "learning_rate": 2.975994489039634e-05,
-      "loss": 52.5438,
-      "step": 123780
-    },
-    {
-      "epoch": 0.500127263985908,
-      "grad_norm": 906.3182373046875,
-      "learning_rate": 2.9756518040039682e-05,
-      "loss": 64.2235,
-      "step": 123790
-    },
-    {
-      "epoch": 0.5001676652512757,
-      "grad_norm": 724.0418701171875,
-      "learning_rate": 2.9753091096952255e-05,
-      "loss": 63.5248,
-      "step": 123800
-    },
-    {
-      "epoch": 0.5002080665166433,
-      "grad_norm": 1490.1466064453125,
-      "learning_rate": 2.9749664061200877e-05,
-      "loss": 38.7024,
-      "step": 123810
-    },
-    {
-      "epoch": 0.5002484677820109,
-      "grad_norm": 262.531494140625,
-      "learning_rate": 2.9746236932852355e-05,
-      "loss": 66.6208,
-      "step": 123820
-    },
-    {
-      "epoch": 0.5002888690473786,
-      "grad_norm": 283.681640625,
-      "learning_rate": 2.974280971197349e-05,
-      "loss": 45.1853,
-      "step": 123830
-    },
-    {
-      "epoch": 0.5003292703127462,
-      "grad_norm": 636.9791259765625,
-      "learning_rate": 2.973938239863111e-05,
-      "loss": 54.2482,
-      "step": 123840
-    },
-    {
-      "epoch": 0.5003696715781139,
-      "grad_norm": 1471.0166015625,
-      "learning_rate": 2.9735954992892033e-05,
-      "loss": 73.4811,
-      "step": 123850
-    },
-    {
-      "epoch": 0.5004100728434815,
-      "grad_norm": 870.6261596679688,
-      "learning_rate": 2.9732527494823083e-05,
-      "loss": 69.1818,
-      "step": 123860
-    },
-    {
-      "epoch": 0.5004504741088491,
-      "grad_norm": 1015.466064453125,
-      "learning_rate": 2.9729099904491058e-05,
-      "loss": 65.9109,
-      "step": 123870
-    },
-    {
-      "epoch": 0.5004908753742168,
-      "grad_norm": 769.5751953125,
-      "learning_rate": 2.97256722219628e-05,
-      "loss": 41.2782,
-      "step": 123880
-    },
-    {
-      "epoch": 0.5005312766395843,
-      "grad_norm": 814.8590087890625,
-      "learning_rate": 2.9722244447305135e-05,
-      "loss": 84.8106,
-      "step": 123890
-    },
-    {
-      "epoch": 0.5005716779049519,
-      "grad_norm": 476.1192932128906,
-      "learning_rate": 2.9718816580584884e-05,
-      "loss": 43.7513,
-      "step": 123900
-    },
-    {
-      "epoch": 0.5006120791703196,
-      "grad_norm": 729.9635620117188,
-      "learning_rate": 2.9715388621868873e-05,
-      "loss": 44.9804,
-      "step": 123910
-    },
-    {
-      "epoch": 0.5006524804356872,
-      "grad_norm": 657.6481323242188,
-      "learning_rate": 2.971196057122393e-05,
-      "loss": 68.3173,
-      "step": 123920
-    },
-    {
-      "epoch": 0.5006928817010549,
-      "grad_norm": 620.4969482421875,
-      "learning_rate": 2.9708532428716883e-05,
-      "loss": 50.9084,
-      "step": 123930
-    },
-    {
-      "epoch": 0.5007332829664225,
-      "grad_norm": 840.12890625,
-      "learning_rate": 2.9705104194414586e-05,
-      "loss": 63.0937,
-      "step": 123940
-    },
-    {
-      "epoch": 0.5007736842317901,
-      "grad_norm": 1180.94970703125,
-      "learning_rate": 2.9701675868383848e-05,
-      "loss": 69.3689,
-      "step": 123950
-    },
-    {
-      "epoch": 0.5008140854971578,
-      "grad_norm": 630.738037109375,
-      "learning_rate": 2.9698247450691525e-05,
-      "loss": 56.1486,
-      "step": 123960
-    },
-    {
-      "epoch": 0.5008544867625254,
-      "grad_norm": 1723.3427734375,
-      "learning_rate": 2.9694818941404444e-05,
-      "loss": 52.5067,
-      "step": 123970
-    },
-    {
-      "epoch": 0.5008948880278931,
-      "grad_norm": 1080.7830810546875,
-      "learning_rate": 2.9691390340589466e-05,
-      "loss": 63.3121,
-      "step": 123980
-    },
-    {
-      "epoch": 0.5009352892932607,
-      "grad_norm": 428.3757629394531,
-      "learning_rate": 2.9687961648313405e-05,
-      "loss": 45.4763,
-      "step": 123990
-    },
-    {
-      "epoch": 0.5009756905586283,
-      "grad_norm": 585.084228515625,
-      "learning_rate": 2.9684532864643122e-05,
-      "loss": 47.8538,
-      "step": 124000
-    },
-    {
-      "epoch": 0.501016091823996,
-      "grad_norm": 951.522705078125,
-      "learning_rate": 2.9681103989645453e-05,
-      "loss": 64.0775,
-      "step": 124010
-    },
-    {
-      "epoch": 0.5010564930893635,
-      "grad_norm": 414.710693359375,
-      "learning_rate": 2.9677675023387258e-05,
-      "loss": 35.3562,
-      "step": 124020
-    },
-    {
-      "epoch": 0.5010968943547311,
-      "grad_norm": 1510.4676513671875,
-      "learning_rate": 2.9674245965935378e-05,
-      "loss": 85.7635,
-      "step": 124030
-    },
-    {
-      "epoch": 0.5011372956200988,
-      "grad_norm": 555.8351440429688,
-      "learning_rate": 2.9670816817356668e-05,
-      "loss": 39.8303,
-      "step": 124040
-    },
-    {
-      "epoch": 0.5011776968854664,
-      "grad_norm": 976.0780639648438,
-      "learning_rate": 2.9667387577717976e-05,
-      "loss": 61.9712,
-      "step": 124050
-    },
-    {
-      "epoch": 0.5012180981508341,
-      "grad_norm": 1387.890869140625,
-      "learning_rate": 2.9663958247086166e-05,
-      "loss": 58.7585,
-      "step": 124060
-    },
-    {
-      "epoch": 0.5012584994162017,
-      "grad_norm": 1036.8333740234375,
-      "learning_rate": 2.966052882552809e-05,
-      "loss": 44.4313,
-      "step": 124070
-    },
-    {
-      "epoch": 0.5012989006815693,
-      "grad_norm": 1022.4268798828125,
-      "learning_rate": 2.9657099313110593e-05,
-      "loss": 57.0276,
-      "step": 124080
-    },
-    {
-      "epoch": 0.501339301946937,
-      "grad_norm": 1398.85546875,
-      "learning_rate": 2.9653669709900555e-05,
-      "loss": 58.4346,
-      "step": 124090
-    },
-    {
-      "epoch": 0.5013797032123046,
-      "grad_norm": 496.8158264160156,
-      "learning_rate": 2.9650240015964825e-05,
-      "loss": 62.5083,
-      "step": 124100
-    },
-    {
-      "epoch": 0.5014201044776723,
-      "grad_norm": 911.7360229492188,
-      "learning_rate": 2.964681023137028e-05,
-      "loss": 52.9003,
-      "step": 124110
-    },
-    {
-      "epoch": 0.5014605057430399,
-      "grad_norm": 932.399658203125,
-      "learning_rate": 2.9643380356183775e-05,
-      "loss": 56.112,
-      "step": 124120
-    },
-    {
-      "epoch": 0.5015009070084075,
-      "grad_norm": 0.0,
-      "learning_rate": 2.9639950390472177e-05,
-      "loss": 45.8005,
-      "step": 124130
-    },
-    {
-      "epoch": 0.5015413082737752,
-      "grad_norm": 789.1591796875,
-      "learning_rate": 2.9636520334302354e-05,
-      "loss": 73.2515,
-      "step": 124140
-    },
-    {
-      "epoch": 0.5015817095391427,
-      "grad_norm": 1213.167236328125,
-      "learning_rate": 2.9633090187741185e-05,
-      "loss": 45.1217,
-      "step": 124150
-    },
-    {
-      "epoch": 0.5016221108045104,
-      "grad_norm": 1115.8275146484375,
-      "learning_rate": 2.9629659950855544e-05,
-      "loss": 48.8063,
-      "step": 124160
-    },
-    {
-      "epoch": 0.501662512069878,
-      "grad_norm": 1056.0667724609375,
-      "learning_rate": 2.9626229623712288e-05,
-      "loss": 59.1212,
-      "step": 124170
-    },
-    {
-      "epoch": 0.5017029133352456,
-      "grad_norm": 269.9258728027344,
-      "learning_rate": 2.9622799206378305e-05,
-      "loss": 83.2247,
-      "step": 124180
-    },
-    {
-      "epoch": 0.5017433146006133,
-      "grad_norm": 539.0936889648438,
-      "learning_rate": 2.961936869892048e-05,
-      "loss": 53.4077,
-      "step": 124190
-    },
-    {
-      "epoch": 0.5017837158659809,
-      "grad_norm": 3030.404052734375,
-      "learning_rate": 2.9615938101405676e-05,
-      "loss": 76.7048,
-      "step": 124200
-    },
-    {
-      "epoch": 0.5018241171313486,
-      "grad_norm": 1794.911376953125,
-      "learning_rate": 2.961250741390078e-05,
-      "loss": 60.769,
-      "step": 124210
-    },
-    {
-      "epoch": 0.5018645183967162,
-      "grad_norm": 1027.421875,
-      "learning_rate": 2.960907663647268e-05,
-      "loss": 43.8524,
-      "step": 124220
-    },
-    {
-      "epoch": 0.5019049196620838,
-      "grad_norm": 726.89697265625,
-      "learning_rate": 2.9605645769188268e-05,
-      "loss": 57.9974,
-      "step": 124230
-    },
-    {
-      "epoch": 0.5019453209274515,
-      "grad_norm": 741.7726440429688,
-      "learning_rate": 2.9602214812114415e-05,
-      "loss": 45.0123,
-      "step": 124240
-    },
-    {
-      "epoch": 0.5019857221928191,
-      "grad_norm": 572.0324096679688,
-      "learning_rate": 2.9598783765318007e-05,
-      "loss": 41.7237,
-      "step": 124250
-    },
-    {
-      "epoch": 0.5020261234581868,
-      "grad_norm": 905.0762329101562,
-      "learning_rate": 2.9595352628865947e-05,
-      "loss": 60.8909,
-      "step": 124260
-    },
-    {
-      "epoch": 0.5020665247235544,
-      "grad_norm": 418.3033447265625,
-      "learning_rate": 2.9591921402825123e-05,
-      "loss": 56.9024,
-      "step": 124270
-    },
-    {
-      "epoch": 0.5021069259889219,
-      "grad_norm": 642.4514770507812,
-      "learning_rate": 2.958849008726242e-05,
-      "loss": 56.9975,
-      "step": 124280
-    },
-    {
-      "epoch": 0.5021473272542896,
-      "grad_norm": 545.2382202148438,
-      "learning_rate": 2.9585058682244748e-05,
-      "loss": 54.7912,
-      "step": 124290
-    },
-    {
-      "epoch": 0.5021877285196572,
-      "grad_norm": 1271.816162109375,
-      "learning_rate": 2.9581627187838994e-05,
-      "loss": 67.879,
-      "step": 124300
-    },
-    {
-      "epoch": 0.5022281297850248,
-      "grad_norm": 404.8785095214844,
-      "learning_rate": 2.9578195604112064e-05,
-      "loss": 68.5172,
-      "step": 124310
-    },
-    {
-      "epoch": 0.5022685310503925,
-      "grad_norm": 1688.64599609375,
-      "learning_rate": 2.9574763931130843e-05,
-      "loss": 53.8717,
-      "step": 124320
-    },
-    {
-      "epoch": 0.5023089323157601,
-      "grad_norm": 574.3252563476562,
-      "learning_rate": 2.9571332168962256e-05,
-      "loss": 66.5893,
-      "step": 124330
-    },
-    {
-      "epoch": 0.5023493335811278,
-      "grad_norm": 418.29864501953125,
-      "learning_rate": 2.956790031767319e-05,
-      "loss": 47.6842,
-      "step": 124340
-    },
-    {
-      "epoch": 0.5023897348464954,
-      "grad_norm": 1550.780029296875,
-      "learning_rate": 2.9564468377330556e-05,
-      "loss": 66.8126,
-      "step": 124350
-    },
-    {
-      "epoch": 0.502430136111863,
-      "grad_norm": 5627.90478515625,
-      "learning_rate": 2.956103634800126e-05,
-      "loss": 86.4269,
-      "step": 124360
-    },
-    {
-      "epoch": 0.5024705373772307,
-      "grad_norm": 2004.539306640625,
-      "learning_rate": 2.9557604229752212e-05,
-      "loss": 57.9298,
-      "step": 124370
-    },
-    {
-      "epoch": 0.5025109386425983,
-      "grad_norm": 531.3135986328125,
-      "learning_rate": 2.9554172022650317e-05,
-      "loss": 53.1231,
-      "step": 124380
-    },
-    {
-      "epoch": 0.502551339907966,
-      "grad_norm": 1106.7730712890625,
-      "learning_rate": 2.9550739726762507e-05,
-      "loss": 62.664,
-      "step": 124390
-    },
-    {
-      "epoch": 0.5025917411733335,
-      "grad_norm": 1226.38818359375,
-      "learning_rate": 2.9547307342155673e-05,
-      "loss": 94.7886,
-      "step": 124400
-    },
-    {
-      "epoch": 0.5026321424387011,
-      "grad_norm": 608.1491088867188,
-      "learning_rate": 2.9543874868896747e-05,
-      "loss": 71.896,
-      "step": 124410
-    },
-    {
-      "epoch": 0.5026725437040688,
-      "grad_norm": 240.57858276367188,
-      "learning_rate": 2.954044230705264e-05,
-      "loss": 96.0706,
-      "step": 124420
-    },
-    {
-      "epoch": 0.5027129449694364,
-      "grad_norm": 596.6292114257812,
-      "learning_rate": 2.9537009656690275e-05,
-      "loss": 50.2395,
-      "step": 124430
-    },
-    {
-      "epoch": 0.502753346234804,
-      "grad_norm": 742.9195556640625,
-      "learning_rate": 2.953357691787656e-05,
-      "loss": 39.4977,
-      "step": 124440
-    },
-    {
-      "epoch": 0.5027937475001717,
-      "grad_norm": 715.0127563476562,
-      "learning_rate": 2.9530144090678435e-05,
-      "loss": 40.7906,
-      "step": 124450
-    },
-    {
-      "epoch": 0.5028341487655393,
-      "grad_norm": 494.5076599121094,
-      "learning_rate": 2.952671117516282e-05,
-      "loss": 57.1695,
-      "step": 124460
-    },
-    {
-      "epoch": 0.502874550030907,
-      "grad_norm": 766.6427001953125,
-      "learning_rate": 2.952327817139664e-05,
-      "loss": 62.9475,
-      "step": 124470
-    },
-    {
-      "epoch": 0.5029149512962746,
-      "grad_norm": 670.2885131835938,
-      "learning_rate": 2.9519845079446823e-05,
-      "loss": 47.9745,
-      "step": 124480
-    },
-    {
-      "epoch": 0.5029553525616423,
-      "grad_norm": 636.1236572265625,
-      "learning_rate": 2.9516411899380296e-05,
-      "loss": 57.113,
-      "step": 124490
-    },
-    {
-      "epoch": 0.5029957538270099,
-      "grad_norm": 434.45477294921875,
-      "learning_rate": 2.9512978631264006e-05,
-      "loss": 44.9944,
-      "step": 124500
-    },
-    {
-      "epoch": 0.5030361550923775,
-      "grad_norm": 687.5653686523438,
-      "learning_rate": 2.950954527516487e-05,
-      "loss": 50.9721,
-      "step": 124510
-    },
-    {
-      "epoch": 0.5030765563577452,
-      "grad_norm": 451.1505126953125,
-      "learning_rate": 2.9506111831149818e-05,
-      "loss": 44.7546,
-      "step": 124520
-    },
-    {
-      "epoch": 0.5031169576231127,
-      "grad_norm": 1212.3389892578125,
-      "learning_rate": 2.9502678299285798e-05,
-      "loss": 44.7764,
-      "step": 124530
-    },
-    {
-      "epoch": 0.5031573588884803,
-      "grad_norm": 572.8779907226562,
-      "learning_rate": 2.949924467963975e-05,
-      "loss": 48.557,
-      "step": 124540
-    },
-    {
-      "epoch": 0.503197760153848,
-      "grad_norm": 520.7388916015625,
-      "learning_rate": 2.949581097227861e-05,
-      "loss": 80.5346,
-      "step": 124550
-    },
-    {
-      "epoch": 0.5032381614192156,
-      "grad_norm": 713.9329833984375,
-      "learning_rate": 2.9492377177269315e-05,
-      "loss": 32.9715,
-      "step": 124560
-    },
-    {
-      "epoch": 0.5032785626845833,
-      "grad_norm": 978.047119140625,
-      "learning_rate": 2.9488943294678818e-05,
-      "loss": 52.3525,
-      "step": 124570
-    },
-    {
-      "epoch": 0.5033189639499509,
-      "grad_norm": 1501.214599609375,
-      "learning_rate": 2.948550932457407e-05,
-      "loss": 55.6765,
-      "step": 124580
-    },
-    {
-      "epoch": 0.5033593652153185,
-      "grad_norm": 625.777587890625,
-      "learning_rate": 2.9482075267021995e-05,
-      "loss": 46.8336,
-      "step": 124590
-    },
-    {
-      "epoch": 0.5033997664806862,
-      "grad_norm": 601.8237915039062,
-      "learning_rate": 2.9478641122089562e-05,
-      "loss": 51.4297,
-      "step": 124600
-    },
-    {
-      "epoch": 0.5034401677460538,
-      "grad_norm": 924.4441528320312,
-      "learning_rate": 2.947520688984371e-05,
-      "loss": 60.7378,
-      "step": 124610
-    },
-    {
-      "epoch": 0.5034805690114215,
-      "grad_norm": 516.689453125,
-      "learning_rate": 2.9471772570351398e-05,
-      "loss": 31.4356,
-      "step": 124620
-    },
-    {
-      "epoch": 0.5035209702767891,
-      "grad_norm": 1193.8798828125,
-      "learning_rate": 2.9468338163679577e-05,
-      "loss": 60.812,
-      "step": 124630
-    },
-    {
-      "epoch": 0.5035613715421567,
-      "grad_norm": 663.6973266601562,
-      "learning_rate": 2.9464903669895205e-05,
-      "loss": 35.3694,
-      "step": 124640
-    },
-    {
-      "epoch": 0.5036017728075244,
-      "grad_norm": 1353.3345947265625,
-      "learning_rate": 2.9461469089065234e-05,
-      "loss": 51.9646,
-      "step": 124650
-    },
-    {
-      "epoch": 0.5036421740728919,
-      "grad_norm": 529.1851806640625,
-      "learning_rate": 2.945803442125663e-05,
-      "loss": 48.2301,
-      "step": 124660
-    },
-    {
-      "epoch": 0.5036825753382596,
-      "grad_norm": 331.9324035644531,
-      "learning_rate": 2.9454599666536347e-05,
-      "loss": 54.444,
-      "step": 124670
-    },
-    {
-      "epoch": 0.5037229766036272,
-      "grad_norm": 620.619384765625,
-      "learning_rate": 2.9451164824971356e-05,
-      "loss": 92.8183,
-      "step": 124680
-    },
-    {
-      "epoch": 0.5037633778689948,
-      "grad_norm": 441.77734375,
-      "learning_rate": 2.9447729896628612e-05,
-      "loss": 47.082,
-      "step": 124690
-    },
-    {
-      "epoch": 0.5038037791343625,
-      "grad_norm": 891.2379150390625,
-      "learning_rate": 2.944429488157508e-05,
-      "loss": 70.2383,
-      "step": 124700
-    },
-    {
-      "epoch": 0.5038441803997301,
-      "grad_norm": 1449.03955078125,
-      "learning_rate": 2.9440859779877728e-05,
-      "loss": 57.3963,
-      "step": 124710
-    },
-    {
-      "epoch": 0.5038845816650978,
-      "grad_norm": 1024.277587890625,
-      "learning_rate": 2.943742459160354e-05,
-      "loss": 48.08,
-      "step": 124720
-    },
-    {
-      "epoch": 0.5039249829304654,
-      "grad_norm": 272.8819885253906,
-      "learning_rate": 2.9433989316819467e-05,
-      "loss": 65.6629,
-      "step": 124730
-    },
-    {
-      "epoch": 0.503965384195833,
-      "grad_norm": 665.0719604492188,
-      "learning_rate": 2.943055395559249e-05,
-      "loss": 60.4535,
-      "step": 124740
-    },
-    {
-      "epoch": 0.5040057854612007,
-      "grad_norm": 1616.3466796875,
-      "learning_rate": 2.9427118507989586e-05,
-      "loss": 88.4401,
-      "step": 124750
-    },
-    {
-      "epoch": 0.5040461867265683,
-      "grad_norm": 1536.4730224609375,
-      "learning_rate": 2.942368297407772e-05,
-      "loss": 67.7257,
-      "step": 124760
-    },
-    {
-      "epoch": 0.504086587991936,
-      "grad_norm": 258.504638671875,
-      "learning_rate": 2.942024735392389e-05,
-      "loss": 45.7606,
-      "step": 124770
-    },
-    {
-      "epoch": 0.5041269892573036,
-      "grad_norm": 1484.1357421875,
-      "learning_rate": 2.9416811647595048e-05,
-      "loss": 65.3528,
-      "step": 124780
-    },
-    {
-      "epoch": 0.5041673905226711,
-      "grad_norm": 933.3783569335938,
-      "learning_rate": 2.94133758551582e-05,
-      "loss": 30.6998,
-      "step": 124790
-    },
-    {
-      "epoch": 0.5042077917880388,
-      "grad_norm": 437.1009216308594,
-      "learning_rate": 2.9409939976680313e-05,
-      "loss": 73.0693,
-      "step": 124800
-    },
-    {
-      "epoch": 0.5042481930534064,
-      "grad_norm": 888.6316528320312,
-      "learning_rate": 2.9406504012228375e-05,
-      "loss": 38.9805,
-      "step": 124810
-    },
-    {
-      "epoch": 0.504288594318774,
-      "grad_norm": 454.36627197265625,
-      "learning_rate": 2.9403067961869367e-05,
-      "loss": 32.8442,
-      "step": 124820
-    },
-    {
-      "epoch": 0.5043289955841417,
-      "grad_norm": 648.9022827148438,
-      "learning_rate": 2.9399631825670292e-05,
-      "loss": 40.4288,
-      "step": 124830
-    },
-    {
-      "epoch": 0.5043693968495093,
-      "grad_norm": 967.1868286132812,
-      "learning_rate": 2.939619560369813e-05,
-      "loss": 63.7735,
-      "step": 124840
-    },
-    {
-      "epoch": 0.504409798114877,
-      "grad_norm": 1663.5084228515625,
-      "learning_rate": 2.9392759296019867e-05,
-      "loss": 57.5468,
-      "step": 124850
-    },
-    {
-      "epoch": 0.5044501993802446,
-      "grad_norm": 1367.6448974609375,
-      "learning_rate": 2.9389322902702497e-05,
-      "loss": 61.3237,
-      "step": 124860
-    },
-    {
-      "epoch": 0.5044906006456122,
-      "grad_norm": 756.0995483398438,
-      "learning_rate": 2.9385886423813024e-05,
-      "loss": 65.3766,
-      "step": 124870
-    },
-    {
-      "epoch": 0.5045310019109799,
-      "grad_norm": 630.2203369140625,
-      "learning_rate": 2.938244985941844e-05,
-      "loss": 42.1302,
-      "step": 124880
-    },
-    {
-      "epoch": 0.5045714031763475,
-      "grad_norm": 451.697998046875,
-      "learning_rate": 2.9379013209585726e-05,
-      "loss": 47.7857,
-      "step": 124890
-    },
-    {
-      "epoch": 0.5046118044417152,
-      "grad_norm": 890.1043701171875,
-      "learning_rate": 2.9375576474381905e-05,
-      "loss": 51.9966,
-      "step": 124900
-    },
-    {
-      "epoch": 0.5046522057070827,
-      "grad_norm": 901.4338989257812,
-      "learning_rate": 2.9372139653873958e-05,
-      "loss": 65.1977,
-      "step": 124910
-    },
-    {
-      "epoch": 0.5046926069724503,
-      "grad_norm": 1069.64208984375,
-      "learning_rate": 2.9368702748128912e-05,
-      "loss": 55.9412,
-      "step": 124920
-    },
-    {
-      "epoch": 0.504733008237818,
-      "grad_norm": 1394.1937255859375,
-      "learning_rate": 2.9365265757213745e-05,
-      "loss": 66.5137,
-      "step": 124930
-    },
-    {
-      "epoch": 0.5047734095031856,
-      "grad_norm": 260.0003967285156,
-      "learning_rate": 2.9361828681195484e-05,
-      "loss": 46.3195,
-      "step": 124940
-    },
-    {
-      "epoch": 0.5048138107685533,
-      "grad_norm": 521.545654296875,
-      "learning_rate": 2.9358391520141122e-05,
-      "loss": 54.6765,
-      "step": 124950
-    },
-    {
-      "epoch": 0.5048542120339209,
-      "grad_norm": 676.0771484375,
-      "learning_rate": 2.935495427411768e-05,
-      "loss": 27.8798,
-      "step": 124960
-    },
-    {
-      "epoch": 0.5048946132992885,
-      "grad_norm": 554.6319580078125,
-      "learning_rate": 2.9351516943192155e-05,
-      "loss": 39.7234,
-      "step": 124970
-    },
-    {
-      "epoch": 0.5049350145646562,
-      "grad_norm": 1820.2799072265625,
-      "learning_rate": 2.9348079527431567e-05,
-      "loss": 47.2992,
-      "step": 124980
-    },
-    {
-      "epoch": 0.5049754158300238,
-      "grad_norm": 730.9981689453125,
-      "learning_rate": 2.9344642026902924e-05,
-      "loss": 58.4385,
-      "step": 124990
-    },
-    {
-      "epoch": 0.5050158170953915,
-      "grad_norm": 232.58773803710938,
-      "learning_rate": 2.9341204441673266e-05,
-      "loss": 51.1445,
-      "step": 125000
-    },
-    {
-      "epoch": 0.5050562183607591,
-      "grad_norm": 744.6976318359375,
-      "learning_rate": 2.9337766771809577e-05,
-      "loss": 56.7142,
-      "step": 125010
-    },
-    {
-      "epoch": 0.5050966196261267,
-      "grad_norm": 807.66748046875,
-      "learning_rate": 2.9334329017378898e-05,
-      "loss": 66.2732,
-      "step": 125020
-    },
-    {
-      "epoch": 0.5051370208914944,
-      "grad_norm": 1156.4183349609375,
-      "learning_rate": 2.933089117844824e-05,
-      "loss": 53.764,
-      "step": 125030
-    },
-    {
-      "epoch": 0.5051774221568619,
-      "grad_norm": 1409.17626953125,
-      "learning_rate": 2.9327453255084638e-05,
-      "loss": 51.1774,
-      "step": 125040
-    },
-    {
-      "epoch": 0.5052178234222295,
-      "grad_norm": 1206.735107421875,
-      "learning_rate": 2.9324015247355098e-05,
-      "loss": 50.6902,
-      "step": 125050
-    },
-    {
-      "epoch": 0.5052582246875972,
-      "grad_norm": 1590.0701904296875,
-      "learning_rate": 2.932057715532665e-05,
-      "loss": 61.7914,
-      "step": 125060
-    },
-    {
-      "epoch": 0.5052986259529648,
-      "grad_norm": 905.3733520507812,
-      "learning_rate": 2.9317138979066327e-05,
-      "loss": 44.9613,
-      "step": 125070
-    },
-    {
-      "epoch": 0.5053390272183325,
-      "grad_norm": 747.811279296875,
-      "learning_rate": 2.9313700718641167e-05,
-      "loss": 65.6674,
-      "step": 125080
-    },
-    {
-      "epoch": 0.5053794284837001,
-      "grad_norm": 0.0,
-      "learning_rate": 2.9310262374118185e-05,
-      "loss": 45.0609,
-      "step": 125090
-    },
-    {
-      "epoch": 0.5054198297490677,
-      "grad_norm": 528.493408203125,
-      "learning_rate": 2.9306823945564422e-05,
-      "loss": 58.0582,
-      "step": 125100
-    },
-    {
-      "epoch": 0.5054602310144354,
-      "grad_norm": 680.7048950195312,
-      "learning_rate": 2.9303385433046902e-05,
-      "loss": 36.582,
-      "step": 125110
-    },
-    {
-      "epoch": 0.505500632279803,
-      "grad_norm": 767.6009521484375,
-      "learning_rate": 2.9299946836632673e-05,
-      "loss": 40.1557,
-      "step": 125120
-    },
-    {
-      "epoch": 0.5055410335451707,
-      "grad_norm": 1023.16015625,
-      "learning_rate": 2.929650815638877e-05,
-      "loss": 47.9998,
-      "step": 125130
-    },
-    {
-      "epoch": 0.5055814348105383,
-      "grad_norm": 588.2609252929688,
-      "learning_rate": 2.9293069392382224e-05,
-      "loss": 50.6491,
-      "step": 125140
-    },
-    {
-      "epoch": 0.5056218360759059,
-      "grad_norm": 831.1854858398438,
-      "learning_rate": 2.9289630544680075e-05,
-      "loss": 68.4844,
-      "step": 125150
-    },
-    {
-      "epoch": 0.5056622373412736,
-      "grad_norm": 1181.1595458984375,
-      "learning_rate": 2.9286191613349374e-05,
-      "loss": 84.0284,
-      "step": 125160
-    },
-    {
-      "epoch": 0.5057026386066411,
-      "grad_norm": 435.8255920410156,
-      "learning_rate": 2.9282752598457165e-05,
-      "loss": 58.2437,
-      "step": 125170
-    },
-    {
-      "epoch": 0.5057430398720087,
-      "grad_norm": 727.278564453125,
-      "learning_rate": 2.9279313500070483e-05,
-      "loss": 66.3914,
-      "step": 125180
-    },
-    {
-      "epoch": 0.5057834411373764,
-      "grad_norm": 302.980224609375,
-      "learning_rate": 2.927587431825639e-05,
-      "loss": 44.221,
-      "step": 125190
-    },
-    {
-      "epoch": 0.505823842402744,
-      "grad_norm": 758.2953491210938,
-      "learning_rate": 2.9272435053081922e-05,
-      "loss": 65.575,
-      "step": 125200
-    },
-    {
-      "epoch": 0.5058642436681117,
-      "grad_norm": 181.88172912597656,
-      "learning_rate": 2.9268995704614132e-05,
-      "loss": 40.2345,
-      "step": 125210
-    },
-    {
-      "epoch": 0.5059046449334793,
-      "grad_norm": 920.4985961914062,
-      "learning_rate": 2.926555627292007e-05,
-      "loss": 51.6828,
-      "step": 125220
-    },
-    {
-      "epoch": 0.505945046198847,
-      "grad_norm": 1029.88330078125,
-      "learning_rate": 2.9262116758066793e-05,
-      "loss": 64.6292,
-      "step": 125230
-    },
-    {
-      "epoch": 0.5059854474642146,
-      "grad_norm": 733.74951171875,
-      "learning_rate": 2.9258677160121352e-05,
-      "loss": 53.2531,
-      "step": 125240
-    },
-    {
-      "epoch": 0.5060258487295822,
-      "grad_norm": 2031.1961669921875,
-      "learning_rate": 2.9255237479150816e-05,
-      "loss": 62.1065,
-      "step": 125250
-    },
-    {
-      "epoch": 0.5060662499949499,
-      "grad_norm": 900.5405883789062,
-      "learning_rate": 2.925179771522223e-05,
-      "loss": 52.0671,
-      "step": 125260
-    },
-    {
-      "epoch": 0.5061066512603175,
-      "grad_norm": 554.5584106445312,
-      "learning_rate": 2.924835786840266e-05,
-      "loss": 38.7164,
-      "step": 125270
-    },
-    {
-      "epoch": 0.5061470525256851,
-      "grad_norm": 386.526123046875,
-      "learning_rate": 2.9244917938759163e-05,
-      "loss": 54.9804,
-      "step": 125280
-    },
-    {
-      "epoch": 0.5061874537910528,
-      "grad_norm": 616.0713500976562,
-      "learning_rate": 2.9241477926358818e-05,
-      "loss": 56.0183,
-      "step": 125290
-    },
-    {
-      "epoch": 0.5062278550564203,
-      "grad_norm": 831.8729858398438,
-      "learning_rate": 2.923803783126866e-05,
-      "loss": 46.269,
-      "step": 125300
-    },
-    {
-      "epoch": 0.506268256321788,
-      "grad_norm": 506.3004455566406,
-      "learning_rate": 2.923459765355578e-05,
-      "loss": 44.9364,
-      "step": 125310
-    },
-    {
-      "epoch": 0.5063086575871556,
-      "grad_norm": 662.8009033203125,
-      "learning_rate": 2.9231157393287234e-05,
-      "loss": 44.294,
-      "step": 125320
-    },
-    {
-      "epoch": 0.5063490588525232,
-      "grad_norm": 1044.3468017578125,
-      "learning_rate": 2.9227717050530107e-05,
-      "loss": 59.7929,
-      "step": 125330
-    },
-    {
-      "epoch": 0.5063894601178909,
-      "grad_norm": 836.830322265625,
-      "learning_rate": 2.922427662535145e-05,
-      "loss": 50.6092,
-      "step": 125340
-    },
-    {
-      "epoch": 0.5064298613832585,
-      "grad_norm": 848.728759765625,
-      "learning_rate": 2.9220836117818344e-05,
-      "loss": 46.637,
-      "step": 125350
-    },
-    {
-      "epoch": 0.5064702626486262,
-      "grad_norm": 456.876953125,
-      "learning_rate": 2.9217395527997875e-05,
-      "loss": 45.2742,
-      "step": 125360
-    },
-    {
-      "epoch": 0.5065106639139938,
-      "grad_norm": 780.4397583007812,
-      "learning_rate": 2.921395485595711e-05,
-      "loss": 47.9569,
-      "step": 125370
-    },
-    {
-      "epoch": 0.5065510651793614,
-      "grad_norm": 851.5180053710938,
-      "learning_rate": 2.9210514101763113e-05,
-      "loss": 74.7248,
-      "step": 125380
-    },
-    {
-      "epoch": 0.5065914664447291,
-      "grad_norm": 776.1827392578125,
-      "learning_rate": 2.9207073265482982e-05,
-      "loss": 49.9978,
-      "step": 125390
-    },
-    {
-      "epoch": 0.5066318677100967,
-      "grad_norm": 528.63330078125,
-      "learning_rate": 2.920363234718379e-05,
-      "loss": 73.2618,
-      "step": 125400
-    },
-    {
-      "epoch": 0.5066722689754644,
-      "grad_norm": 1279.852294921875,
-      "learning_rate": 2.9200191346932627e-05,
-      "loss": 48.9025,
-      "step": 125410
-    },
-    {
-      "epoch": 0.506712670240832,
-      "grad_norm": 1353.2442626953125,
-      "learning_rate": 2.919675026479656e-05,
-      "loss": 51.6571,
-      "step": 125420
-    },
-    {
-      "epoch": 0.5067530715061995,
-      "grad_norm": 1191.01416015625,
-      "learning_rate": 2.9193309100842693e-05,
-      "loss": 46.1766,
-      "step": 125430
-    },
-    {
-      "epoch": 0.5067934727715672,
-      "grad_norm": 1029.6173095703125,
-      "learning_rate": 2.9189867855138103e-05,
-      "loss": 58.42,
-      "step": 125440
-    },
-    {
-      "epoch": 0.5068338740369348,
-      "grad_norm": 575.3421630859375,
-      "learning_rate": 2.918642652774989e-05,
-      "loss": 49.3698,
-      "step": 125450
-    },
-    {
-      "epoch": 0.5068742753023024,
-      "grad_norm": 710.7742919921875,
-      "learning_rate": 2.9182985118745136e-05,
-      "loss": 46.0317,
-      "step": 125460
-    },
-    {
-      "epoch": 0.5069146765676701,
-      "grad_norm": 794.8711547851562,
-      "learning_rate": 2.9179543628190925e-05,
-      "loss": 61.0076,
-      "step": 125470
-    },
-    {
-      "epoch": 0.5069550778330377,
-      "grad_norm": 621.3440551757812,
-      "learning_rate": 2.9176102056154363e-05,
-      "loss": 58.3854,
-      "step": 125480
-    },
-    {
-      "epoch": 0.5069954790984054,
-      "grad_norm": 617.9365844726562,
-      "learning_rate": 2.9172660402702546e-05,
-      "loss": 47.3791,
-      "step": 125490
-    },
-    {
-      "epoch": 0.507035880363773,
-      "grad_norm": 902.1868286132812,
-      "learning_rate": 2.916921866790256e-05,
-      "loss": 49.7716,
-      "step": 125500
-    },
-    {
-      "epoch": 0.5070762816291406,
-      "grad_norm": 882.5235595703125,
-      "learning_rate": 2.9165776851821508e-05,
-      "loss": 44.4464,
-      "step": 125510
-    },
-    {
-      "epoch": 0.5071166828945083,
-      "grad_norm": 893.1301879882812,
-      "learning_rate": 2.9162334954526493e-05,
-      "loss": 40.7801,
-      "step": 125520
-    },
-    {
-      "epoch": 0.5071570841598759,
-      "grad_norm": 457.0363464355469,
-      "learning_rate": 2.915889297608462e-05,
-      "loss": 43.9219,
-      "step": 125530
-    },
-    {
-      "epoch": 0.5071974854252436,
-      "grad_norm": 504.5679626464844,
-      "learning_rate": 2.9155450916562994e-05,
-      "loss": 33.9357,
-      "step": 125540
-    },
-    {
-      "epoch": 0.5072378866906111,
-      "grad_norm": 866.373046875,
-      "learning_rate": 2.91520087760287e-05,
-      "loss": 72.3561,
-      "step": 125550
-    },
-    {
-      "epoch": 0.5072782879559787,
-      "grad_norm": 599.7020874023438,
-      "learning_rate": 2.9148566554548857e-05,
-      "loss": 53.9173,
-      "step": 125560
-    },
-    {
-      "epoch": 0.5073186892213464,
-      "grad_norm": 567.0419921875,
-      "learning_rate": 2.914512425219058e-05,
-      "loss": 71.651,
-      "step": 125570
-    },
-    {
-      "epoch": 0.507359090486714,
-      "grad_norm": 628.0565795898438,
-      "learning_rate": 2.914168186902097e-05,
-      "loss": 56.5678,
-      "step": 125580
-    },
-    {
-      "epoch": 0.5073994917520817,
-      "grad_norm": 528.4154663085938,
-      "learning_rate": 2.9138239405107136e-05,
-      "loss": 42.4691,
-      "step": 125590
-    },
-    {
-      "epoch": 0.5074398930174493,
-      "grad_norm": 829.2301635742188,
-      "learning_rate": 2.9134796860516194e-05,
-      "loss": 49.0979,
-      "step": 125600
-    },
-    {
-      "epoch": 0.5074802942828169,
-      "grad_norm": 1135.746826171875,
-      "learning_rate": 2.9131354235315268e-05,
-      "loss": 60.1448,
-      "step": 125610
-    },
-    {
-      "epoch": 0.5075206955481846,
-      "grad_norm": 402.7625427246094,
-      "learning_rate": 2.912791152957145e-05,
-      "loss": 69.6794,
-      "step": 125620
-    },
-    {
-      "epoch": 0.5075610968135522,
-      "grad_norm": 654.6475219726562,
-      "learning_rate": 2.9124468743351884e-05,
-      "loss": 58.2971,
-      "step": 125630
-    },
-    {
-      "epoch": 0.5076014980789199,
-      "grad_norm": 552.7247924804688,
-      "learning_rate": 2.9121025876723674e-05,
-      "loss": 66.4654,
-      "step": 125640
-    },
-    {
-      "epoch": 0.5076418993442875,
-      "grad_norm": 887.902099609375,
-      "learning_rate": 2.9117582929753932e-05,
-      "loss": 76.861,
-      "step": 125650
-    },
-    {
-      "epoch": 0.5076823006096551,
-      "grad_norm": 966.7318725585938,
-      "learning_rate": 2.9114139902509807e-05,
-      "loss": 57.0813,
-      "step": 125660
-    },
-    {
-      "epoch": 0.5077227018750228,
-      "grad_norm": 1011.2301635742188,
-      "learning_rate": 2.9110696795058394e-05,
-      "loss": 45.7711,
-      "step": 125670
-    },
-    {
-      "epoch": 0.5077631031403903,
-      "grad_norm": 557.7918090820312,
-      "learning_rate": 2.9107253607466832e-05,
-      "loss": 55.3871,
-      "step": 125680
-    },
-    {
-      "epoch": 0.5078035044057579,
-      "grad_norm": 3118.763916015625,
-      "learning_rate": 2.910381033980225e-05,
-      "loss": 50.6585,
-      "step": 125690
-    },
-    {
-      "epoch": 0.5078439056711256,
-      "grad_norm": 863.99658203125,
-      "learning_rate": 2.910036699213178e-05,
-      "loss": 68.7242,
-      "step": 125700
-    },
-    {
-      "epoch": 0.5078843069364932,
-      "grad_norm": 1471.3134765625,
-      "learning_rate": 2.909692356452254e-05,
-      "loss": 52.9448,
-      "step": 125710
-    },
-    {
-      "epoch": 0.5079247082018609,
-      "grad_norm": 553.6021118164062,
-      "learning_rate": 2.9093480057041662e-05,
-      "loss": 55.443,
-      "step": 125720
-    },
-    {
-      "epoch": 0.5079651094672285,
-      "grad_norm": 329.5192565917969,
-      "learning_rate": 2.9090036469756276e-05,
-      "loss": 67.455,
-      "step": 125730
-    },
-    {
-      "epoch": 0.5080055107325961,
-      "grad_norm": 1218.230712890625,
-      "learning_rate": 2.9086592802733536e-05,
-      "loss": 84.1644,
-      "step": 125740
-    },
-    {
-      "epoch": 0.5080459119979638,
-      "grad_norm": 140.51890563964844,
-      "learning_rate": 2.908314905604056e-05,
-      "loss": 58.1322,
-      "step": 125750
-    },
-    {
-      "epoch": 0.5080863132633314,
-      "grad_norm": 557.7669677734375,
-      "learning_rate": 2.9079705229744493e-05,
-      "loss": 58.6665,
-      "step": 125760
-    },
-    {
-      "epoch": 0.5081267145286991,
-      "grad_norm": 357.09234619140625,
-      "learning_rate": 2.907626132391246e-05,
-      "loss": 33.4991,
-      "step": 125770
-    },
-    {
-      "epoch": 0.5081671157940667,
-      "grad_norm": 404.6373596191406,
-      "learning_rate": 2.9072817338611636e-05,
-      "loss": 89.8573,
-      "step": 125780
-    },
-    {
-      "epoch": 0.5082075170594343,
-      "grad_norm": 606.2252807617188,
-      "learning_rate": 2.9069373273909123e-05,
-      "loss": 61.2002,
-      "step": 125790
-    },
-    {
-      "epoch": 0.508247918324802,
-      "grad_norm": 458.4631652832031,
-      "learning_rate": 2.9065929129872094e-05,
-      "loss": 46.1727,
-      "step": 125800
-    },
-    {
-      "epoch": 0.5082883195901695,
-      "grad_norm": 901.8555908203125,
-      "learning_rate": 2.906248490656768e-05,
-      "loss": 48.2973,
-      "step": 125810
-    },
-    {
-      "epoch": 0.5083287208555372,
-      "grad_norm": 676.2325439453125,
-      "learning_rate": 2.905904060406303e-05,
-      "loss": 57.2065,
-      "step": 125820
-    },
-    {
-      "epoch": 0.5083691221209048,
-      "grad_norm": 1410.12158203125,
-      "learning_rate": 2.905559622242529e-05,
-      "loss": 45.1312,
-      "step": 125830
-    },
-    {
-      "epoch": 0.5084095233862724,
-      "grad_norm": 386.5525207519531,
-      "learning_rate": 2.9052151761721617e-05,
-      "loss": 73.7721,
-      "step": 125840
-    },
-    {
-      "epoch": 0.5084499246516401,
-      "grad_norm": 1108.298583984375,
-      "learning_rate": 2.9048707222019154e-05,
-      "loss": 66.7324,
-      "step": 125850
-    },
-    {
-      "epoch": 0.5084903259170077,
-      "grad_norm": 752.5873413085938,
-      "learning_rate": 2.904526260338507e-05,
-      "loss": 52.5618,
-      "step": 125860
-    },
-    {
-      "epoch": 0.5085307271823754,
-      "grad_norm": 393.9751281738281,
-      "learning_rate": 2.9041817905886504e-05,
-      "loss": 55.0743,
-      "step": 125870
-    },
-    {
-      "epoch": 0.508571128447743,
-      "grad_norm": 509.97735595703125,
-      "learning_rate": 2.9038373129590622e-05,
-      "loss": 51.6777,
-      "step": 125880
-    },
-    {
-      "epoch": 0.5086115297131106,
-      "grad_norm": 1160.9713134765625,
-      "learning_rate": 2.903492827456457e-05,
-      "loss": 59.5145,
-      "step": 125890
-    },
-    {
-      "epoch": 0.5086519309784783,
-      "grad_norm": 510.33514404296875,
-      "learning_rate": 2.903148334087552e-05,
-      "loss": 34.4883,
-      "step": 125900
-    },
-    {
-      "epoch": 0.5086923322438459,
-      "grad_norm": 778.451171875,
-      "learning_rate": 2.9028038328590617e-05,
-      "loss": 57.813,
-      "step": 125910
-    },
-    {
-      "epoch": 0.5087327335092136,
-      "grad_norm": 604.14990234375,
-      "learning_rate": 2.9024593237777037e-05,
-      "loss": 44.8038,
-      "step": 125920
-    },
-    {
-      "epoch": 0.5087731347745812,
-      "grad_norm": 1083.8634033203125,
-      "learning_rate": 2.902114806850194e-05,
-      "loss": 66.7281,
-      "step": 125930
-    },
-    {
-      "epoch": 0.5088135360399487,
-      "grad_norm": 626.0941772460938,
-      "learning_rate": 2.9017702820832498e-05,
-      "loss": 64.0597,
-      "step": 125940
-    },
-    {
-      "epoch": 0.5088539373053164,
-      "grad_norm": 2549.89453125,
-      "learning_rate": 2.9014257494835862e-05,
-      "loss": 83.0733,
-      "step": 125950
-    },
-    {
-      "epoch": 0.508894338570684,
-      "grad_norm": 271.8528137207031,
-      "learning_rate": 2.901081209057921e-05,
-      "loss": 72.4616,
-      "step": 125960
-    },
-    {
-      "epoch": 0.5089347398360516,
-      "grad_norm": 418.9285583496094,
-      "learning_rate": 2.900736660812972e-05,
-      "loss": 32.0242,
-      "step": 125970
-    },
-    {
-      "epoch": 0.5089751411014193,
-      "grad_norm": 748.1174926757812,
-      "learning_rate": 2.900392104755455e-05,
-      "loss": 62.5196,
-      "step": 125980
-    },
-    {
-      "epoch": 0.5090155423667869,
-      "grad_norm": 1057.87060546875,
-      "learning_rate": 2.900047540892088e-05,
-      "loss": 66.0844,
-      "step": 125990
-    },
-    {
-      "epoch": 0.5090559436321546,
-      "grad_norm": 1053.06689453125,
-      "learning_rate": 2.8997029692295874e-05,
-      "loss": 59.8133,
-      "step": 126000
-    },
-    {
-      "epoch": 0.5090963448975222,
-      "grad_norm": 1008.4521484375,
-      "learning_rate": 2.8993583897746717e-05,
-      "loss": 67.719,
-      "step": 126010
-    },
-    {
-      "epoch": 0.5091367461628898,
-      "grad_norm": 400.8494567871094,
-      "learning_rate": 2.8990138025340596e-05,
-      "loss": 55.9787,
-      "step": 126020
-    },
-    {
-      "epoch": 0.5091771474282575,
-      "grad_norm": 776.619140625,
-      "learning_rate": 2.8986692075144673e-05,
-      "loss": 53.733,
-      "step": 126030
-    },
-    {
-      "epoch": 0.5092175486936251,
-      "grad_norm": 344.3357238769531,
-      "learning_rate": 2.8983246047226135e-05,
-      "loss": 49.7655,
-      "step": 126040
-    },
-    {
-      "epoch": 0.5092579499589928,
-      "grad_norm": 461.5896911621094,
-      "learning_rate": 2.897979994165217e-05,
-      "loss": 66.4076,
-      "step": 126050
-    },
-    {
-      "epoch": 0.5092983512243604,
-      "grad_norm": 0.0,
-      "learning_rate": 2.8976353758489955e-05,
-      "loss": 44.4132,
-      "step": 126060
-    },
-    {
-      "epoch": 0.5093387524897279,
-      "grad_norm": 446.7651062011719,
-      "learning_rate": 2.897290749780667e-05,
-      "loss": 48.4824,
-      "step": 126070
-    },
-    {
-      "epoch": 0.5093791537550956,
-      "grad_norm": 1131.740966796875,
-      "learning_rate": 2.8969461159669513e-05,
-      "loss": 45.0098,
-      "step": 126080
-    },
-    {
-      "epoch": 0.5094195550204632,
-      "grad_norm": 1675.897705078125,
-      "learning_rate": 2.8966014744145663e-05,
-      "loss": 93.9438,
-      "step": 126090
-    },
-    {
-      "epoch": 0.5094599562858309,
-      "grad_norm": 665.18994140625,
-      "learning_rate": 2.8962568251302324e-05,
-      "loss": 52.6255,
-      "step": 126100
-    },
-    {
-      "epoch": 0.5095003575511985,
-      "grad_norm": 404.408203125,
-      "learning_rate": 2.895912168120667e-05,
-      "loss": 57.1605,
-      "step": 126110
-    },
-    {
-      "epoch": 0.5095407588165661,
-      "grad_norm": 690.7555541992188,
-      "learning_rate": 2.8955675033925895e-05,
-      "loss": 63.0719,
-      "step": 126120
-    },
-    {
-      "epoch": 0.5095811600819338,
-      "grad_norm": 937.3909912109375,
-      "learning_rate": 2.89522283095272e-05,
-      "loss": 61.4134,
-      "step": 126130
-    },
-    {
-      "epoch": 0.5096215613473014,
-      "grad_norm": 1157.58642578125,
-      "learning_rate": 2.8948781508077786e-05,
-      "loss": 43.2224,
-      "step": 126140
-    },
-    {
-      "epoch": 0.509661962612669,
-      "grad_norm": 766.99609375,
-      "learning_rate": 2.894533462964485e-05,
-      "loss": 45.9795,
-      "step": 126150
-    },
-    {
-      "epoch": 0.5097023638780367,
-      "grad_norm": 1816.978515625,
-      "learning_rate": 2.894188767429557e-05,
-      "loss": 53.0026,
-      "step": 126160
-    },
-    {
-      "epoch": 0.5097427651434043,
-      "grad_norm": 813.2989501953125,
-      "learning_rate": 2.8938440642097164e-05,
-      "loss": 37.4644,
-      "step": 126170
-    },
-    {
-      "epoch": 0.509783166408772,
-      "grad_norm": 691.785888671875,
-      "learning_rate": 2.893499353311683e-05,
-      "loss": 51.4019,
-      "step": 126180
-    },
-    {
-      "epoch": 0.5098235676741395,
-      "grad_norm": 1490.9388427734375,
-      "learning_rate": 2.8931546347421773e-05,
-      "loss": 45.6698,
-      "step": 126190
-    },
-    {
-      "epoch": 0.5098639689395071,
-      "grad_norm": 548.1686401367188,
-      "learning_rate": 2.8928099085079197e-05,
-      "loss": 41.4219,
-      "step": 126200
-    },
-    {
-      "epoch": 0.5099043702048748,
-      "grad_norm": 1110.064697265625,
-      "learning_rate": 2.89246517461563e-05,
-      "loss": 71.6463,
-      "step": 126210
-    },
-    {
-      "epoch": 0.5099447714702424,
-      "grad_norm": 549.9130859375,
-      "learning_rate": 2.892120433072031e-05,
-      "loss": 45.8994,
-      "step": 126220
-    },
-    {
-      "epoch": 0.5099851727356101,
-      "grad_norm": 4628.6025390625,
-      "learning_rate": 2.8917756838838418e-05,
-      "loss": 75.2251,
-      "step": 126230
-    },
-    {
-      "epoch": 0.5100255740009777,
-      "grad_norm": 518.2777099609375,
-      "learning_rate": 2.8914309270577834e-05,
-      "loss": 48.5844,
-      "step": 126240
-    },
-    {
-      "epoch": 0.5100659752663453,
-      "grad_norm": 677.185302734375,
-      "learning_rate": 2.8910861626005776e-05,
-      "loss": 61.1778,
-      "step": 126250
-    },
-    {
-      "epoch": 0.510106376531713,
-      "grad_norm": 1070.945556640625,
-      "learning_rate": 2.8907413905189456e-05,
-      "loss": 43.0625,
-      "step": 126260
-    },
-    {
-      "epoch": 0.5101467777970806,
-      "grad_norm": 1414.0343017578125,
-      "learning_rate": 2.8903966108196096e-05,
-      "loss": 69.5534,
-      "step": 126270
-    },
-    {
-      "epoch": 0.5101871790624483,
-      "grad_norm": 1134.17041015625,
-      "learning_rate": 2.8900518235092905e-05,
-      "loss": 73.4212,
-      "step": 126280
-    },
-    {
-      "epoch": 0.5102275803278159,
-      "grad_norm": 970.3851318359375,
-      "learning_rate": 2.8897070285947098e-05,
-      "loss": 58.0343,
-      "step": 126290
-    },
-    {
-      "epoch": 0.5102679815931835,
-      "grad_norm": 340.81231689453125,
-      "learning_rate": 2.8893622260825904e-05,
-      "loss": 76.9128,
-      "step": 126300
-    },
-    {
-      "epoch": 0.5103083828585512,
-      "grad_norm": 637.6676025390625,
-      "learning_rate": 2.889017415979654e-05,
-      "loss": 68.674,
-      "step": 126310
-    },
-    {
-      "epoch": 0.5103487841239187,
-      "grad_norm": 643.8228759765625,
-      "learning_rate": 2.8886725982926232e-05,
-      "loss": 81.6261,
-      "step": 126320
-    },
-    {
-      "epoch": 0.5103891853892863,
-      "grad_norm": 378.8822021484375,
-      "learning_rate": 2.8883277730282194e-05,
-      "loss": 46.2597,
-      "step": 126330
-    },
-    {
-      "epoch": 0.510429586654654,
-      "grad_norm": 1168.7510986328125,
-      "learning_rate": 2.8879829401931652e-05,
-      "loss": 71.7926,
-      "step": 126340
-    },
-    {
-      "epoch": 0.5104699879200216,
-      "grad_norm": 522.0606689453125,
-      "learning_rate": 2.8876380997941847e-05,
-      "loss": 54.6862,
-      "step": 126350
-    },
-    {
-      "epoch": 0.5105103891853893,
-      "grad_norm": 396.95263671875,
-      "learning_rate": 2.8872932518379997e-05,
-      "loss": 44.9638,
-      "step": 126360
-    },
-    {
-      "epoch": 0.5105507904507569,
-      "grad_norm": 794.6138305664062,
-      "learning_rate": 2.886948396331333e-05,
-      "loss": 76.7646,
-      "step": 126370
-    },
-    {
-      "epoch": 0.5105911917161245,
-      "grad_norm": 1046.785888671875,
-      "learning_rate": 2.8866035332809084e-05,
-      "loss": 53.4936,
-      "step": 126380
-    },
-    {
-      "epoch": 0.5106315929814922,
-      "grad_norm": 869.3177490234375,
-      "learning_rate": 2.886258662693449e-05,
-      "loss": 43.9115,
-      "step": 126390
-    },
-    {
-      "epoch": 0.5106719942468598,
-      "grad_norm": 724.6814575195312,
-      "learning_rate": 2.8859137845756784e-05,
-      "loss": 45.5164,
-      "step": 126400
-    },
-    {
-      "epoch": 0.5107123955122275,
-      "grad_norm": 411.1709289550781,
-      "learning_rate": 2.8855688989343193e-05,
-      "loss": 57.5156,
-      "step": 126410
-    },
-    {
-      "epoch": 0.5107527967775951,
-      "grad_norm": 326.8053283691406,
-      "learning_rate": 2.885224005776096e-05,
-      "loss": 51.5223,
-      "step": 126420
-    },
-    {
-      "epoch": 0.5107931980429627,
-      "grad_norm": 843.7996215820312,
-      "learning_rate": 2.884879105107733e-05,
-      "loss": 44.6353,
-      "step": 126430
-    },
-    {
-      "epoch": 0.5108335993083304,
-      "grad_norm": 1049.7962646484375,
-      "learning_rate": 2.884534196935953e-05,
-      "loss": 58.139,
-      "step": 126440
-    },
-    {
-      "epoch": 0.5108740005736979,
-      "grad_norm": 2668.488037109375,
-      "learning_rate": 2.8841892812674808e-05,
-      "loss": 37.7585,
-      "step": 126450
-    },
-    {
-      "epoch": 0.5109144018390656,
-      "grad_norm": 1897.6781005859375,
-      "learning_rate": 2.8838443581090412e-05,
-      "loss": 50.773,
-      "step": 126460
-    },
-    {
-      "epoch": 0.5109548031044332,
-      "grad_norm": 556.2504272460938,
-      "learning_rate": 2.8834994274673582e-05,
-      "loss": 56.0956,
-      "step": 126470
-    },
-    {
-      "epoch": 0.5109952043698008,
-      "grad_norm": 292.6050109863281,
-      "learning_rate": 2.8831544893491563e-05,
-      "loss": 49.7071,
-      "step": 126480
-    },
-    {
-      "epoch": 0.5110356056351685,
-      "grad_norm": 442.0137939453125,
-      "learning_rate": 2.882809543761161e-05,
-      "loss": 51.5451,
-      "step": 126490
-    },
-    {
-      "epoch": 0.5110760069005361,
-      "grad_norm": 810.0142211914062,
-      "learning_rate": 2.8824645907100954e-05,
-      "loss": 55.3643,
-      "step": 126500
-    },
-    {
-      "epoch": 0.5111164081659038,
-      "grad_norm": 1782.1546630859375,
-      "learning_rate": 2.8821196302026863e-05,
-      "loss": 68.3655,
-      "step": 126510
-    },
-    {
-      "epoch": 0.5111568094312714,
-      "grad_norm": 900.7822875976562,
-      "learning_rate": 2.881774662245658e-05,
-      "loss": 48.6357,
-      "step": 126520
-    },
-    {
-      "epoch": 0.511197210696639,
-      "grad_norm": 648.9537963867188,
-      "learning_rate": 2.8814296868457364e-05,
-      "loss": 51.299,
-      "step": 126530
-    },
-    {
-      "epoch": 0.5112376119620067,
-      "grad_norm": 949.2380981445312,
-      "learning_rate": 2.8810847040096467e-05,
-      "loss": 66.241,
-      "step": 126540
-    },
-    {
-      "epoch": 0.5112780132273743,
-      "grad_norm": 227.76019287109375,
-      "learning_rate": 2.8807397137441145e-05,
-      "loss": 32.5676,
-      "step": 126550
-    },
-    {
-      "epoch": 0.511318414492742,
-      "grad_norm": 545.3878173828125,
-      "learning_rate": 2.8803947160558652e-05,
-      "loss": 77.758,
-      "step": 126560
-    },
-    {
-      "epoch": 0.5113588157581096,
-      "grad_norm": 621.6892700195312,
-      "learning_rate": 2.8800497109516263e-05,
-      "loss": 58.1789,
-      "step": 126570
-    },
-    {
-      "epoch": 0.5113992170234771,
-      "grad_norm": 1468.47509765625,
-      "learning_rate": 2.8797046984381208e-05,
-      "loss": 72.7109,
-      "step": 126580
-    },
-    {
-      "epoch": 0.5114396182888448,
-      "grad_norm": 406.4592590332031,
-      "learning_rate": 2.8793596785220783e-05,
-      "loss": 64.9677,
-      "step": 126590
-    },
-    {
-      "epoch": 0.5114800195542124,
-      "grad_norm": 290.8427429199219,
-      "learning_rate": 2.879014651210223e-05,
-      "loss": 47.5232,
-      "step": 126600
-    },
-    {
-      "epoch": 0.51152042081958,
-      "grad_norm": 1523.126220703125,
-      "learning_rate": 2.8786696165092812e-05,
-      "loss": 42.1534,
-      "step": 126610
-    },
-    {
-      "epoch": 0.5115608220849477,
-      "grad_norm": 1009.9312133789062,
-      "learning_rate": 2.8783245744259806e-05,
-      "loss": 61.5785,
-      "step": 126620
-    },
-    {
-      "epoch": 0.5116012233503153,
-      "grad_norm": 527.927490234375,
-      "learning_rate": 2.877979524967048e-05,
-      "loss": 61.9816,
-      "step": 126630
-    },
-    {
-      "epoch": 0.511641624615683,
-      "grad_norm": 575.1068725585938,
-      "learning_rate": 2.8776344681392105e-05,
-      "loss": 60.9859,
-      "step": 126640
-    },
-    {
-      "epoch": 0.5116820258810506,
-      "grad_norm": 338.4177551269531,
-      "learning_rate": 2.877289403949194e-05,
-      "loss": 54.107,
-      "step": 126650
-    },
-    {
-      "epoch": 0.5117224271464182,
-      "grad_norm": 354.56005859375,
-      "learning_rate": 2.876944332403726e-05,
-      "loss": 54.2842,
-      "step": 126660
-    },
-    {
-      "epoch": 0.5117628284117859,
-      "grad_norm": 989.342041015625,
-      "learning_rate": 2.8765992535095345e-05,
-      "loss": 100.1021,
-      "step": 126670
-    },
-    {
-      "epoch": 0.5118032296771535,
-      "grad_norm": 547.06982421875,
-      "learning_rate": 2.8762541672733472e-05,
-      "loss": 48.5291,
-      "step": 126680
-    },
-    {
-      "epoch": 0.5118436309425212,
-      "grad_norm": 1068.9014892578125,
-      "learning_rate": 2.8759090737018902e-05,
-      "loss": 58.7899,
-      "step": 126690
-    },
-    {
-      "epoch": 0.5118840322078888,
-      "grad_norm": 1199.883544921875,
-      "learning_rate": 2.875563972801893e-05,
-      "loss": 43.4314,
-      "step": 126700
-    },
-    {
-      "epoch": 0.5119244334732563,
-      "grad_norm": 616.8649291992188,
-      "learning_rate": 2.8752188645800822e-05,
-      "loss": 46.0097,
-      "step": 126710
-    },
-    {
-      "epoch": 0.511964834738624,
-      "grad_norm": 2177.439697265625,
-      "learning_rate": 2.874873749043187e-05,
-      "loss": 33.2443,
-      "step": 126720
-    },
-    {
-      "epoch": 0.5120052360039916,
-      "grad_norm": 3227.6328125,
-      "learning_rate": 2.8745286261979348e-05,
-      "loss": 64.0699,
-      "step": 126730
-    },
-    {
-      "epoch": 0.5120456372693593,
-      "grad_norm": 754.943115234375,
-      "learning_rate": 2.874183496051055e-05,
-      "loss": 40.789,
-      "step": 126740
-    },
-    {
-      "epoch": 0.5120860385347269,
-      "grad_norm": 560.1809692382812,
-      "learning_rate": 2.8738383586092745e-05,
-      "loss": 42.9098,
-      "step": 126750
-    },
-    {
-      "epoch": 0.5121264398000945,
-      "grad_norm": 1200.2210693359375,
-      "learning_rate": 2.8734932138793225e-05,
-      "loss": 58.1281,
-      "step": 126760
-    },
-    {
-      "epoch": 0.5121668410654622,
-      "grad_norm": 569.3571166992188,
-      "learning_rate": 2.8731480618679285e-05,
-      "loss": 71.4217,
-      "step": 126770
-    },
-    {
-      "epoch": 0.5122072423308298,
-      "grad_norm": 1209.4700927734375,
-      "learning_rate": 2.8728029025818204e-05,
-      "loss": 51.1106,
-      "step": 126780
-    },
-    {
-      "epoch": 0.5122476435961975,
-      "grad_norm": 541.9495849609375,
-      "learning_rate": 2.872457736027728e-05,
-      "loss": 42.31,
-      "step": 126790
-    },
-    {
-      "epoch": 0.5122880448615651,
-      "grad_norm": 1008.5130004882812,
-      "learning_rate": 2.8721125622123806e-05,
-      "loss": 101.0686,
-      "step": 126800
-    },
-    {
-      "epoch": 0.5123284461269327,
-      "grad_norm": 794.3250732421875,
-      "learning_rate": 2.8717673811425072e-05,
-      "loss": 56.3601,
-      "step": 126810
-    },
-    {
-      "epoch": 0.5123688473923004,
-      "grad_norm": 796.3021240234375,
-      "learning_rate": 2.8714221928248368e-05,
-      "loss": 50.3433,
-      "step": 126820
-    },
-    {
-      "epoch": 0.5124092486576679,
-      "grad_norm": 536.2626953125,
-      "learning_rate": 2.8710769972661e-05,
-      "loss": 53.0116,
-      "step": 126830
-    },
-    {
-      "epoch": 0.5124496499230355,
-      "grad_norm": 816.8209228515625,
-      "learning_rate": 2.8707317944730268e-05,
-      "loss": 42.8337,
-      "step": 126840
-    },
-    {
-      "epoch": 0.5124900511884032,
-      "grad_norm": 1031.8271484375,
-      "learning_rate": 2.8703865844523452e-05,
-      "loss": 47.6963,
-      "step": 126850
-    },
-    {
-      "epoch": 0.5125304524537708,
-      "grad_norm": 1931.7418212890625,
-      "learning_rate": 2.8700413672107866e-05,
-      "loss": 52.6271,
-      "step": 126860
-    },
-    {
-      "epoch": 0.5125708537191385,
-      "grad_norm": 747.9002075195312,
-      "learning_rate": 2.869696142755081e-05,
-      "loss": 57.4997,
-      "step": 126870
-    },
-    {
-      "epoch": 0.5126112549845061,
-      "grad_norm": 460.6169128417969,
-      "learning_rate": 2.8693509110919598e-05,
-      "loss": 49.3172,
-      "step": 126880
-    },
-    {
-      "epoch": 0.5126516562498737,
-      "grad_norm": 395.9313659667969,
-      "learning_rate": 2.8690056722281513e-05,
-      "loss": 50.4048,
-      "step": 126890
-    },
-    {
-      "epoch": 0.5126920575152414,
-      "grad_norm": 673.880859375,
-      "learning_rate": 2.8686604261703875e-05,
-      "loss": 44.8503,
-      "step": 126900
-    },
-    {
-      "epoch": 0.512732458780609,
-      "grad_norm": 745.4364624023438,
-      "learning_rate": 2.8683151729253994e-05,
-      "loss": 50.778,
-      "step": 126910
-    },
-    {
-      "epoch": 0.5127728600459767,
-      "grad_norm": 887.1217041015625,
-      "learning_rate": 2.8679699124999166e-05,
-      "loss": 44.0796,
-      "step": 126920
-    },
-    {
-      "epoch": 0.5128132613113443,
-      "grad_norm": 565.948486328125,
-      "learning_rate": 2.8676246449006715e-05,
-      "loss": 79.4067,
-      "step": 126930
-    },
-    {
-      "epoch": 0.5128536625767119,
-      "grad_norm": 602.2057495117188,
-      "learning_rate": 2.8672793701343946e-05,
-      "loss": 53.1035,
-      "step": 126940
-    },
-    {
-      "epoch": 0.5128940638420796,
-      "grad_norm": 670.5474243164062,
-      "learning_rate": 2.8669340882078166e-05,
-      "loss": 60.5767,
-      "step": 126950
-    },
-    {
-      "epoch": 0.5129344651074471,
-      "grad_norm": 447.6339111328125,
-      "learning_rate": 2.866588799127671e-05,
-      "loss": 67.3426,
-      "step": 126960
-    },
-    {
-      "epoch": 0.5129748663728148,
-      "grad_norm": 728.9623413085938,
-      "learning_rate": 2.8662435029006868e-05,
-      "loss": 72.0813,
-      "step": 126970
-    },
-    {
-      "epoch": 0.5130152676381824,
-      "grad_norm": 600.4077758789062,
-      "learning_rate": 2.865898199533597e-05,
-      "loss": 54.8805,
-      "step": 126980
-    },
-    {
-      "epoch": 0.51305566890355,
-      "grad_norm": 580.8352661132812,
-      "learning_rate": 2.865552889033134e-05,
-      "loss": 55.582,
-      "step": 126990
-    },
-    {
-      "epoch": 0.5130960701689177,
-      "grad_norm": 1138.0733642578125,
-      "learning_rate": 2.8652075714060295e-05,
-      "loss": 64.3382,
-      "step": 127000
-    },
-    {
-      "epoch": 0.5131364714342853,
-      "grad_norm": 281.50640869140625,
-      "learning_rate": 2.864862246659015e-05,
-      "loss": 45.8135,
-      "step": 127010
-    },
-    {
-      "epoch": 0.513176872699653,
-      "grad_norm": 1133.2386474609375,
-      "learning_rate": 2.8645169147988226e-05,
-      "loss": 56.7795,
-      "step": 127020
-    },
-    {
-      "epoch": 0.5132172739650206,
-      "grad_norm": 846.5953979492188,
-      "learning_rate": 2.8641715758321857e-05,
-      "loss": 31.9575,
-      "step": 127030
-    },
-    {
-      "epoch": 0.5132576752303882,
-      "grad_norm": 3238.64013671875,
-      "learning_rate": 2.8638262297658368e-05,
-      "loss": 78.6456,
-      "step": 127040
-    },
-    {
-      "epoch": 0.5132980764957559,
-      "grad_norm": 1550.1759033203125,
-      "learning_rate": 2.863480876606508e-05,
-      "loss": 56.8611,
-      "step": 127050
-    },
-    {
-      "epoch": 0.5133384777611235,
-      "grad_norm": 442.9187927246094,
-      "learning_rate": 2.863135516360932e-05,
-      "loss": 41.2573,
-      "step": 127060
-    },
-    {
-      "epoch": 0.5133788790264912,
-      "grad_norm": 1137.1116943359375,
-      "learning_rate": 2.8627901490358422e-05,
-      "loss": 46.5267,
-      "step": 127070
-    },
-    {
-      "epoch": 0.5134192802918588,
-      "grad_norm": 1043.6861572265625,
-      "learning_rate": 2.8624447746379722e-05,
-      "loss": 60.2835,
-      "step": 127080
-    },
-    {
-      "epoch": 0.5134596815572263,
-      "grad_norm": 739.7015380859375,
-      "learning_rate": 2.862099393174055e-05,
-      "loss": 48.624,
-      "step": 127090
-    },
-    {
-      "epoch": 0.513500082822594,
-      "grad_norm": 945.224853515625,
-      "learning_rate": 2.861754004650823e-05,
-      "loss": 63.0741,
-      "step": 127100
-    },
-    {
-      "epoch": 0.5135404840879616,
-      "grad_norm": 425.7527160644531,
-      "learning_rate": 2.8614086090750103e-05,
-      "loss": 46.5734,
-      "step": 127110
-    },
-    {
-      "epoch": 0.5135808853533292,
-      "grad_norm": 1409.0302734375,
-      "learning_rate": 2.8610632064533517e-05,
-      "loss": 47.2388,
-      "step": 127120
-    },
-    {
-      "epoch": 0.5136212866186969,
-      "grad_norm": 581.88916015625,
-      "learning_rate": 2.8607177967925792e-05,
-      "loss": 69.2831,
-      "step": 127130
-    },
-    {
-      "epoch": 0.5136616878840645,
-      "grad_norm": 478.259765625,
-      "learning_rate": 2.8603723800994275e-05,
-      "loss": 60.1264,
-      "step": 127140
-    },
-    {
-      "epoch": 0.5137020891494322,
-      "grad_norm": 2480.73095703125,
-      "learning_rate": 2.8600269563806302e-05,
-      "loss": 68.6383,
-      "step": 127150
-    },
-    {
-      "epoch": 0.5137424904147998,
-      "grad_norm": 673.7091064453125,
-      "learning_rate": 2.859681525642923e-05,
-      "loss": 65.6581,
-      "step": 127160
-    },
-    {
-      "epoch": 0.5137828916801674,
-      "grad_norm": 675.0384521484375,
-      "learning_rate": 2.8593360878930392e-05,
-      "loss": 57.0633,
-      "step": 127170
-    },
-    {
-      "epoch": 0.5138232929455351,
-      "grad_norm": 613.3917236328125,
-      "learning_rate": 2.8589906431377134e-05,
-      "loss": 50.5051,
-      "step": 127180
-    },
-    {
-      "epoch": 0.5138636942109027,
-      "grad_norm": 358.6791687011719,
-      "learning_rate": 2.8586451913836797e-05,
-      "loss": 48.4163,
-      "step": 127190
-    },
-    {
-      "epoch": 0.5139040954762704,
-      "grad_norm": 1500.3143310546875,
-      "learning_rate": 2.858299732637674e-05,
-      "loss": 69.6514,
-      "step": 127200
-    },
-    {
-      "epoch": 0.513944496741638,
-      "grad_norm": 303.2386474609375,
-      "learning_rate": 2.8579542669064296e-05,
-      "loss": 46.2417,
-      "step": 127210
-    },
-    {
-      "epoch": 0.5139848980070055,
-      "grad_norm": 936.28662109375,
-      "learning_rate": 2.8576087941966835e-05,
-      "loss": 59.2663,
-      "step": 127220
-    },
-    {
-      "epoch": 0.5140252992723732,
-      "grad_norm": 550.0284423828125,
-      "learning_rate": 2.857263314515169e-05,
-      "loss": 46.5445,
-      "step": 127230
-    },
-    {
-      "epoch": 0.5140657005377408,
-      "grad_norm": 1154.9842529296875,
-      "learning_rate": 2.856917827868622e-05,
-      "loss": 48.4179,
-      "step": 127240
-    },
-    {
-      "epoch": 0.5141061018031085,
-      "grad_norm": 400.9862365722656,
-      "learning_rate": 2.8565723342637796e-05,
-      "loss": 84.6768,
-      "step": 127250
-    },
-    {
-      "epoch": 0.5141465030684761,
-      "grad_norm": 1014.5621948242188,
-      "learning_rate": 2.856226833707375e-05,
-      "loss": 60.3505,
-      "step": 127260
-    },
-    {
-      "epoch": 0.5141869043338437,
-      "grad_norm": 351.1904602050781,
-      "learning_rate": 2.855881326206145e-05,
-      "loss": 59.1501,
-      "step": 127270
-    },
-    {
-      "epoch": 0.5142273055992114,
-      "grad_norm": 520.452392578125,
-      "learning_rate": 2.855535811766825e-05,
-      "loss": 60.3091,
-      "step": 127280
-    },
-    {
-      "epoch": 0.514267706864579,
-      "grad_norm": 717.9920043945312,
-      "learning_rate": 2.8551902903961526e-05,
-      "loss": 40.0301,
-      "step": 127290
-    },
-    {
-      "epoch": 0.5143081081299467,
-      "grad_norm": 560.2972412109375,
-      "learning_rate": 2.854844762100861e-05,
-      "loss": 50.7043,
-      "step": 127300
-    },
-    {
-      "epoch": 0.5143485093953143,
-      "grad_norm": 1860.7342529296875,
-      "learning_rate": 2.854499226887689e-05,
-      "loss": 63.0763,
-      "step": 127310
-    },
-    {
-      "epoch": 0.5143889106606819,
-      "grad_norm": 135.43519592285156,
-      "learning_rate": 2.8541536847633717e-05,
-      "loss": 45.6506,
-      "step": 127320
-    },
-    {
-      "epoch": 0.5144293119260496,
-      "grad_norm": 766.4346313476562,
-      "learning_rate": 2.8538081357346465e-05,
-      "loss": 72.0964,
-      "step": 127330
-    },
-    {
-      "epoch": 0.5144697131914172,
-      "grad_norm": 778.751220703125,
-      "learning_rate": 2.8534625798082488e-05,
-      "loss": 65.8823,
-      "step": 127340
-    },
-    {
-      "epoch": 0.5145101144567847,
-      "grad_norm": 490.54461669921875,
-      "learning_rate": 2.853117016990917e-05,
-      "loss": 41.0561,
-      "step": 127350
-    },
-    {
-      "epoch": 0.5145505157221524,
-      "grad_norm": 450.0224914550781,
-      "learning_rate": 2.8527714472893862e-05,
-      "loss": 61.406,
-      "step": 127360
-    },
-    {
-      "epoch": 0.51459091698752,
-      "grad_norm": 298.4858093261719,
-      "learning_rate": 2.8524258707103957e-05,
-      "loss": 88.5857,
-      "step": 127370
-    },
-    {
-      "epoch": 0.5146313182528877,
-      "grad_norm": 1860.1826171875,
-      "learning_rate": 2.85208028726068e-05,
-      "loss": 108.9013,
-      "step": 127380
-    },
-    {
-      "epoch": 0.5146717195182553,
-      "grad_norm": 948.806396484375,
-      "learning_rate": 2.8517346969469782e-05,
-      "loss": 47.6132,
-      "step": 127390
-    },
-    {
-      "epoch": 0.5147121207836229,
-      "grad_norm": 1325.517578125,
-      "learning_rate": 2.8513890997760272e-05,
-      "loss": 45.2896,
-      "step": 127400
-    },
-    {
-      "epoch": 0.5147525220489906,
-      "grad_norm": 605.529052734375,
-      "learning_rate": 2.851043495754566e-05,
-      "loss": 43.9705,
-      "step": 127410
-    },
-    {
-      "epoch": 0.5147929233143582,
-      "grad_norm": 426.92193603515625,
-      "learning_rate": 2.8506978848893302e-05,
-      "loss": 39.5349,
-      "step": 127420
-    },
-    {
-      "epoch": 0.5148333245797259,
-      "grad_norm": 1157.5235595703125,
-      "learning_rate": 2.8503522671870585e-05,
-      "loss": 56.4158,
-      "step": 127430
-    },
-    {
-      "epoch": 0.5148737258450935,
-      "grad_norm": 782.9813232421875,
-      "learning_rate": 2.8500066426544896e-05,
-      "loss": 85.5858,
-      "step": 127440
-    },
-    {
-      "epoch": 0.5149141271104611,
-      "grad_norm": 1053.0633544921875,
-      "learning_rate": 2.849661011298361e-05,
-      "loss": 64.2638,
-      "step": 127450
-    },
-    {
-      "epoch": 0.5149545283758288,
-      "grad_norm": 652.576171875,
-      "learning_rate": 2.8493153731254102e-05,
-      "loss": 43.2801,
-      "step": 127460
-    },
-    {
-      "epoch": 0.5149949296411963,
-      "grad_norm": 1399.2457275390625,
-      "learning_rate": 2.8489697281423767e-05,
-      "loss": 57.9567,
-      "step": 127470
-    },
-    {
-      "epoch": 0.515035330906564,
-      "grad_norm": 1636.6573486328125,
-      "learning_rate": 2.8486240763559986e-05,
-      "loss": 87.6768,
-      "step": 127480
-    },
-    {
-      "epoch": 0.5150757321719316,
-      "grad_norm": 0.0,
-      "learning_rate": 2.848278417773015e-05,
-      "loss": 68.869,
-      "step": 127490
-    },
-    {
-      "epoch": 0.5151161334372992,
-      "grad_norm": 499.08416748046875,
-      "learning_rate": 2.8479327524001636e-05,
-      "loss": 44.9667,
-      "step": 127500
-    },
-    {
-      "epoch": 0.5151565347026669,
-      "grad_norm": 905.2075805664062,
-      "learning_rate": 2.8475870802441844e-05,
-      "loss": 51.4555,
-      "step": 127510
-    },
-    {
-      "epoch": 0.5151969359680345,
-      "grad_norm": 567.349365234375,
-      "learning_rate": 2.847241401311817e-05,
-      "loss": 37.061,
-      "step": 127520
-    },
-    {
-      "epoch": 0.5152373372334021,
-      "grad_norm": 499.4542541503906,
-      "learning_rate": 2.846895715609799e-05,
-      "loss": 46.9687,
-      "step": 127530
-    },
-    {
-      "epoch": 0.5152777384987698,
-      "grad_norm": 1374.189208984375,
-      "learning_rate": 2.8465500231448704e-05,
-      "loss": 74.0274,
-      "step": 127540
-    },
-    {
-      "epoch": 0.5153181397641374,
-      "grad_norm": 1301.90673828125,
-      "learning_rate": 2.8462043239237707e-05,
-      "loss": 51.4661,
-      "step": 127550
-    },
-    {
-      "epoch": 0.5153585410295051,
-      "grad_norm": 290.5340576171875,
-      "learning_rate": 2.845858617953239e-05,
-      "loss": 49.4092,
-      "step": 127560
-    },
-    {
-      "epoch": 0.5153989422948727,
-      "grad_norm": 617.203857421875,
-      "learning_rate": 2.8455129052400166e-05,
-      "loss": 48.6922,
-      "step": 127570
-    },
-    {
-      "epoch": 0.5154393435602403,
-      "grad_norm": 877.1514282226562,
-      "learning_rate": 2.8451671857908415e-05,
-      "loss": 57.4519,
-      "step": 127580
-    },
-    {
-      "epoch": 0.515479744825608,
-      "grad_norm": 1293.734375,
-      "learning_rate": 2.844821459612454e-05,
-      "loss": 97.9196,
-      "step": 127590
-    },
-    {
-      "epoch": 0.5155201460909755,
-      "grad_norm": 628.9718017578125,
-      "learning_rate": 2.844475726711595e-05,
-      "loss": 44.7918,
-      "step": 127600
-    },
-    {
-      "epoch": 0.5155605473563432,
-      "grad_norm": 645.7540283203125,
-      "learning_rate": 2.844129987095005e-05,
-      "loss": 42.1978,
-      "step": 127610
-    },
-    {
-      "epoch": 0.5156009486217108,
-      "grad_norm": 470.0179138183594,
-      "learning_rate": 2.8437842407694236e-05,
-      "loss": 59.6109,
-      "step": 127620
-    },
-    {
-      "epoch": 0.5156413498870784,
-      "grad_norm": 548.3775024414062,
-      "learning_rate": 2.843438487741591e-05,
-      "loss": 53.7288,
-      "step": 127630
-    },
-    {
-      "epoch": 0.5156817511524461,
-      "grad_norm": 294.67889404296875,
-      "learning_rate": 2.843092728018248e-05,
-      "loss": 49.3608,
-      "step": 127640
-    },
-    {
-      "epoch": 0.5157221524178137,
-      "grad_norm": 2212.821533203125,
-      "learning_rate": 2.8427469616061364e-05,
-      "loss": 59.0455,
-      "step": 127650
-    },
-    {
-      "epoch": 0.5157625536831814,
-      "grad_norm": 849.8643798828125,
-      "learning_rate": 2.8424011885119954e-05,
-      "loss": 44.9502,
-      "step": 127660
-    },
-    {
-      "epoch": 0.515802954948549,
-      "grad_norm": 565.5281372070312,
-      "learning_rate": 2.842055408742567e-05,
-      "loss": 53.8755,
-      "step": 127670
-    },
-    {
-      "epoch": 0.5158433562139166,
-      "grad_norm": 874.5379638671875,
-      "learning_rate": 2.8417096223045925e-05,
-      "loss": 53.4064,
-      "step": 127680
-    },
-    {
-      "epoch": 0.5158837574792843,
-      "grad_norm": 761.3839111328125,
-      "learning_rate": 2.841363829204814e-05,
-      "loss": 40.8466,
-      "step": 127690
-    },
-    {
-      "epoch": 0.5159241587446519,
-      "grad_norm": 409.4591369628906,
-      "learning_rate": 2.841018029449971e-05,
-      "loss": 40.9471,
-      "step": 127700
-    },
-    {
-      "epoch": 0.5159645600100196,
-      "grad_norm": 896.9284057617188,
-      "learning_rate": 2.8406722230468063e-05,
-      "loss": 69.344,
-      "step": 127710
-    },
-    {
-      "epoch": 0.5160049612753872,
-      "grad_norm": 248.2589569091797,
-      "learning_rate": 2.840326410002061e-05,
-      "loss": 53.1701,
-      "step": 127720
-    },
-    {
-      "epoch": 0.5160453625407547,
-      "grad_norm": 770.6051025390625,
-      "learning_rate": 2.839980590322477e-05,
-      "loss": 37.5784,
-      "step": 127730
-    },
-    {
-      "epoch": 0.5160857638061224,
-      "grad_norm": 1029.7786865234375,
-      "learning_rate": 2.8396347640147962e-05,
-      "loss": 57.3108,
-      "step": 127740
-    },
-    {
-      "epoch": 0.51612616507149,
-      "grad_norm": 450.9689636230469,
-      "learning_rate": 2.8392889310857612e-05,
-      "loss": 47.8311,
-      "step": 127750
-    },
-    {
-      "epoch": 0.5161665663368576,
-      "grad_norm": 471.0567932128906,
-      "learning_rate": 2.8389430915421132e-05,
-      "loss": 58.49,
-      "step": 127760
-    },
-    {
-      "epoch": 0.5162069676022253,
-      "grad_norm": 991.8955688476562,
-      "learning_rate": 2.8385972453905958e-05,
-      "loss": 43.4331,
-      "step": 127770
-    },
-    {
-      "epoch": 0.5162473688675929,
-      "grad_norm": 0.0,
-      "learning_rate": 2.8382513926379504e-05,
-      "loss": 66.81,
-      "step": 127780
-    },
-    {
-      "epoch": 0.5162877701329606,
-      "grad_norm": 1275.402587890625,
-      "learning_rate": 2.837905533290921e-05,
-      "loss": 33.7486,
-      "step": 127790
-    },
-    {
-      "epoch": 0.5163281713983282,
-      "grad_norm": 921.3496704101562,
-      "learning_rate": 2.8375596673562482e-05,
-      "loss": 66.0726,
-      "step": 127800
-    },
-    {
-      "epoch": 0.5163685726636958,
-      "grad_norm": 1107.1654052734375,
-      "learning_rate": 2.8372137948406762e-05,
-      "loss": 64.4679,
-      "step": 127810
-    },
-    {
-      "epoch": 0.5164089739290635,
-      "grad_norm": 978.436279296875,
-      "learning_rate": 2.8368679157509477e-05,
-      "loss": 55.1411,
-      "step": 127820
-    },
-    {
-      "epoch": 0.5164493751944311,
-      "grad_norm": 794.9774169921875,
-      "learning_rate": 2.8365220300938055e-05,
-      "loss": 44.3256,
-      "step": 127830
-    },
-    {
-      "epoch": 0.5164897764597988,
-      "grad_norm": 2009.2308349609375,
-      "learning_rate": 2.8361761378759934e-05,
-      "loss": 52.7043,
-      "step": 127840
-    },
-    {
-      "epoch": 0.5165301777251664,
-      "grad_norm": 989.152099609375,
-      "learning_rate": 2.8358302391042536e-05,
-      "loss": 37.631,
-      "step": 127850
-    },
-    {
-      "epoch": 0.5165705789905339,
-      "grad_norm": 356.483642578125,
-      "learning_rate": 2.8354843337853314e-05,
-      "loss": 77.6594,
-      "step": 127860
-    },
-    {
-      "epoch": 0.5166109802559016,
-      "grad_norm": 1363.230224609375,
-      "learning_rate": 2.835138421925969e-05,
-      "loss": 79.6859,
-      "step": 127870
-    },
-    {
-      "epoch": 0.5166513815212692,
-      "grad_norm": 811.1392211914062,
-      "learning_rate": 2.834792503532911e-05,
-      "loss": 75.7948,
-      "step": 127880
-    },
-    {
-      "epoch": 0.5166917827866369,
-      "grad_norm": 1689.7584228515625,
-      "learning_rate": 2.8344465786129e-05,
-      "loss": 79.6354,
-      "step": 127890
-    },
-    {
-      "epoch": 0.5167321840520045,
-      "grad_norm": 251.05508422851562,
-      "learning_rate": 2.8341006471726816e-05,
-      "loss": 38.7622,
-      "step": 127900
-    },
-    {
-      "epoch": 0.5167725853173721,
-      "grad_norm": 1307.02880859375,
-      "learning_rate": 2.833754709218998e-05,
-      "loss": 75.8891,
-      "step": 127910
-    },
-    {
-      "epoch": 0.5168129865827398,
-      "grad_norm": 743.4976196289062,
-      "learning_rate": 2.833408764758595e-05,
-      "loss": 37.3798,
-      "step": 127920
-    },
-    {
-      "epoch": 0.5168533878481074,
-      "grad_norm": 395.01214599609375,
-      "learning_rate": 2.833062813798216e-05,
-      "loss": 53.7051,
-      "step": 127930
-    },
-    {
-      "epoch": 0.516893789113475,
-      "grad_norm": 1006.5745849609375,
-      "learning_rate": 2.832716856344607e-05,
-      "loss": 53.2251,
-      "step": 127940
-    },
-    {
-      "epoch": 0.5169341903788427,
-      "grad_norm": 649.6722412109375,
-      "learning_rate": 2.832370892404511e-05,
-      "loss": 47.6598,
-      "step": 127950
-    },
-    {
-      "epoch": 0.5169745916442103,
-      "grad_norm": 952.0791625976562,
-      "learning_rate": 2.832024921984674e-05,
-      "loss": 46.0533,
-      "step": 127960
-    },
-    {
-      "epoch": 0.517014992909578,
-      "grad_norm": 663.2807006835938,
-      "learning_rate": 2.8316789450918396e-05,
-      "loss": 60.6055,
-      "step": 127970
-    },
-    {
-      "epoch": 0.5170553941749455,
-      "grad_norm": 1854.5562744140625,
-      "learning_rate": 2.8313329617327537e-05,
-      "loss": 104.9597,
-      "step": 127980
-    },
-    {
-      "epoch": 0.5170957954403131,
-      "grad_norm": 399.9706726074219,
-      "learning_rate": 2.8309869719141608e-05,
-      "loss": 73.0087,
-      "step": 127990
-    },
-    {
-      "epoch": 0.5171361967056808,
-      "grad_norm": 742.0582275390625,
-      "learning_rate": 2.8306409756428064e-05,
-      "loss": 63.6711,
-      "step": 128000
-    },
-    {
-      "epoch": 0.5171765979710484,
-      "grad_norm": 426.78814697265625,
-      "learning_rate": 2.8302949729254358e-05,
-      "loss": 44.5783,
-      "step": 128010
-    },
-    {
-      "epoch": 0.5172169992364161,
-      "grad_norm": 537.81494140625,
-      "learning_rate": 2.8299489637687954e-05,
-      "loss": 46.2207,
-      "step": 128020
-    },
-    {
-      "epoch": 0.5172574005017837,
-      "grad_norm": 545.3671264648438,
-      "learning_rate": 2.8296029481796292e-05,
-      "loss": 52.6649,
-      "step": 128030
-    },
-    {
-      "epoch": 0.5172978017671513,
-      "grad_norm": 1167.154541015625,
-      "learning_rate": 2.829256926164685e-05,
-      "loss": 68.5006,
-      "step": 128040
-    },
-    {
-      "epoch": 0.517338203032519,
-      "grad_norm": 743.8500366210938,
-      "learning_rate": 2.8289108977307067e-05,
-      "loss": 49.9936,
-      "step": 128050
-    },
-    {
-      "epoch": 0.5173786042978866,
-      "grad_norm": 708.7972412109375,
-      "learning_rate": 2.8285648628844413e-05,
-      "loss": 58.3343,
-      "step": 128060
-    },
-    {
-      "epoch": 0.5174190055632543,
-      "grad_norm": 952.4853515625,
-      "learning_rate": 2.8282188216326345e-05,
-      "loss": 51.3421,
-      "step": 128070
-    },
-    {
-      "epoch": 0.5174594068286219,
-      "grad_norm": 761.9249267578125,
-      "learning_rate": 2.8278727739820333e-05,
-      "loss": 47.8939,
-      "step": 128080
-    },
-    {
-      "epoch": 0.5174998080939895,
-      "grad_norm": 1164.4027099609375,
-      "learning_rate": 2.827526719939383e-05,
-      "loss": 112.3727,
-      "step": 128090
-    },
-    {
-      "epoch": 0.5175402093593572,
-      "grad_norm": 1486.0279541015625,
-      "learning_rate": 2.827180659511431e-05,
-      "loss": 54.5274,
-      "step": 128100
-    },
-    {
-      "epoch": 0.5175806106247247,
-      "grad_norm": 1370.9776611328125,
-      "learning_rate": 2.8268345927049234e-05,
-      "loss": 65.6802,
-      "step": 128110
-    },
-    {
-      "epoch": 0.5176210118900924,
-      "grad_norm": 796.0413818359375,
-      "learning_rate": 2.8264885195266065e-05,
-      "loss": 58.2697,
-      "step": 128120
-    },
-    {
-      "epoch": 0.51766141315546,
-      "grad_norm": 703.8768310546875,
-      "learning_rate": 2.8261424399832293e-05,
-      "loss": 49.1636,
-      "step": 128130
-    },
-    {
-      "epoch": 0.5177018144208276,
-      "grad_norm": 419.2386779785156,
-      "learning_rate": 2.825796354081537e-05,
-      "loss": 39.2348,
-      "step": 128140
-    },
-    {
-      "epoch": 0.5177422156861953,
-      "grad_norm": 687.80859375,
-      "learning_rate": 2.8254502618282763e-05,
-      "loss": 33.7805,
-      "step": 128150
-    },
-    {
-      "epoch": 0.5177826169515629,
-      "grad_norm": 0.0,
-      "learning_rate": 2.8251041632301957e-05,
-      "loss": 54.0625,
-      "step": 128160
-    },
-    {
-      "epoch": 0.5178230182169306,
-      "grad_norm": 458.79937744140625,
-      "learning_rate": 2.8247580582940413e-05,
-      "loss": 60.1953,
-      "step": 128170
-    },
-    {
-      "epoch": 0.5178634194822982,
-      "grad_norm": 739.254638671875,
-      "learning_rate": 2.824411947026563e-05,
-      "loss": 67.9315,
-      "step": 128180
-    },
-    {
-      "epoch": 0.5179038207476658,
-      "grad_norm": 998.814208984375,
-      "learning_rate": 2.824065829434505e-05,
-      "loss": 52.8274,
-      "step": 128190
-    },
-    {
-      "epoch": 0.5179442220130335,
-      "grad_norm": 862.947265625,
-      "learning_rate": 2.8237197055246172e-05,
-      "loss": 47.5776,
-      "step": 128200
-    },
-    {
-      "epoch": 0.5179846232784011,
-      "grad_norm": 0.0,
-      "learning_rate": 2.8233735753036484e-05,
-      "loss": 58.8695,
-      "step": 128210
-    },
-    {
-      "epoch": 0.5180250245437688,
-      "grad_norm": 1318.5653076171875,
-      "learning_rate": 2.823027438778344e-05,
-      "loss": 46.4017,
-      "step": 128220
-    },
-    {
-      "epoch": 0.5180654258091364,
-      "grad_norm": 1210.0972900390625,
-      "learning_rate": 2.8226812959554537e-05,
-      "loss": 61.901,
-      "step": 128230
-    },
-    {
-      "epoch": 0.5181058270745039,
-      "grad_norm": 756.2396850585938,
-      "learning_rate": 2.8223351468417254e-05,
-      "loss": 49.0368,
-      "step": 128240
-    },
-    {
-      "epoch": 0.5181462283398716,
-      "grad_norm": 419.0620422363281,
-      "learning_rate": 2.8219889914439074e-05,
-      "loss": 52.058,
-      "step": 128250
-    },
-    {
-      "epoch": 0.5181866296052392,
-      "grad_norm": 868.4646606445312,
-      "learning_rate": 2.821642829768748e-05,
-      "loss": 78.654,
-      "step": 128260
-    },
-    {
-      "epoch": 0.5182270308706068,
-      "grad_norm": 424.6618957519531,
-      "learning_rate": 2.8212966618229964e-05,
-      "loss": 94.6811,
-      "step": 128270
-    },
-    {
-      "epoch": 0.5182674321359745,
-      "grad_norm": 535.59228515625,
-      "learning_rate": 2.8209504876134007e-05,
-      "loss": 44.2588,
-      "step": 128280
-    },
-    {
-      "epoch": 0.5183078334013421,
-      "grad_norm": 578.2286987304688,
-      "learning_rate": 2.8206043071467102e-05,
-      "loss": 38.9842,
-      "step": 128290
-    },
-    {
-      "epoch": 0.5183482346667098,
-      "grad_norm": 948.8009033203125,
-      "learning_rate": 2.8202581204296742e-05,
-      "loss": 66.6254,
-      "step": 128300
-    },
-    {
-      "epoch": 0.5183886359320774,
-      "grad_norm": 956.4691772460938,
-      "learning_rate": 2.819911927469041e-05,
-      "loss": 83.3204,
-      "step": 128310
-    },
-    {
-      "epoch": 0.518429037197445,
-      "grad_norm": 662.3033447265625,
-      "learning_rate": 2.8195657282715594e-05,
-      "loss": 33.0041,
-      "step": 128320
-    },
-    {
-      "epoch": 0.5184694384628127,
-      "grad_norm": 841.9630737304688,
-      "learning_rate": 2.81921952284398e-05,
-      "loss": 40.0045,
-      "step": 128330
-    },
-    {
-      "epoch": 0.5185098397281803,
-      "grad_norm": 644.8792114257812,
-      "learning_rate": 2.818873311193051e-05,
-      "loss": 49.6458,
-      "step": 128340
-    },
-    {
-      "epoch": 0.518550240993548,
-      "grad_norm": 1007.34765625,
-      "learning_rate": 2.8185270933255237e-05,
-      "loss": 45.5391,
-      "step": 128350
-    },
-    {
-      "epoch": 0.5185906422589156,
-      "grad_norm": 412.73883056640625,
-      "learning_rate": 2.8181808692481453e-05,
-      "loss": 45.3399,
-      "step": 128360
-    },
-    {
-      "epoch": 0.5186310435242831,
-      "grad_norm": 775.2849731445312,
-      "learning_rate": 2.817834638967668e-05,
-      "loss": 56.8275,
-      "step": 128370
-    },
-    {
-      "epoch": 0.5186714447896508,
-      "grad_norm": 880.3590698242188,
-      "learning_rate": 2.817488402490841e-05,
-      "loss": 62.0417,
-      "step": 128380
-    },
-    {
-      "epoch": 0.5187118460550184,
-      "grad_norm": 0.0,
-      "learning_rate": 2.8171421598244134e-05,
-      "loss": 100.0851,
-      "step": 128390
-    },
-    {
-      "epoch": 0.518752247320386,
-      "grad_norm": 757.4847412109375,
-      "learning_rate": 2.816795910975137e-05,
-      "loss": 81.3571,
-      "step": 128400
-    },
-    {
-      "epoch": 0.5187926485857537,
-      "grad_norm": 986.7409057617188,
-      "learning_rate": 2.8164496559497605e-05,
-      "loss": 99.9801,
-      "step": 128410
-    },
-    {
-      "epoch": 0.5188330498511213,
-      "grad_norm": 1231.8388671875,
-      "learning_rate": 2.816103394755035e-05,
-      "loss": 36.7881,
-      "step": 128420
-    },
-    {
-      "epoch": 0.518873451116489,
-      "grad_norm": 551.5934448242188,
-      "learning_rate": 2.8157571273977117e-05,
-      "loss": 48.0908,
-      "step": 128430
-    },
-    {
-      "epoch": 0.5189138523818566,
-      "grad_norm": 957.9308471679688,
-      "learning_rate": 2.8154108538845404e-05,
-      "loss": 54.7105,
-      "step": 128440
-    },
-    {
-      "epoch": 0.5189542536472243,
-      "grad_norm": 746.921875,
-      "learning_rate": 2.8150645742222714e-05,
-      "loss": 47.9935,
-      "step": 128450
-    },
-    {
-      "epoch": 0.5189946549125919,
-      "grad_norm": 817.0953369140625,
-      "learning_rate": 2.814718288417657e-05,
-      "loss": 34.7214,
-      "step": 128460
-    },
-    {
-      "epoch": 0.5190350561779595,
-      "grad_norm": 652.5723876953125,
-      "learning_rate": 2.814371996477448e-05,
-      "loss": 51.5591,
-      "step": 128470
-    },
-    {
-      "epoch": 0.5190754574433272,
-      "grad_norm": 1426.8663330078125,
-      "learning_rate": 2.8140256984083947e-05,
-      "loss": 53.6987,
-      "step": 128480
-    },
-    {
-      "epoch": 0.5191158587086948,
-      "grad_norm": 1349.2745361328125,
-      "learning_rate": 2.8136793942172483e-05,
-      "loss": 67.8195,
-      "step": 128490
-    },
-    {
-      "epoch": 0.5191562599740623,
-      "grad_norm": 1255.9248046875,
-      "learning_rate": 2.8133330839107608e-05,
-      "loss": 89.4671,
-      "step": 128500
-    },
-    {
-      "epoch": 0.51919666123943,
-      "grad_norm": 772.8133544921875,
-      "learning_rate": 2.8129867674956838e-05,
-      "loss": 99.116,
-      "step": 128510
-    },
-    {
-      "epoch": 0.5192370625047976,
-      "grad_norm": 799.4025268554688,
-      "learning_rate": 2.8126404449787685e-05,
-      "loss": 37.4048,
-      "step": 128520
-    },
-    {
-      "epoch": 0.5192774637701653,
-      "grad_norm": 767.2107543945312,
-      "learning_rate": 2.8122941163667667e-05,
-      "loss": 53.6676,
-      "step": 128530
-    },
-    {
-      "epoch": 0.5193178650355329,
-      "grad_norm": 717.8088989257812,
-      "learning_rate": 2.8119477816664296e-05,
-      "loss": 53.9604,
-      "step": 128540
-    },
-    {
-      "epoch": 0.5193582663009005,
-      "grad_norm": 568.8512573242188,
-      "learning_rate": 2.8116014408845116e-05,
-      "loss": 52.336,
-      "step": 128550
-    },
-    {
-      "epoch": 0.5193986675662682,
-      "grad_norm": 1033.583984375,
-      "learning_rate": 2.8112550940277616e-05,
-      "loss": 60.4758,
-      "step": 128560
-    },
-    {
-      "epoch": 0.5194390688316358,
-      "grad_norm": 267.0958251953125,
-      "learning_rate": 2.810908741102934e-05,
-      "loss": 87.3894,
-      "step": 128570
-    },
-    {
-      "epoch": 0.5194794700970035,
-      "grad_norm": 966.0076293945312,
-      "learning_rate": 2.8105623821167804e-05,
-      "loss": 64.1264,
-      "step": 128580
-    },
-    {
-      "epoch": 0.5195198713623711,
-      "grad_norm": 888.2479858398438,
-      "learning_rate": 2.810216017076053e-05,
-      "loss": 37.3439,
-      "step": 128590
-    },
-    {
-      "epoch": 0.5195602726277387,
-      "grad_norm": 1010.2241821289062,
-      "learning_rate": 2.8098696459875046e-05,
-      "loss": 59.4499,
-      "step": 128600
-    },
-    {
-      "epoch": 0.5196006738931064,
-      "grad_norm": 672.6810302734375,
-      "learning_rate": 2.8095232688578883e-05,
-      "loss": 59.5572,
-      "step": 128610
-    },
-    {
-      "epoch": 0.5196410751584739,
-      "grad_norm": 1322.0794677734375,
-      "learning_rate": 2.809176885693956e-05,
-      "loss": 75.077,
-      "step": 128620
-    },
-    {
-      "epoch": 0.5196814764238415,
-      "grad_norm": 647.8616943359375,
-      "learning_rate": 2.8088304965024614e-05,
-      "loss": 54.4772,
-      "step": 128630
-    },
-    {
-      "epoch": 0.5197218776892092,
-      "grad_norm": 895.9517211914062,
-      "learning_rate": 2.8084841012901574e-05,
-      "loss": 32.6104,
-      "step": 128640
-    },
-    {
-      "epoch": 0.5197622789545768,
-      "grad_norm": 346.7025451660156,
-      "learning_rate": 2.808137700063797e-05,
-      "loss": 39.3208,
-      "step": 128650
-    },
-    {
-      "epoch": 0.5198026802199445,
-      "grad_norm": 911.7136840820312,
-      "learning_rate": 2.807791292830133e-05,
-      "loss": 58.0953,
-      "step": 128660
-    },
-    {
-      "epoch": 0.5198430814853121,
-      "grad_norm": 522.1300659179688,
-      "learning_rate": 2.8074448795959203e-05,
-      "loss": 73.4923,
-      "step": 128670
-    },
-    {
-      "epoch": 0.5198834827506797,
-      "grad_norm": 657.18505859375,
-      "learning_rate": 2.8070984603679107e-05,
-      "loss": 42.6445,
-      "step": 128680
-    },
-    {
-      "epoch": 0.5199238840160474,
-      "grad_norm": 389.8962707519531,
-      "learning_rate": 2.8067520351528587e-05,
-      "loss": 43.4139,
-      "step": 128690
-    },
-    {
-      "epoch": 0.519964285281415,
-      "grad_norm": 851.1239013671875,
-      "learning_rate": 2.806405603957517e-05,
-      "loss": 72.5371,
-      "step": 128700
-    },
-    {
-      "epoch": 0.5200046865467827,
-      "grad_norm": 3148.775634765625,
-      "learning_rate": 2.8060591667886416e-05,
-      "loss": 56.5808,
-      "step": 128710
-    },
-    {
-      "epoch": 0.5200450878121503,
-      "grad_norm": 1202.496826171875,
-      "learning_rate": 2.8057127236529844e-05,
-      "loss": 46.4309,
-      "step": 128720
-    },
-    {
-      "epoch": 0.520085489077518,
-      "grad_norm": 962.0496826171875,
-      "learning_rate": 2.805366274557301e-05,
-      "loss": 50.884,
-      "step": 128730
-    },
-    {
-      "epoch": 0.5201258903428856,
-      "grad_norm": 453.456298828125,
-      "learning_rate": 2.8050198195083444e-05,
-      "loss": 59.2252,
-      "step": 128740
-    },
-    {
-      "epoch": 0.5201662916082531,
-      "grad_norm": 239.76718139648438,
-      "learning_rate": 2.8046733585128687e-05,
-      "loss": 70.7534,
-      "step": 128750
-    },
-    {
-      "epoch": 0.5202066928736208,
-      "grad_norm": 457.1228942871094,
-      "learning_rate": 2.80432689157763e-05,
-      "loss": 52.9828,
-      "step": 128760
-    },
-    {
-      "epoch": 0.5202470941389884,
-      "grad_norm": 877.8587036132812,
-      "learning_rate": 2.8039804187093816e-05,
-      "loss": 52.8952,
-      "step": 128770
-    },
-    {
-      "epoch": 0.520287495404356,
-      "grad_norm": 1153.672119140625,
-      "learning_rate": 2.803633939914878e-05,
-      "loss": 60.0604,
-      "step": 128780
-    },
-    {
-      "epoch": 0.5203278966697237,
-      "grad_norm": 120.61048126220703,
-      "learning_rate": 2.803287455200875e-05,
-      "loss": 47.4182,
-      "step": 128790
-    },
-    {
-      "epoch": 0.5203682979350913,
-      "grad_norm": 1153.82080078125,
-      "learning_rate": 2.8029409645741267e-05,
-      "loss": 59.4981,
-      "step": 128800
-    },
-    {
-      "epoch": 0.520408699200459,
-      "grad_norm": 418.24920654296875,
-      "learning_rate": 2.8025944680413878e-05,
-      "loss": 37.4401,
-      "step": 128810
-    },
-    {
-      "epoch": 0.5204491004658266,
-      "grad_norm": 781.8113403320312,
-      "learning_rate": 2.8022479656094154e-05,
-      "loss": 48.7074,
-      "step": 128820
-    },
-    {
-      "epoch": 0.5204895017311942,
-      "grad_norm": 836.5515747070312,
-      "learning_rate": 2.801901457284962e-05,
-      "loss": 51.3323,
-      "step": 128830
-    },
-    {
-      "epoch": 0.5205299029965619,
-      "grad_norm": 1007.6831665039062,
-      "learning_rate": 2.8015549430747852e-05,
-      "loss": 63.1766,
-      "step": 128840
-    },
-    {
-      "epoch": 0.5205703042619295,
-      "grad_norm": 1059.2291259765625,
-      "learning_rate": 2.8012084229856382e-05,
-      "loss": 48.2031,
-      "step": 128850
-    },
-    {
-      "epoch": 0.5206107055272972,
-      "grad_norm": 435.08837890625,
-      "learning_rate": 2.800861897024279e-05,
-      "loss": 61.0167,
-      "step": 128860
-    },
-    {
-      "epoch": 0.5206511067926648,
-      "grad_norm": 520.9664916992188,
-      "learning_rate": 2.8005153651974614e-05,
-      "loss": 68.8364,
-      "step": 128870
-    },
-    {
-      "epoch": 0.5206915080580323,
-      "grad_norm": 1375.3497314453125,
-      "learning_rate": 2.8001688275119432e-05,
-      "loss": 73.5849,
-      "step": 128880
-    },
-    {
-      "epoch": 0.5207319093234,
-      "grad_norm": 570.526611328125,
-      "learning_rate": 2.799822283974478e-05,
-      "loss": 45.2284,
-      "step": 128890
-    },
-    {
-      "epoch": 0.5207723105887676,
-      "grad_norm": 3453.25927734375,
-      "learning_rate": 2.7994757345918244e-05,
-      "loss": 65.3527,
-      "step": 128900
-    },
-    {
-      "epoch": 0.5208127118541352,
-      "grad_norm": 807.7914428710938,
-      "learning_rate": 2.7991291793707357e-05,
-      "loss": 37.759,
-      "step": 128910
-    },
-    {
-      "epoch": 0.5208531131195029,
-      "grad_norm": 2692.131103515625,
-      "learning_rate": 2.7987826183179712e-05,
-      "loss": 70.0045,
-      "step": 128920
-    },
-    {
-      "epoch": 0.5208935143848705,
-      "grad_norm": 316.426025390625,
-      "learning_rate": 2.798436051440284e-05,
-      "loss": 50.6802,
-      "step": 128930
-    },
-    {
-      "epoch": 0.5209339156502382,
-      "grad_norm": 1145.9749755859375,
-      "learning_rate": 2.7980894787444334e-05,
-      "loss": 44.7317,
-      "step": 128940
-    },
-    {
-      "epoch": 0.5209743169156058,
-      "grad_norm": 571.5775146484375,
-      "learning_rate": 2.7977429002371747e-05,
-      "loss": 39.6738,
-      "step": 128950
-    },
-    {
-      "epoch": 0.5210147181809734,
-      "grad_norm": 625.548828125,
-      "learning_rate": 2.797396315925265e-05,
-      "loss": 34.7972,
-      "step": 128960
-    },
-    {
-      "epoch": 0.5210551194463411,
-      "grad_norm": 733.4710083007812,
-      "learning_rate": 2.7970497258154603e-05,
-      "loss": 38.4996,
-      "step": 128970
-    },
-    {
-      "epoch": 0.5210955207117087,
-      "grad_norm": 1038.670654296875,
-      "learning_rate": 2.7967031299145193e-05,
-      "loss": 49.1648,
-      "step": 128980
-    },
-    {
-      "epoch": 0.5211359219770764,
-      "grad_norm": 545.0449829101562,
-      "learning_rate": 2.7963565282291977e-05,
-      "loss": 48.0368,
-      "step": 128990
-    },
-    {
-      "epoch": 0.521176323242444,
-      "grad_norm": 428.3600769042969,
-      "learning_rate": 2.7960099207662532e-05,
-      "loss": 60.4256,
-      "step": 129000
-    },
-    {
-      "epoch": 0.5212167245078115,
-      "grad_norm": 714.7028198242188,
-      "learning_rate": 2.7956633075324424e-05,
-      "loss": 60.8093,
-      "step": 129010
-    },
-    {
-      "epoch": 0.5212571257731792,
-      "grad_norm": 1208.1380615234375,
-      "learning_rate": 2.795316688534523e-05,
-      "loss": 74.678,
-      "step": 129020
-    },
-    {
-      "epoch": 0.5212975270385468,
-      "grad_norm": 534.9729614257812,
-      "learning_rate": 2.794970063779253e-05,
-      "loss": 51.2989,
-      "step": 129030
-    },
-    {
-      "epoch": 0.5213379283039145,
-      "grad_norm": 756.8164672851562,
-      "learning_rate": 2.79462343327339e-05,
-      "loss": 66.086,
-      "step": 129040
-    },
-    {
-      "epoch": 0.5213783295692821,
-      "grad_norm": 575.9217529296875,
-      "learning_rate": 2.794276797023691e-05,
-      "loss": 49.4426,
-      "step": 129050
-    },
-    {
-      "epoch": 0.5214187308346497,
-      "grad_norm": 498.71441650390625,
-      "learning_rate": 2.7939301550369146e-05,
-      "loss": 48.8029,
-      "step": 129060
-    },
-    {
-      "epoch": 0.5214591321000174,
-      "grad_norm": 594.4951782226562,
-      "learning_rate": 2.7935835073198192e-05,
-      "loss": 34.5576,
-      "step": 129070
-    },
-    {
-      "epoch": 0.521499533365385,
-      "grad_norm": 1014.3102416992188,
-      "learning_rate": 2.793236853879161e-05,
-      "loss": 58.8379,
-      "step": 129080
-    },
-    {
-      "epoch": 0.5215399346307527,
-      "grad_norm": 560.2127685546875,
-      "learning_rate": 2.7928901947217008e-05,
-      "loss": 63.1494,
-      "step": 129090
-    },
-    {
-      "epoch": 0.5215803358961203,
-      "grad_norm": 666.1370239257812,
-      "learning_rate": 2.792543529854194e-05,
-      "loss": 53.1794,
-      "step": 129100
-    },
-    {
-      "epoch": 0.5216207371614879,
-      "grad_norm": 636.662353515625,
-      "learning_rate": 2.7921968592834006e-05,
-      "loss": 25.9828,
-      "step": 129110
-    },
-    {
-      "epoch": 0.5216611384268556,
-      "grad_norm": 835.5521850585938,
-      "learning_rate": 2.79185018301608e-05,
-      "loss": 57.4081,
-      "step": 129120
-    },
-    {
-      "epoch": 0.5217015396922232,
-      "grad_norm": 992.4476928710938,
-      "learning_rate": 2.791503501058989e-05,
-      "loss": 52.8603,
-      "step": 129130
-    },
-    {
-      "epoch": 0.5217419409575907,
-      "grad_norm": 328.57794189453125,
-      "learning_rate": 2.7911568134188875e-05,
-      "loss": 88.1504,
-      "step": 129140
-    },
-    {
-      "epoch": 0.5217823422229584,
-      "grad_norm": 1140.0035400390625,
-      "learning_rate": 2.7908101201025337e-05,
-      "loss": 48.8337,
-      "step": 129150
-    },
-    {
-      "epoch": 0.521822743488326,
-      "grad_norm": 745.549072265625,
-      "learning_rate": 2.7904634211166876e-05,
-      "loss": 55.3195,
-      "step": 129160
-    },
-    {
-      "epoch": 0.5218631447536937,
-      "grad_norm": 969.3555297851562,
-      "learning_rate": 2.7901167164681073e-05,
-      "loss": 55.3426,
-      "step": 129170
-    },
-    {
-      "epoch": 0.5219035460190613,
-      "grad_norm": 726.4353637695312,
-      "learning_rate": 2.7897700061635517e-05,
-      "loss": 49.0015,
-      "step": 129180
-    },
-    {
-      "epoch": 0.5219439472844289,
-      "grad_norm": 374.95843505859375,
-      "learning_rate": 2.7894232902097813e-05,
-      "loss": 46.3873,
-      "step": 129190
-    },
-    {
-      "epoch": 0.5219843485497966,
-      "grad_norm": 511.00885009765625,
-      "learning_rate": 2.7890765686135544e-05,
-      "loss": 47.9072,
-      "step": 129200
-    },
-    {
-      "epoch": 0.5220247498151642,
-      "grad_norm": 512.494873046875,
-      "learning_rate": 2.788729841381631e-05,
-      "loss": 29.6466,
-      "step": 129210
-    },
-    {
-      "epoch": 0.5220651510805319,
-      "grad_norm": 551.0260620117188,
-      "learning_rate": 2.7883831085207707e-05,
-      "loss": 65.4076,
-      "step": 129220
-    },
-    {
-      "epoch": 0.5221055523458995,
-      "grad_norm": 621.8154907226562,
-      "learning_rate": 2.788036370037733e-05,
-      "loss": 55.3526,
-      "step": 129230
-    },
-    {
-      "epoch": 0.5221459536112671,
-      "grad_norm": 477.7518310546875,
-      "learning_rate": 2.7876896259392788e-05,
-      "loss": 70.7479,
-      "step": 129240
-    },
-    {
-      "epoch": 0.5221863548766348,
-      "grad_norm": 228.69644165039062,
-      "learning_rate": 2.787342876232167e-05,
-      "loss": 45.7491,
-      "step": 129250
-    },
-    {
-      "epoch": 0.5222267561420023,
-      "grad_norm": 406.46026611328125,
-      "learning_rate": 2.7869961209231577e-05,
-      "loss": 40.5636,
-      "step": 129260
-    },
-    {
-      "epoch": 0.52226715740737,
-      "grad_norm": 1248.6951904296875,
-      "learning_rate": 2.7866493600190107e-05,
-      "loss": 43.6552,
-      "step": 129270
-    },
-    {
-      "epoch": 0.5223075586727376,
-      "grad_norm": 780.5540161132812,
-      "learning_rate": 2.7863025935264875e-05,
-      "loss": 42.3198,
-      "step": 129280
-    },
-    {
-      "epoch": 0.5223479599381052,
-      "grad_norm": 830.3927612304688,
-      "learning_rate": 2.785955821452348e-05,
-      "loss": 50.129,
-      "step": 129290
-    },
-    {
-      "epoch": 0.5223883612034729,
-      "grad_norm": 162.7823028564453,
-      "learning_rate": 2.7856090438033522e-05,
-      "loss": 35.8086,
-      "step": 129300
-    },
-    {
-      "epoch": 0.5224287624688405,
-      "grad_norm": 577.8681030273438,
-      "learning_rate": 2.785262260586261e-05,
-      "loss": 48.4552,
-      "step": 129310
-    },
-    {
-      "epoch": 0.5224691637342082,
-      "grad_norm": 495.5697021484375,
-      "learning_rate": 2.7849154718078346e-05,
-      "loss": 41.2047,
-      "step": 129320
-    },
-    {
-      "epoch": 0.5225095649995758,
-      "grad_norm": 650.5645141601562,
-      "learning_rate": 2.784568677474836e-05,
-      "loss": 65.8662,
-      "step": 129330
-    },
-    {
-      "epoch": 0.5225499662649434,
-      "grad_norm": 276.66241455078125,
-      "learning_rate": 2.7842218775940237e-05,
-      "loss": 33.7441,
-      "step": 129340
-    },
-    {
-      "epoch": 0.5225903675303111,
-      "grad_norm": 840.7539672851562,
-      "learning_rate": 2.783875072172159e-05,
-      "loss": 49.3532,
-      "step": 129350
-    },
-    {
-      "epoch": 0.5226307687956787,
-      "grad_norm": 315.2930603027344,
-      "learning_rate": 2.783528261216004e-05,
-      "loss": 37.2258,
-      "step": 129360
-    },
-    {
-      "epoch": 0.5226711700610464,
-      "grad_norm": 398.0323791503906,
-      "learning_rate": 2.78318144473232e-05,
-      "loss": 85.9904,
-      "step": 129370
-    },
-    {
-      "epoch": 0.522711571326414,
-      "grad_norm": 779.0153198242188,
-      "learning_rate": 2.7828346227278674e-05,
-      "loss": 53.5858,
-      "step": 129380
-    },
-    {
-      "epoch": 0.5227519725917815,
-      "grad_norm": 941.3084106445312,
-      "learning_rate": 2.782487795209408e-05,
-      "loss": 77.1186,
-      "step": 129390
-    },
-    {
-      "epoch": 0.5227923738571492,
-      "grad_norm": 1087.8790283203125,
-      "learning_rate": 2.782140962183704e-05,
-      "loss": 75.8238,
-      "step": 129400
-    },
-    {
-      "epoch": 0.5228327751225168,
-      "grad_norm": 836.9957275390625,
-      "learning_rate": 2.7817941236575173e-05,
-      "loss": 60.6787,
-      "step": 129410
-    },
-    {
-      "epoch": 0.5228731763878844,
-      "grad_norm": 542.893798828125,
-      "learning_rate": 2.781447279637608e-05,
-      "loss": 54.1359,
-      "step": 129420
-    },
-    {
-      "epoch": 0.5229135776532521,
-      "grad_norm": 706.9591064453125,
-      "learning_rate": 2.7811004301307403e-05,
-      "loss": 42.0238,
-      "step": 129430
-    },
-    {
-      "epoch": 0.5229539789186197,
-      "grad_norm": 975.5703735351562,
-      "learning_rate": 2.7807535751436738e-05,
-      "loss": 42.8671,
-      "step": 129440
-    },
-    {
-      "epoch": 0.5229943801839874,
-      "grad_norm": 1424.2943115234375,
-      "learning_rate": 2.7804067146831725e-05,
-      "loss": 69.0784,
-      "step": 129450
-    },
-    {
-      "epoch": 0.523034781449355,
-      "grad_norm": 1137.7862548828125,
-      "learning_rate": 2.7800598487559975e-05,
-      "loss": 46.7197,
-      "step": 129460
-    },
-    {
-      "epoch": 0.5230751827147226,
-      "grad_norm": 195.2952423095703,
-      "learning_rate": 2.7797129773689118e-05,
-      "loss": 55.5766,
-      "step": 129470
-    },
-    {
-      "epoch": 0.5231155839800903,
-      "grad_norm": 810.8615112304688,
-      "learning_rate": 2.7793661005286774e-05,
-      "loss": 42.1326,
-      "step": 129480
-    },
-    {
-      "epoch": 0.5231559852454579,
-      "grad_norm": 259.7433166503906,
-      "learning_rate": 2.7790192182420578e-05,
-      "loss": 40.0814,
-      "step": 129490
-    },
-    {
-      "epoch": 0.5231963865108256,
-      "grad_norm": 465.7834777832031,
-      "learning_rate": 2.7786723305158136e-05,
-      "loss": 41.8416,
-      "step": 129500
-    },
-    {
-      "epoch": 0.5232367877761932,
-      "grad_norm": 309.6441650390625,
-      "learning_rate": 2.7783254373567103e-05,
-      "loss": 92.4435,
-      "step": 129510
-    },
-    {
-      "epoch": 0.5232771890415607,
-      "grad_norm": 689.2532348632812,
-      "learning_rate": 2.7779785387715078e-05,
-      "loss": 50.8395,
-      "step": 129520
-    },
-    {
-      "epoch": 0.5233175903069284,
-      "grad_norm": 551.0718994140625,
-      "learning_rate": 2.7776316347669722e-05,
-      "loss": 53.9811,
-      "step": 129530
-    },
-    {
-      "epoch": 0.523357991572296,
-      "grad_norm": 792.376953125,
-      "learning_rate": 2.7772847253498636e-05,
-      "loss": 38.2251,
-      "step": 129540
-    },
-    {
-      "epoch": 0.5233983928376637,
-      "grad_norm": 687.2542724609375,
-      "learning_rate": 2.7769378105269467e-05,
-      "loss": 72.9509,
-      "step": 129550
-    },
-    {
-      "epoch": 0.5234387941030313,
-      "grad_norm": 856.1763916015625,
-      "learning_rate": 2.7765908903049848e-05,
-      "loss": 42.2359,
-      "step": 129560
-    },
-    {
-      "epoch": 0.5234791953683989,
-      "grad_norm": 865.5512084960938,
-      "learning_rate": 2.7762439646907417e-05,
-      "loss": 46.5808,
-      "step": 129570
-    },
-    {
-      "epoch": 0.5235195966337666,
-      "grad_norm": 218.27801513671875,
-      "learning_rate": 2.7758970336909795e-05,
-      "loss": 58.7573,
-      "step": 129580
-    },
-    {
-      "epoch": 0.5235599978991342,
-      "grad_norm": 797.5692749023438,
-      "learning_rate": 2.7755500973124625e-05,
-      "loss": 64.8211,
-      "step": 129590
-    },
-    {
-      "epoch": 0.5236003991645019,
-      "grad_norm": 1161.782470703125,
-      "learning_rate": 2.7752031555619555e-05,
-      "loss": 62.3158,
-      "step": 129600
-    },
-    {
-      "epoch": 0.5236408004298695,
-      "grad_norm": 495.8021545410156,
-      "learning_rate": 2.774856208446221e-05,
-      "loss": 34.1338,
-      "step": 129610
-    },
-    {
-      "epoch": 0.5236812016952371,
-      "grad_norm": 538.5377807617188,
-      "learning_rate": 2.7745092559720227e-05,
-      "loss": 84.3193,
-      "step": 129620
-    },
-    {
-      "epoch": 0.5237216029606048,
-      "grad_norm": 472.7118225097656,
-      "learning_rate": 2.7741622981461253e-05,
-      "loss": 88.2121,
-      "step": 129630
-    },
-    {
-      "epoch": 0.5237620042259724,
-      "grad_norm": 680.1412353515625,
-      "learning_rate": 2.773815334975292e-05,
-      "loss": 61.683,
-      "step": 129640
-    },
-    {
-      "epoch": 0.5238024054913399,
-      "grad_norm": 194.98385620117188,
-      "learning_rate": 2.7734683664662892e-05,
-      "loss": 35.9615,
-      "step": 129650
-    },
-    {
-      "epoch": 0.5238428067567076,
-      "grad_norm": 664.1810913085938,
-      "learning_rate": 2.7731213926258794e-05,
-      "loss": 76.9602,
-      "step": 129660
-    },
-    {
-      "epoch": 0.5238832080220752,
-      "grad_norm": 675.6226806640625,
-      "learning_rate": 2.7727744134608263e-05,
-      "loss": 49.6265,
-      "step": 129670
-    },
-    {
-      "epoch": 0.5239236092874429,
-      "grad_norm": 424.91912841796875,
-      "learning_rate": 2.7724274289778974e-05,
-      "loss": 49.0146,
-      "step": 129680
-    },
-    {
-      "epoch": 0.5239640105528105,
-      "grad_norm": 1368.484619140625,
-      "learning_rate": 2.7720804391838544e-05,
-      "loss": 53.2469,
-      "step": 129690
-    },
-    {
-      "epoch": 0.5240044118181781,
-      "grad_norm": 988.5595092773438,
-      "learning_rate": 2.771733444085463e-05,
-      "loss": 72.9639,
-      "step": 129700
-    },
-    {
-      "epoch": 0.5240448130835458,
-      "grad_norm": 1024.283447265625,
-      "learning_rate": 2.771386443689489e-05,
-      "loss": 35.8563,
-      "step": 129710
-    },
-    {
-      "epoch": 0.5240852143489134,
-      "grad_norm": 2501.96142578125,
-      "learning_rate": 2.7710394380026954e-05,
-      "loss": 93.8633,
-      "step": 129720
-    },
-    {
-      "epoch": 0.5241256156142811,
-      "grad_norm": 397.55413818359375,
-      "learning_rate": 2.7706924270318496e-05,
-      "loss": 51.6667,
-      "step": 129730
-    },
-    {
-      "epoch": 0.5241660168796487,
-      "grad_norm": 401.48687744140625,
-      "learning_rate": 2.770345410783715e-05,
-      "loss": 53.5041,
-      "step": 129740
-    },
-    {
-      "epoch": 0.5242064181450163,
-      "grad_norm": 561.6734619140625,
-      "learning_rate": 2.7699983892650573e-05,
-      "loss": 60.5054,
-      "step": 129750
-    },
-    {
-      "epoch": 0.524246819410384,
-      "grad_norm": 1459.1732177734375,
-      "learning_rate": 2.769651362482642e-05,
-      "loss": 80.3514,
-      "step": 129760
-    },
-    {
-      "epoch": 0.5242872206757516,
-      "grad_norm": 211.4439239501953,
-      "learning_rate": 2.7693043304432354e-05,
-      "loss": 53.1344,
-      "step": 129770
-    },
-    {
-      "epoch": 0.5243276219411191,
-      "grad_norm": 535.6337890625,
-      "learning_rate": 2.7689572931536017e-05,
-      "loss": 59.1036,
-      "step": 129780
-    },
-    {
-      "epoch": 0.5243680232064868,
-      "grad_norm": 212.34646606445312,
-      "learning_rate": 2.7686102506205068e-05,
-      "loss": 50.72,
-      "step": 129790
-    },
-    {
-      "epoch": 0.5244084244718544,
-      "grad_norm": 1038.571044921875,
-      "learning_rate": 2.7682632028507167e-05,
-      "loss": 43.0176,
-      "step": 129800
-    },
-    {
-      "epoch": 0.5244488257372221,
-      "grad_norm": 621.1820678710938,
-      "learning_rate": 2.7679161498509976e-05,
-      "loss": 25.86,
-      "step": 129810
-    },
-    {
-      "epoch": 0.5244892270025897,
-      "grad_norm": 441.5834045410156,
-      "learning_rate": 2.7675690916281156e-05,
-      "loss": 54.9453,
-      "step": 129820
-    },
-    {
-      "epoch": 0.5245296282679573,
-      "grad_norm": 317.9010925292969,
-      "learning_rate": 2.7672220281888357e-05,
-      "loss": 95.5096,
-      "step": 129830
-    },
-    {
-      "epoch": 0.524570029533325,
-      "grad_norm": 862.2459106445312,
-      "learning_rate": 2.766874959539925e-05,
-      "loss": 69.8221,
-      "step": 129840
-    },
-    {
-      "epoch": 0.5246104307986926,
-      "grad_norm": 531.95263671875,
-      "learning_rate": 2.76652788568815e-05,
-      "loss": 49.4422,
-      "step": 129850
-    },
-    {
-      "epoch": 0.5246508320640603,
-      "grad_norm": 694.7820434570312,
-      "learning_rate": 2.7661808066402767e-05,
-      "loss": 46.5116,
-      "step": 129860
-    },
-    {
-      "epoch": 0.5246912333294279,
-      "grad_norm": 1010.5025024414062,
-      "learning_rate": 2.765833722403071e-05,
-      "loss": 71.7905,
-      "step": 129870
-    },
-    {
-      "epoch": 0.5247316345947955,
-      "grad_norm": 285.8402099609375,
-      "learning_rate": 2.7654866329833002e-05,
-      "loss": 37.7266,
-      "step": 129880
-    },
-    {
-      "epoch": 0.5247720358601632,
-      "grad_norm": 556.6085205078125,
-      "learning_rate": 2.7651395383877304e-05,
-      "loss": 54.9936,
-      "step": 129890
-    },
-    {
-      "epoch": 0.5248124371255307,
-      "grad_norm": 1335.543701171875,
-      "learning_rate": 2.76479243862313e-05,
-      "loss": 71.0658,
-      "step": 129900
-    },
-    {
-      "epoch": 0.5248528383908984,
-      "grad_norm": 510.7142639160156,
-      "learning_rate": 2.7644453336962633e-05,
-      "loss": 86.5262,
-      "step": 129910
-    },
-    {
-      "epoch": 0.524893239656266,
-      "grad_norm": 629.845458984375,
-      "learning_rate": 2.7640982236138992e-05,
-      "loss": 47.5349,
-      "step": 129920
-    },
-    {
-      "epoch": 0.5249336409216336,
-      "grad_norm": 1113.2122802734375,
-      "learning_rate": 2.7637511083828043e-05,
-      "loss": 66.3521,
-      "step": 129930
-    },
-    {
-      "epoch": 0.5249740421870013,
-      "grad_norm": 365.0244140625,
-      "learning_rate": 2.763403988009746e-05,
-      "loss": 54.905,
-      "step": 129940
-    },
-    {
-      "epoch": 0.5250144434523689,
-      "grad_norm": 543.3900756835938,
-      "learning_rate": 2.7630568625014917e-05,
-      "loss": 48.9247,
-      "step": 129950
-    },
-    {
-      "epoch": 0.5250548447177366,
-      "grad_norm": 381.6798400878906,
-      "learning_rate": 2.7627097318648076e-05,
-      "loss": 39.9984,
-      "step": 129960
-    },
-    {
-      "epoch": 0.5250952459831042,
-      "grad_norm": 1278.4371337890625,
-      "learning_rate": 2.7623625961064618e-05,
-      "loss": 60.6886,
-      "step": 129970
-    },
-    {
-      "epoch": 0.5251356472484718,
-      "grad_norm": 500.2679443359375,
-      "learning_rate": 2.7620154552332232e-05,
-      "loss": 54.9953,
-      "step": 129980
-    },
-    {
-      "epoch": 0.5251760485138395,
-      "grad_norm": 482.51458740234375,
-      "learning_rate": 2.7616683092518576e-05,
-      "loss": 55.6544,
-      "step": 129990
-    },
-    {
-      "epoch": 0.5252164497792071,
-      "grad_norm": 359.9608154296875,
-      "learning_rate": 2.761321158169134e-05,
-      "loss": 64.5541,
-      "step": 130000
-    },
-    {
-      "epoch": 0.5252568510445748,
-      "grad_norm": 398.5904541015625,
-      "learning_rate": 2.7609740019918197e-05,
-      "loss": 53.5339,
-      "step": 130010
-    },
-    {
-      "epoch": 0.5252972523099424,
-      "grad_norm": 698.3560791015625,
-      "learning_rate": 2.7606268407266827e-05,
-      "loss": 61.4191,
-      "step": 130020
-    },
-    {
-      "epoch": 0.5253376535753099,
-      "grad_norm": 768.4830932617188,
-      "learning_rate": 2.7602796743804922e-05,
-      "loss": 47.3355,
-      "step": 130030
-    },
-    {
-      "epoch": 0.5253780548406776,
-      "grad_norm": 775.7195434570312,
-      "learning_rate": 2.7599325029600143e-05,
-      "loss": 25.0439,
-      "step": 130040
-    },
-    {
-      "epoch": 0.5254184561060452,
-      "grad_norm": 1195.4205322265625,
-      "learning_rate": 2.7595853264720184e-05,
-      "loss": 70.0329,
-      "step": 130050
-    },
-    {
-      "epoch": 0.5254588573714128,
-      "grad_norm": 606.2637329101562,
-      "learning_rate": 2.759238144923274e-05,
-      "loss": 53.4816,
-      "step": 130060
-    },
-    {
-      "epoch": 0.5254992586367805,
-      "grad_norm": 469.8358459472656,
-      "learning_rate": 2.7588909583205475e-05,
-      "loss": 49.5405,
-      "step": 130070
-    },
-    {
-      "epoch": 0.5255396599021481,
-      "grad_norm": 648.6143188476562,
-      "learning_rate": 2.7585437666706087e-05,
-      "loss": 69.9888,
-      "step": 130080
-    },
-    {
-      "epoch": 0.5255800611675158,
-      "grad_norm": 612.1096801757812,
-      "learning_rate": 2.758196569980226e-05,
-      "loss": 51.9348,
-      "step": 130090
-    },
-    {
-      "epoch": 0.5256204624328834,
-      "grad_norm": 1045.827392578125,
-      "learning_rate": 2.7578493682561685e-05,
-      "loss": 75.3227,
-      "step": 130100
-    },
-    {
-      "epoch": 0.525660863698251,
-      "grad_norm": 990.9442138671875,
-      "learning_rate": 2.757502161505205e-05,
-      "loss": 73.4528,
-      "step": 130110
-    },
-    {
-      "epoch": 0.5257012649636187,
-      "grad_norm": 510.2374572753906,
-      "learning_rate": 2.7571549497341042e-05,
-      "loss": 60.0831,
-      "step": 130120
-    },
-    {
-      "epoch": 0.5257416662289863,
-      "grad_norm": 1088.753662109375,
-      "learning_rate": 2.756807732949635e-05,
-      "loss": 63.8334,
-      "step": 130130
-    },
-    {
-      "epoch": 0.525782067494354,
-      "grad_norm": 1749.695068359375,
-      "learning_rate": 2.756460511158567e-05,
-      "loss": 54.9062,
-      "step": 130140
-    },
-    {
-      "epoch": 0.5258224687597216,
-      "grad_norm": 992.9537963867188,
-      "learning_rate": 2.756113284367669e-05,
-      "loss": 55.7644,
-      "step": 130150
-    },
-    {
-      "epoch": 0.5258628700250891,
-      "grad_norm": 679.5419311523438,
-      "learning_rate": 2.7557660525837108e-05,
-      "loss": 51.0159,
-      "step": 130160
-    },
-    {
-      "epoch": 0.5259032712904568,
-      "grad_norm": 444.13031005859375,
-      "learning_rate": 2.7554188158134616e-05,
-      "loss": 19.1531,
-      "step": 130170
-    },
-    {
-      "epoch": 0.5259436725558244,
-      "grad_norm": 416.7135314941406,
-      "learning_rate": 2.7550715740636917e-05,
-      "loss": 46.4561,
-      "step": 130180
-    },
-    {
-      "epoch": 0.525984073821192,
-      "grad_norm": 933.5376586914062,
-      "learning_rate": 2.7547243273411695e-05,
-      "loss": 66.3492,
-      "step": 130190
-    },
-    {
-      "epoch": 0.5260244750865597,
-      "grad_norm": 444.9652404785156,
-      "learning_rate": 2.754377075652666e-05,
-      "loss": 50.9244,
-      "step": 130200
-    },
-    {
-      "epoch": 0.5260648763519273,
-      "grad_norm": 652.8956298828125,
-      "learning_rate": 2.7540298190049503e-05,
-      "loss": 59.4898,
-      "step": 130210
-    },
-    {
-      "epoch": 0.526105277617295,
-      "grad_norm": 645.0548095703125,
-      "learning_rate": 2.7536825574047925e-05,
-      "loss": 35.8312,
-      "step": 130220
-    },
-    {
-      "epoch": 0.5261456788826626,
-      "grad_norm": 534.1641845703125,
-      "learning_rate": 2.7533352908589622e-05,
-      "loss": 40.8292,
-      "step": 130230
-    },
-    {
-      "epoch": 0.5261860801480303,
-      "grad_norm": 420.173583984375,
-      "learning_rate": 2.7529880193742297e-05,
-      "loss": 30.2268,
-      "step": 130240
-    },
-    {
-      "epoch": 0.5262264814133979,
-      "grad_norm": 465.4990234375,
-      "learning_rate": 2.7526407429573657e-05,
-      "loss": 64.3663,
-      "step": 130250
-    },
-    {
-      "epoch": 0.5262668826787655,
-      "grad_norm": 511.3827819824219,
-      "learning_rate": 2.7522934616151414e-05,
-      "loss": 56.9852,
-      "step": 130260
-    },
-    {
-      "epoch": 0.5263072839441332,
-      "grad_norm": 624.65087890625,
-      "learning_rate": 2.751946175354325e-05,
-      "loss": 63.1493,
-      "step": 130270
-    },
-    {
-      "epoch": 0.5263476852095008,
-      "grad_norm": 751.966064453125,
-      "learning_rate": 2.7515988841816887e-05,
-      "loss": 69.2688,
-      "step": 130280
-    },
-    {
-      "epoch": 0.5263880864748683,
-      "grad_norm": 907.3563842773438,
-      "learning_rate": 2.7512515881040028e-05,
-      "loss": 54.0088,
-      "step": 130290
-    },
-    {
-      "epoch": 0.526428487740236,
-      "grad_norm": 893.8932495117188,
-      "learning_rate": 2.7509042871280372e-05,
-      "loss": 54.6191,
-      "step": 130300
-    },
-    {
-      "epoch": 0.5264688890056036,
-      "grad_norm": 1132.3380126953125,
-      "learning_rate": 2.750556981260564e-05,
-      "loss": 51.4256,
-      "step": 130310
-    },
-    {
-      "epoch": 0.5265092902709713,
-      "grad_norm": 408.7311096191406,
-      "learning_rate": 2.7502096705083535e-05,
-      "loss": 44.0828,
-      "step": 130320
-    },
-    {
-      "epoch": 0.5265496915363389,
-      "grad_norm": 569.1919555664062,
-      "learning_rate": 2.749862354878176e-05,
-      "loss": 34.8458,
-      "step": 130330
-    },
-    {
-      "epoch": 0.5265900928017065,
-      "grad_norm": 759.1968383789062,
-      "learning_rate": 2.7495150343768034e-05,
-      "loss": 59.1251,
-      "step": 130340
-    },
-    {
-      "epoch": 0.5266304940670742,
-      "grad_norm": 563.669677734375,
-      "learning_rate": 2.7491677090110076e-05,
-      "loss": 48.0987,
-      "step": 130350
-    },
-    {
-      "epoch": 0.5266708953324418,
-      "grad_norm": 1401.8712158203125,
-      "learning_rate": 2.7488203787875577e-05,
-      "loss": 67.7033,
-      "step": 130360
-    },
-    {
-      "epoch": 0.5267112965978095,
-      "grad_norm": 496.8099060058594,
-      "learning_rate": 2.7484730437132278e-05,
-      "loss": 44.0139,
-      "step": 130370
-    },
-    {
-      "epoch": 0.5267516978631771,
-      "grad_norm": 533.447998046875,
-      "learning_rate": 2.7481257037947872e-05,
-      "loss": 47.0373,
-      "step": 130380
-    },
-    {
-      "epoch": 0.5267920991285447,
-      "grad_norm": 825.5570068359375,
-      "learning_rate": 2.7477783590390082e-05,
-      "loss": 63.7042,
-      "step": 130390
-    },
-    {
-      "epoch": 0.5268325003939124,
-      "grad_norm": 577.7359619140625,
-      "learning_rate": 2.747431009452663e-05,
-      "loss": 43.8473,
-      "step": 130400
-    },
-    {
-      "epoch": 0.52687290165928,
-      "grad_norm": 2243.215087890625,
-      "learning_rate": 2.747083655042522e-05,
-      "loss": 58.3573,
-      "step": 130410
-    },
-    {
-      "epoch": 0.5269133029246476,
-      "grad_norm": 897.6618041992188,
-      "learning_rate": 2.7467362958153587e-05,
-      "loss": 85.0911,
-      "step": 130420
-    },
-    {
-      "epoch": 0.5269537041900152,
-      "grad_norm": 976.666259765625,
-      "learning_rate": 2.7463889317779446e-05,
-      "loss": 55.5755,
-      "step": 130430
-    },
-    {
-      "epoch": 0.5269941054553828,
-      "grad_norm": 1133.8804931640625,
-      "learning_rate": 2.7460415629370508e-05,
-      "loss": 59.8554,
-      "step": 130440
-    },
-    {
-      "epoch": 0.5270345067207505,
-      "grad_norm": 757.4534912109375,
-      "learning_rate": 2.7456941892994497e-05,
-      "loss": 43.9233,
-      "step": 130450
-    },
-    {
-      "epoch": 0.5270749079861181,
-      "grad_norm": 962.9010009765625,
-      "learning_rate": 2.7453468108719145e-05,
-      "loss": 65.6795,
-      "step": 130460
-    },
-    {
-      "epoch": 0.5271153092514858,
-      "grad_norm": 394.7574157714844,
-      "learning_rate": 2.744999427661217e-05,
-      "loss": 69.4007,
-      "step": 130470
-    },
-    {
-      "epoch": 0.5271557105168534,
-      "grad_norm": 762.9472045898438,
-      "learning_rate": 2.744652039674129e-05,
-      "loss": 49.1858,
-      "step": 130480
-    },
-    {
-      "epoch": 0.527196111782221,
-      "grad_norm": 2199.121337890625,
-      "learning_rate": 2.7443046469174237e-05,
-      "loss": 77.0622,
-      "step": 130490
-    },
-    {
-      "epoch": 0.5272365130475887,
-      "grad_norm": 370.2516784667969,
-      "learning_rate": 2.7439572493978736e-05,
-      "loss": 51.3727,
-      "step": 130500
-    },
-    {
-      "epoch": 0.5272769143129563,
-      "grad_norm": 1177.574951171875,
-      "learning_rate": 2.7436098471222522e-05,
-      "loss": 47.053,
-      "step": 130510
-    },
-    {
-      "epoch": 0.527317315578324,
-      "grad_norm": 1359.2659912109375,
-      "learning_rate": 2.74326244009733e-05,
-      "loss": 61.0058,
-      "step": 130520
-    },
-    {
-      "epoch": 0.5273577168436916,
-      "grad_norm": 1003.060546875,
-      "learning_rate": 2.7429150283298817e-05,
-      "loss": 52.1394,
-      "step": 130530
-    },
-    {
-      "epoch": 0.5273981181090591,
-      "grad_norm": 1656.4512939453125,
-      "learning_rate": 2.7425676118266808e-05,
-      "loss": 46.5512,
-      "step": 130540
-    },
-    {
-      "epoch": 0.5274385193744268,
-      "grad_norm": 509.01220703125,
-      "learning_rate": 2.7422201905944982e-05,
-      "loss": 52.2351,
-      "step": 130550
-    },
-    {
-      "epoch": 0.5274789206397944,
-      "grad_norm": 1386.0987548828125,
-      "learning_rate": 2.7418727646401094e-05,
-      "loss": 44.8062,
-      "step": 130560
-    },
-    {
-      "epoch": 0.527519321905162,
-      "grad_norm": 708.5176391601562,
-      "learning_rate": 2.741525333970285e-05,
-      "loss": 36.5751,
-      "step": 130570
-    },
-    {
-      "epoch": 0.5275597231705297,
-      "grad_norm": 3067.193115234375,
-      "learning_rate": 2.7411778985918006e-05,
-      "loss": 98.0832,
-      "step": 130580
-    },
-    {
-      "epoch": 0.5276001244358973,
-      "grad_norm": 598.675537109375,
-      "learning_rate": 2.7408304585114298e-05,
-      "loss": 60.4569,
-      "step": 130590
-    },
-    {
-      "epoch": 0.527640525701265,
-      "grad_norm": 821.6907348632812,
-      "learning_rate": 2.7404830137359444e-05,
-      "loss": 71.1968,
-      "step": 130600
-    },
-    {
-      "epoch": 0.5276809269666326,
-      "grad_norm": 796.5547485351562,
-      "learning_rate": 2.740135564272119e-05,
-      "loss": 64.8786,
-      "step": 130610
-    },
-    {
-      "epoch": 0.5277213282320002,
-      "grad_norm": 587.248291015625,
-      "learning_rate": 2.7397881101267263e-05,
-      "loss": 61.1327,
-      "step": 130620
-    },
-    {
-      "epoch": 0.5277617294973679,
-      "grad_norm": 603.5186157226562,
-      "learning_rate": 2.7394406513065423e-05,
-      "loss": 54.6609,
-      "step": 130630
-    },
-    {
-      "epoch": 0.5278021307627355,
-      "grad_norm": 1664.0540771484375,
-      "learning_rate": 2.739093187818339e-05,
-      "loss": 69.9045,
-      "step": 130640
-    },
-    {
-      "epoch": 0.5278425320281032,
-      "grad_norm": 847.181640625,
-      "learning_rate": 2.7387457196688908e-05,
-      "loss": 60.1619,
-      "step": 130650
-    },
-    {
-      "epoch": 0.5278829332934708,
-      "grad_norm": 1435.385498046875,
-      "learning_rate": 2.7383982468649714e-05,
-      "loss": 48.167,
-      "step": 130660
-    },
-    {
-      "epoch": 0.5279233345588383,
-      "grad_norm": 884.9189453125,
-      "learning_rate": 2.738050769413357e-05,
-      "loss": 57.8182,
-      "step": 130670
-    },
-    {
-      "epoch": 0.527963735824206,
-      "grad_norm": 568.2256469726562,
-      "learning_rate": 2.7377032873208186e-05,
-      "loss": 66.162,
-      "step": 130680
-    },
-    {
-      "epoch": 0.5280041370895736,
-      "grad_norm": 744.9805908203125,
-      "learning_rate": 2.737355800594133e-05,
-      "loss": 60.4696,
-      "step": 130690
-    },
-    {
-      "epoch": 0.5280445383549413,
-      "grad_norm": 907.050537109375,
-      "learning_rate": 2.7370083092400735e-05,
-      "loss": 43.196,
-      "step": 130700
-    },
-    {
-      "epoch": 0.5280849396203089,
-      "grad_norm": 778.4954833984375,
-      "learning_rate": 2.7366608132654154e-05,
-      "loss": 51.7738,
-      "step": 130710
-    },
-    {
-      "epoch": 0.5281253408856765,
-      "grad_norm": 405.15338134765625,
-      "learning_rate": 2.7363133126769325e-05,
-      "loss": 64.0481,
-      "step": 130720
-    },
-    {
-      "epoch": 0.5281657421510442,
-      "grad_norm": 591.9571533203125,
-      "learning_rate": 2.735965807481401e-05,
-      "loss": 45.1595,
-      "step": 130730
-    },
-    {
-      "epoch": 0.5282061434164118,
-      "grad_norm": 985.9591064453125,
-      "learning_rate": 2.7356182976855934e-05,
-      "loss": 35.7408,
-      "step": 130740
-    },
-    {
-      "epoch": 0.5282465446817795,
-      "grad_norm": 460.90887451171875,
-      "learning_rate": 2.7352707832962865e-05,
-      "loss": 52.3368,
-      "step": 130750
-    },
-    {
-      "epoch": 0.5282869459471471,
-      "grad_norm": 900.5870971679688,
-      "learning_rate": 2.734923264320254e-05,
-      "loss": 31.7685,
-      "step": 130760
-    },
-    {
-      "epoch": 0.5283273472125147,
-      "grad_norm": 851.9434814453125,
-      "learning_rate": 2.7345757407642714e-05,
-      "loss": 41.7662,
-      "step": 130770
-    },
-    {
-      "epoch": 0.5283677484778824,
-      "grad_norm": 2293.946533203125,
-      "learning_rate": 2.7342282126351144e-05,
-      "loss": 96.9833,
-      "step": 130780
-    },
-    {
-      "epoch": 0.52840814974325,
-      "grad_norm": 837.9186401367188,
-      "learning_rate": 2.7338806799395577e-05,
-      "loss": 53.9366,
-      "step": 130790
-    },
-    {
-      "epoch": 0.5284485510086175,
-      "grad_norm": 314.82135009765625,
-      "learning_rate": 2.733533142684377e-05,
-      "loss": 56.3716,
-      "step": 130800
-    },
-    {
-      "epoch": 0.5284889522739852,
-      "grad_norm": 918.69287109375,
-      "learning_rate": 2.7331856008763472e-05,
-      "loss": 50.4389,
-      "step": 130810
-    },
-    {
-      "epoch": 0.5285293535393528,
-      "grad_norm": 775.286865234375,
-      "learning_rate": 2.7328380545222436e-05,
-      "loss": 64.7548,
-      "step": 130820
-    },
-    {
-      "epoch": 0.5285697548047205,
-      "grad_norm": 461.6344909667969,
-      "learning_rate": 2.732490503628843e-05,
-      "loss": 53.2063,
-      "step": 130830
-    },
-    {
-      "epoch": 0.5286101560700881,
-      "grad_norm": 415.2874755859375,
-      "learning_rate": 2.73214294820292e-05,
-      "loss": 94.1432,
-      "step": 130840
-    },
-    {
-      "epoch": 0.5286505573354557,
-      "grad_norm": 718.537109375,
-      "learning_rate": 2.7317953882512504e-05,
-      "loss": 47.0089,
-      "step": 130850
-    },
-    {
-      "epoch": 0.5286909586008234,
-      "grad_norm": 524.0968017578125,
-      "learning_rate": 2.7314478237806107e-05,
-      "loss": 50.3915,
-      "step": 130860
-    },
-    {
-      "epoch": 0.528731359866191,
-      "grad_norm": 685.7463989257812,
-      "learning_rate": 2.7311002547977766e-05,
-      "loss": 47.7177,
-      "step": 130870
-    },
-    {
-      "epoch": 0.5287717611315587,
-      "grad_norm": 658.6242065429688,
-      "learning_rate": 2.730752681309524e-05,
-      "loss": 44.9115,
-      "step": 130880
-    },
-    {
-      "epoch": 0.5288121623969263,
-      "grad_norm": 831.5072021484375,
-      "learning_rate": 2.730405103322629e-05,
-      "loss": 36.1585,
-      "step": 130890
-    },
-    {
-      "epoch": 0.5288525636622939,
-      "grad_norm": 2113.5791015625,
-      "learning_rate": 2.7300575208438683e-05,
-      "loss": 64.7953,
-      "step": 130900
-    },
-    {
-      "epoch": 0.5288929649276616,
-      "grad_norm": 1694.7403564453125,
-      "learning_rate": 2.729709933880017e-05,
-      "loss": 64.9504,
-      "step": 130910
-    },
-    {
-      "epoch": 0.5289333661930292,
-      "grad_norm": 959.0403442382812,
-      "learning_rate": 2.7293623424378535e-05,
-      "loss": 62.3716,
-      "step": 130920
-    },
-    {
-      "epoch": 0.5289737674583967,
-      "grad_norm": 2491.924072265625,
-      "learning_rate": 2.7290147465241517e-05,
-      "loss": 52.1772,
-      "step": 130930
-    },
-    {
-      "epoch": 0.5290141687237644,
-      "grad_norm": 243.4221954345703,
-      "learning_rate": 2.7286671461456897e-05,
-      "loss": 38.3009,
-      "step": 130940
-    },
-    {
-      "epoch": 0.529054569989132,
-      "grad_norm": 1660.8577880859375,
-      "learning_rate": 2.7283195413092445e-05,
-      "loss": 60.126,
-      "step": 130950
-    },
-    {
-      "epoch": 0.5290949712544997,
-      "grad_norm": 609.1519165039062,
-      "learning_rate": 2.7279719320215924e-05,
-      "loss": 59.0399,
-      "step": 130960
-    },
-    {
-      "epoch": 0.5291353725198673,
-      "grad_norm": 379.8237609863281,
-      "learning_rate": 2.7276243182895094e-05,
-      "loss": 64.3225,
-      "step": 130970
-    },
-    {
-      "epoch": 0.529175773785235,
-      "grad_norm": 535.1643676757812,
-      "learning_rate": 2.7272767001197742e-05,
-      "loss": 37.2136,
-      "step": 130980
-    },
-    {
-      "epoch": 0.5292161750506026,
-      "grad_norm": 587.9727783203125,
-      "learning_rate": 2.726929077519162e-05,
-      "loss": 41.9734,
-      "step": 130990
-    },
-    {
-      "epoch": 0.5292565763159702,
-      "grad_norm": 652.3465576171875,
-      "learning_rate": 2.726581450494451e-05,
-      "loss": 83.3366,
-      "step": 131000
-    },
-    {
-      "epoch": 0.5292969775813379,
-      "grad_norm": 395.1925964355469,
-      "learning_rate": 2.7262338190524173e-05,
-      "loss": 70.538,
-      "step": 131010
-    },
-    {
-      "epoch": 0.5293373788467055,
-      "grad_norm": 602.4820556640625,
-      "learning_rate": 2.7258861831998388e-05,
-      "loss": 55.8564,
-      "step": 131020
-    },
-    {
-      "epoch": 0.5293777801120731,
-      "grad_norm": 2387.6884765625,
-      "learning_rate": 2.7255385429434932e-05,
-      "loss": 77.701,
-      "step": 131030
-    },
-    {
-      "epoch": 0.5294181813774408,
-      "grad_norm": 626.4151000976562,
-      "learning_rate": 2.725190898290158e-05,
-      "loss": 71.4265,
-      "step": 131040
-    },
-    {
-      "epoch": 0.5294585826428083,
-      "grad_norm": 1227.336669921875,
-      "learning_rate": 2.7248432492466096e-05,
-      "loss": 63.7168,
-      "step": 131050
-    },
-    {
-      "epoch": 0.529498983908176,
-      "grad_norm": 506.81475830078125,
-      "learning_rate": 2.7244955958196265e-05,
-      "loss": 38.2177,
-      "step": 131060
-    },
-    {
-      "epoch": 0.5295393851735436,
-      "grad_norm": 550.5950317382812,
-      "learning_rate": 2.7241479380159868e-05,
-      "loss": 69.3767,
-      "step": 131070
-    },
-    {
-      "epoch": 0.5295797864389112,
-      "grad_norm": 751.3677368164062,
-      "learning_rate": 2.723800275842468e-05,
-      "loss": 45.0376,
-      "step": 131080
-    },
-    {
-      "epoch": 0.5296201877042789,
-      "grad_norm": 459.6380310058594,
-      "learning_rate": 2.7234526093058464e-05,
-      "loss": 59.1021,
-      "step": 131090
-    },
-    {
-      "epoch": 0.5296605889696465,
-      "grad_norm": 463.3095703125,
-      "learning_rate": 2.7231049384129016e-05,
-      "loss": 37.5911,
-      "step": 131100
-    },
-    {
-      "epoch": 0.5297009902350142,
-      "grad_norm": 0.0,
-      "learning_rate": 2.7227572631704107e-05,
-      "loss": 52.0239,
-      "step": 131110
-    },
-    {
-      "epoch": 0.5297413915003818,
-      "grad_norm": 1287.0667724609375,
-      "learning_rate": 2.7224095835851525e-05,
-      "loss": 85.9998,
-      "step": 131120
-    },
-    {
-      "epoch": 0.5297817927657494,
-      "grad_norm": 710.130615234375,
-      "learning_rate": 2.722061899663905e-05,
-      "loss": 41.3392,
-      "step": 131130
-    },
-    {
-      "epoch": 0.5298221940311171,
-      "grad_norm": 241.11277770996094,
-      "learning_rate": 2.7217142114134463e-05,
-      "loss": 75.5654,
-      "step": 131140
-    },
-    {
-      "epoch": 0.5298625952964847,
-      "grad_norm": 2309.976806640625,
-      "learning_rate": 2.7213665188405556e-05,
-      "loss": 62.305,
-      "step": 131150
-    },
-    {
-      "epoch": 0.5299029965618524,
-      "grad_norm": 592.4625244140625,
-      "learning_rate": 2.721018821952011e-05,
-      "loss": 49.6338,
-      "step": 131160
-    },
-    {
-      "epoch": 0.52994339782722,
-      "grad_norm": 3443.605224609375,
-      "learning_rate": 2.7206711207545893e-05,
-      "loss": 65.5017,
-      "step": 131170
-    },
-    {
-      "epoch": 0.5299837990925875,
-      "grad_norm": 924.45654296875,
-      "learning_rate": 2.7203234152550712e-05,
-      "loss": 52.5658,
-      "step": 131180
-    },
-    {
-      "epoch": 0.5300242003579552,
-      "grad_norm": 385.39910888671875,
-      "learning_rate": 2.719975705460234e-05,
-      "loss": 35.2366,
-      "step": 131190
-    },
-    {
-      "epoch": 0.5300646016233228,
-      "grad_norm": 709.4951171875,
-      "learning_rate": 2.7196279913768584e-05,
-      "loss": 56.7616,
-      "step": 131200
-    },
-    {
-      "epoch": 0.5301050028886904,
-      "grad_norm": 849.27294921875,
-      "learning_rate": 2.719280273011721e-05,
-      "loss": 62.0372,
-      "step": 131210
-    },
-    {
-      "epoch": 0.5301454041540581,
-      "grad_norm": 943.2574462890625,
-      "learning_rate": 2.7189325503716022e-05,
-      "loss": 81.1754,
-      "step": 131220
-    },
-    {
-      "epoch": 0.5301858054194257,
-      "grad_norm": 839.4765014648438,
-      "learning_rate": 2.7185848234632803e-05,
-      "loss": 65.2041,
-      "step": 131230
-    },
-    {
-      "epoch": 0.5302262066847934,
-      "grad_norm": 879.865966796875,
-      "learning_rate": 2.7182370922935353e-05,
-      "loss": 57.2173,
-      "step": 131240
-    },
-    {
-      "epoch": 0.530266607950161,
-      "grad_norm": 980.9358520507812,
-      "learning_rate": 2.717889356869146e-05,
-      "loss": 47.6376,
-      "step": 131250
-    },
-    {
-      "epoch": 0.5303070092155286,
-      "grad_norm": 680.2994384765625,
-      "learning_rate": 2.717541617196891e-05,
-      "loss": 35.0096,
-      "step": 131260
-    },
-    {
-      "epoch": 0.5303474104808963,
-      "grad_norm": 870.2879638671875,
-      "learning_rate": 2.71719387328355e-05,
-      "loss": 54.6478,
-      "step": 131270
-    },
-    {
-      "epoch": 0.5303878117462639,
-      "grad_norm": 456.5863952636719,
-      "learning_rate": 2.716846125135903e-05,
-      "loss": 66.5173,
-      "step": 131280
-    },
-    {
-      "epoch": 0.5304282130116316,
-      "grad_norm": 1249.5938720703125,
-      "learning_rate": 2.716498372760729e-05,
-      "loss": 44.1854,
-      "step": 131290
-    },
-    {
-      "epoch": 0.5304686142769992,
-      "grad_norm": 364.5703430175781,
-      "learning_rate": 2.7161506161648076e-05,
-      "loss": 58.9766,
-      "step": 131300
-    },
-    {
-      "epoch": 0.5305090155423667,
-      "grad_norm": 890.297119140625,
-      "learning_rate": 2.7158028553549187e-05,
-      "loss": 56.6682,
-      "step": 131310
-    },
-    {
-      "epoch": 0.5305494168077344,
-      "grad_norm": 925.9620361328125,
-      "learning_rate": 2.715455090337842e-05,
-      "loss": 43.9666,
-      "step": 131320
-    },
-    {
-      "epoch": 0.530589818073102,
-      "grad_norm": 511.55609130859375,
-      "learning_rate": 2.715107321120358e-05,
-      "loss": 51.9629,
-      "step": 131330
-    },
-    {
-      "epoch": 0.5306302193384697,
-      "grad_norm": 547.1328125,
-      "learning_rate": 2.7147595477092457e-05,
-      "loss": 43.9675,
-      "step": 131340
-    },
-    {
-      "epoch": 0.5306706206038373,
-      "grad_norm": 784.9970703125,
-      "learning_rate": 2.7144117701112846e-05,
-      "loss": 65.4148,
-      "step": 131350
-    },
-    {
-      "epoch": 0.5307110218692049,
-      "grad_norm": 653.8527221679688,
-      "learning_rate": 2.7140639883332564e-05,
-      "loss": 64.9822,
-      "step": 131360
-    },
-    {
-      "epoch": 0.5307514231345726,
-      "grad_norm": 480.8045959472656,
-      "learning_rate": 2.713716202381941e-05,
-      "loss": 53.8297,
-      "step": 131370
-    },
-    {
-      "epoch": 0.5307918243999402,
-      "grad_norm": 251.93240356445312,
-      "learning_rate": 2.713368412264118e-05,
-      "loss": 55.8464,
-      "step": 131380
-    },
-    {
-      "epoch": 0.5308322256653079,
-      "grad_norm": 609.9501953125,
-      "learning_rate": 2.713020617986567e-05,
-      "loss": 38.5014,
-      "step": 131390
-    },
-    {
-      "epoch": 0.5308726269306755,
-      "grad_norm": 921.5916748046875,
-      "learning_rate": 2.7126728195560702e-05,
-      "loss": 52.8066,
-      "step": 131400
-    },
-    {
-      "epoch": 0.5309130281960431,
-      "grad_norm": 912.7449951171875,
-      "learning_rate": 2.7123250169794075e-05,
-      "loss": 54.2252,
-      "step": 131410
-    },
-    {
-      "epoch": 0.5309534294614108,
-      "grad_norm": 718.9556274414062,
-      "learning_rate": 2.711977210263359e-05,
-      "loss": 57.7856,
-      "step": 131420
-    },
-    {
-      "epoch": 0.5309938307267784,
-      "grad_norm": 457.2162170410156,
-      "learning_rate": 2.7116293994147053e-05,
-      "loss": 64.9568,
-      "step": 131430
-    },
-    {
-      "epoch": 0.5310342319921459,
-      "grad_norm": 621.4869995117188,
-      "learning_rate": 2.711281584440228e-05,
-      "loss": 62.9255,
-      "step": 131440
-    },
-    {
-      "epoch": 0.5310746332575136,
-      "grad_norm": 792.2979125976562,
-      "learning_rate": 2.710933765346707e-05,
-      "loss": 47.9132,
-      "step": 131450
-    },
-    {
-      "epoch": 0.5311150345228812,
-      "grad_norm": 539.8186645507812,
-      "learning_rate": 2.710585942140924e-05,
-      "loss": 54.2183,
-      "step": 131460
-    },
-    {
-      "epoch": 0.5311554357882489,
-      "grad_norm": 558.9265747070312,
-      "learning_rate": 2.710238114829659e-05,
-      "loss": 49.2721,
-      "step": 131470
-    },
-    {
-      "epoch": 0.5311958370536165,
-      "grad_norm": 709.4071655273438,
-      "learning_rate": 2.7098902834196943e-05,
-      "loss": 58.9437,
-      "step": 131480
-    },
-    {
-      "epoch": 0.5312362383189841,
-      "grad_norm": 654.2186889648438,
-      "learning_rate": 2.7095424479178106e-05,
-      "loss": 36.4761,
-      "step": 131490
-    },
-    {
-      "epoch": 0.5312766395843518,
-      "grad_norm": 889.7582397460938,
-      "learning_rate": 2.7091946083307896e-05,
-      "loss": 68.6755,
-      "step": 131500
-    },
-    {
-      "epoch": 0.5313170408497194,
-      "grad_norm": 1404.284912109375,
-      "learning_rate": 2.708846764665411e-05,
-      "loss": 67.0801,
-      "step": 131510
-    },
-    {
-      "epoch": 0.5313574421150871,
-      "grad_norm": 774.0541381835938,
-      "learning_rate": 2.7084989169284568e-05,
-      "loss": 70.2885,
-      "step": 131520
-    },
-    {
-      "epoch": 0.5313978433804547,
-      "grad_norm": 676.1631469726562,
-      "learning_rate": 2.70815106512671e-05,
-      "loss": 57.3858,
-      "step": 131530
-    },
-    {
-      "epoch": 0.5314382446458223,
-      "grad_norm": 288.2684631347656,
-      "learning_rate": 2.7078032092669502e-05,
-      "loss": 57.0647,
-      "step": 131540
-    },
-    {
-      "epoch": 0.53147864591119,
-      "grad_norm": 692.8833618164062,
-      "learning_rate": 2.70745534935596e-05,
-      "loss": 52.4784,
-      "step": 131550
-    },
-    {
-      "epoch": 0.5315190471765576,
-      "grad_norm": 360.534912109375,
-      "learning_rate": 2.707107485400521e-05,
-      "loss": 76.2931,
-      "step": 131560
-    },
-    {
-      "epoch": 0.5315594484419252,
-      "grad_norm": 902.7174682617188,
-      "learning_rate": 2.7067596174074155e-05,
-      "loss": 66.0437,
-      "step": 131570
-    },
-    {
-      "epoch": 0.5315998497072928,
-      "grad_norm": 1684.62451171875,
-      "learning_rate": 2.7064117453834243e-05,
-      "loss": 48.1203,
-      "step": 131580
-    },
-    {
-      "epoch": 0.5316402509726604,
-      "grad_norm": 74.62276458740234,
-      "learning_rate": 2.70606386933533e-05,
-      "loss": 69.9313,
-      "step": 131590
-    },
-    {
-      "epoch": 0.5316806522380281,
-      "grad_norm": 347.76605224609375,
-      "learning_rate": 2.705715989269914e-05,
-      "loss": 67.9091,
-      "step": 131600
-    },
-    {
-      "epoch": 0.5317210535033957,
-      "grad_norm": 2573.411865234375,
-      "learning_rate": 2.70536810519396e-05,
-      "loss": 58.865,
-      "step": 131610
-    },
-    {
-      "epoch": 0.5317614547687634,
-      "grad_norm": 265.0232849121094,
-      "learning_rate": 2.705020217114248e-05,
-      "loss": 47.0783,
-      "step": 131620
-    },
-    {
-      "epoch": 0.531801856034131,
-      "grad_norm": 801.228515625,
-      "learning_rate": 2.7046723250375617e-05,
-      "loss": 51.892,
-      "step": 131630
-    },
-    {
-      "epoch": 0.5318422572994986,
-      "grad_norm": 833.2240600585938,
-      "learning_rate": 2.7043244289706826e-05,
-      "loss": 49.0188,
-      "step": 131640
-    },
-    {
-      "epoch": 0.5318826585648663,
-      "grad_norm": 984.8556518554688,
-      "learning_rate": 2.7039765289203946e-05,
-      "loss": 55.2805,
-      "step": 131650
-    },
-    {
-      "epoch": 0.5319230598302339,
-      "grad_norm": 779.8922119140625,
-      "learning_rate": 2.703628624893478e-05,
-      "loss": 40.7662,
-      "step": 131660
-    },
-    {
-      "epoch": 0.5319634610956016,
-      "grad_norm": 687.851318359375,
-      "learning_rate": 2.703280716896717e-05,
-      "loss": 71.506,
-      "step": 131670
-    },
-    {
-      "epoch": 0.5320038623609692,
-      "grad_norm": 561.60888671875,
-      "learning_rate": 2.702932804936894e-05,
-      "loss": 76.0556,
-      "step": 131680
-    },
-    {
-      "epoch": 0.5320442636263367,
-      "grad_norm": 597.3526611328125,
-      "learning_rate": 2.7025848890207917e-05,
-      "loss": 51.0563,
-      "step": 131690
-    },
-    {
-      "epoch": 0.5320846648917044,
-      "grad_norm": 1162.3173828125,
-      "learning_rate": 2.7022369691551917e-05,
-      "loss": 67.7418,
-      "step": 131700
-    },
-    {
-      "epoch": 0.532125066157072,
-      "grad_norm": 1007.927734375,
-      "learning_rate": 2.7018890453468788e-05,
-      "loss": 55.7545,
-      "step": 131710
-    },
-    {
-      "epoch": 0.5321654674224396,
-      "grad_norm": 583.6637573242188,
-      "learning_rate": 2.7015411176026344e-05,
-      "loss": 59.1643,
-      "step": 131720
-    },
-    {
-      "epoch": 0.5322058686878073,
-      "grad_norm": 503.95025634765625,
-      "learning_rate": 2.7011931859292427e-05,
-      "loss": 57.7244,
-      "step": 131730
-    },
-    {
-      "epoch": 0.5322462699531749,
-      "grad_norm": 545.329345703125,
-      "learning_rate": 2.7008452503334858e-05,
-      "loss": 46.0308,
-      "step": 131740
-    },
-    {
-      "epoch": 0.5322866712185426,
-      "grad_norm": 380.2509765625,
-      "learning_rate": 2.7004973108221472e-05,
-      "loss": 40.5332,
-      "step": 131750
-    },
-    {
-      "epoch": 0.5323270724839102,
-      "grad_norm": 740.1156005859375,
-      "learning_rate": 2.700149367402011e-05,
-      "loss": 60.6691,
-      "step": 131760
-    },
-    {
-      "epoch": 0.5323674737492778,
-      "grad_norm": 1089.05712890625,
-      "learning_rate": 2.69980142007986e-05,
-      "loss": 65.5629,
-      "step": 131770
-    },
-    {
-      "epoch": 0.5324078750146455,
-      "grad_norm": 537.0314331054688,
-      "learning_rate": 2.699453468862477e-05,
-      "loss": 55.9601,
-      "step": 131780
-    },
-    {
-      "epoch": 0.5324482762800131,
-      "grad_norm": 266.5636291503906,
-      "learning_rate": 2.699105513756645e-05,
-      "loss": 59.2656,
-      "step": 131790
-    },
-    {
-      "epoch": 0.5324886775453808,
-      "grad_norm": 889.6873779296875,
-      "learning_rate": 2.6987575547691497e-05,
-      "loss": 76.0885,
-      "step": 131800
-    },
-    {
-      "epoch": 0.5325290788107484,
-      "grad_norm": 1049.7486572265625,
-      "learning_rate": 2.698409591906773e-05,
-      "loss": 50.9139,
-      "step": 131810
-    },
-    {
-      "epoch": 0.5325694800761159,
-      "grad_norm": 820.5618286132812,
-      "learning_rate": 2.6980616251762997e-05,
-      "loss": 36.7931,
-      "step": 131820
-    },
-    {
-      "epoch": 0.5326098813414836,
-      "grad_norm": 643.490966796875,
-      "learning_rate": 2.6977136545845122e-05,
-      "loss": 56.8342,
-      "step": 131830
-    },
-    {
-      "epoch": 0.5326502826068512,
-      "grad_norm": 853.44970703125,
-      "learning_rate": 2.6973656801381963e-05,
-      "loss": 60.0717,
-      "step": 131840
-    },
-    {
-      "epoch": 0.5326906838722189,
-      "grad_norm": 440.1711730957031,
-      "learning_rate": 2.697017701844134e-05,
-      "loss": 45.1691,
-      "step": 131850
-    },
-    {
-      "epoch": 0.5327310851375865,
-      "grad_norm": 468.0743408203125,
-      "learning_rate": 2.6966697197091108e-05,
-      "loss": 54.0596,
-      "step": 131860
-    },
-    {
-      "epoch": 0.5327714864029541,
-      "grad_norm": 1050.264892578125,
-      "learning_rate": 2.69632173373991e-05,
-      "loss": 55.2322,
-      "step": 131870
-    },
-    {
-      "epoch": 0.5328118876683218,
-      "grad_norm": 720.6689453125,
-      "learning_rate": 2.695973743943315e-05,
-      "loss": 55.2848,
-      "step": 131880
-    },
-    {
-      "epoch": 0.5328522889336894,
-      "grad_norm": 1108.028564453125,
-      "learning_rate": 2.6956257503261116e-05,
-      "loss": 40.9585,
-      "step": 131890
-    },
-    {
-      "epoch": 0.532892690199057,
-      "grad_norm": 1092.5072021484375,
-      "learning_rate": 2.695277752895084e-05,
-      "loss": 44.4573,
-      "step": 131900
-    },
-    {
-      "epoch": 0.5329330914644247,
-      "grad_norm": 625.7965087890625,
-      "learning_rate": 2.6949297516570156e-05,
-      "loss": 28.5109,
-      "step": 131910
-    },
-    {
-      "epoch": 0.5329734927297923,
-      "grad_norm": 681.8804931640625,
-      "learning_rate": 2.6945817466186912e-05,
-      "loss": 43.6512,
-      "step": 131920
-    },
-    {
-      "epoch": 0.53301389399516,
-      "grad_norm": 756.3519897460938,
-      "learning_rate": 2.694233737786896e-05,
-      "loss": 76.8343,
-      "step": 131930
-    },
-    {
-      "epoch": 0.5330542952605276,
-      "grad_norm": 1092.2066650390625,
-      "learning_rate": 2.693885725168414e-05,
-      "loss": 46.8568,
-      "step": 131940
-    },
-    {
-      "epoch": 0.5330946965258951,
-      "grad_norm": 988.5387573242188,
-      "learning_rate": 2.6935377087700297e-05,
-      "loss": 73.0625,
-      "step": 131950
-    },
-    {
-      "epoch": 0.5331350977912628,
-      "grad_norm": 940.03076171875,
-      "learning_rate": 2.693189688598528e-05,
-      "loss": 63.5997,
-      "step": 131960
-    },
-    {
-      "epoch": 0.5331754990566304,
-      "grad_norm": 372.99151611328125,
-      "learning_rate": 2.6928416646606936e-05,
-      "loss": 57.4221,
-      "step": 131970
-    },
-    {
-      "epoch": 0.5332159003219981,
-      "grad_norm": 1288.4371337890625,
-      "learning_rate": 2.6924936369633125e-05,
-      "loss": 48.3105,
-      "step": 131980
-    },
-    {
-      "epoch": 0.5332563015873657,
-      "grad_norm": 2535.275634765625,
-      "learning_rate": 2.6921456055131683e-05,
-      "loss": 67.3763,
-      "step": 131990
-    },
-    {
-      "epoch": 0.5332967028527333,
-      "grad_norm": 3868.3837890625,
-      "learning_rate": 2.6917975703170466e-05,
-      "loss": 65.5701,
-      "step": 132000
-    },
-    {
-      "epoch": 0.533337104118101,
-      "grad_norm": 810.7564697265625,
-      "learning_rate": 2.691449531381733e-05,
-      "loss": 64.9525,
-      "step": 132010
-    },
-    {
-      "epoch": 0.5333775053834686,
-      "grad_norm": 1662.32373046875,
-      "learning_rate": 2.6911014887140122e-05,
-      "loss": 57.4445,
-      "step": 132020
-    },
-    {
-      "epoch": 0.5334179066488363,
-      "grad_norm": 528.4577026367188,
-      "learning_rate": 2.6907534423206692e-05,
-      "loss": 49.0796,
-      "step": 132030
-    },
-    {
-      "epoch": 0.5334583079142039,
-      "grad_norm": 798.1748657226562,
-      "learning_rate": 2.6904053922084895e-05,
-      "loss": 51.6413,
-      "step": 132040
-    },
-    {
-      "epoch": 0.5334987091795715,
-      "grad_norm": 589.7070922851562,
-      "learning_rate": 2.6900573383842583e-05,
-      "loss": 33.4644,
-      "step": 132050
-    },
-    {
-      "epoch": 0.5335391104449392,
-      "grad_norm": 757.95458984375,
-      "learning_rate": 2.689709280854762e-05,
-      "loss": 52.9057,
-      "step": 132060
-    },
-    {
-      "epoch": 0.5335795117103068,
-      "grad_norm": 567.9570922851562,
-      "learning_rate": 2.6893612196267853e-05,
-      "loss": 46.4571,
-      "step": 132070
-    },
-    {
-      "epoch": 0.5336199129756743,
-      "grad_norm": 732.6300048828125,
-      "learning_rate": 2.6890131547071147e-05,
-      "loss": 66.3549,
-      "step": 132080
-    },
-    {
-      "epoch": 0.533660314241042,
-      "grad_norm": 1564.1744384765625,
-      "learning_rate": 2.6886650861025343e-05,
-      "loss": 82.0941,
-      "step": 132090
-    },
-    {
-      "epoch": 0.5337007155064096,
-      "grad_norm": 6092.22314453125,
-      "learning_rate": 2.6883170138198323e-05,
-      "loss": 73.8271,
-      "step": 132100
-    },
-    {
-      "epoch": 0.5337411167717773,
-      "grad_norm": 3677.495361328125,
-      "learning_rate": 2.6879689378657923e-05,
-      "loss": 83.4729,
-      "step": 132110
-    },
-    {
-      "epoch": 0.5337815180371449,
-      "grad_norm": 830.318603515625,
-      "learning_rate": 2.6876208582472012e-05,
-      "loss": 43.1947,
-      "step": 132120
-    },
-    {
-      "epoch": 0.5338219193025125,
-      "grad_norm": 861.20849609375,
-      "learning_rate": 2.687272774970845e-05,
-      "loss": 79.1685,
-      "step": 132130
-    },
-    {
-      "epoch": 0.5338623205678802,
-      "grad_norm": 541.7510375976562,
-      "learning_rate": 2.6869246880435095e-05,
-      "loss": 50.6851,
-      "step": 132140
-    },
-    {
-      "epoch": 0.5339027218332478,
-      "grad_norm": 2233.982177734375,
-      "learning_rate": 2.686576597471981e-05,
-      "loss": 64.9713,
-      "step": 132150
-    },
-    {
-      "epoch": 0.5339431230986155,
-      "grad_norm": 801.9125366210938,
-      "learning_rate": 2.686228503263045e-05,
-      "loss": 67.6367,
-      "step": 132160
-    },
-    {
-      "epoch": 0.5339835243639831,
-      "grad_norm": 577.6140747070312,
-      "learning_rate": 2.685880405423489e-05,
-      "loss": 56.8992,
-      "step": 132170
-    },
-    {
-      "epoch": 0.5340239256293507,
-      "grad_norm": 303.9769592285156,
-      "learning_rate": 2.6855323039601e-05,
-      "loss": 55.4841,
-      "step": 132180
-    },
-    {
-      "epoch": 0.5340643268947184,
-      "grad_norm": 0.0,
-      "learning_rate": 2.685184198879662e-05,
-      "loss": 79.864,
-      "step": 132190
-    },
-    {
-      "epoch": 0.534104728160086,
-      "grad_norm": 1271.53759765625,
-      "learning_rate": 2.684836090188963e-05,
-      "loss": 53.1909,
-      "step": 132200
-    },
-    {
-      "epoch": 0.5341451294254536,
-      "grad_norm": 390.48504638671875,
-      "learning_rate": 2.6844879778947884e-05,
-      "loss": 45.1817,
-      "step": 132210
-    },
-    {
-      "epoch": 0.5341855306908212,
-      "grad_norm": 1045.8829345703125,
-      "learning_rate": 2.6841398620039273e-05,
-      "loss": 57.4035,
-      "step": 132220
-    },
-    {
-      "epoch": 0.5342259319561888,
-      "grad_norm": 211.4333038330078,
-      "learning_rate": 2.6837917425231633e-05,
-      "loss": 93.9413,
-      "step": 132230
-    },
-    {
-      "epoch": 0.5342663332215565,
-      "grad_norm": 362.2946472167969,
-      "learning_rate": 2.6834436194592853e-05,
-      "loss": 33.1701,
-      "step": 132240
-    },
-    {
-      "epoch": 0.5343067344869241,
-      "grad_norm": 875.4388427734375,
-      "learning_rate": 2.6830954928190794e-05,
-      "loss": 76.5484,
-      "step": 132250
-    },
-    {
-      "epoch": 0.5343471357522918,
-      "grad_norm": 546.877197265625,
-      "learning_rate": 2.682747362609333e-05,
-      "loss": 38.9788,
-      "step": 132260
-    },
-    {
-      "epoch": 0.5343875370176594,
-      "grad_norm": 674.6320190429688,
-      "learning_rate": 2.6823992288368322e-05,
-      "loss": 70.6211,
-      "step": 132270
-    },
-    {
-      "epoch": 0.534427938283027,
-      "grad_norm": 868.1038818359375,
-      "learning_rate": 2.6820510915083648e-05,
-      "loss": 49.9378,
-      "step": 132280
-    },
-    {
-      "epoch": 0.5344683395483947,
-      "grad_norm": 355.66741943359375,
-      "learning_rate": 2.681702950630717e-05,
-      "loss": 34.1495,
-      "step": 132290
-    },
-    {
-      "epoch": 0.5345087408137623,
-      "grad_norm": 460.1276550292969,
-      "learning_rate": 2.6813548062106775e-05,
-      "loss": 64.0979,
-      "step": 132300
-    },
-    {
-      "epoch": 0.53454914207913,
-      "grad_norm": 0.0,
-      "learning_rate": 2.6810066582550324e-05,
-      "loss": 59.0976,
-      "step": 132310
-    },
-    {
-      "epoch": 0.5345895433444976,
-      "grad_norm": 310.49627685546875,
-      "learning_rate": 2.6806585067705692e-05,
-      "loss": 64.1751,
-      "step": 132320
-    },
-    {
-      "epoch": 0.5346299446098651,
-      "grad_norm": 862.5602416992188,
-      "learning_rate": 2.680310351764075e-05,
-      "loss": 63.3662,
-      "step": 132330
-    },
-    {
-      "epoch": 0.5346703458752328,
-      "grad_norm": 802.3466186523438,
-      "learning_rate": 2.679962193242338e-05,
-      "loss": 71.8971,
-      "step": 132340
-    },
-    {
-      "epoch": 0.5347107471406004,
-      "grad_norm": 1650.000244140625,
-      "learning_rate": 2.6796140312121458e-05,
-      "loss": 59.4168,
-      "step": 132350
-    },
-    {
-      "epoch": 0.534751148405968,
-      "grad_norm": 832.1773071289062,
-      "learning_rate": 2.6792658656802856e-05,
-      "loss": 45.569,
-      "step": 132360
-    },
-    {
-      "epoch": 0.5347915496713357,
-      "grad_norm": 917.97412109375,
-      "learning_rate": 2.6789176966535444e-05,
-      "loss": 45.4459,
-      "step": 132370
-    },
-    {
-      "epoch": 0.5348319509367033,
-      "grad_norm": 583.6998901367188,
-      "learning_rate": 2.678569524138711e-05,
-      "loss": 66.7321,
-      "step": 132380
-    },
-    {
-      "epoch": 0.534872352202071,
-      "grad_norm": 418.53680419921875,
-      "learning_rate": 2.678221348142573e-05,
-      "loss": 34.0593,
-      "step": 132390
-    },
-    {
-      "epoch": 0.5349127534674386,
-      "grad_norm": 402.7685241699219,
-      "learning_rate": 2.6778731686719178e-05,
-      "loss": 55.3928,
-      "step": 132400
-    },
-    {
-      "epoch": 0.5349531547328062,
-      "grad_norm": 372.7942199707031,
-      "learning_rate": 2.6775249857335333e-05,
-      "loss": 30.0867,
-      "step": 132410
-    },
-    {
-      "epoch": 0.5349935559981739,
-      "grad_norm": 1119.2479248046875,
-      "learning_rate": 2.677176799334208e-05,
-      "loss": 58.895,
-      "step": 132420
-    },
-    {
-      "epoch": 0.5350339572635415,
-      "grad_norm": 407.2303771972656,
-      "learning_rate": 2.6768286094807298e-05,
-      "loss": 64.0234,
-      "step": 132430
-    },
-    {
-      "epoch": 0.5350743585289092,
-      "grad_norm": 855.4847412109375,
-      "learning_rate": 2.6764804161798867e-05,
-      "loss": 89.9496,
-      "step": 132440
-    },
-    {
-      "epoch": 0.5351147597942768,
-      "grad_norm": 1015.7791748046875,
-      "learning_rate": 2.6761322194384674e-05,
-      "loss": 60.1903,
-      "step": 132450
-    },
-    {
-      "epoch": 0.5351551610596443,
-      "grad_norm": 1254.9368896484375,
-      "learning_rate": 2.6757840192632598e-05,
-      "loss": 61.4563,
-      "step": 132460
-    },
-    {
-      "epoch": 0.535195562325012,
-      "grad_norm": 299.9359436035156,
-      "learning_rate": 2.6754358156610525e-05,
-      "loss": 38.3524,
-      "step": 132470
-    },
-    {
-      "epoch": 0.5352359635903796,
-      "grad_norm": 811.1012573242188,
-      "learning_rate": 2.6750876086386328e-05,
-      "loss": 47.9699,
-      "step": 132480
-    },
-    {
-      "epoch": 0.5352763648557473,
-      "grad_norm": 893.4951171875,
-      "learning_rate": 2.6747393982027903e-05,
-      "loss": 48.4848,
-      "step": 132490
-    },
-    {
-      "epoch": 0.5353167661211149,
-      "grad_norm": 727.3240966796875,
-      "learning_rate": 2.674391184360313e-05,
-      "loss": 34.6639,
-      "step": 132500
-    },
-    {
-      "epoch": 0.5353571673864825,
-      "grad_norm": 2415.583251953125,
-      "learning_rate": 2.6740429671179907e-05,
-      "loss": 73.8259,
-      "step": 132510
-    },
-    {
-      "epoch": 0.5353975686518502,
-      "grad_norm": 443.5478515625,
-      "learning_rate": 2.6736947464826108e-05,
-      "loss": 48.6447,
-      "step": 132520
-    },
-    {
-      "epoch": 0.5354379699172178,
-      "grad_norm": 1031.6845703125,
-      "learning_rate": 2.6733465224609622e-05,
-      "loss": 54.8037,
-      "step": 132530
-    },
-    {
-      "epoch": 0.5354783711825855,
-      "grad_norm": 564.192138671875,
-      "learning_rate": 2.6729982950598338e-05,
-      "loss": 51.8397,
-      "step": 132540
-    },
-    {
-      "epoch": 0.5355187724479531,
-      "grad_norm": 472.5278625488281,
-      "learning_rate": 2.6726500642860154e-05,
-      "loss": 49.6563,
-      "step": 132550
-    },
-    {
-      "epoch": 0.5355591737133207,
-      "grad_norm": 551.7787475585938,
-      "learning_rate": 2.6723018301462937e-05,
-      "loss": 43.9799,
-      "step": 132560
-    },
-    {
-      "epoch": 0.5355995749786884,
-      "grad_norm": 1193.7701416015625,
-      "learning_rate": 2.67195359264746e-05,
-      "loss": 64.9433,
-      "step": 132570
-    },
-    {
-      "epoch": 0.535639976244056,
-      "grad_norm": 799.9347534179688,
-      "learning_rate": 2.671605351796302e-05,
-      "loss": 41.734,
-      "step": 132580
-    },
-    {
-      "epoch": 0.5356803775094235,
-      "grad_norm": 759.9945068359375,
-      "learning_rate": 2.67125710759961e-05,
-      "loss": 76.4194,
-      "step": 132590
-    },
-    {
-      "epoch": 0.5357207787747912,
-      "grad_norm": 647.0215454101562,
-      "learning_rate": 2.6709088600641717e-05,
-      "loss": 43.6254,
-      "step": 132600
-    },
-    {
-      "epoch": 0.5357611800401588,
-      "grad_norm": 720.9840698242188,
-      "learning_rate": 2.6705606091967778e-05,
-      "loss": 49.2521,
-      "step": 132610
-    },
-    {
-      "epoch": 0.5358015813055265,
-      "grad_norm": 1699.47216796875,
-      "learning_rate": 2.670212355004217e-05,
-      "loss": 48.1515,
-      "step": 132620
-    },
-    {
-      "epoch": 0.5358419825708941,
-      "grad_norm": 912.8600463867188,
-      "learning_rate": 2.6698640974932793e-05,
-      "loss": 60.5469,
-      "step": 132630
-    },
-    {
-      "epoch": 0.5358823838362617,
-      "grad_norm": 562.8499755859375,
-      "learning_rate": 2.6695158366707522e-05,
-      "loss": 53.9552,
-      "step": 132640
-    },
-    {
-      "epoch": 0.5359227851016294,
-      "grad_norm": 778.0060424804688,
-      "learning_rate": 2.6691675725434272e-05,
-      "loss": 52.4513,
-      "step": 132650
-    },
-    {
-      "epoch": 0.535963186366997,
-      "grad_norm": 773.4092407226562,
-      "learning_rate": 2.6688193051180933e-05,
-      "loss": 39.1972,
-      "step": 132660
-    },
-    {
-      "epoch": 0.5360035876323647,
-      "grad_norm": 608.3802490234375,
-      "learning_rate": 2.66847103440154e-05,
-      "loss": 37.7895,
-      "step": 132670
-    },
-    {
-      "epoch": 0.5360439888977323,
-      "grad_norm": 775.6356811523438,
-      "learning_rate": 2.6681227604005576e-05,
-      "loss": 41.3886,
-      "step": 132680
-    },
-    {
-      "epoch": 0.5360843901630999,
-      "grad_norm": 1359.913818359375,
-      "learning_rate": 2.6677744831219348e-05,
-      "loss": 69.0232,
-      "step": 132690
-    },
-    {
-      "epoch": 0.5361247914284676,
-      "grad_norm": 540.3826293945312,
-      "learning_rate": 2.6674262025724627e-05,
-      "loss": 32.8396,
-      "step": 132700
-    },
-    {
-      "epoch": 0.5361651926938352,
-      "grad_norm": 862.3062744140625,
-      "learning_rate": 2.66707791875893e-05,
-      "loss": 64.9376,
-      "step": 132710
-    },
-    {
-      "epoch": 0.5362055939592028,
-      "grad_norm": 557.792236328125,
-      "learning_rate": 2.666729631688128e-05,
-      "loss": 53.6906,
-      "step": 132720
-    },
-    {
-      "epoch": 0.5362459952245704,
-      "grad_norm": 711.6216430664062,
-      "learning_rate": 2.6663813413668455e-05,
-      "loss": 55.813,
-      "step": 132730
-    },
-    {
-      "epoch": 0.536286396489938,
-      "grad_norm": 795.2600708007812,
-      "learning_rate": 2.6660330478018726e-05,
-      "loss": 57.3366,
-      "step": 132740
-    },
-    {
-      "epoch": 0.5363267977553057,
-      "grad_norm": 375.18975830078125,
-      "learning_rate": 2.6656847510000012e-05,
-      "loss": 64.0476,
-      "step": 132750
-    },
-    {
-      "epoch": 0.5363671990206733,
-      "grad_norm": 765.218994140625,
-      "learning_rate": 2.6653364509680188e-05,
-      "loss": 52.8319,
-      "step": 132760
-    },
-    {
-      "epoch": 0.536407600286041,
-      "grad_norm": 609.1416015625,
-      "learning_rate": 2.664988147712718e-05,
-      "loss": 34.5201,
-      "step": 132770
-    },
-    {
-      "epoch": 0.5364480015514086,
-      "grad_norm": 629.1380615234375,
-      "learning_rate": 2.664639841240888e-05,
-      "loss": 30.9847,
-      "step": 132780
-    },
-    {
-      "epoch": 0.5364884028167762,
-      "grad_norm": 0.0,
-      "learning_rate": 2.6642915315593204e-05,
-      "loss": 41.8538,
-      "step": 132790
-    },
-    {
-      "epoch": 0.5365288040821439,
-      "grad_norm": 751.7651977539062,
-      "learning_rate": 2.6639432186748043e-05,
-      "loss": 53.989,
-      "step": 132800
-    },
-    {
-      "epoch": 0.5365692053475115,
-      "grad_norm": 829.0548095703125,
-      "learning_rate": 2.6635949025941303e-05,
-      "loss": 55.0998,
-      "step": 132810
-    },
-    {
-      "epoch": 0.5366096066128792,
-      "grad_norm": 1614.1934814453125,
-      "learning_rate": 2.6632465833240893e-05,
-      "loss": 67.0182,
-      "step": 132820
-    },
-    {
-      "epoch": 0.5366500078782468,
-      "grad_norm": 1317.6927490234375,
-      "learning_rate": 2.662898260871473e-05,
-      "loss": 61.6694,
-      "step": 132830
-    },
-    {
-      "epoch": 0.5366904091436144,
-      "grad_norm": 427.5072937011719,
-      "learning_rate": 2.662549935243071e-05,
-      "loss": 50.9423,
-      "step": 132840
-    },
-    {
-      "epoch": 0.536730810408982,
-      "grad_norm": 1174.3800048828125,
-      "learning_rate": 2.6622016064456738e-05,
-      "loss": 72.6584,
-      "step": 132850
-    },
-    {
-      "epoch": 0.5367712116743496,
-      "grad_norm": 379.9467468261719,
-      "learning_rate": 2.661853274486073e-05,
-      "loss": 64.3494,
-      "step": 132860
-    },
-    {
-      "epoch": 0.5368116129397172,
-      "grad_norm": 1121.682861328125,
-      "learning_rate": 2.6615049393710596e-05,
-      "loss": 87.2399,
-      "step": 132870
-    },
-    {
-      "epoch": 0.5368520142050849,
-      "grad_norm": 1015.9825439453125,
-      "learning_rate": 2.661156601107424e-05,
-      "loss": 76.0801,
-      "step": 132880
-    },
-    {
-      "epoch": 0.5368924154704525,
-      "grad_norm": 364.2757568359375,
-      "learning_rate": 2.660808259701958e-05,
-      "loss": 44.1075,
-      "step": 132890
-    },
-    {
-      "epoch": 0.5369328167358202,
-      "grad_norm": 350.0528259277344,
-      "learning_rate": 2.6604599151614513e-05,
-      "loss": 41.6351,
-      "step": 132900
-    },
-    {
-      "epoch": 0.5369732180011878,
-      "grad_norm": 742.4920654296875,
-      "learning_rate": 2.660111567492696e-05,
-      "loss": 47.2528,
-      "step": 132910
-    },
-    {
-      "epoch": 0.5370136192665554,
-      "grad_norm": 588.64599609375,
-      "learning_rate": 2.6597632167024843e-05,
-      "loss": 49.0769,
-      "step": 132920
-    },
-    {
-      "epoch": 0.5370540205319231,
-      "grad_norm": 519.329345703125,
-      "learning_rate": 2.6594148627976056e-05,
-      "loss": 27.8517,
-      "step": 132930
-    },
-    {
-      "epoch": 0.5370944217972907,
-      "grad_norm": 620.1698608398438,
-      "learning_rate": 2.659066505784852e-05,
-      "loss": 30.6305,
-      "step": 132940
-    },
-    {
-      "epoch": 0.5371348230626584,
-      "grad_norm": 615.6215209960938,
-      "learning_rate": 2.6587181456710153e-05,
-      "loss": 70.2314,
-      "step": 132950
-    },
-    {
-      "epoch": 0.537175224328026,
-      "grad_norm": 766.043212890625,
-      "learning_rate": 2.6583697824628868e-05,
-      "loss": 59.5871,
-      "step": 132960
-    },
-    {
-      "epoch": 0.5372156255933935,
-      "grad_norm": 719.8123168945312,
-      "learning_rate": 2.6580214161672577e-05,
-      "loss": 92.2357,
-      "step": 132970
-    },
-    {
-      "epoch": 0.5372560268587612,
-      "grad_norm": 0.0,
-      "learning_rate": 2.65767304679092e-05,
-      "loss": 56.5093,
-      "step": 132980
-    },
-    {
-      "epoch": 0.5372964281241288,
-      "grad_norm": 615.7906494140625,
-      "learning_rate": 2.6573246743406643e-05,
-      "loss": 82.8496,
-      "step": 132990
-    },
-    {
-      "epoch": 0.5373368293894965,
-      "grad_norm": 377.28985595703125,
-      "learning_rate": 2.656976298823284e-05,
-      "loss": 61.0799,
-      "step": 133000
-    },
-    {
-      "epoch": 0.5373772306548641,
-      "grad_norm": 839.7172241210938,
-      "learning_rate": 2.656627920245569e-05,
-      "loss": 57.0245,
-      "step": 133010
-    },
-    {
-      "epoch": 0.5374176319202317,
-      "grad_norm": 296.6985168457031,
-      "learning_rate": 2.6562795386143126e-05,
-      "loss": 83.4302,
-      "step": 133020
-    },
-    {
-      "epoch": 0.5374580331855994,
-      "grad_norm": 615.6727294921875,
-      "learning_rate": 2.6559311539363057e-05,
-      "loss": 83.2225,
-      "step": 133030
-    },
-    {
-      "epoch": 0.537498434450967,
-      "grad_norm": 518.0780029296875,
-      "learning_rate": 2.6555827662183414e-05,
-      "loss": 61.7653,
-      "step": 133040
-    },
-    {
-      "epoch": 0.5375388357163347,
-      "grad_norm": 585.662841796875,
-      "learning_rate": 2.6552343754672103e-05,
-      "loss": 30.4691,
-      "step": 133050
-    },
-    {
-      "epoch": 0.5375792369817023,
-      "grad_norm": 763.933349609375,
-      "learning_rate": 2.654885981689706e-05,
-      "loss": 47.9403,
-      "step": 133060
-    },
-    {
-      "epoch": 0.5376196382470699,
-      "grad_norm": 4562.193359375,
-      "learning_rate": 2.654537584892619e-05,
-      "loss": 94.3141,
-      "step": 133070
-    },
-    {
-      "epoch": 0.5376600395124376,
-      "grad_norm": 2078.553466796875,
-      "learning_rate": 2.6541891850827427e-05,
-      "loss": 59.5855,
-      "step": 133080
-    },
-    {
-      "epoch": 0.5377004407778052,
-      "grad_norm": 882.0210571289062,
-      "learning_rate": 2.653840782266868e-05,
-      "loss": 48.3455,
-      "step": 133090
-    },
-    {
-      "epoch": 0.5377408420431727,
-      "grad_norm": 643.2105712890625,
-      "learning_rate": 2.653492376451789e-05,
-      "loss": 50.6237,
-      "step": 133100
-    },
-    {
-      "epoch": 0.5377812433085404,
-      "grad_norm": 468.3760986328125,
-      "learning_rate": 2.6531439676442966e-05,
-      "loss": 50.7116,
-      "step": 133110
-    },
-    {
-      "epoch": 0.537821644573908,
-      "grad_norm": 763.0037841796875,
-      "learning_rate": 2.6527955558511842e-05,
-      "loss": 78.7773,
-      "step": 133120
-    },
-    {
-      "epoch": 0.5378620458392757,
-      "grad_norm": 289.5901184082031,
-      "learning_rate": 2.652447141079243e-05,
-      "loss": 46.9934,
-      "step": 133130
-    },
-    {
-      "epoch": 0.5379024471046433,
-      "grad_norm": 776.5145874023438,
-      "learning_rate": 2.6520987233352668e-05,
-      "loss": 65.4828,
-      "step": 133140
-    },
-    {
-      "epoch": 0.5379428483700109,
-      "grad_norm": 2137.632568359375,
-      "learning_rate": 2.6517503026260477e-05,
-      "loss": 51.402,
-      "step": 133150
-    },
-    {
-      "epoch": 0.5379832496353786,
-      "grad_norm": 1670.868896484375,
-      "learning_rate": 2.6514018789583784e-05,
-      "loss": 58.1956,
-      "step": 133160
-    },
-    {
-      "epoch": 0.5380236509007462,
-      "grad_norm": 451.3445739746094,
-      "learning_rate": 2.651053452339051e-05,
-      "loss": 47.8556,
-      "step": 133170
-    },
-    {
-      "epoch": 0.5380640521661139,
-      "grad_norm": 993.9606323242188,
-      "learning_rate": 2.650705022774859e-05,
-      "loss": 53.2049,
-      "step": 133180
-    },
-    {
-      "epoch": 0.5381044534314815,
-      "grad_norm": 836.609130859375,
-      "learning_rate": 2.6503565902725945e-05,
-      "loss": 43.5089,
-      "step": 133190
-    },
-    {
-      "epoch": 0.5381448546968491,
-      "grad_norm": 999.381103515625,
-      "learning_rate": 2.650008154839052e-05,
-      "loss": 61.5361,
-      "step": 133200
-    },
-    {
-      "epoch": 0.5381852559622168,
-      "grad_norm": 806.6781616210938,
-      "learning_rate": 2.6496597164810228e-05,
-      "loss": 82.2622,
-      "step": 133210
-    },
-    {
-      "epoch": 0.5382256572275844,
-      "grad_norm": 576.73974609375,
-      "learning_rate": 2.6493112752053e-05,
-      "loss": 32.6687,
-      "step": 133220
-    },
-    {
-      "epoch": 0.538266058492952,
-      "grad_norm": 362.6260986328125,
-      "learning_rate": 2.6489628310186777e-05,
-      "loss": 27.9188,
-      "step": 133230
-    },
-    {
-      "epoch": 0.5383064597583196,
-      "grad_norm": 422.7489013671875,
-      "learning_rate": 2.6486143839279487e-05,
-      "loss": 54.0248,
-      "step": 133240
-    },
-    {
-      "epoch": 0.5383468610236872,
-      "grad_norm": 578.5451049804688,
-      "learning_rate": 2.6482659339399045e-05,
-      "loss": 60.191,
-      "step": 133250
-    },
-    {
-      "epoch": 0.5383872622890549,
-      "grad_norm": 2537.281005859375,
-      "learning_rate": 2.64791748106134e-05,
-      "loss": 58.6423,
-      "step": 133260
-    },
-    {
-      "epoch": 0.5384276635544225,
-      "grad_norm": 537.8567504882812,
-      "learning_rate": 2.647569025299048e-05,
-      "loss": 39.9818,
-      "step": 133270
-    },
-    {
-      "epoch": 0.5384680648197901,
-      "grad_norm": 415.552001953125,
-      "learning_rate": 2.647220566659822e-05,
-      "loss": 41.9125,
-      "step": 133280
-    },
-    {
-      "epoch": 0.5385084660851578,
-      "grad_norm": 603.8895874023438,
-      "learning_rate": 2.6468721051504554e-05,
-      "loss": 74.081,
-      "step": 133290
-    },
-    {
-      "epoch": 0.5385488673505254,
-      "grad_norm": 918.3076171875,
-      "learning_rate": 2.646523640777741e-05,
-      "loss": 67.8838,
-      "step": 133300
-    },
-    {
-      "epoch": 0.5385892686158931,
-      "grad_norm": 644.3486938476562,
-      "learning_rate": 2.646175173548474e-05,
-      "loss": 39.718,
-      "step": 133310
-    },
-    {
-      "epoch": 0.5386296698812607,
-      "grad_norm": 1127.198486328125,
-      "learning_rate": 2.6458267034694463e-05,
-      "loss": 50.1396,
-      "step": 133320
-    },
-    {
-      "epoch": 0.5386700711466283,
-      "grad_norm": 1032.7508544921875,
-      "learning_rate": 2.645478230547451e-05,
-      "loss": 60.5964,
-      "step": 133330
-    },
-    {
-      "epoch": 0.538710472411996,
-      "grad_norm": 2308.70654296875,
-      "learning_rate": 2.6451297547892834e-05,
-      "loss": 86.4414,
-      "step": 133340
-    },
-    {
-      "epoch": 0.5387508736773636,
-      "grad_norm": 250.09999084472656,
-      "learning_rate": 2.644781276201736e-05,
-      "loss": 38.5809,
-      "step": 133350
-    },
-    {
-      "epoch": 0.5387912749427312,
-      "grad_norm": 892.7601318359375,
-      "learning_rate": 2.6444327947916036e-05,
-      "loss": 54.9328,
-      "step": 133360
-    },
-    {
-      "epoch": 0.5388316762080988,
-      "grad_norm": 742.9862670898438,
-      "learning_rate": 2.6440843105656793e-05,
-      "loss": 53.7666,
-      "step": 133370
-    },
-    {
-      "epoch": 0.5388720774734664,
-      "grad_norm": 580.2630004882812,
-      "learning_rate": 2.6437358235307576e-05,
-      "loss": 42.9191,
-      "step": 133380
-    },
-    {
-      "epoch": 0.5389124787388341,
-      "grad_norm": 657.9290161132812,
-      "learning_rate": 2.643387333693631e-05,
-      "loss": 48.353,
-      "step": 133390
-    },
-    {
-      "epoch": 0.5389528800042017,
-      "grad_norm": 278.8210144042969,
-      "learning_rate": 2.6430388410610955e-05,
-      "loss": 28.9319,
-      "step": 133400
-    },
-    {
-      "epoch": 0.5389932812695694,
-      "grad_norm": 4367.41015625,
-      "learning_rate": 2.6426903456399442e-05,
-      "loss": 80.4687,
-      "step": 133410
-    },
-    {
-      "epoch": 0.539033682534937,
-      "grad_norm": 192.4734649658203,
-      "learning_rate": 2.6423418474369704e-05,
-      "loss": 34.9511,
-      "step": 133420
-    },
-    {
-      "epoch": 0.5390740838003046,
-      "grad_norm": 2244.14892578125,
-      "learning_rate": 2.6419933464589695e-05,
-      "loss": 88.4328,
-      "step": 133430
-    },
-    {
-      "epoch": 0.5391144850656723,
-      "grad_norm": 520.8280639648438,
-      "learning_rate": 2.641644842712735e-05,
-      "loss": 60.3455,
-      "step": 133440
-    },
-    {
-      "epoch": 0.5391548863310399,
-      "grad_norm": 356.8963623046875,
-      "learning_rate": 2.6412963362050618e-05,
-      "loss": 42.0201,
-      "step": 133450
-    },
-    {
-      "epoch": 0.5391952875964076,
-      "grad_norm": 578.19873046875,
-      "learning_rate": 2.640947826942743e-05,
-      "loss": 80.3492,
-      "step": 133460
-    },
-    {
-      "epoch": 0.5392356888617752,
-      "grad_norm": 855.69677734375,
-      "learning_rate": 2.640599314932574e-05,
-      "loss": 64.0816,
-      "step": 133470
-    },
-    {
-      "epoch": 0.5392760901271428,
-      "grad_norm": 1241.7418212890625,
-      "learning_rate": 2.6402508001813496e-05,
-      "loss": 59.2113,
-      "step": 133480
-    },
-    {
-      "epoch": 0.5393164913925104,
-      "grad_norm": 245.85995483398438,
-      "learning_rate": 2.6399022826958635e-05,
-      "loss": 40.4469,
-      "step": 133490
-    },
-    {
-      "epoch": 0.539356892657878,
-      "grad_norm": 1086.880615234375,
-      "learning_rate": 2.6395537624829096e-05,
-      "loss": 47.158,
-      "step": 133500
-    },
-    {
-      "epoch": 0.5393972939232456,
-      "grad_norm": 1283.5257568359375,
-      "learning_rate": 2.639205239549284e-05,
-      "loss": 75.8677,
-      "step": 133510
-    },
-    {
-      "epoch": 0.5394376951886133,
-      "grad_norm": 973.451904296875,
-      "learning_rate": 2.63885671390178e-05,
-      "loss": 60.6037,
-      "step": 133520
-    },
-    {
-      "epoch": 0.5394780964539809,
-      "grad_norm": 743.983642578125,
-      "learning_rate": 2.6385081855471937e-05,
-      "loss": 46.7652,
-      "step": 133530
-    },
-    {
-      "epoch": 0.5395184977193486,
-      "grad_norm": 338.3520812988281,
-      "learning_rate": 2.638159654492318e-05,
-      "loss": 84.5076,
-      "step": 133540
-    },
-    {
-      "epoch": 0.5395588989847162,
-      "grad_norm": 460.87103271484375,
-      "learning_rate": 2.6378111207439494e-05,
-      "loss": 51.0171,
-      "step": 133550
-    },
-    {
-      "epoch": 0.5395993002500838,
-      "grad_norm": 0.0,
-      "learning_rate": 2.637462584308881e-05,
-      "loss": 49.7129,
-      "step": 133560
-    },
-    {
-      "epoch": 0.5396397015154515,
-      "grad_norm": 1022.263427734375,
-      "learning_rate": 2.6371140451939103e-05,
-      "loss": 69.8698,
-      "step": 133570
-    },
-    {
-      "epoch": 0.5396801027808191,
-      "grad_norm": 1827.470947265625,
-      "learning_rate": 2.6367655034058302e-05,
-      "loss": 79.5532,
-      "step": 133580
-    },
-    {
-      "epoch": 0.5397205040461868,
-      "grad_norm": 608.9415893554688,
-      "learning_rate": 2.6364169589514358e-05,
-      "loss": 61.6891,
-      "step": 133590
-    },
-    {
-      "epoch": 0.5397609053115544,
-      "grad_norm": 1445.496826171875,
-      "learning_rate": 2.636068411837523e-05,
-      "loss": 67.8332,
-      "step": 133600
-    },
-    {
-      "epoch": 0.5398013065769219,
-      "grad_norm": 633.453369140625,
-      "learning_rate": 2.6357198620708868e-05,
-      "loss": 41.0135,
-      "step": 133610
-    },
-    {
-      "epoch": 0.5398417078422896,
-      "grad_norm": 1123.755615234375,
-      "learning_rate": 2.635371309658321e-05,
-      "loss": 65.9337,
-      "step": 133620
-    },
-    {
-      "epoch": 0.5398821091076572,
-      "grad_norm": 412.05029296875,
-      "learning_rate": 2.6350227546066218e-05,
-      "loss": 57.8581,
-      "step": 133630
-    },
-    {
-      "epoch": 0.5399225103730249,
-      "grad_norm": 794.2384643554688,
-      "learning_rate": 2.634674196922585e-05,
-      "loss": 47.7789,
-      "step": 133640
-    },
-    {
-      "epoch": 0.5399629116383925,
-      "grad_norm": 839.2406616210938,
-      "learning_rate": 2.6343256366130054e-05,
-      "loss": 49.6203,
-      "step": 133650
-    },
-    {
-      "epoch": 0.5400033129037601,
-      "grad_norm": 474.2958068847656,
-      "learning_rate": 2.633977073684679e-05,
-      "loss": 73.9534,
-      "step": 133660
-    },
-    {
-      "epoch": 0.5400437141691278,
-      "grad_norm": 611.3136596679688,
-      "learning_rate": 2.6336285081443996e-05,
-      "loss": 90.5097,
-      "step": 133670
-    },
-    {
-      "epoch": 0.5400841154344954,
-      "grad_norm": 1800.31884765625,
-      "learning_rate": 2.633279939998964e-05,
-      "loss": 66.3581,
-      "step": 133680
-    },
-    {
-      "epoch": 0.540124516699863,
-      "grad_norm": 437.4395751953125,
-      "learning_rate": 2.6329313692551672e-05,
-      "loss": 54.8113,
-      "step": 133690
-    },
-    {
-      "epoch": 0.5401649179652307,
-      "grad_norm": 891.3457641601562,
-      "learning_rate": 2.6325827959198045e-05,
-      "loss": 59.0138,
-      "step": 133700
-    },
-    {
-      "epoch": 0.5402053192305983,
-      "grad_norm": 971.3360595703125,
-      "learning_rate": 2.6322342199996726e-05,
-      "loss": 42.0446,
-      "step": 133710
-    },
-    {
-      "epoch": 0.540245720495966,
-      "grad_norm": 1470.095703125,
-      "learning_rate": 2.6318856415015664e-05,
-      "loss": 67.5125,
-      "step": 133720
-    },
-    {
-      "epoch": 0.5402861217613336,
-      "grad_norm": 553.43408203125,
-      "learning_rate": 2.631537060432282e-05,
-      "loss": 45.4572,
-      "step": 133730
-    },
-    {
-      "epoch": 0.5403265230267011,
-      "grad_norm": 1151.4344482421875,
-      "learning_rate": 2.631188476798614e-05,
-      "loss": 41.0355,
-      "step": 133740
-    },
-    {
-      "epoch": 0.5403669242920688,
-      "grad_norm": 1145.021728515625,
-      "learning_rate": 2.63083989060736e-05,
-      "loss": 48.7941,
-      "step": 133750
-    },
-    {
-      "epoch": 0.5404073255574364,
-      "grad_norm": 640.4161987304688,
-      "learning_rate": 2.6304913018653144e-05,
-      "loss": 58.756,
-      "step": 133760
-    },
-    {
-      "epoch": 0.5404477268228041,
-      "grad_norm": 887.5857543945312,
-      "learning_rate": 2.630142710579274e-05,
-      "loss": 54.0097,
-      "step": 133770
-    },
-    {
-      "epoch": 0.5404881280881717,
-      "grad_norm": 817.3486328125,
-      "learning_rate": 2.6297941167560346e-05,
-      "loss": 45.1971,
-      "step": 133780
-    },
-    {
-      "epoch": 0.5405285293535393,
-      "grad_norm": 959.50732421875,
-      "learning_rate": 2.6294455204023915e-05,
-      "loss": 43.2953,
-      "step": 133790
-    },
-    {
-      "epoch": 0.540568930618907,
-      "grad_norm": 315.968017578125,
-      "learning_rate": 2.6290969215251416e-05,
-      "loss": 76.2952,
-      "step": 133800
-    },
-    {
-      "epoch": 0.5406093318842746,
-      "grad_norm": 1908.3179931640625,
-      "learning_rate": 2.628748320131081e-05,
-      "loss": 71.9718,
-      "step": 133810
-    },
-    {
-      "epoch": 0.5406497331496423,
-      "grad_norm": 869.3490600585938,
-      "learning_rate": 2.6283997162270052e-05,
-      "loss": 48.5477,
-      "step": 133820
-    },
-    {
-      "epoch": 0.5406901344150099,
-      "grad_norm": 1390.5584716796875,
-      "learning_rate": 2.6280511098197113e-05,
-      "loss": 57.5297,
-      "step": 133830
-    },
-    {
-      "epoch": 0.5407305356803775,
-      "grad_norm": 1634.8409423828125,
-      "learning_rate": 2.627702500915995e-05,
-      "loss": 56.8467,
-      "step": 133840
-    },
-    {
-      "epoch": 0.5407709369457452,
-      "grad_norm": 796.09716796875,
-      "learning_rate": 2.6273538895226522e-05,
-      "loss": 43.6418,
-      "step": 133850
-    },
-    {
-      "epoch": 0.5408113382111128,
-      "grad_norm": 676.2175903320312,
-      "learning_rate": 2.6270052756464803e-05,
-      "loss": 61.2231,
-      "step": 133860
-    },
-    {
-      "epoch": 0.5408517394764804,
-      "grad_norm": 843.5305786132812,
-      "learning_rate": 2.626656659294275e-05,
-      "loss": 39.5511,
-      "step": 133870
-    },
-    {
-      "epoch": 0.540892140741848,
-      "grad_norm": 584.4107055664062,
-      "learning_rate": 2.6263080404728325e-05,
-      "loss": 40.8211,
-      "step": 133880
-    },
-    {
-      "epoch": 0.5409325420072156,
-      "grad_norm": 617.6050415039062,
-      "learning_rate": 2.62595941918895e-05,
-      "loss": 46.0637,
-      "step": 133890
-    },
-    {
-      "epoch": 0.5409729432725833,
-      "grad_norm": 323.1057434082031,
-      "learning_rate": 2.6256107954494242e-05,
-      "loss": 59.272,
-      "step": 133900
-    },
-    {
-      "epoch": 0.5410133445379509,
-      "grad_norm": 534.6497802734375,
-      "learning_rate": 2.6252621692610507e-05,
-      "loss": 47.1891,
-      "step": 133910
-    },
-    {
-      "epoch": 0.5410537458033186,
-      "grad_norm": 1699.9815673828125,
-      "learning_rate": 2.6249135406306273e-05,
-      "loss": 34.6852,
-      "step": 133920
-    },
-    {
-      "epoch": 0.5410941470686862,
-      "grad_norm": 595.4637451171875,
-      "learning_rate": 2.6245649095649494e-05,
-      "loss": 54.051,
-      "step": 133930
-    },
-    {
-      "epoch": 0.5411345483340538,
-      "grad_norm": 538.2552490234375,
-      "learning_rate": 2.6242162760708154e-05,
-      "loss": 35.2021,
-      "step": 133940
-    },
-    {
-      "epoch": 0.5411749495994215,
-      "grad_norm": 1009.6656494140625,
-      "learning_rate": 2.6238676401550207e-05,
-      "loss": 40.208,
-      "step": 133950
-    },
-    {
-      "epoch": 0.5412153508647891,
-      "grad_norm": 941.228515625,
-      "learning_rate": 2.623519001824362e-05,
-      "loss": 48.1016,
-      "step": 133960
-    },
-    {
-      "epoch": 0.5412557521301568,
-      "grad_norm": 884.5780029296875,
-      "learning_rate": 2.6231703610856373e-05,
-      "loss": 57.1634,
-      "step": 133970
-    },
-    {
-      "epoch": 0.5412961533955244,
-      "grad_norm": 670.623291015625,
-      "learning_rate": 2.6228217179456433e-05,
-      "loss": 56.2344,
-      "step": 133980
-    },
-    {
-      "epoch": 0.541336554660892,
-      "grad_norm": 432.82928466796875,
-      "learning_rate": 2.6224730724111758e-05,
-      "loss": 37.2939,
-      "step": 133990
-    },
-    {
-      "epoch": 0.5413769559262596,
-      "grad_norm": 505.16387939453125,
-      "learning_rate": 2.6221244244890336e-05,
-      "loss": 44.5701,
-      "step": 134000
-    },
-    {
-      "epoch": 0.5414173571916272,
-      "grad_norm": 661.7396240234375,
-      "learning_rate": 2.6217757741860123e-05,
-      "loss": 52.0883,
-      "step": 134010
-    },
-    {
-      "epoch": 0.5414577584569948,
-      "grad_norm": 419.75286865234375,
-      "learning_rate": 2.6214271215089104e-05,
-      "loss": 51.607,
-      "step": 134020
-    },
-    {
-      "epoch": 0.5414981597223625,
-      "grad_norm": 303.2333984375,
-      "learning_rate": 2.621078466464523e-05,
-      "loss": 23.728,
-      "step": 134030
-    },
-    {
-      "epoch": 0.5415385609877301,
-      "grad_norm": 1355.118408203125,
-      "learning_rate": 2.6207298090596493e-05,
-      "loss": 61.6765,
-      "step": 134040
-    },
-    {
-      "epoch": 0.5415789622530978,
-      "grad_norm": 1755.49853515625,
-      "learning_rate": 2.6203811493010856e-05,
-      "loss": 44.694,
-      "step": 134050
-    },
-    {
-      "epoch": 0.5416193635184654,
-      "grad_norm": 396.1255798339844,
-      "learning_rate": 2.6200324871956295e-05,
-      "loss": 76.6947,
-      "step": 134060
-    },
-    {
-      "epoch": 0.541659764783833,
-      "grad_norm": 1570.49072265625,
-      "learning_rate": 2.619683822750078e-05,
-      "loss": 64.72,
-      "step": 134070
-    },
-    {
-      "epoch": 0.5417001660492007,
-      "grad_norm": 632.6025390625,
-      "learning_rate": 2.6193351559712292e-05,
-      "loss": 69.9561,
-      "step": 134080
-    },
-    {
-      "epoch": 0.5417405673145683,
-      "grad_norm": 579.8509521484375,
-      "learning_rate": 2.6189864868658803e-05,
-      "loss": 69.4796,
-      "step": 134090
-    },
-    {
-      "epoch": 0.541780968579936,
-      "grad_norm": 274.0668640136719,
-      "learning_rate": 2.6186378154408286e-05,
-      "loss": 46.6143,
-      "step": 134100
-    },
-    {
-      "epoch": 0.5418213698453036,
-      "grad_norm": 633.9044189453125,
-      "learning_rate": 2.618289141702871e-05,
-      "loss": 52.109,
-      "step": 134110
-    },
-    {
-      "epoch": 0.5418617711106711,
-      "grad_norm": 948.8427124023438,
-      "learning_rate": 2.6179404656588058e-05,
-      "loss": 71.229,
-      "step": 134120
-    },
-    {
-      "epoch": 0.5419021723760388,
-      "grad_norm": 1424.39208984375,
-      "learning_rate": 2.6175917873154303e-05,
-      "loss": 61.8993,
-      "step": 134130
-    },
-    {
-      "epoch": 0.5419425736414064,
-      "grad_norm": 464.7524108886719,
-      "learning_rate": 2.6172431066795428e-05,
-      "loss": 60.2696,
-      "step": 134140
-    },
-    {
-      "epoch": 0.541982974906774,
-      "grad_norm": 422.7701721191406,
-      "learning_rate": 2.6168944237579406e-05,
-      "loss": 57.3853,
-      "step": 134150
-    },
-    {
-      "epoch": 0.5420233761721417,
-      "grad_norm": 1085.177001953125,
-      "learning_rate": 2.616545738557421e-05,
-      "loss": 81.7053,
-      "step": 134160
-    },
-    {
-      "epoch": 0.5420637774375093,
-      "grad_norm": 518.3070068359375,
-      "learning_rate": 2.6161970510847826e-05,
-      "loss": 62.6892,
-      "step": 134170
-    },
-    {
-      "epoch": 0.542104178702877,
-      "grad_norm": 746.0762939453125,
-      "learning_rate": 2.6158483613468227e-05,
-      "loss": 48.352,
-      "step": 134180
-    },
-    {
-      "epoch": 0.5421445799682446,
-      "grad_norm": 437.72576904296875,
-      "learning_rate": 2.6154996693503396e-05,
-      "loss": 47.7975,
-      "step": 134190
-    },
-    {
-      "epoch": 0.5421849812336123,
-      "grad_norm": 1170.718505859375,
-      "learning_rate": 2.615150975102131e-05,
-      "loss": 53.6286,
-      "step": 134200
-    },
-    {
-      "epoch": 0.5422253824989799,
-      "grad_norm": 435.82586669921875,
-      "learning_rate": 2.6148022786089944e-05,
-      "loss": 47.8883,
-      "step": 134210
-    },
-    {
-      "epoch": 0.5422657837643475,
-      "grad_norm": 860.9259643554688,
-      "learning_rate": 2.6144535798777286e-05,
-      "loss": 53.4346,
-      "step": 134220
-    },
-    {
-      "epoch": 0.5423061850297152,
-      "grad_norm": 888.4954223632812,
-      "learning_rate": 2.6141048789151314e-05,
-      "loss": 52.0218,
-      "step": 134230
-    },
-    {
-      "epoch": 0.5423465862950828,
-      "grad_norm": 998.63623046875,
-      "learning_rate": 2.6137561757280003e-05,
-      "loss": 33.0164,
-      "step": 134240
-    },
-    {
-      "epoch": 0.5423869875604503,
-      "grad_norm": 950.3190307617188,
-      "learning_rate": 2.6134074703231344e-05,
-      "loss": 58.7313,
-      "step": 134250
-    },
-    {
-      "epoch": 0.542427388825818,
-      "grad_norm": 1152.3165283203125,
-      "learning_rate": 2.6130587627073315e-05,
-      "loss": 65.6751,
-      "step": 134260
-    },
-    {
-      "epoch": 0.5424677900911856,
-      "grad_norm": 518.1432495117188,
-      "learning_rate": 2.6127100528873904e-05,
-      "loss": 42.0353,
-      "step": 134270
-    },
-    {
-      "epoch": 0.5425081913565533,
-      "grad_norm": 858.6574096679688,
-      "learning_rate": 2.6123613408701082e-05,
-      "loss": 64.6581,
-      "step": 134280
-    },
-    {
-      "epoch": 0.5425485926219209,
-      "grad_norm": 709.0899658203125,
-      "learning_rate": 2.6120126266622836e-05,
-      "loss": 41.2389,
-      "step": 134290
-    },
-    {
-      "epoch": 0.5425889938872885,
-      "grad_norm": 1239.24365234375,
-      "learning_rate": 2.6116639102707156e-05,
-      "loss": 35.4557,
-      "step": 134300
-    },
-    {
-      "epoch": 0.5426293951526562,
-      "grad_norm": 486.78656005859375,
-      "learning_rate": 2.6113151917022018e-05,
-      "loss": 28.9147,
-      "step": 134310
-    },
-    {
-      "epoch": 0.5426697964180238,
-      "grad_norm": 784.9171142578125,
-      "learning_rate": 2.610966470963541e-05,
-      "loss": 31.0144,
-      "step": 134320
-    },
-    {
-      "epoch": 0.5427101976833915,
-      "grad_norm": 698.8657836914062,
-      "learning_rate": 2.6106177480615318e-05,
-      "loss": 67.4756,
-      "step": 134330
-    },
-    {
-      "epoch": 0.5427505989487591,
-      "grad_norm": 612.57958984375,
-      "learning_rate": 2.610269023002973e-05,
-      "loss": 34.029,
-      "step": 134340
-    },
-    {
-      "epoch": 0.5427910002141267,
-      "grad_norm": 658.6572265625,
-      "learning_rate": 2.6099202957946624e-05,
-      "loss": 44.2772,
-      "step": 134350
-    },
-    {
-      "epoch": 0.5428314014794944,
-      "grad_norm": 1145.214111328125,
-      "learning_rate": 2.6095715664433995e-05,
-      "loss": 62.3332,
-      "step": 134360
-    },
-    {
-      "epoch": 0.542871802744862,
-      "grad_norm": 735.1185913085938,
-      "learning_rate": 2.6092228349559822e-05,
-      "loss": 43.975,
-      "step": 134370
-    },
-    {
-      "epoch": 0.5429122040102295,
-      "grad_norm": 706.6088256835938,
-      "learning_rate": 2.6088741013392098e-05,
-      "loss": 65.3609,
-      "step": 134380
-    },
-    {
-      "epoch": 0.5429526052755972,
-      "grad_norm": 851.6122436523438,
-      "learning_rate": 2.6085253655998805e-05,
-      "loss": 52.3161,
-      "step": 134390
-    },
-    {
-      "epoch": 0.5429930065409648,
-      "grad_norm": 286.50323486328125,
-      "learning_rate": 2.6081766277447927e-05,
-      "loss": 57.4317,
-      "step": 134400
-    },
-    {
-      "epoch": 0.5430334078063325,
-      "grad_norm": 781.5896606445312,
-      "learning_rate": 2.6078278877807467e-05,
-      "loss": 39.7626,
-      "step": 134410
-    },
-    {
-      "epoch": 0.5430738090717001,
-      "grad_norm": 825.2921142578125,
-      "learning_rate": 2.6074791457145398e-05,
-      "loss": 48.5062,
-      "step": 134420
-    },
-    {
-      "epoch": 0.5431142103370677,
-      "grad_norm": 396.2430114746094,
-      "learning_rate": 2.6071304015529726e-05,
-      "loss": 53.2996,
-      "step": 134430
-    },
-    {
-      "epoch": 0.5431546116024354,
-      "grad_norm": 781.8477172851562,
-      "learning_rate": 2.606781655302843e-05,
-      "loss": 43.2636,
-      "step": 134440
-    },
-    {
-      "epoch": 0.543195012867803,
-      "grad_norm": 1481.7069091796875,
-      "learning_rate": 2.6064329069709493e-05,
-      "loss": 64.267,
-      "step": 134450
-    },
-    {
-      "epoch": 0.5432354141331707,
-      "grad_norm": 272.7846984863281,
-      "learning_rate": 2.6060841565640913e-05,
-      "loss": 41.5866,
-      "step": 134460
-    },
-    {
-      "epoch": 0.5432758153985383,
-      "grad_norm": 696.9764404296875,
-      "learning_rate": 2.6057354040890685e-05,
-      "loss": 53.8912,
-      "step": 134470
-    },
-    {
-      "epoch": 0.543316216663906,
-      "grad_norm": 426.2699279785156,
-      "learning_rate": 2.605386649552679e-05,
-      "loss": 39.0158,
-      "step": 134480
-    },
-    {
-      "epoch": 0.5433566179292736,
-      "grad_norm": 325.85888671875,
-      "learning_rate": 2.605037892961723e-05,
-      "loss": 55.8296,
-      "step": 134490
-    },
-    {
-      "epoch": 0.5433970191946412,
-      "grad_norm": 485.76611328125,
-      "learning_rate": 2.604689134322999e-05,
-      "loss": 85.8357,
-      "step": 134500
-    },
-    {
-      "epoch": 0.5434374204600088,
-      "grad_norm": 769.1482543945312,
-      "learning_rate": 2.604340373643307e-05,
-      "loss": 47.1949,
-      "step": 134510
-    },
-    {
-      "epoch": 0.5434778217253764,
-      "grad_norm": 2402.829833984375,
-      "learning_rate": 2.603991610929445e-05,
-      "loss": 55.1604,
-      "step": 134520
-    },
-    {
-      "epoch": 0.543518222990744,
-      "grad_norm": 572.4566040039062,
-      "learning_rate": 2.6036428461882133e-05,
-      "loss": 46.891,
-      "step": 134530
-    },
-    {
-      "epoch": 0.5435586242561117,
-      "grad_norm": 387.22119140625,
-      "learning_rate": 2.603294079426411e-05,
-      "loss": 61.2975,
-      "step": 134540
-    },
-    {
-      "epoch": 0.5435990255214793,
-      "grad_norm": 427.8232116699219,
-      "learning_rate": 2.602945310650838e-05,
-      "loss": 30.9293,
-      "step": 134550
-    },
-    {
-      "epoch": 0.543639426786847,
-      "grad_norm": 595.3522338867188,
-      "learning_rate": 2.6025965398682916e-05,
-      "loss": 42.5073,
-      "step": 134560
-    },
-    {
-      "epoch": 0.5436798280522146,
-      "grad_norm": 946.1700439453125,
-      "learning_rate": 2.602247767085574e-05,
-      "loss": 60.556,
-      "step": 134570
-    },
-    {
-      "epoch": 0.5437202293175822,
-      "grad_norm": 1055.9263916015625,
-      "learning_rate": 2.6018989923094828e-05,
-      "loss": 50.0367,
-      "step": 134580
-    },
-    {
-      "epoch": 0.5437606305829499,
-      "grad_norm": 1703.814453125,
-      "learning_rate": 2.6015502155468192e-05,
-      "loss": 41.7789,
-      "step": 134590
-    },
-    {
-      "epoch": 0.5438010318483175,
-      "grad_norm": 632.1807250976562,
-      "learning_rate": 2.6012014368043814e-05,
-      "loss": 72.0978,
-      "step": 134600
-    },
-    {
-      "epoch": 0.5438414331136852,
-      "grad_norm": 1356.0809326171875,
-      "learning_rate": 2.6008526560889694e-05,
-      "loss": 66.3761,
-      "step": 134610
-    },
-    {
-      "epoch": 0.5438818343790528,
-      "grad_norm": 691.0104370117188,
-      "learning_rate": 2.6005038734073833e-05,
-      "loss": 52.5915,
-      "step": 134620
-    },
-    {
-      "epoch": 0.5439222356444204,
-      "grad_norm": 687.1522216796875,
-      "learning_rate": 2.600155088766423e-05,
-      "loss": 40.2666,
-      "step": 134630
-    },
-    {
-      "epoch": 0.543962636909788,
-      "grad_norm": 682.845947265625,
-      "learning_rate": 2.5998063021728865e-05,
-      "loss": 49.6002,
-      "step": 134640
-    },
-    {
-      "epoch": 0.5440030381751556,
-      "grad_norm": 1438.562255859375,
-      "learning_rate": 2.599457513633575e-05,
-      "loss": 60.5318,
-      "step": 134650
-    },
-    {
-      "epoch": 0.5440434394405232,
-      "grad_norm": 388.2558288574219,
-      "learning_rate": 2.599108723155288e-05,
-      "loss": 37.3879,
-      "step": 134660
-    },
-    {
-      "epoch": 0.5440838407058909,
-      "grad_norm": 543.9114379882812,
-      "learning_rate": 2.5987599307448256e-05,
-      "loss": 63.803,
-      "step": 134670
-    },
-    {
-      "epoch": 0.5441242419712585,
-      "grad_norm": 2017.16796875,
-      "learning_rate": 2.5984111364089876e-05,
-      "loss": 64.5238,
-      "step": 134680
-    },
-    {
-      "epoch": 0.5441646432366262,
-      "grad_norm": 572.1366577148438,
-      "learning_rate": 2.598062340154574e-05,
-      "loss": 50.1777,
-      "step": 134690
-    },
-    {
-      "epoch": 0.5442050445019938,
-      "grad_norm": 289.9861145019531,
-      "learning_rate": 2.5977135419883842e-05,
-      "loss": 53.5399,
-      "step": 134700
-    },
-    {
-      "epoch": 0.5442454457673614,
-      "grad_norm": 889.2007446289062,
-      "learning_rate": 2.597364741917219e-05,
-      "loss": 72.6208,
-      "step": 134710
-    },
-    {
-      "epoch": 0.5442858470327291,
-      "grad_norm": 1161.2286376953125,
-      "learning_rate": 2.597015939947878e-05,
-      "loss": 59.4138,
-      "step": 134720
-    },
-    {
-      "epoch": 0.5443262482980967,
-      "grad_norm": 737.5440063476562,
-      "learning_rate": 2.59666713608716e-05,
-      "loss": 67.8319,
-      "step": 134730
-    },
-    {
-      "epoch": 0.5443666495634644,
-      "grad_norm": 346.41070556640625,
-      "learning_rate": 2.596318330341868e-05,
-      "loss": 59.8536,
-      "step": 134740
-    },
-    {
-      "epoch": 0.544407050828832,
-      "grad_norm": 390.0060119628906,
-      "learning_rate": 2.5959695227188004e-05,
-      "loss": 38.0195,
-      "step": 134750
-    },
-    {
-      "epoch": 0.5444474520941995,
-      "grad_norm": 577.295166015625,
-      "learning_rate": 2.595620713224757e-05,
-      "loss": 59.1009,
-      "step": 134760
-    },
-    {
-      "epoch": 0.5444878533595672,
-      "grad_norm": 509.3887939453125,
-      "learning_rate": 2.5952719018665382e-05,
-      "loss": 47.7892,
-      "step": 134770
-    },
-    {
-      "epoch": 0.5445282546249348,
-      "grad_norm": 715.0154418945312,
-      "learning_rate": 2.5949230886509457e-05,
-      "loss": 66.1531,
-      "step": 134780
-    },
-    {
-      "epoch": 0.5445686558903025,
-      "grad_norm": 1432.1541748046875,
-      "learning_rate": 2.594574273584779e-05,
-      "loss": 52.1664,
-      "step": 134790
-    },
-    {
-      "epoch": 0.5446090571556701,
-      "grad_norm": 882.5562133789062,
-      "learning_rate": 2.594225456674837e-05,
-      "loss": 72.5866,
-      "step": 134800
-    },
-    {
-      "epoch": 0.5446494584210377,
-      "grad_norm": 658.062744140625,
-      "learning_rate": 2.5938766379279212e-05,
-      "loss": 62.3709,
-      "step": 134810
-    },
-    {
-      "epoch": 0.5446898596864054,
-      "grad_norm": 738.4263916015625,
-      "learning_rate": 2.593527817350832e-05,
-      "loss": 64.2618,
-      "step": 134820
-    },
-    {
-      "epoch": 0.544730260951773,
-      "grad_norm": 459.119140625,
-      "learning_rate": 2.593178994950371e-05,
-      "loss": 47.4513,
-      "step": 134830
-    },
-    {
-      "epoch": 0.5447706622171407,
-      "grad_norm": 598.9742431640625,
-      "learning_rate": 2.5928301707333365e-05,
-      "loss": 37.9745,
-      "step": 134840
-    },
-    {
-      "epoch": 0.5448110634825083,
-      "grad_norm": 628.7976684570312,
-      "learning_rate": 2.59248134470653e-05,
-      "loss": 29.9893,
-      "step": 134850
-    },
-    {
-      "epoch": 0.5448514647478759,
-      "grad_norm": 485.0265808105469,
-      "learning_rate": 2.592132516876753e-05,
-      "loss": 62.2994,
-      "step": 134860
-    },
-    {
-      "epoch": 0.5448918660132436,
-      "grad_norm": 485.9800720214844,
-      "learning_rate": 2.591783687250804e-05,
-      "loss": 63.8224,
-      "step": 134870
-    },
-    {
-      "epoch": 0.5449322672786112,
-      "grad_norm": 473.7646484375,
-      "learning_rate": 2.5914348558354857e-05,
-      "loss": 37.5881,
-      "step": 134880
-    },
-    {
-      "epoch": 0.5449726685439787,
-      "grad_norm": 760.0999755859375,
-      "learning_rate": 2.5910860226375972e-05,
-      "loss": 70.4306,
-      "step": 134890
-    },
-    {
-      "epoch": 0.5450130698093464,
-      "grad_norm": 886.9143676757812,
-      "learning_rate": 2.5907371876639398e-05,
-      "loss": 59.3818,
-      "step": 134900
-    },
-    {
-      "epoch": 0.545053471074714,
-      "grad_norm": 765.5799560546875,
-      "learning_rate": 2.5903883509213146e-05,
-      "loss": 26.8561,
-      "step": 134910
-    },
-    {
-      "epoch": 0.5450938723400817,
-      "grad_norm": 939.5110473632812,
-      "learning_rate": 2.5900395124165218e-05,
-      "loss": 54.5256,
-      "step": 134920
-    },
-    {
-      "epoch": 0.5451342736054493,
-      "grad_norm": 1170.457275390625,
-      "learning_rate": 2.589690672156362e-05,
-      "loss": 57.3966,
-      "step": 134930
-    },
-    {
-      "epoch": 0.5451746748708169,
-      "grad_norm": 346.5357666015625,
-      "learning_rate": 2.5893418301476364e-05,
-      "loss": 63.307,
-      "step": 134940
-    },
-    {
-      "epoch": 0.5452150761361846,
-      "grad_norm": 643.980224609375,
-      "learning_rate": 2.5889929863971462e-05,
-      "loss": 49.9001,
-      "step": 134950
-    },
-    {
-      "epoch": 0.5452554774015522,
-      "grad_norm": 1780.751220703125,
-      "learning_rate": 2.5886441409116923e-05,
-      "loss": 78.8434,
-      "step": 134960
-    },
-    {
-      "epoch": 0.5452958786669199,
-      "grad_norm": 705.0413208007812,
-      "learning_rate": 2.5882952936980746e-05,
-      "loss": 74.3941,
-      "step": 134970
-    },
-    {
-      "epoch": 0.5453362799322875,
-      "grad_norm": 1106.158935546875,
-      "learning_rate": 2.5879464447630946e-05,
-      "loss": 45.0385,
-      "step": 134980
-    },
-    {
-      "epoch": 0.5453766811976551,
-      "grad_norm": 502.71795654296875,
-      "learning_rate": 2.587597594113554e-05,
-      "loss": 59.7987,
-      "step": 134990
-    },
-    {
-      "epoch": 0.5454170824630228,
-      "grad_norm": 651.170654296875,
-      "learning_rate": 2.587248741756253e-05,
-      "loss": 51.0282,
-      "step": 135000
-    },
-    {
-      "epoch": 0.5454574837283904,
-      "grad_norm": 539.2788696289062,
-      "learning_rate": 2.5868998876979923e-05,
-      "loss": 86.3156,
-      "step": 135010
-    },
-    {
-      "epoch": 0.545497884993758,
-      "grad_norm": 836.768310546875,
-      "learning_rate": 2.5865510319455737e-05,
-      "loss": 73.081,
-      "step": 135020
-    },
-    {
-      "epoch": 0.5455382862591256,
-      "grad_norm": 370.10723876953125,
-      "learning_rate": 2.586202174505799e-05,
-      "loss": 62.4055,
-      "step": 135030
-    },
-    {
-      "epoch": 0.5455786875244932,
-      "grad_norm": 620.5315551757812,
-      "learning_rate": 2.5858533153854675e-05,
-      "loss": 49.814,
-      "step": 135040
-    },
-    {
-      "epoch": 0.5456190887898609,
-      "grad_norm": 597.864501953125,
-      "learning_rate": 2.5855044545913825e-05,
-      "loss": 35.0934,
-      "step": 135050
-    },
-    {
-      "epoch": 0.5456594900552285,
-      "grad_norm": 2643.165283203125,
-      "learning_rate": 2.5851555921303434e-05,
-      "loss": 59.2937,
-      "step": 135060
-    },
-    {
-      "epoch": 0.5456998913205962,
-      "grad_norm": 0.0,
-      "learning_rate": 2.584806728009152e-05,
-      "loss": 39.2349,
-      "step": 135070
-    },
-    {
-      "epoch": 0.5457402925859638,
-      "grad_norm": 1795.3609619140625,
-      "learning_rate": 2.58445786223461e-05,
-      "loss": 59.5827,
-      "step": 135080
-    },
-    {
-      "epoch": 0.5457806938513314,
-      "grad_norm": 647.521484375,
-      "learning_rate": 2.584108994813518e-05,
-      "loss": 39.9268,
-      "step": 135090
-    },
-    {
-      "epoch": 0.5458210951166991,
-      "grad_norm": 280.2913818359375,
-      "learning_rate": 2.583760125752679e-05,
-      "loss": 34.3207,
-      "step": 135100
-    },
-    {
-      "epoch": 0.5458614963820667,
-      "grad_norm": 788.9907836914062,
-      "learning_rate": 2.5834112550588922e-05,
-      "loss": 70.1502,
-      "step": 135110
-    },
-    {
-      "epoch": 0.5459018976474344,
-      "grad_norm": 1395.0435791015625,
-      "learning_rate": 2.5830623827389612e-05,
-      "loss": 68.4634,
-      "step": 135120
-    },
-    {
-      "epoch": 0.545942298912802,
-      "grad_norm": 507.60723876953125,
-      "learning_rate": 2.5827135087996858e-05,
-      "loss": 60.2177,
-      "step": 135130
-    },
-    {
-      "epoch": 0.5459827001781696,
-      "grad_norm": 657.4443969726562,
-      "learning_rate": 2.5823646332478674e-05,
-      "loss": 115.6672,
-      "step": 135140
-    },
-    {
-      "epoch": 0.5460231014435372,
-      "grad_norm": 0.0,
-      "learning_rate": 2.582015756090308e-05,
-      "loss": 91.5032,
-      "step": 135150
-    },
-    {
-      "epoch": 0.5460635027089048,
-      "grad_norm": 770.5996704101562,
-      "learning_rate": 2.5816668773338098e-05,
-      "loss": 49.6053,
-      "step": 135160
-    },
-    {
-      "epoch": 0.5461039039742724,
-      "grad_norm": 915.6427612304688,
-      "learning_rate": 2.5813179969851737e-05,
-      "loss": 74.4677,
-      "step": 135170
-    },
-    {
-      "epoch": 0.5461443052396401,
-      "grad_norm": 642.6882934570312,
-      "learning_rate": 2.5809691150512012e-05,
-      "loss": 39.5173,
-      "step": 135180
-    },
-    {
-      "epoch": 0.5461847065050077,
-      "grad_norm": 1886.410888671875,
-      "learning_rate": 2.5806202315386934e-05,
-      "loss": 62.8264,
-      "step": 135190
-    },
-    {
-      "epoch": 0.5462251077703754,
-      "grad_norm": 546.2448120117188,
-      "learning_rate": 2.5802713464544542e-05,
-      "loss": 45.3419,
-      "step": 135200
-    },
-    {
-      "epoch": 0.546265509035743,
-      "grad_norm": 10494.7939453125,
-      "learning_rate": 2.5799224598052828e-05,
-      "loss": 71.9984,
-      "step": 135210
-    },
-    {
-      "epoch": 0.5463059103011106,
-      "grad_norm": 660.2236938476562,
-      "learning_rate": 2.5795735715979823e-05,
-      "loss": 71.6231,
-      "step": 135220
-    },
-    {
-      "epoch": 0.5463463115664783,
-      "grad_norm": 1856.3646240234375,
-      "learning_rate": 2.579224681839354e-05,
-      "loss": 60.5392,
-      "step": 135230
-    },
-    {
-      "epoch": 0.5463867128318459,
-      "grad_norm": 680.1478881835938,
-      "learning_rate": 2.5788757905362e-05,
-      "loss": 38.0885,
-      "step": 135240
-    },
-    {
-      "epoch": 0.5464271140972136,
-      "grad_norm": 603.8659057617188,
-      "learning_rate": 2.578526897695321e-05,
-      "loss": 76.2062,
-      "step": 135250
-    },
-    {
-      "epoch": 0.5464675153625812,
-      "grad_norm": 587.8827514648438,
-      "learning_rate": 2.5781780033235204e-05,
-      "loss": 47.7149,
-      "step": 135260
-    },
-    {
-      "epoch": 0.5465079166279488,
-      "grad_norm": 704.3698120117188,
-      "learning_rate": 2.5778291074275983e-05,
-      "loss": 38.3459,
-      "step": 135270
-    },
-    {
-      "epoch": 0.5465483178933164,
-      "grad_norm": 840.8921508789062,
-      "learning_rate": 2.5774802100143592e-05,
-      "loss": 47.8546,
-      "step": 135280
-    },
-    {
-      "epoch": 0.546588719158684,
-      "grad_norm": 855.8980102539062,
-      "learning_rate": 2.5771313110906026e-05,
-      "loss": 59.4118,
-      "step": 135290
-    },
-    {
-      "epoch": 0.5466291204240517,
-      "grad_norm": 342.91058349609375,
-      "learning_rate": 2.576782410663132e-05,
-      "loss": 68.2463,
-      "step": 135300
-    },
-    {
-      "epoch": 0.5466695216894193,
-      "grad_norm": 579.0650634765625,
-      "learning_rate": 2.576433508738748e-05,
-      "loss": 44.4399,
-      "step": 135310
-    },
-    {
-      "epoch": 0.5467099229547869,
-      "grad_norm": 1014.5051879882812,
-      "learning_rate": 2.5760846053242544e-05,
-      "loss": 61.9376,
-      "step": 135320
-    },
-    {
-      "epoch": 0.5467503242201546,
-      "grad_norm": 702.8917236328125,
-      "learning_rate": 2.5757357004264514e-05,
-      "loss": 84.4671,
-      "step": 135330
-    },
-    {
-      "epoch": 0.5467907254855222,
-      "grad_norm": 780.0779418945312,
-      "learning_rate": 2.575386794052142e-05,
-      "loss": 59.8046,
-      "step": 135340
-    },
-    {
-      "epoch": 0.5468311267508899,
-      "grad_norm": 1222.5645751953125,
-      "learning_rate": 2.575037886208128e-05,
-      "loss": 53.4124,
-      "step": 135350
-    },
-    {
-      "epoch": 0.5468715280162575,
-      "grad_norm": 512.1873779296875,
-      "learning_rate": 2.5746889769012128e-05,
-      "loss": 50.8429,
-      "step": 135360
-    },
-    {
-      "epoch": 0.5469119292816251,
-      "grad_norm": 329.0351257324219,
-      "learning_rate": 2.5743400661381968e-05,
-      "loss": 33.7403,
-      "step": 135370
-    },
-    {
-      "epoch": 0.5469523305469928,
-      "grad_norm": 1148.149169921875,
-      "learning_rate": 2.573991153925883e-05,
-      "loss": 71.7041,
-      "step": 135380
-    },
-    {
-      "epoch": 0.5469927318123604,
-      "grad_norm": 1018.9400634765625,
-      "learning_rate": 2.5736422402710742e-05,
-      "loss": 63.8832,
-      "step": 135390
-    },
-    {
-      "epoch": 0.5470331330777279,
-      "grad_norm": 339.0023193359375,
-      "learning_rate": 2.5732933251805713e-05,
-      "loss": 63.8842,
-      "step": 135400
-    },
-    {
-      "epoch": 0.5470735343430956,
-      "grad_norm": 316.14666748046875,
-      "learning_rate": 2.5729444086611777e-05,
-      "loss": 35.9186,
-      "step": 135410
-    },
-    {
-      "epoch": 0.5471139356084632,
-      "grad_norm": 304.40130615234375,
-      "learning_rate": 2.5725954907196947e-05,
-      "loss": 32.5816,
-      "step": 135420
-    },
-    {
-      "epoch": 0.5471543368738309,
-      "grad_norm": 640.4146118164062,
-      "learning_rate": 2.572246571362925e-05,
-      "loss": 56.8944,
-      "step": 135430
-    },
-    {
-      "epoch": 0.5471947381391985,
-      "grad_norm": 1118.9725341796875,
-      "learning_rate": 2.5718976505976717e-05,
-      "loss": 42.6715,
-      "step": 135440
-    },
-    {
-      "epoch": 0.5472351394045661,
-      "grad_norm": 3967.9638671875,
-      "learning_rate": 2.571548728430737e-05,
-      "loss": 68.9002,
-      "step": 135450
-    },
-    {
-      "epoch": 0.5472755406699338,
-      "grad_norm": 647.3598022460938,
-      "learning_rate": 2.5711998048689227e-05,
-      "loss": 69.0035,
-      "step": 135460
-    },
-    {
-      "epoch": 0.5473159419353014,
-      "grad_norm": 724.192138671875,
-      "learning_rate": 2.5708508799190322e-05,
-      "loss": 59.7795,
-      "step": 135470
-    },
-    {
-      "epoch": 0.5473563432006691,
-      "grad_norm": 251.8446807861328,
-      "learning_rate": 2.5705019535878668e-05,
-      "loss": 53.0616,
-      "step": 135480
-    },
-    {
-      "epoch": 0.5473967444660367,
-      "grad_norm": 777.9346923828125,
-      "learning_rate": 2.5701530258822294e-05,
-      "loss": 75.1509,
-      "step": 135490
-    },
-    {
-      "epoch": 0.5474371457314043,
-      "grad_norm": 483.5498352050781,
-      "learning_rate": 2.5698040968089225e-05,
-      "loss": 53.8124,
-      "step": 135500
-    },
-    {
-      "epoch": 0.547477546996772,
-      "grad_norm": 1032.5994873046875,
-      "learning_rate": 2.569455166374749e-05,
-      "loss": 44.2035,
-      "step": 135510
-    },
-    {
-      "epoch": 0.5475179482621396,
-      "grad_norm": 527.7037353515625,
-      "learning_rate": 2.569106234586511e-05,
-      "loss": 67.0334,
-      "step": 135520
-    },
-    {
-      "epoch": 0.5475583495275071,
-      "grad_norm": 494.0249328613281,
-      "learning_rate": 2.568757301451012e-05,
-      "loss": 34.6893,
-      "step": 135530
-    },
-    {
-      "epoch": 0.5475987507928748,
-      "grad_norm": 351.9414978027344,
-      "learning_rate": 2.5684083669750537e-05,
-      "loss": 59.7537,
-      "step": 135540
-    },
-    {
-      "epoch": 0.5476391520582424,
-      "grad_norm": 922.2720336914062,
-      "learning_rate": 2.568059431165438e-05,
-      "loss": 64.9178,
-      "step": 135550
-    },
-    {
-      "epoch": 0.5476795533236101,
-      "grad_norm": 570.0916137695312,
-      "learning_rate": 2.5677104940289702e-05,
-      "loss": 40.3571,
-      "step": 135560
-    },
-    {
-      "epoch": 0.5477199545889777,
-      "grad_norm": 652.1029663085938,
-      "learning_rate": 2.567361555572452e-05,
-      "loss": 49.9205,
-      "step": 135570
-    },
-    {
-      "epoch": 0.5477603558543453,
-      "grad_norm": 826.6941528320312,
-      "learning_rate": 2.5670126158026842e-05,
-      "loss": 54.1507,
-      "step": 135580
-    },
-    {
-      "epoch": 0.547800757119713,
-      "grad_norm": 547.8671875,
-      "learning_rate": 2.5666636747264705e-05,
-      "loss": 59.982,
-      "step": 135590
-    },
-    {
-      "epoch": 0.5478411583850806,
-      "grad_norm": 757.22998046875,
-      "learning_rate": 2.566314732350615e-05,
-      "loss": 56.2098,
-      "step": 135600
-    },
-    {
-      "epoch": 0.5478815596504483,
-      "grad_norm": 351.811767578125,
-      "learning_rate": 2.56596578868192e-05,
-      "loss": 65.6569,
-      "step": 135610
-    },
-    {
-      "epoch": 0.5479219609158159,
-      "grad_norm": 1225.1767578125,
-      "learning_rate": 2.565616843727187e-05,
-      "loss": 53.0704,
-      "step": 135620
-    },
-    {
-      "epoch": 0.5479623621811835,
-      "grad_norm": 1242.2264404296875,
-      "learning_rate": 2.5652678974932204e-05,
-      "loss": 71.0538,
-      "step": 135630
-    },
-    {
-      "epoch": 0.5480027634465512,
-      "grad_norm": 195.21212768554688,
-      "learning_rate": 2.5649189499868232e-05,
-      "loss": 52.8797,
-      "step": 135640
-    },
-    {
-      "epoch": 0.5480431647119188,
-      "grad_norm": 1128.5284423828125,
-      "learning_rate": 2.564570001214797e-05,
-      "loss": 51.5018,
-      "step": 135650
-    },
-    {
-      "epoch": 0.5480835659772864,
-      "grad_norm": 694.2415161132812,
-      "learning_rate": 2.5642210511839452e-05,
-      "loss": 61.8394,
-      "step": 135660
-    },
-    {
-      "epoch": 0.548123967242654,
-      "grad_norm": 901.8828735351562,
-      "learning_rate": 2.5638720999010713e-05,
-      "loss": 48.585,
-      "step": 135670
-    },
-    {
-      "epoch": 0.5481643685080216,
-      "grad_norm": 640.2681274414062,
-      "learning_rate": 2.5635231473729772e-05,
-      "loss": 43.3395,
-      "step": 135680
-    },
-    {
-      "epoch": 0.5482047697733893,
-      "grad_norm": 672.07421875,
-      "learning_rate": 2.5631741936064675e-05,
-      "loss": 46.9696,
-      "step": 135690
-    },
-    {
-      "epoch": 0.5482451710387569,
-      "grad_norm": 883.359130859375,
-      "learning_rate": 2.562825238608344e-05,
-      "loss": 43.4834,
-      "step": 135700
-    },
-    {
-      "epoch": 0.5482855723041246,
-      "grad_norm": 1151.069091796875,
-      "learning_rate": 2.56247628238541e-05,
-      "loss": 52.7021,
-      "step": 135710
-    },
-    {
-      "epoch": 0.5483259735694922,
-      "grad_norm": 1026.870361328125,
-      "learning_rate": 2.562127324944469e-05,
-      "loss": 67.7404,
-      "step": 135720
-    },
-    {
-      "epoch": 0.5483663748348598,
-      "grad_norm": 348.47821044921875,
-      "learning_rate": 2.561778366292324e-05,
-      "loss": 36.8606,
-      "step": 135730
-    },
-    {
-      "epoch": 0.5484067761002275,
-      "grad_norm": 1066.57666015625,
-      "learning_rate": 2.5614294064357773e-05,
-      "loss": 95.5207,
-      "step": 135740
-    },
-    {
-      "epoch": 0.5484471773655951,
-      "grad_norm": 504.4042053222656,
-      "learning_rate": 2.5610804453816333e-05,
-      "loss": 50.8941,
-      "step": 135750
-    },
-    {
-      "epoch": 0.5484875786309628,
-      "grad_norm": 811.9468383789062,
-      "learning_rate": 2.560731483136694e-05,
-      "loss": 39.8252,
-      "step": 135760
-    },
-    {
-      "epoch": 0.5485279798963304,
-      "grad_norm": 0.0,
-      "learning_rate": 2.560382519707764e-05,
-      "loss": 78.3535,
-      "step": 135770
-    },
-    {
-      "epoch": 0.548568381161698,
-      "grad_norm": 522.0183715820312,
-      "learning_rate": 2.5600335551016445e-05,
-      "loss": 57.2499,
-      "step": 135780
-    },
-    {
-      "epoch": 0.5486087824270656,
-      "grad_norm": 385.5660705566406,
-      "learning_rate": 2.559684589325141e-05,
-      "loss": 64.1449,
-      "step": 135790
-    },
-    {
-      "epoch": 0.5486491836924332,
-      "grad_norm": 1208.748779296875,
-      "learning_rate": 2.559335622385055e-05,
-      "loss": 71.5,
-      "step": 135800
-    },
-    {
-      "epoch": 0.5486895849578008,
-      "grad_norm": 794.9710083007812,
-      "learning_rate": 2.5589866542881912e-05,
-      "loss": 72.0629,
-      "step": 135810
-    },
-    {
-      "epoch": 0.5487299862231685,
-      "grad_norm": 589.7503662109375,
-      "learning_rate": 2.5586376850413517e-05,
-      "loss": 50.9222,
-      "step": 135820
-    },
-    {
-      "epoch": 0.5487703874885361,
-      "grad_norm": 743.7083129882812,
-      "learning_rate": 2.5582887146513406e-05,
-      "loss": 36.8839,
-      "step": 135830
-    },
-    {
-      "epoch": 0.5488107887539038,
-      "grad_norm": 826.81787109375,
-      "learning_rate": 2.5579397431249606e-05,
-      "loss": 52.2726,
-      "step": 135840
-    },
-    {
-      "epoch": 0.5488511900192714,
-      "grad_norm": 924.5673828125,
-      "learning_rate": 2.557590770469016e-05,
-      "loss": 45.644,
-      "step": 135850
-    },
-    {
-      "epoch": 0.548891591284639,
-      "grad_norm": 1268.26123046875,
-      "learning_rate": 2.5572417966903094e-05,
-      "loss": 78.6158,
-      "step": 135860
-    },
-    {
-      "epoch": 0.5489319925500067,
-      "grad_norm": 822.70703125,
-      "learning_rate": 2.5568928217956446e-05,
-      "loss": 36.7586,
-      "step": 135870
-    },
-    {
-      "epoch": 0.5489723938153743,
-      "grad_norm": 612.1898193359375,
-      "learning_rate": 2.5565438457918244e-05,
-      "loss": 35.3985,
-      "step": 135880
-    },
-    {
-      "epoch": 0.549012795080742,
-      "grad_norm": 3193.71435546875,
-      "learning_rate": 2.5561948686856535e-05,
-      "loss": 47.6024,
-      "step": 135890
-    },
-    {
-      "epoch": 0.5490531963461096,
-      "grad_norm": 145.80606079101562,
-      "learning_rate": 2.5558458904839345e-05,
-      "loss": 25.6671,
-      "step": 135900
-    },
-    {
-      "epoch": 0.5490935976114772,
-      "grad_norm": 1195.0611572265625,
-      "learning_rate": 2.5554969111934713e-05,
-      "loss": 62.6013,
-      "step": 135910
-    },
-    {
-      "epoch": 0.5491339988768448,
-      "grad_norm": 651.7173461914062,
-      "learning_rate": 2.5551479308210668e-05,
-      "loss": 56.0175,
-      "step": 135920
-    },
-    {
-      "epoch": 0.5491744001422124,
-      "grad_norm": 658.3231201171875,
-      "learning_rate": 2.5547989493735258e-05,
-      "loss": 45.2132,
-      "step": 135930
-    },
-    {
-      "epoch": 0.54921480140758,
-      "grad_norm": 629.660400390625,
-      "learning_rate": 2.5544499668576505e-05,
-      "loss": 56.837,
-      "step": 135940
-    },
-    {
-      "epoch": 0.5492552026729477,
-      "grad_norm": 558.2518920898438,
-      "learning_rate": 2.5541009832802448e-05,
-      "loss": 47.7196,
-      "step": 135950
-    },
-    {
-      "epoch": 0.5492956039383153,
-      "grad_norm": 552.365478515625,
-      "learning_rate": 2.5537519986481122e-05,
-      "loss": 78.8988,
-      "step": 135960
-    },
-    {
-      "epoch": 0.549336005203683,
-      "grad_norm": 1013.0728149414062,
-      "learning_rate": 2.5534030129680577e-05,
-      "loss": 47.0896,
-      "step": 135970
-    },
-    {
-      "epoch": 0.5493764064690506,
-      "grad_norm": 625.7086181640625,
-      "learning_rate": 2.5530540262468837e-05,
-      "loss": 45.671,
-      "step": 135980
-    },
-    {
-      "epoch": 0.5494168077344183,
-      "grad_norm": 4387.12646484375,
-      "learning_rate": 2.552705038491394e-05,
-      "loss": 62.9142,
-      "step": 135990
-    },
-    {
-      "epoch": 0.5494572089997859,
-      "grad_norm": 943.5322875976562,
-      "learning_rate": 2.5523560497083926e-05,
-      "loss": 55.8262,
-      "step": 136000
-    },
-    {
-      "epoch": 0.5494976102651535,
-      "grad_norm": 213.67581176757812,
-      "learning_rate": 2.552007059904683e-05,
-      "loss": 48.1262,
-      "step": 136010
-    },
-    {
-      "epoch": 0.5495380115305212,
-      "grad_norm": 1800.0438232421875,
-      "learning_rate": 2.5516580690870695e-05,
-      "loss": 46.4066,
-      "step": 136020
-    },
-    {
-      "epoch": 0.5495784127958888,
-      "grad_norm": 747.0396728515625,
-      "learning_rate": 2.5513090772623543e-05,
-      "loss": 72.1643,
-      "step": 136030
-    },
-    {
-      "epoch": 0.5496188140612563,
-      "grad_norm": 852.1857299804688,
-      "learning_rate": 2.5509600844373427e-05,
-      "loss": 48.1462,
-      "step": 136040
-    },
-    {
-      "epoch": 0.549659215326624,
-      "grad_norm": 584.1296997070312,
-      "learning_rate": 2.550611090618838e-05,
-      "loss": 64.4517,
-      "step": 136050
-    },
-    {
-      "epoch": 0.5496996165919916,
-      "grad_norm": 639.3566284179688,
-      "learning_rate": 2.5502620958136443e-05,
-      "loss": 55.9694,
-      "step": 136060
-    },
-    {
-      "epoch": 0.5497400178573593,
-      "grad_norm": 634.5128173828125,
-      "learning_rate": 2.5499131000285646e-05,
-      "loss": 44.5816,
-      "step": 136070
-    },
-    {
-      "epoch": 0.5497804191227269,
-      "grad_norm": 403.07159423828125,
-      "learning_rate": 2.5495641032704043e-05,
-      "loss": 63.8988,
-      "step": 136080
-    },
-    {
-      "epoch": 0.5498208203880945,
-      "grad_norm": 770.9432373046875,
-      "learning_rate": 2.549215105545965e-05,
-      "loss": 61.3341,
-      "step": 136090
-    },
-    {
-      "epoch": 0.5498612216534622,
-      "grad_norm": 0.0,
-      "learning_rate": 2.5488661068620533e-05,
-      "loss": 37.3569,
-      "step": 136100
-    },
-    {
-      "epoch": 0.5499016229188298,
-      "grad_norm": 493.512939453125,
-      "learning_rate": 2.5485171072254704e-05,
-      "loss": 54.7347,
-      "step": 136110
-    },
-    {
-      "epoch": 0.5499420241841975,
-      "grad_norm": 0.0,
-      "learning_rate": 2.5481681066430217e-05,
-      "loss": 82.7196,
-      "step": 136120
-    },
-    {
-      "epoch": 0.5499824254495651,
-      "grad_norm": 815.2086791992188,
-      "learning_rate": 2.5478191051215117e-05,
-      "loss": 39.4943,
-      "step": 136130
-    },
-    {
-      "epoch": 0.5500228267149327,
-      "grad_norm": 294.3896789550781,
-      "learning_rate": 2.5474701026677433e-05,
-      "loss": 37.5106,
-      "step": 136140
-    },
-    {
-      "epoch": 0.5500632279803004,
-      "grad_norm": 645.8425903320312,
-      "learning_rate": 2.547121099288521e-05,
-      "loss": 51.6582,
-      "step": 136150
-    },
-    {
-      "epoch": 0.550103629245668,
-      "grad_norm": 259.2788391113281,
-      "learning_rate": 2.5467720949906483e-05,
-      "loss": 63.0887,
-      "step": 136160
-    },
-    {
-      "epoch": 0.5501440305110356,
-      "grad_norm": 341.1513977050781,
-      "learning_rate": 2.5464230897809294e-05,
-      "loss": 37.9722,
-      "step": 136170
-    },
-    {
-      "epoch": 0.5501844317764032,
-      "grad_norm": 1073.13427734375,
-      "learning_rate": 2.546074083666169e-05,
-      "loss": 63.717,
-      "step": 136180
-    },
-    {
-      "epoch": 0.5502248330417708,
-      "grad_norm": 564.9509887695312,
-      "learning_rate": 2.54572507665317e-05,
-      "loss": 50.2804,
-      "step": 136190
-    },
-    {
-      "epoch": 0.5502652343071385,
-      "grad_norm": 547.4306640625,
-      "learning_rate": 2.545376068748737e-05,
-      "loss": 52.0443,
-      "step": 136200
-    },
-    {
-      "epoch": 0.5503056355725061,
-      "grad_norm": 2279.78662109375,
-      "learning_rate": 2.545027059959674e-05,
-      "loss": 59.4126,
-      "step": 136210
-    },
-    {
-      "epoch": 0.5503460368378738,
-      "grad_norm": 826.3348388671875,
-      "learning_rate": 2.544678050292787e-05,
-      "loss": 47.2906,
-      "step": 136220
-    },
-    {
-      "epoch": 0.5503864381032414,
-      "grad_norm": 681.75927734375,
-      "learning_rate": 2.5443290397548768e-05,
-      "loss": 44.3834,
-      "step": 136230
-    },
-    {
-      "epoch": 0.550426839368609,
-      "grad_norm": 677.0407104492188,
-      "learning_rate": 2.5439800283527494e-05,
-      "loss": 78.2347,
-      "step": 136240
-    },
-    {
-      "epoch": 0.5504672406339767,
-      "grad_norm": 516.7835693359375,
-      "learning_rate": 2.5436310160932092e-05,
-      "loss": 52.4863,
-      "step": 136250
-    },
-    {
-      "epoch": 0.5505076418993443,
-      "grad_norm": 830.3655395507812,
-      "learning_rate": 2.54328200298306e-05,
-      "loss": 44.6782,
-      "step": 136260
-    },
-    {
-      "epoch": 0.550548043164712,
-      "grad_norm": 475.61456298828125,
-      "learning_rate": 2.542932989029105e-05,
-      "loss": 47.868,
-      "step": 136270
-    },
-    {
-      "epoch": 0.5505884444300796,
-      "grad_norm": 869.2997436523438,
-      "learning_rate": 2.5425839742381498e-05,
-      "loss": 58.4036,
-      "step": 136280
-    },
-    {
-      "epoch": 0.5506288456954472,
-      "grad_norm": 452.5263671875,
-      "learning_rate": 2.5422349586169974e-05,
-      "loss": 54.134,
-      "step": 136290
-    },
-    {
-      "epoch": 0.5506692469608148,
-      "grad_norm": 624.853759765625,
-      "learning_rate": 2.5418859421724538e-05,
-      "loss": 67.5689,
-      "step": 136300
-    },
-    {
-      "epoch": 0.5507096482261824,
-      "grad_norm": 680.9571533203125,
-      "learning_rate": 2.5415369249113212e-05,
-      "loss": 68.5904,
-      "step": 136310
-    },
-    {
-      "epoch": 0.55075004949155,
-      "grad_norm": 326.5666198730469,
-      "learning_rate": 2.5411879068404056e-05,
-      "loss": 84.7915,
-      "step": 136320
-    },
-    {
-      "epoch": 0.5507904507569177,
-      "grad_norm": 239.2113494873047,
-      "learning_rate": 2.5408388879665108e-05,
-      "loss": 37.1209,
-      "step": 136330
-    },
-    {
-      "epoch": 0.5508308520222853,
-      "grad_norm": 844.7891845703125,
-      "learning_rate": 2.54048986829644e-05,
-      "loss": 43.2222,
-      "step": 136340
-    },
-    {
-      "epoch": 0.550871253287653,
-      "grad_norm": 908.1921997070312,
-      "learning_rate": 2.540140847836999e-05,
-      "loss": 35.1715,
-      "step": 136350
-    },
-    {
-      "epoch": 0.5509116545530206,
-      "grad_norm": 649.4369506835938,
-      "learning_rate": 2.539791826594991e-05,
-      "loss": 35.0707,
-      "step": 136360
-    },
-    {
-      "epoch": 0.5509520558183882,
-      "grad_norm": 816.3155517578125,
-      "learning_rate": 2.5394428045772212e-05,
-      "loss": 46.9021,
-      "step": 136370
-    },
-    {
-      "epoch": 0.5509924570837559,
-      "grad_norm": 1066.6396484375,
-      "learning_rate": 2.539093781790494e-05,
-      "loss": 49.8077,
-      "step": 136380
-    },
-    {
-      "epoch": 0.5510328583491235,
-      "grad_norm": 581.2196044921875,
-      "learning_rate": 2.5387447582416123e-05,
-      "loss": 113.6221,
-      "step": 136390
-    },
-    {
-      "epoch": 0.5510732596144912,
-      "grad_norm": 908.7518310546875,
-      "learning_rate": 2.5383957339373825e-05,
-      "loss": 55.2377,
-      "step": 136400
-    },
-    {
-      "epoch": 0.5511136608798588,
-      "grad_norm": 683.6449584960938,
-      "learning_rate": 2.5380467088846077e-05,
-      "loss": 51.5513,
-      "step": 136410
-    },
-    {
-      "epoch": 0.5511540621452264,
-      "grad_norm": 1066.43994140625,
-      "learning_rate": 2.537697683090093e-05,
-      "loss": 54.2596,
-      "step": 136420
-    },
-    {
-      "epoch": 0.551194463410594,
-      "grad_norm": 1037.6993408203125,
-      "learning_rate": 2.537348656560643e-05,
-      "loss": 97.0623,
-      "step": 136430
-    },
-    {
-      "epoch": 0.5512348646759616,
-      "grad_norm": 524.144287109375,
-      "learning_rate": 2.5369996293030606e-05,
-      "loss": 64.2785,
-      "step": 136440
-    },
-    {
-      "epoch": 0.5512752659413293,
-      "grad_norm": 847.4541015625,
-      "learning_rate": 2.536650601324152e-05,
-      "loss": 52.7042,
-      "step": 136450
-    },
-    {
-      "epoch": 0.5513156672066969,
-      "grad_norm": 603.0610961914062,
-      "learning_rate": 2.536301572630721e-05,
-      "loss": 36.3441,
-      "step": 136460
-    },
-    {
-      "epoch": 0.5513560684720645,
-      "grad_norm": 818.6026000976562,
-      "learning_rate": 2.535952543229572e-05,
-      "loss": 49.2163,
-      "step": 136470
-    },
-    {
-      "epoch": 0.5513964697374322,
-      "grad_norm": 1117.1014404296875,
-      "learning_rate": 2.5356035131275096e-05,
-      "loss": 54.204,
-      "step": 136480
-    },
-    {
-      "epoch": 0.5514368710027998,
-      "grad_norm": 1214.039794921875,
-      "learning_rate": 2.5352544823313384e-05,
-      "loss": 67.4448,
-      "step": 136490
-    },
-    {
-      "epoch": 0.5514772722681675,
-      "grad_norm": 671.1990356445312,
-      "learning_rate": 2.5349054508478637e-05,
-      "loss": 52.2546,
-      "step": 136500
-    },
-    {
-      "epoch": 0.5515176735335351,
-      "grad_norm": 832.948974609375,
-      "learning_rate": 2.534556418683888e-05,
-      "loss": 52.6752,
-      "step": 136510
-    },
-    {
-      "epoch": 0.5515580747989027,
-      "grad_norm": 684.7549438476562,
-      "learning_rate": 2.5342073858462185e-05,
-      "loss": 67.4169,
-      "step": 136520
-    },
-    {
-      "epoch": 0.5515984760642704,
-      "grad_norm": 284.89532470703125,
-      "learning_rate": 2.5338583523416575e-05,
-      "loss": 66.7467,
-      "step": 136530
-    },
-    {
-      "epoch": 0.551638877329638,
-      "grad_norm": 538.3557739257812,
-      "learning_rate": 2.5335093181770104e-05,
-      "loss": 44.1986,
-      "step": 136540
-    },
-    {
-      "epoch": 0.5516792785950057,
-      "grad_norm": 1377.850341796875,
-      "learning_rate": 2.5331602833590824e-05,
-      "loss": 59.193,
-      "step": 136550
-    },
-    {
-      "epoch": 0.5517196798603732,
-      "grad_norm": 873.0044555664062,
-      "learning_rate": 2.532811247894677e-05,
-      "loss": 55.7959,
-      "step": 136560
-    },
-    {
-      "epoch": 0.5517600811257408,
-      "grad_norm": 1132.6865234375,
-      "learning_rate": 2.5324622117906e-05,
-      "loss": 46.0268,
-      "step": 136570
-    },
-    {
-      "epoch": 0.5518004823911085,
-      "grad_norm": 4735.5791015625,
-      "learning_rate": 2.5321131750536547e-05,
-      "loss": 54.9213,
-      "step": 136580
-    },
-    {
-      "epoch": 0.5518408836564761,
-      "grad_norm": 715.4901123046875,
-      "learning_rate": 2.5317641376906475e-05,
-      "loss": 56.9693,
-      "step": 136590
-    },
-    {
-      "epoch": 0.5518812849218437,
-      "grad_norm": 998.7701416015625,
-      "learning_rate": 2.531415099708382e-05,
-      "loss": 52.3184,
-      "step": 136600
-    },
-    {
-      "epoch": 0.5519216861872114,
-      "grad_norm": 2349.027099609375,
-      "learning_rate": 2.5310660611136628e-05,
-      "loss": 64.0717,
-      "step": 136610
-    },
-    {
-      "epoch": 0.551962087452579,
-      "grad_norm": 1127.0347900390625,
-      "learning_rate": 2.530717021913294e-05,
-      "loss": 43.0908,
-      "step": 136620
-    },
-    {
-      "epoch": 0.5520024887179467,
-      "grad_norm": 574.6187133789062,
-      "learning_rate": 2.530367982114082e-05,
-      "loss": 40.4501,
-      "step": 136630
-    },
-    {
-      "epoch": 0.5520428899833143,
-      "grad_norm": 2183.83544921875,
-      "learning_rate": 2.53001894172283e-05,
-      "loss": 64.5596,
-      "step": 136640
-    },
-    {
-      "epoch": 0.5520832912486819,
-      "grad_norm": 508.4299621582031,
-      "learning_rate": 2.5296699007463433e-05,
-      "loss": 48.9093,
-      "step": 136650
-    },
-    {
-      "epoch": 0.5521236925140496,
-      "grad_norm": 1008.5830688476562,
-      "learning_rate": 2.5293208591914263e-05,
-      "loss": 71.5341,
-      "step": 136660
-    },
-    {
-      "epoch": 0.5521640937794172,
-      "grad_norm": 573.5001220703125,
-      "learning_rate": 2.5289718170648853e-05,
-      "loss": 35.5604,
-      "step": 136670
-    },
-    {
-      "epoch": 0.5522044950447847,
-      "grad_norm": 1256.3994140625,
-      "learning_rate": 2.5286227743735225e-05,
-      "loss": 60.7903,
-      "step": 136680
-    },
-    {
-      "epoch": 0.5522448963101524,
-      "grad_norm": 427.8146057128906,
-      "learning_rate": 2.528273731124145e-05,
-      "loss": 62.5258,
-      "step": 136690
-    },
-    {
-      "epoch": 0.55228529757552,
-      "grad_norm": 1044.406982421875,
-      "learning_rate": 2.527924687323556e-05,
-      "loss": 71.0137,
-      "step": 136700
-    },
-    {
-      "epoch": 0.5523256988408877,
-      "grad_norm": 916.89599609375,
-      "learning_rate": 2.5275756429785608e-05,
-      "loss": 52.149,
-      "step": 136710
-    },
-    {
-      "epoch": 0.5523661001062553,
-      "grad_norm": 551.5916137695312,
-      "learning_rate": 2.527226598095964e-05,
-      "loss": 71.3226,
-      "step": 136720
-    },
-    {
-      "epoch": 0.552406501371623,
-      "grad_norm": 668.67431640625,
-      "learning_rate": 2.526877552682571e-05,
-      "loss": 66.0503,
-      "step": 136730
-    },
-    {
-      "epoch": 0.5524469026369906,
-      "grad_norm": 1805.9869384765625,
-      "learning_rate": 2.5265285067451862e-05,
-      "loss": 63.5316,
-      "step": 136740
-    },
-    {
-      "epoch": 0.5524873039023582,
-      "grad_norm": 1308.774169921875,
-      "learning_rate": 2.5261794602906145e-05,
-      "loss": 45.1674,
-      "step": 136750
-    },
-    {
-      "epoch": 0.5525277051677259,
-      "grad_norm": 242.80426025390625,
-      "learning_rate": 2.5258304133256612e-05,
-      "loss": 59.7236,
-      "step": 136760
-    },
-    {
-      "epoch": 0.5525681064330935,
-      "grad_norm": 341.7113952636719,
-      "learning_rate": 2.52548136585713e-05,
-      "loss": 42.4413,
-      "step": 136770
-    },
-    {
-      "epoch": 0.5526085076984611,
-      "grad_norm": 571.553955078125,
-      "learning_rate": 2.5251323178918268e-05,
-      "loss": 47.814,
-      "step": 136780
-    },
-    {
-      "epoch": 0.5526489089638288,
-      "grad_norm": 2131.88525390625,
-      "learning_rate": 2.5247832694365565e-05,
-      "loss": 65.054,
-      "step": 136790
-    },
-    {
-      "epoch": 0.5526893102291964,
-      "grad_norm": 412.0196533203125,
-      "learning_rate": 2.524434220498123e-05,
-      "loss": 57.1556,
-      "step": 136800
-    },
-    {
-      "epoch": 0.552729711494564,
-      "grad_norm": 347.2409362792969,
-      "learning_rate": 2.524085171083332e-05,
-      "loss": 51.0059,
-      "step": 136810
-    },
-    {
-      "epoch": 0.5527701127599316,
-      "grad_norm": 1224.644287109375,
-      "learning_rate": 2.523736121198988e-05,
-      "loss": 40.1427,
-      "step": 136820
-    },
-    {
-      "epoch": 0.5528105140252992,
-      "grad_norm": 1251.9407958984375,
-      "learning_rate": 2.5233870708518964e-05,
-      "loss": 45.218,
-      "step": 136830
-    },
-    {
-      "epoch": 0.5528509152906669,
-      "grad_norm": 1171.9188232421875,
-      "learning_rate": 2.5230380200488613e-05,
-      "loss": 57.0013,
-      "step": 136840
-    },
-    {
-      "epoch": 0.5528913165560345,
-      "grad_norm": 820.2431640625,
-      "learning_rate": 2.5226889687966882e-05,
-      "loss": 61.3478,
-      "step": 136850
-    },
-    {
-      "epoch": 0.5529317178214022,
-      "grad_norm": 1104.1824951171875,
-      "learning_rate": 2.5223399171021827e-05,
-      "loss": 47.4096,
-      "step": 136860
-    },
-    {
-      "epoch": 0.5529721190867698,
-      "grad_norm": 1183.5810546875,
-      "learning_rate": 2.5219908649721492e-05,
-      "loss": 62.6051,
-      "step": 136870
-    },
-    {
-      "epoch": 0.5530125203521374,
-      "grad_norm": 736.5234375,
-      "learning_rate": 2.5216418124133916e-05,
-      "loss": 48.401,
-      "step": 136880
-    },
-    {
-      "epoch": 0.5530529216175051,
-      "grad_norm": 710.7758178710938,
-      "learning_rate": 2.5212927594327157e-05,
-      "loss": 60.8935,
-      "step": 136890
-    },
-    {
-      "epoch": 0.5530933228828727,
-      "grad_norm": 590.540771484375,
-      "learning_rate": 2.520943706036927e-05,
-      "loss": 46.2608,
-      "step": 136900
-    },
-    {
-      "epoch": 0.5531337241482404,
-      "grad_norm": 2185.54833984375,
-      "learning_rate": 2.52059465223283e-05,
-      "loss": 63.3829,
-      "step": 136910
-    },
-    {
-      "epoch": 0.553174125413608,
-      "grad_norm": 604.2235107421875,
-      "learning_rate": 2.520245598027229e-05,
-      "loss": 40.5135,
-      "step": 136920
-    },
-    {
-      "epoch": 0.5532145266789756,
-      "grad_norm": 400.40631103515625,
-      "learning_rate": 2.5198965434269294e-05,
-      "loss": 40.3837,
-      "step": 136930
-    },
-    {
-      "epoch": 0.5532549279443432,
-      "grad_norm": 770.1949462890625,
-      "learning_rate": 2.5195474884387376e-05,
-      "loss": 59.8826,
-      "step": 136940
-    },
-    {
-      "epoch": 0.5532953292097108,
-      "grad_norm": 817.746337890625,
-      "learning_rate": 2.5191984330694573e-05,
-      "loss": 66.2542,
-      "step": 136950
-    },
-    {
-      "epoch": 0.5533357304750784,
-      "grad_norm": 758.0955810546875,
-      "learning_rate": 2.518849377325893e-05,
-      "loss": 62.4526,
-      "step": 136960
-    },
-    {
-      "epoch": 0.5533761317404461,
-      "grad_norm": 451.2418518066406,
-      "learning_rate": 2.5185003212148507e-05,
-      "loss": 53.2829,
-      "step": 136970
-    },
-    {
-      "epoch": 0.5534165330058137,
-      "grad_norm": 2254.761474609375,
-      "learning_rate": 2.518151264743135e-05,
-      "loss": 50.6571,
-      "step": 136980
-    },
-    {
-      "epoch": 0.5534569342711814,
-      "grad_norm": 1203.5234375,
-      "learning_rate": 2.5178022079175506e-05,
-      "loss": 69.3263,
-      "step": 136990
-    },
-    {
-      "epoch": 0.553497335536549,
-      "grad_norm": 707.7672119140625,
-      "learning_rate": 2.517453150744904e-05,
-      "loss": 65.1655,
-      "step": 137000
-    },
-    {
-      "epoch": 0.5535377368019166,
-      "grad_norm": 330.6707763671875,
-      "learning_rate": 2.517104093231999e-05,
-      "loss": 59.2168,
-      "step": 137010
-    },
-    {
-      "epoch": 0.5535781380672843,
-      "grad_norm": 910.3479614257812,
-      "learning_rate": 2.5167550353856402e-05,
-      "loss": 35.873,
-      "step": 137020
-    },
-    {
-      "epoch": 0.5536185393326519,
-      "grad_norm": 888.7249755859375,
-      "learning_rate": 2.516405977212634e-05,
-      "loss": 48.7685,
-      "step": 137030
-    },
-    {
-      "epoch": 0.5536589405980196,
-      "grad_norm": 696.96533203125,
-      "learning_rate": 2.5160569187197852e-05,
-      "loss": 31.9463,
-      "step": 137040
-    },
-    {
-      "epoch": 0.5536993418633872,
-      "grad_norm": 561.5834350585938,
-      "learning_rate": 2.5157078599138977e-05,
-      "loss": 59.4334,
-      "step": 137050
-    },
-    {
-      "epoch": 0.5537397431287548,
-      "grad_norm": 636.3404541015625,
-      "learning_rate": 2.5153588008017776e-05,
-      "loss": 45.7352,
-      "step": 137060
-    },
-    {
-      "epoch": 0.5537801443941224,
-      "grad_norm": 775.7879028320312,
-      "learning_rate": 2.5150097413902297e-05,
-      "loss": 56.8609,
-      "step": 137070
-    },
-    {
-      "epoch": 0.55382054565949,
-      "grad_norm": 1065.6773681640625,
-      "learning_rate": 2.5146606816860597e-05,
-      "loss": 42.274,
-      "step": 137080
-    },
-    {
-      "epoch": 0.5538609469248577,
-      "grad_norm": 1046.0736083984375,
-      "learning_rate": 2.5143116216960717e-05,
-      "loss": 47.2468,
-      "step": 137090
-    },
-    {
-      "epoch": 0.5539013481902253,
-      "grad_norm": 508.81201171875,
-      "learning_rate": 2.5139625614270706e-05,
-      "loss": 74.2703,
-      "step": 137100
-    },
-    {
-      "epoch": 0.5539417494555929,
-      "grad_norm": 254.36724853515625,
-      "learning_rate": 2.5136135008858637e-05,
-      "loss": 40.8546,
-      "step": 137110
-    },
-    {
-      "epoch": 0.5539821507209606,
-      "grad_norm": 481.7911682128906,
-      "learning_rate": 2.513264440079254e-05,
-      "loss": 61.7229,
-      "step": 137120
-    },
-    {
-      "epoch": 0.5540225519863282,
-      "grad_norm": 1409.0474853515625,
-      "learning_rate": 2.5129153790140463e-05,
-      "loss": 63.4276,
-      "step": 137130
-    },
-    {
-      "epoch": 0.5540629532516959,
-      "grad_norm": 1046.4349365234375,
-      "learning_rate": 2.5125663176970476e-05,
-      "loss": 41.4423,
-      "step": 137140
-    },
-    {
-      "epoch": 0.5541033545170635,
-      "grad_norm": 860.8171997070312,
-      "learning_rate": 2.5122172561350616e-05,
-      "loss": 61.4118,
-      "step": 137150
-    },
-    {
-      "epoch": 0.5541437557824311,
-      "grad_norm": 733.4225463867188,
-      "learning_rate": 2.5118681943348944e-05,
-      "loss": 36.8318,
-      "step": 137160
-    },
-    {
-      "epoch": 0.5541841570477988,
-      "grad_norm": 582.643310546875,
-      "learning_rate": 2.51151913230335e-05,
-      "loss": 37.8533,
-      "step": 137170
-    },
-    {
-      "epoch": 0.5542245583131664,
-      "grad_norm": 1445.744873046875,
-      "learning_rate": 2.5111700700472346e-05,
-      "loss": 70.5182,
-      "step": 137180
-    },
-    {
-      "epoch": 0.5542649595785339,
-      "grad_norm": 709.0853271484375,
-      "learning_rate": 2.5108210075733523e-05,
-      "loss": 45.407,
-      "step": 137190
-    },
-    {
-      "epoch": 0.5543053608439016,
-      "grad_norm": 413.23663330078125,
-      "learning_rate": 2.51047194488851e-05,
-      "loss": 57.8371,
-      "step": 137200
-    },
-    {
-      "epoch": 0.5543457621092692,
-      "grad_norm": 660.6998901367188,
-      "learning_rate": 2.5101228819995116e-05,
-      "loss": 75.2915,
-      "step": 137210
-    },
-    {
-      "epoch": 0.5543861633746369,
-      "grad_norm": 959.9091796875,
-      "learning_rate": 2.5097738189131614e-05,
-      "loss": 69.5816,
-      "step": 137220
-    },
-    {
-      "epoch": 0.5544265646400045,
-      "grad_norm": 401.2200012207031,
-      "learning_rate": 2.509424755636266e-05,
-      "loss": 53.4176,
-      "step": 137230
-    },
-    {
-      "epoch": 0.5544669659053721,
-      "grad_norm": 673.7756958007812,
-      "learning_rate": 2.509075692175631e-05,
-      "loss": 47.1198,
-      "step": 137240
-    },
-    {
-      "epoch": 0.5545073671707398,
-      "grad_norm": 1097.22998046875,
-      "learning_rate": 2.5087266285380596e-05,
-      "loss": 68.0358,
-      "step": 137250
-    },
-    {
-      "epoch": 0.5545477684361074,
-      "grad_norm": 893.8514404296875,
-      "learning_rate": 2.508377564730358e-05,
-      "loss": 75.7999,
-      "step": 137260
-    },
-    {
-      "epoch": 0.5545881697014751,
-      "grad_norm": 618.3114624023438,
-      "learning_rate": 2.508028500759332e-05,
-      "loss": 61.6705,
-      "step": 137270
-    },
-    {
-      "epoch": 0.5546285709668427,
-      "grad_norm": 694.9136962890625,
-      "learning_rate": 2.5076794366317867e-05,
-      "loss": 71.5496,
-      "step": 137280
-    },
-    {
-      "epoch": 0.5546689722322103,
-      "grad_norm": 558.5074462890625,
-      "learning_rate": 2.5073303723545265e-05,
-      "loss": 68.4956,
-      "step": 137290
-    },
-    {
-      "epoch": 0.554709373497578,
-      "grad_norm": 609.846435546875,
-      "learning_rate": 2.506981307934357e-05,
-      "loss": 63.5097,
-      "step": 137300
-    },
-    {
-      "epoch": 0.5547497747629456,
-      "grad_norm": 332.0170593261719,
-      "learning_rate": 2.5066322433780827e-05,
-      "loss": 33.8037,
-      "step": 137310
-    },
-    {
-      "epoch": 0.5547901760283132,
-      "grad_norm": 1042.158447265625,
-      "learning_rate": 2.5062831786925102e-05,
-      "loss": 43.3467,
-      "step": 137320
-    },
-    {
-      "epoch": 0.5548305772936808,
-      "grad_norm": 1328.8170166015625,
-      "learning_rate": 2.505934113884443e-05,
-      "loss": 59.4562,
-      "step": 137330
-    },
-    {
-      "epoch": 0.5548709785590484,
-      "grad_norm": 1468.5960693359375,
-      "learning_rate": 2.5055850489606875e-05,
-      "loss": 46.9289,
-      "step": 137340
-    },
-    {
-      "epoch": 0.5549113798244161,
-      "grad_norm": 968.5849609375,
-      "learning_rate": 2.505235983928048e-05,
-      "loss": 54.1335,
-      "step": 137350
-    },
-    {
-      "epoch": 0.5549517810897837,
-      "grad_norm": 1135.44091796875,
-      "learning_rate": 2.5048869187933316e-05,
-      "loss": 44.6849,
-      "step": 137360
-    },
-    {
-      "epoch": 0.5549921823551514,
-      "grad_norm": 748.2118530273438,
-      "learning_rate": 2.504537853563342e-05,
-      "loss": 50.3875,
-      "step": 137370
-    },
-    {
-      "epoch": 0.555032583620519,
-      "grad_norm": 301.2623596191406,
-      "learning_rate": 2.5041887882448844e-05,
-      "loss": 67.9134,
-      "step": 137380
-    },
-    {
-      "epoch": 0.5550729848858866,
-      "grad_norm": 1220.4659423828125,
-      "learning_rate": 2.5038397228447634e-05,
-      "loss": 26.7308,
-      "step": 137390
-    },
-    {
-      "epoch": 0.5551133861512543,
-      "grad_norm": 1018.2291259765625,
-      "learning_rate": 2.5034906573697864e-05,
-      "loss": 49.0374,
-      "step": 137400
-    },
-    {
-      "epoch": 0.5551537874166219,
-      "grad_norm": 808.0350952148438,
-      "learning_rate": 2.5031415918267564e-05,
-      "loss": 77.4769,
-      "step": 137410
-    },
-    {
-      "epoch": 0.5551941886819896,
-      "grad_norm": 1768.087646484375,
-      "learning_rate": 2.5027925262224795e-05,
-      "loss": 77.0356,
-      "step": 137420
-    },
-    {
-      "epoch": 0.5552345899473572,
-      "grad_norm": 568.6273803710938,
-      "learning_rate": 2.5024434605637604e-05,
-      "loss": 68.0419,
-      "step": 137430
-    },
-    {
-      "epoch": 0.5552749912127248,
-      "grad_norm": 688.5547485351562,
-      "learning_rate": 2.5020943948574055e-05,
-      "loss": 62.4521,
-      "step": 137440
-    },
-    {
-      "epoch": 0.5553153924780924,
-      "grad_norm": 561.1484985351562,
-      "learning_rate": 2.501745329110219e-05,
-      "loss": 47.788,
-      "step": 137450
-    },
-    {
-      "epoch": 0.55535579374346,
-      "grad_norm": 1244.0697021484375,
-      "learning_rate": 2.5013962633290072e-05,
-      "loss": 47.7701,
-      "step": 137460
-    },
-    {
-      "epoch": 0.5553961950088276,
-      "grad_norm": 647.6009521484375,
-      "learning_rate": 2.5010471975205733e-05,
-      "loss": 69.8529,
-      "step": 137470
-    },
-    {
-      "epoch": 0.5554365962741953,
-      "grad_norm": 2588.730712890625,
-      "learning_rate": 2.500698131691725e-05,
-      "loss": 54.0912,
-      "step": 137480
-    },
-    {
-      "epoch": 0.5554769975395629,
-      "grad_norm": 540.1988525390625,
-      "learning_rate": 2.500349065849265e-05,
-      "loss": 78.9863,
-      "step": 137490
-    },
-    {
-      "epoch": 0.5555173988049306,
-      "grad_norm": 1182.00732421875,
-      "learning_rate": 2.5e-05,
-      "loss": 77.0684,
-      "step": 137500
-    },
-    {
-      "epoch": 0.5555578000702982,
-      "grad_norm": 668.05517578125,
-      "learning_rate": 2.499650934150736e-05,
-      "loss": 35.7313,
-      "step": 137510
-    },
-    {
-      "epoch": 0.5555982013356658,
-      "grad_norm": 337.5390930175781,
-      "learning_rate": 2.499301868308276e-05,
-      "loss": 47.2015,
-      "step": 137520
-    },
-    {
-      "epoch": 0.5556386026010335,
-      "grad_norm": 656.7407836914062,
-      "learning_rate": 2.498952802479427e-05,
-      "loss": 43.5451,
-      "step": 137530
-    },
-    {
-      "epoch": 0.5556790038664011,
-      "grad_norm": 710.9033813476562,
-      "learning_rate": 2.4986037366709937e-05,
-      "loss": 49.4604,
-      "step": 137540
-    },
-    {
-      "epoch": 0.5557194051317688,
-      "grad_norm": 1154.9649658203125,
-      "learning_rate": 2.4982546708897814e-05,
-      "loss": 59.851,
-      "step": 137550
-    },
-    {
-      "epoch": 0.5557598063971364,
-      "grad_norm": 548.6576538085938,
-      "learning_rate": 2.4979056051425954e-05,
-      "loss": 92.8117,
-      "step": 137560
-    },
-    {
-      "epoch": 0.555800207662504,
-      "grad_norm": 622.8199462890625,
-      "learning_rate": 2.4975565394362395e-05,
-      "loss": 37.9003,
-      "step": 137570
-    },
-    {
-      "epoch": 0.5558406089278716,
-      "grad_norm": 559.3489379882812,
-      "learning_rate": 2.4972074737775214e-05,
-      "loss": 48.2217,
-      "step": 137580
-    },
-    {
-      "epoch": 0.5558810101932392,
-      "grad_norm": 1215.4371337890625,
-      "learning_rate": 2.4968584081732448e-05,
-      "loss": 83.8556,
-      "step": 137590
-    },
-    {
-      "epoch": 0.5559214114586069,
-      "grad_norm": 371.1655578613281,
-      "learning_rate": 2.496509342630214e-05,
-      "loss": 40.4336,
-      "step": 137600
-    },
-    {
-      "epoch": 0.5559618127239745,
-      "grad_norm": 3215.786376953125,
-      "learning_rate": 2.496160277155237e-05,
-      "loss": 81.3515,
-      "step": 137610
-    },
-    {
-      "epoch": 0.5560022139893421,
-      "grad_norm": 420.7957458496094,
-      "learning_rate": 2.4958112117551165e-05,
-      "loss": 84.0413,
-      "step": 137620
-    },
-    {
-      "epoch": 0.5560426152547098,
-      "grad_norm": 889.9630737304688,
-      "learning_rate": 2.495462146436659e-05,
-      "loss": 46.8617,
-      "step": 137630
-    },
-    {
-      "epoch": 0.5560830165200774,
-      "grad_norm": 1820.2559814453125,
-      "learning_rate": 2.4951130812066686e-05,
-      "loss": 71.0172,
-      "step": 137640
-    },
-    {
-      "epoch": 0.556123417785445,
-      "grad_norm": 1237.7259521484375,
-      "learning_rate": 2.4947640160719515e-05,
-      "loss": 27.0079,
-      "step": 137650
-    },
-    {
-      "epoch": 0.5561638190508127,
-      "grad_norm": 641.0753784179688,
-      "learning_rate": 2.4944149510393134e-05,
-      "loss": 57.1568,
-      "step": 137660
-    },
-    {
-      "epoch": 0.5562042203161803,
-      "grad_norm": 450.4702453613281,
-      "learning_rate": 2.4940658861155584e-05,
-      "loss": 50.1699,
-      "step": 137670
-    },
-    {
-      "epoch": 0.556244621581548,
-      "grad_norm": 1086.3056640625,
-      "learning_rate": 2.4937168213074907e-05,
-      "loss": 46.7683,
-      "step": 137680
-    },
-    {
-      "epoch": 0.5562850228469156,
-      "grad_norm": 404.1825866699219,
-      "learning_rate": 2.493367756621918e-05,
-      "loss": 35.0012,
-      "step": 137690
-    },
-    {
-      "epoch": 0.5563254241122833,
-      "grad_norm": 943.6931762695312,
-      "learning_rate": 2.493018692065644e-05,
-      "loss": 54.1753,
-      "step": 137700
-    },
-    {
-      "epoch": 0.5563658253776508,
-      "grad_norm": 530.2384033203125,
-      "learning_rate": 2.4926696276454737e-05,
-      "loss": 41.3772,
-      "step": 137710
-    },
-    {
-      "epoch": 0.5564062266430184,
-      "grad_norm": 347.5144348144531,
-      "learning_rate": 2.492320563368214e-05,
-      "loss": 44.261,
-      "step": 137720
-    },
-    {
-      "epoch": 0.5564466279083861,
-      "grad_norm": 996.7993774414062,
-      "learning_rate": 2.4919714992406677e-05,
-      "loss": 74.4061,
-      "step": 137730
-    },
-    {
-      "epoch": 0.5564870291737537,
-      "grad_norm": 922.4503173828125,
-      "learning_rate": 2.491622435269642e-05,
-      "loss": 62.489,
-      "step": 137740
-    },
-    {
-      "epoch": 0.5565274304391213,
-      "grad_norm": 500.00689697265625,
-      "learning_rate": 2.4912733714619417e-05,
-      "loss": 28.2802,
-      "step": 137750
-    },
-    {
-      "epoch": 0.556567831704489,
-      "grad_norm": 333.4855041503906,
-      "learning_rate": 2.4909243078243696e-05,
-      "loss": 47.8612,
-      "step": 137760
-    },
-    {
-      "epoch": 0.5566082329698566,
-      "grad_norm": 898.7628173828125,
-      "learning_rate": 2.4905752443637345e-05,
-      "loss": 46.9569,
-      "step": 137770
-    },
-    {
-      "epoch": 0.5566486342352243,
-      "grad_norm": 303.6540832519531,
-      "learning_rate": 2.490226181086838e-05,
-      "loss": 44.0417,
-      "step": 137780
-    },
-    {
-      "epoch": 0.5566890355005919,
-      "grad_norm": 8116.3369140625,
-      "learning_rate": 2.489877118000489e-05,
-      "loss": 56.3838,
-      "step": 137790
-    },
-    {
-      "epoch": 0.5567294367659595,
-      "grad_norm": 549.66943359375,
-      "learning_rate": 2.4895280551114907e-05,
-      "loss": 50.9485,
-      "step": 137800
-    },
-    {
-      "epoch": 0.5567698380313272,
-      "grad_norm": 622.78564453125,
-      "learning_rate": 2.4891789924266476e-05,
-      "loss": 44.1438,
-      "step": 137810
-    },
-    {
-      "epoch": 0.5568102392966948,
-      "grad_norm": 762.0851440429688,
-      "learning_rate": 2.4888299299527663e-05,
-      "loss": 62.1743,
-      "step": 137820
-    },
-    {
-      "epoch": 0.5568506405620623,
-      "grad_norm": 685.5193481445312,
-      "learning_rate": 2.488480867696651e-05,
-      "loss": 33.5156,
-      "step": 137830
-    },
-    {
-      "epoch": 0.55689104182743,
-      "grad_norm": 534.5140991210938,
-      "learning_rate": 2.4881318056651062e-05,
-      "loss": 50.1773,
-      "step": 137840
-    },
-    {
-      "epoch": 0.5569314430927976,
-      "grad_norm": 592.9160766601562,
-      "learning_rate": 2.4877827438649393e-05,
-      "loss": 36.2395,
-      "step": 137850
-    },
-    {
-      "epoch": 0.5569718443581653,
-      "grad_norm": 296.64471435546875,
-      "learning_rate": 2.4874336823029526e-05,
-      "loss": 79.5094,
-      "step": 137860
-    },
-    {
-      "epoch": 0.5570122456235329,
-      "grad_norm": 933.6554565429688,
-      "learning_rate": 2.487084620985954e-05,
-      "loss": 67.565,
-      "step": 137870
-    },
-    {
-      "epoch": 0.5570526468889005,
-      "grad_norm": 498.0602722167969,
-      "learning_rate": 2.4867355599207474e-05,
-      "loss": 55.1443,
-      "step": 137880
-    },
-    {
-      "epoch": 0.5570930481542682,
-      "grad_norm": 635.6414794921875,
-      "learning_rate": 2.4863864991141372e-05,
-      "loss": 50.4595,
-      "step": 137890
-    },
-    {
-      "epoch": 0.5571334494196358,
-      "grad_norm": 509.6726379394531,
-      "learning_rate": 2.48603743857293e-05,
-      "loss": 45.6835,
-      "step": 137900
-    },
-    {
-      "epoch": 0.5571738506850035,
-      "grad_norm": 458.8603515625,
-      "learning_rate": 2.48568837830393e-05,
-      "loss": 52.0281,
-      "step": 137910
-    },
-    {
-      "epoch": 0.5572142519503711,
-      "grad_norm": 1398.501220703125,
-      "learning_rate": 2.4853393183139412e-05,
-      "loss": 74.1148,
-      "step": 137920
-    },
-    {
-      "epoch": 0.5572546532157387,
-      "grad_norm": 501.7352600097656,
-      "learning_rate": 2.484990258609771e-05,
-      "loss": 38.4496,
-      "step": 137930
-    },
-    {
-      "epoch": 0.5572950544811064,
-      "grad_norm": 354.11016845703125,
-      "learning_rate": 2.4846411991982223e-05,
-      "loss": 52.2043,
-      "step": 137940
-    },
-    {
-      "epoch": 0.557335455746474,
-      "grad_norm": 889.2261352539062,
-      "learning_rate": 2.484292140086103e-05,
-      "loss": 52.7851,
-      "step": 137950
-    },
-    {
-      "epoch": 0.5573758570118416,
-      "grad_norm": 566.2061767578125,
-      "learning_rate": 2.4839430812802157e-05,
-      "loss": 75.0353,
-      "step": 137960
-    },
-    {
-      "epoch": 0.5574162582772092,
-      "grad_norm": 1404.2596435546875,
-      "learning_rate": 2.4835940227873662e-05,
-      "loss": 66.9802,
-      "step": 137970
-    },
-    {
-      "epoch": 0.5574566595425768,
-      "grad_norm": 673.5670166015625,
-      "learning_rate": 2.4832449646143604e-05,
-      "loss": 63.724,
-      "step": 137980
-    },
-    {
-      "epoch": 0.5574970608079445,
-      "grad_norm": 693.6373291015625,
-      "learning_rate": 2.482895906768002e-05,
-      "loss": 52.0077,
-      "step": 137990
-    },
-    {
-      "epoch": 0.5575374620733121,
-      "grad_norm": 672.2484741210938,
-      "learning_rate": 2.4825468492550964e-05,
-      "loss": 39.796,
-      "step": 138000
-    },
-    {
-      "epoch": 0.5575778633386798,
-      "grad_norm": 1082.0235595703125,
-      "learning_rate": 2.4821977920824497e-05,
-      "loss": 39.0499,
-      "step": 138010
-    },
-    {
-      "epoch": 0.5576182646040474,
-      "grad_norm": 1062.3084716796875,
-      "learning_rate": 2.481848735256865e-05,
-      "loss": 76.0777,
-      "step": 138020
-    },
-    {
-      "epoch": 0.557658665869415,
-      "grad_norm": 1710.125732421875,
-      "learning_rate": 2.4814996787851495e-05,
-      "loss": 61.158,
-      "step": 138030
-    },
-    {
-      "epoch": 0.5576990671347827,
-      "grad_norm": 868.0357055664062,
-      "learning_rate": 2.481150622674108e-05,
-      "loss": 42.1739,
-      "step": 138040
-    },
-    {
-      "epoch": 0.5577394684001503,
-      "grad_norm": 746.6460571289062,
-      "learning_rate": 2.4808015669305433e-05,
-      "loss": 50.9672,
-      "step": 138050
-    },
-    {
-      "epoch": 0.557779869665518,
-      "grad_norm": 444.3233337402344,
-      "learning_rate": 2.480452511561263e-05,
-      "loss": 37.3236,
-      "step": 138060
-    },
-    {
-      "epoch": 0.5578202709308856,
-      "grad_norm": 914.9209594726562,
-      "learning_rate": 2.48010345657307e-05,
-      "loss": 51.8856,
-      "step": 138070
-    },
-    {
-      "epoch": 0.5578606721962532,
-      "grad_norm": 663.5612182617188,
-      "learning_rate": 2.4797544019727717e-05,
-      "loss": 28.8472,
-      "step": 138080
-    },
-    {
-      "epoch": 0.5579010734616208,
-      "grad_norm": 731.248046875,
-      "learning_rate": 2.4794053477671713e-05,
-      "loss": 43.9837,
-      "step": 138090
-    },
-    {
-      "epoch": 0.5579414747269884,
-      "grad_norm": 371.97589111328125,
-      "learning_rate": 2.4790562939630734e-05,
-      "loss": 35.7051,
-      "step": 138100
-    },
-    {
-      "epoch": 0.557981875992356,
-      "grad_norm": 582.62353515625,
-      "learning_rate": 2.478707240567285e-05,
-      "loss": 46.4093,
-      "step": 138110
-    },
-    {
-      "epoch": 0.5580222772577237,
-      "grad_norm": 626.8383178710938,
-      "learning_rate": 2.4783581875866097e-05,
-      "loss": 62.2491,
-      "step": 138120
-    },
-    {
-      "epoch": 0.5580626785230913,
-      "grad_norm": 956.4906005859375,
-      "learning_rate": 2.4780091350278514e-05,
-      "loss": 45.8993,
-      "step": 138130
-    },
-    {
-      "epoch": 0.558103079788459,
-      "grad_norm": 1804.482421875,
-      "learning_rate": 2.477660082897818e-05,
-      "loss": 69.7101,
-      "step": 138140
-    },
-    {
-      "epoch": 0.5581434810538266,
-      "grad_norm": 856.2649536132812,
-      "learning_rate": 2.4773110312033117e-05,
-      "loss": 50.3831,
-      "step": 138150
-    },
-    {
-      "epoch": 0.5581838823191942,
-      "grad_norm": 671.4500732421875,
-      "learning_rate": 2.4769619799511393e-05,
-      "loss": 64.6485,
-      "step": 138160
-    },
-    {
-      "epoch": 0.5582242835845619,
-      "grad_norm": 861.5789184570312,
-      "learning_rate": 2.4766129291481048e-05,
-      "loss": 77.0027,
-      "step": 138170
-    },
-    {
-      "epoch": 0.5582646848499295,
-      "grad_norm": 462.4314880371094,
-      "learning_rate": 2.4762638788010122e-05,
-      "loss": 60.2575,
-      "step": 138180
-    },
-    {
-      "epoch": 0.5583050861152972,
-      "grad_norm": 970.7669067382812,
-      "learning_rate": 2.4759148289166687e-05,
-      "loss": 98.2652,
-      "step": 138190
-    },
-    {
-      "epoch": 0.5583454873806648,
-      "grad_norm": 1482.3714599609375,
-      "learning_rate": 2.475565779501878e-05,
-      "loss": 41.1083,
-      "step": 138200
-    },
-    {
-      "epoch": 0.5583858886460324,
-      "grad_norm": 703.8884887695312,
-      "learning_rate": 2.475216730563444e-05,
-      "loss": 49.8838,
-      "step": 138210
-    },
-    {
-      "epoch": 0.5584262899114,
-      "grad_norm": 150.86387634277344,
-      "learning_rate": 2.4748676821081738e-05,
-      "loss": 40.4376,
-      "step": 138220
-    },
-    {
-      "epoch": 0.5584666911767676,
-      "grad_norm": 944.3568725585938,
-      "learning_rate": 2.47451863414287e-05,
-      "loss": 57.4413,
-      "step": 138230
-    },
-    {
-      "epoch": 0.5585070924421353,
-      "grad_norm": 659.5052490234375,
-      "learning_rate": 2.4741695866743397e-05,
-      "loss": 47.3132,
-      "step": 138240
-    },
-    {
-      "epoch": 0.5585474937075029,
-      "grad_norm": 506.6811828613281,
-      "learning_rate": 2.4738205397093864e-05,
-      "loss": 78.2513,
-      "step": 138250
-    },
-    {
-      "epoch": 0.5585878949728705,
-      "grad_norm": 689.1748046875,
-      "learning_rate": 2.473471493254814e-05,
-      "loss": 60.5565,
-      "step": 138260
-    },
-    {
-      "epoch": 0.5586282962382382,
-      "grad_norm": 548.3529052734375,
-      "learning_rate": 2.4731224473174295e-05,
-      "loss": 54.7061,
-      "step": 138270
-    },
-    {
-      "epoch": 0.5586686975036058,
-      "grad_norm": 626.2839965820312,
-      "learning_rate": 2.472773401904037e-05,
-      "loss": 59.1142,
-      "step": 138280
-    },
-    {
-      "epoch": 0.5587090987689735,
-      "grad_norm": 472.8750305175781,
-      "learning_rate": 2.4724243570214398e-05,
-      "loss": 56.8832,
-      "step": 138290
-    },
-    {
-      "epoch": 0.5587495000343411,
-      "grad_norm": 623.64794921875,
-      "learning_rate": 2.4720753126764447e-05,
-      "loss": 41.724,
-      "step": 138300
-    },
-    {
-      "epoch": 0.5587899012997087,
-      "grad_norm": 493.74017333984375,
-      "learning_rate": 2.4717262688758557e-05,
-      "loss": 35.2544,
-      "step": 138310
-    },
-    {
-      "epoch": 0.5588303025650764,
-      "grad_norm": 706.80126953125,
-      "learning_rate": 2.471377225626478e-05,
-      "loss": 64.1044,
-      "step": 138320
-    },
-    {
-      "epoch": 0.558870703830444,
-      "grad_norm": 702.6170043945312,
-      "learning_rate": 2.4710281829351156e-05,
-      "loss": 26.0041,
-      "step": 138330
-    },
-    {
-      "epoch": 0.5589111050958117,
-      "grad_norm": 1678.392333984375,
-      "learning_rate": 2.4706791408085736e-05,
-      "loss": 70.0253,
-      "step": 138340
-    },
-    {
-      "epoch": 0.5589515063611792,
-      "grad_norm": 1137.38818359375,
-      "learning_rate": 2.4703300992536573e-05,
-      "loss": 63.3032,
-      "step": 138350
-    },
-    {
-      "epoch": 0.5589919076265468,
-      "grad_norm": 1046.39990234375,
-      "learning_rate": 2.4699810582771713e-05,
-      "loss": 52.2509,
-      "step": 138360
-    },
-    {
-      "epoch": 0.5590323088919145,
-      "grad_norm": 657.4580078125,
-      "learning_rate": 2.4696320178859187e-05,
-      "loss": 52.0369,
-      "step": 138370
-    },
-    {
-      "epoch": 0.5590727101572821,
-      "grad_norm": 866.4102172851562,
-      "learning_rate": 2.4692829780867065e-05,
-      "loss": 56.7212,
-      "step": 138380
-    },
-    {
-      "epoch": 0.5591131114226497,
-      "grad_norm": 531.6983642578125,
-      "learning_rate": 2.4689339388863374e-05,
-      "loss": 41.7753,
-      "step": 138390
-    },
-    {
-      "epoch": 0.5591535126880174,
-      "grad_norm": 506.9232482910156,
-      "learning_rate": 2.4685849002916183e-05,
-      "loss": 34.2688,
-      "step": 138400
-    },
-    {
-      "epoch": 0.559193913953385,
-      "grad_norm": 803.9929809570312,
-      "learning_rate": 2.468235862309353e-05,
-      "loss": 52.5372,
-      "step": 138410
-    },
-    {
-      "epoch": 0.5592343152187527,
-      "grad_norm": 1219.44970703125,
-      "learning_rate": 2.4678868249463452e-05,
-      "loss": 62.4074,
-      "step": 138420
-    },
-    {
-      "epoch": 0.5592747164841203,
-      "grad_norm": 802.236328125,
-      "learning_rate": 2.4675377882094007e-05,
-      "loss": 39.6712,
-      "step": 138430
-    },
-    {
-      "epoch": 0.5593151177494879,
-      "grad_norm": 560.7493896484375,
-      "learning_rate": 2.4671887521053237e-05,
-      "loss": 71.1677,
-      "step": 138440
-    },
-    {
-      "epoch": 0.5593555190148556,
-      "grad_norm": 443.1597595214844,
-      "learning_rate": 2.466839716640918e-05,
-      "loss": 43.8879,
-      "step": 138450
-    },
-    {
-      "epoch": 0.5593959202802232,
-      "grad_norm": 4036.05810546875,
-      "learning_rate": 2.46649068182299e-05,
-      "loss": 84.4937,
-      "step": 138460
-    },
-    {
-      "epoch": 0.5594363215455908,
-      "grad_norm": 1169.473876953125,
-      "learning_rate": 2.4661416476583428e-05,
-      "loss": 52.9824,
-      "step": 138470
-    },
-    {
-      "epoch": 0.5594767228109584,
-      "grad_norm": 186.98561096191406,
-      "learning_rate": 2.465792614153782e-05,
-      "loss": 74.6176,
-      "step": 138480
-    },
-    {
-      "epoch": 0.559517124076326,
-      "grad_norm": 1336.8724365234375,
-      "learning_rate": 2.465443581316112e-05,
-      "loss": 56.0161,
-      "step": 138490
-    },
-    {
-      "epoch": 0.5595575253416937,
-      "grad_norm": 1130.03564453125,
-      "learning_rate": 2.4650945491521372e-05,
-      "loss": 76.923,
-      "step": 138500
-    },
-    {
-      "epoch": 0.5595979266070613,
-      "grad_norm": 351.6864318847656,
-      "learning_rate": 2.4647455176686622e-05,
-      "loss": 51.3765,
-      "step": 138510
-    },
-    {
-      "epoch": 0.559638327872429,
-      "grad_norm": 799.4815063476562,
-      "learning_rate": 2.4643964868724914e-05,
-      "loss": 77.3449,
-      "step": 138520
-    },
-    {
-      "epoch": 0.5596787291377966,
-      "grad_norm": 691.9085693359375,
-      "learning_rate": 2.4640474567704285e-05,
-      "loss": 38.7629,
-      "step": 138530
-    },
-    {
-      "epoch": 0.5597191304031642,
-      "grad_norm": 446.3203125,
-      "learning_rate": 2.46369842736928e-05,
-      "loss": 55.0333,
-      "step": 138540
-    },
-    {
-      "epoch": 0.5597595316685319,
-      "grad_norm": 991.9611206054688,
-      "learning_rate": 2.4633493986758484e-05,
-      "loss": 54.1677,
-      "step": 138550
-    },
-    {
-      "epoch": 0.5597999329338995,
-      "grad_norm": 1506.22607421875,
-      "learning_rate": 2.46300037069694e-05,
-      "loss": 37.269,
-      "step": 138560
-    },
-    {
-      "epoch": 0.5598403341992672,
-      "grad_norm": 915.8712768554688,
-      "learning_rate": 2.4626513434393584e-05,
-      "loss": 71.686,
-      "step": 138570
-    },
-    {
-      "epoch": 0.5598807354646348,
-      "grad_norm": 507.74920654296875,
-      "learning_rate": 2.4623023169099073e-05,
-      "loss": 52.8719,
-      "step": 138580
-    },
-    {
-      "epoch": 0.5599211367300024,
-      "grad_norm": 471.32745361328125,
-      "learning_rate": 2.4619532911153932e-05,
-      "loss": 57.2638,
-      "step": 138590
-    },
-    {
-      "epoch": 0.55996153799537,
-      "grad_norm": 660.4644775390625,
-      "learning_rate": 2.4616042660626177e-05,
-      "loss": 38.7556,
-      "step": 138600
-    },
-    {
-      "epoch": 0.5600019392607376,
-      "grad_norm": 1198.232421875,
-      "learning_rate": 2.461255241758388e-05,
-      "loss": 34.7362,
-      "step": 138610
-    },
-    {
-      "epoch": 0.5600423405261052,
-      "grad_norm": 1170.1473388671875,
-      "learning_rate": 2.4609062182095072e-05,
-      "loss": 56.8247,
-      "step": 138620
-    },
-    {
-      "epoch": 0.5600827417914729,
-      "grad_norm": 505.6723327636719,
-      "learning_rate": 2.4605571954227787e-05,
-      "loss": 62.388,
-      "step": 138630
-    },
-    {
-      "epoch": 0.5601231430568405,
-      "grad_norm": 983.4881591796875,
-      "learning_rate": 2.4602081734050093e-05,
-      "loss": 29.2795,
-      "step": 138640
-    },
-    {
-      "epoch": 0.5601635443222082,
-      "grad_norm": 346.18719482421875,
-      "learning_rate": 2.459859152163002e-05,
-      "loss": 65.6756,
-      "step": 138650
-    },
-    {
-      "epoch": 0.5602039455875758,
-      "grad_norm": 1048.971923828125,
-      "learning_rate": 2.4595101317035603e-05,
-      "loss": 60.1032,
-      "step": 138660
-    },
-    {
-      "epoch": 0.5602443468529434,
-      "grad_norm": 1085.2611083984375,
-      "learning_rate": 2.4591611120334898e-05,
-      "loss": 61.6514,
-      "step": 138670
-    },
-    {
-      "epoch": 0.5602847481183111,
-      "grad_norm": 320.9248046875,
-      "learning_rate": 2.4588120931595947e-05,
-      "loss": 38.1592,
-      "step": 138680
-    },
-    {
-      "epoch": 0.5603251493836787,
-      "grad_norm": 478.6922302246094,
-      "learning_rate": 2.458463075088679e-05,
-      "loss": 59.3947,
-      "step": 138690
-    },
-    {
-      "epoch": 0.5603655506490464,
-      "grad_norm": 874.609130859375,
-      "learning_rate": 2.458114057827547e-05,
-      "loss": 61.9768,
-      "step": 138700
-    },
-    {
-      "epoch": 0.560405951914414,
-      "grad_norm": 894.510498046875,
-      "learning_rate": 2.4577650413830025e-05,
-      "loss": 46.5881,
-      "step": 138710
-    },
-    {
-      "epoch": 0.5604463531797816,
-      "grad_norm": 621.612548828125,
-      "learning_rate": 2.4574160257618508e-05,
-      "loss": 43.6552,
-      "step": 138720
-    },
-    {
-      "epoch": 0.5604867544451492,
-      "grad_norm": 1093.0335693359375,
-      "learning_rate": 2.457067010970896e-05,
-      "loss": 50.9287,
-      "step": 138730
-    },
-    {
-      "epoch": 0.5605271557105168,
-      "grad_norm": 519.4285278320312,
-      "learning_rate": 2.4567179970169407e-05,
-      "loss": 45.5831,
-      "step": 138740
-    },
-    {
-      "epoch": 0.5605675569758845,
-      "grad_norm": 280.7900695800781,
-      "learning_rate": 2.4563689839067913e-05,
-      "loss": 70.7343,
-      "step": 138750
-    },
-    {
-      "epoch": 0.5606079582412521,
-      "grad_norm": 411.0301513671875,
-      "learning_rate": 2.4560199716472508e-05,
-      "loss": 34.1959,
-      "step": 138760
-    },
-    {
-      "epoch": 0.5606483595066197,
-      "grad_norm": 399.39208984375,
-      "learning_rate": 2.4556709602451238e-05,
-      "loss": 54.2914,
-      "step": 138770
-    },
-    {
-      "epoch": 0.5606887607719874,
-      "grad_norm": 739.2390747070312,
-      "learning_rate": 2.4553219497072143e-05,
-      "loss": 58.249,
-      "step": 138780
-    },
-    {
-      "epoch": 0.560729162037355,
-      "grad_norm": 522.2506713867188,
-      "learning_rate": 2.4549729400403254e-05,
-      "loss": 38.7297,
-      "step": 138790
-    },
-    {
-      "epoch": 0.5607695633027227,
-      "grad_norm": 320.0365295410156,
-      "learning_rate": 2.4546239312512635e-05,
-      "loss": 33.2641,
-      "step": 138800
-    },
-    {
-      "epoch": 0.5608099645680903,
-      "grad_norm": 449.9502258300781,
-      "learning_rate": 2.454274923346831e-05,
-      "loss": 29.3585,
-      "step": 138810
-    },
-    {
-      "epoch": 0.5608503658334579,
-      "grad_norm": 849.8306884765625,
-      "learning_rate": 2.4539259163338315e-05,
-      "loss": 35.6265,
-      "step": 138820
-    },
-    {
-      "epoch": 0.5608907670988256,
-      "grad_norm": 693.4956665039062,
-      "learning_rate": 2.4535769102190712e-05,
-      "loss": 39.2643,
-      "step": 138830
-    },
-    {
-      "epoch": 0.5609311683641932,
-      "grad_norm": 631.84423828125,
-      "learning_rate": 2.4532279050093523e-05,
-      "loss": 42.8962,
-      "step": 138840
-    },
-    {
-      "epoch": 0.5609715696295609,
-      "grad_norm": 846.31298828125,
-      "learning_rate": 2.45287890071148e-05,
-      "loss": 58.0521,
-      "step": 138850
-    },
-    {
-      "epoch": 0.5610119708949284,
-      "grad_norm": 468.14959716796875,
-      "learning_rate": 2.4525298973322576e-05,
-      "loss": 60.8873,
-      "step": 138860
-    },
-    {
-      "epoch": 0.561052372160296,
-      "grad_norm": 1028.050048828125,
-      "learning_rate": 2.4521808948784886e-05,
-      "loss": 40.6721,
-      "step": 138870
-    },
-    {
-      "epoch": 0.5610927734256637,
-      "grad_norm": 2670.3349609375,
-      "learning_rate": 2.4518318933569785e-05,
-      "loss": 65.057,
-      "step": 138880
-    },
-    {
-      "epoch": 0.5611331746910313,
-      "grad_norm": 1479.5867919921875,
-      "learning_rate": 2.4514828927745305e-05,
-      "loss": 71.9667,
-      "step": 138890
-    },
-    {
-      "epoch": 0.5611735759563989,
-      "grad_norm": 1349.51123046875,
-      "learning_rate": 2.4511338931379473e-05,
-      "loss": 71.3655,
-      "step": 138900
-    },
-    {
-      "epoch": 0.5612139772217666,
-      "grad_norm": 686.8629760742188,
-      "learning_rate": 2.4507848944540355e-05,
-      "loss": 52.6041,
-      "step": 138910
-    },
-    {
-      "epoch": 0.5612543784871342,
-      "grad_norm": 901.5208740234375,
-      "learning_rate": 2.4504358967295966e-05,
-      "loss": 75.5176,
-      "step": 138920
-    },
-    {
-      "epoch": 0.5612947797525019,
-      "grad_norm": 729.52197265625,
-      "learning_rate": 2.450086899971436e-05,
-      "loss": 47.1792,
-      "step": 138930
-    },
-    {
-      "epoch": 0.5613351810178695,
-      "grad_norm": 546.037841796875,
-      "learning_rate": 2.449737904186357e-05,
-      "loss": 43.953,
-      "step": 138940
-    },
-    {
-      "epoch": 0.5613755822832371,
-      "grad_norm": 1656.893798828125,
-      "learning_rate": 2.4493889093811622e-05,
-      "loss": 48.267,
-      "step": 138950
-    },
-    {
-      "epoch": 0.5614159835486048,
-      "grad_norm": 861.8642578125,
-      "learning_rate": 2.449039915562658e-05,
-      "loss": 74.9551,
-      "step": 138960
-    },
-    {
-      "epoch": 0.5614563848139724,
-      "grad_norm": 749.0470581054688,
-      "learning_rate": 2.4486909227376466e-05,
-      "loss": 65.9108,
-      "step": 138970
-    },
-    {
-      "epoch": 0.5614967860793401,
-      "grad_norm": 1193.1160888671875,
-      "learning_rate": 2.4483419309129315e-05,
-      "loss": 53.8905,
-      "step": 138980
-    },
-    {
-      "epoch": 0.5615371873447076,
-      "grad_norm": 3202.59521484375,
-      "learning_rate": 2.4479929400953178e-05,
-      "loss": 63.3941,
-      "step": 138990
-    },
-    {
-      "epoch": 0.5615775886100752,
-      "grad_norm": 3087.826171875,
-      "learning_rate": 2.447643950291608e-05,
-      "loss": 85.127,
-      "step": 139000
-    },
-    {
-      "epoch": 0.5616179898754429,
-      "grad_norm": 636.1074829101562,
-      "learning_rate": 2.447294961508606e-05,
-      "loss": 50.8728,
-      "step": 139010
-    },
-    {
-      "epoch": 0.5616583911408105,
-      "grad_norm": 771.0210571289062,
-      "learning_rate": 2.446945973753117e-05,
-      "loss": 43.0814,
-      "step": 139020
-    },
-    {
-      "epoch": 0.5616987924061781,
-      "grad_norm": 304.8282165527344,
-      "learning_rate": 2.4465969870319426e-05,
-      "loss": 57.8161,
-      "step": 139030
-    },
-    {
-      "epoch": 0.5617391936715458,
-      "grad_norm": 921.5098266601562,
-      "learning_rate": 2.4462480013518883e-05,
-      "loss": 58.7815,
-      "step": 139040
-    },
-    {
-      "epoch": 0.5617795949369134,
-      "grad_norm": 516.6439819335938,
-      "learning_rate": 2.4458990167197555e-05,
-      "loss": 74.0419,
-      "step": 139050
-    },
-    {
-      "epoch": 0.5618199962022811,
-      "grad_norm": 490.185546875,
-      "learning_rate": 2.4455500331423504e-05,
-      "loss": 43.1365,
-      "step": 139060
-    },
-    {
-      "epoch": 0.5618603974676487,
-      "grad_norm": 716.62109375,
-      "learning_rate": 2.4452010506264755e-05,
-      "loss": 50.1602,
-      "step": 139070
-    },
-    {
-      "epoch": 0.5619007987330163,
-      "grad_norm": 493.0600891113281,
-      "learning_rate": 2.444852069178933e-05,
-      "loss": 54.9955,
-      "step": 139080
-    },
-    {
-      "epoch": 0.561941199998384,
-      "grad_norm": 525.10205078125,
-      "learning_rate": 2.4445030888065293e-05,
-      "loss": 69.9999,
-      "step": 139090
-    },
-    {
-      "epoch": 0.5619816012637516,
-      "grad_norm": 700.7877197265625,
-      "learning_rate": 2.444154109516066e-05,
-      "loss": 60.9567,
-      "step": 139100
-    },
-    {
-      "epoch": 0.5620220025291192,
-      "grad_norm": 408.17535400390625,
-      "learning_rate": 2.443805131314347e-05,
-      "loss": 56.1129,
-      "step": 139110
-    },
-    {
-      "epoch": 0.5620624037944868,
-      "grad_norm": 700.475341796875,
-      "learning_rate": 2.4434561542081762e-05,
-      "loss": 78.0615,
-      "step": 139120
-    },
-    {
-      "epoch": 0.5621028050598544,
-      "grad_norm": 1132.6651611328125,
-      "learning_rate": 2.4431071782043556e-05,
-      "loss": 44.449,
-      "step": 139130
-    },
-    {
-      "epoch": 0.5621432063252221,
-      "grad_norm": 405.23724365234375,
-      "learning_rate": 2.442758203309691e-05,
-      "loss": 65.6864,
-      "step": 139140
-    },
-    {
-      "epoch": 0.5621836075905897,
-      "grad_norm": 628.817138671875,
-      "learning_rate": 2.442409229530985e-05,
-      "loss": 58.2262,
-      "step": 139150
-    },
-    {
-      "epoch": 0.5622240088559574,
-      "grad_norm": 616.2658081054688,
-      "learning_rate": 2.4420602568750393e-05,
-      "loss": 56.0243,
-      "step": 139160
-    },
-    {
-      "epoch": 0.562264410121325,
-      "grad_norm": 520.2138061523438,
-      "learning_rate": 2.44171128534866e-05,
-      "loss": 50.8846,
-      "step": 139170
-    },
-    {
-      "epoch": 0.5623048113866926,
-      "grad_norm": 615.4083251953125,
-      "learning_rate": 2.441362314958649e-05,
-      "loss": 56.1158,
-      "step": 139180
-    },
-    {
-      "epoch": 0.5623452126520603,
-      "grad_norm": 764.9922485351562,
-      "learning_rate": 2.4410133457118097e-05,
-      "loss": 66.4539,
-      "step": 139190
-    },
-    {
-      "epoch": 0.5623856139174279,
-      "grad_norm": 628.9678344726562,
-      "learning_rate": 2.4406643776149458e-05,
-      "loss": 83.326,
-      "step": 139200
-    },
-    {
-      "epoch": 0.5624260151827956,
-      "grad_norm": 1019.9813232421875,
-      "learning_rate": 2.4403154106748592e-05,
-      "loss": 61.2499,
-      "step": 139210
-    },
-    {
-      "epoch": 0.5624664164481632,
-      "grad_norm": 714.0093994140625,
-      "learning_rate": 2.4399664448983557e-05,
-      "loss": 47.1922,
-      "step": 139220
-    },
-    {
-      "epoch": 0.5625068177135308,
-      "grad_norm": 1170.500732421875,
-      "learning_rate": 2.4396174802922372e-05,
-      "loss": 68.0282,
-      "step": 139230
-    },
-    {
-      "epoch": 0.5625472189788984,
-      "grad_norm": 788.6015014648438,
-      "learning_rate": 2.439268516863306e-05,
-      "loss": 47.13,
-      "step": 139240
-    },
-    {
-      "epoch": 0.562587620244266,
-      "grad_norm": 572.7946166992188,
-      "learning_rate": 2.4389195546183673e-05,
-      "loss": 64.8227,
-      "step": 139250
-    },
-    {
-      "epoch": 0.5626280215096336,
-      "grad_norm": 511.4493713378906,
-      "learning_rate": 2.4385705935642232e-05,
-      "loss": 68.7031,
-      "step": 139260
-    },
-    {
-      "epoch": 0.5626684227750013,
-      "grad_norm": 736.9781494140625,
-      "learning_rate": 2.4382216337076767e-05,
-      "loss": 52.4168,
-      "step": 139270
-    },
-    {
-      "epoch": 0.5627088240403689,
-      "grad_norm": 686.5897827148438,
-      "learning_rate": 2.437872675055532e-05,
-      "loss": 33.2072,
-      "step": 139280
-    },
-    {
-      "epoch": 0.5627492253057366,
-      "grad_norm": 851.3128662109375,
-      "learning_rate": 2.43752371761459e-05,
-      "loss": 70.3393,
-      "step": 139290
-    },
-    {
-      "epoch": 0.5627896265711042,
-      "grad_norm": 363.2201232910156,
-      "learning_rate": 2.4371747613916566e-05,
-      "loss": 51.8346,
-      "step": 139300
-    },
-    {
-      "epoch": 0.5628300278364718,
-      "grad_norm": 997.1961669921875,
-      "learning_rate": 2.4368258063935335e-05,
-      "loss": 56.4316,
-      "step": 139310
-    },
-    {
-      "epoch": 0.5628704291018395,
-      "grad_norm": 683.9410400390625,
-      "learning_rate": 2.4364768526270227e-05,
-      "loss": 31.9099,
-      "step": 139320
-    },
-    {
-      "epoch": 0.5629108303672071,
-      "grad_norm": 339.04949951171875,
-      "learning_rate": 2.4361279000989296e-05,
-      "loss": 56.1524,
-      "step": 139330
-    },
-    {
-      "epoch": 0.5629512316325748,
-      "grad_norm": 1158.9278564453125,
-      "learning_rate": 2.4357789488160557e-05,
-      "loss": 67.9605,
-      "step": 139340
-    },
-    {
-      "epoch": 0.5629916328979424,
-      "grad_norm": 823.2213134765625,
-      "learning_rate": 2.4354299987852035e-05,
-      "loss": 77.956,
-      "step": 139350
-    },
-    {
-      "epoch": 0.56303203416331,
-      "grad_norm": 238.20631408691406,
-      "learning_rate": 2.4350810500131777e-05,
-      "loss": 76.2513,
-      "step": 139360
-    },
-    {
-      "epoch": 0.5630724354286776,
-      "grad_norm": 540.9037475585938,
-      "learning_rate": 2.4347321025067795e-05,
-      "loss": 59.1482,
-      "step": 139370
-    },
-    {
-      "epoch": 0.5631128366940452,
-      "grad_norm": 968.451904296875,
-      "learning_rate": 2.4343831562728135e-05,
-      "loss": 58.3164,
-      "step": 139380
-    },
-    {
-      "epoch": 0.5631532379594129,
-      "grad_norm": 870.2232666015625,
-      "learning_rate": 2.434034211318081e-05,
-      "loss": 71.8574,
-      "step": 139390
-    },
-    {
-      "epoch": 0.5631936392247805,
-      "grad_norm": 329.17626953125,
-      "learning_rate": 2.4336852676493847e-05,
-      "loss": 49.5366,
-      "step": 139400
-    },
-    {
-      "epoch": 0.5632340404901481,
-      "grad_norm": 534.2448120117188,
-      "learning_rate": 2.4333363252735297e-05,
-      "loss": 44.9457,
-      "step": 139410
-    },
-    {
-      "epoch": 0.5632744417555158,
-      "grad_norm": 509.2497863769531,
-      "learning_rate": 2.432987384197317e-05,
-      "loss": 75.4781,
-      "step": 139420
-    },
-    {
-      "epoch": 0.5633148430208834,
-      "grad_norm": 333.1900634765625,
-      "learning_rate": 2.432638444427549e-05,
-      "loss": 47.3443,
-      "step": 139430
-    },
-    {
-      "epoch": 0.5633552442862511,
-      "grad_norm": 676.5568237304688,
-      "learning_rate": 2.43228950597103e-05,
-      "loss": 42.6256,
-      "step": 139440
-    },
-    {
-      "epoch": 0.5633956455516187,
-      "grad_norm": 794.4973754882812,
-      "learning_rate": 2.4319405688345614e-05,
-      "loss": 51.409,
-      "step": 139450
-    },
-    {
-      "epoch": 0.5634360468169863,
-      "grad_norm": 753.9666137695312,
-      "learning_rate": 2.431591633024947e-05,
-      "loss": 55.2452,
-      "step": 139460
-    },
-    {
-      "epoch": 0.563476448082354,
-      "grad_norm": 753.4974975585938,
-      "learning_rate": 2.431242698548989e-05,
-      "loss": 51.5466,
-      "step": 139470
-    },
-    {
-      "epoch": 0.5635168493477216,
-      "grad_norm": 870.044189453125,
-      "learning_rate": 2.4308937654134893e-05,
-      "loss": 40.7293,
-      "step": 139480
-    },
-    {
-      "epoch": 0.5635572506130893,
-      "grad_norm": 375.4700012207031,
-      "learning_rate": 2.4305448336252518e-05,
-      "loss": 55.643,
-      "step": 139490
-    },
-    {
-      "epoch": 0.5635976518784568,
-      "grad_norm": 647.6185913085938,
-      "learning_rate": 2.4301959031910784e-05,
-      "loss": 44.7589,
-      "step": 139500
-    },
-    {
-      "epoch": 0.5636380531438244,
-      "grad_norm": 535.95947265625,
-      "learning_rate": 2.4298469741177708e-05,
-      "loss": 40.7305,
-      "step": 139510
-    },
-    {
-      "epoch": 0.5636784544091921,
-      "grad_norm": 1032.9677734375,
-      "learning_rate": 2.429498046412134e-05,
-      "loss": 43.6444,
-      "step": 139520
-    },
-    {
-      "epoch": 0.5637188556745597,
-      "grad_norm": 699.4422607421875,
-      "learning_rate": 2.4291491200809684e-05,
-      "loss": 42.5152,
-      "step": 139530
-    },
-    {
-      "epoch": 0.5637592569399273,
-      "grad_norm": 357.033935546875,
-      "learning_rate": 2.4288001951310775e-05,
-      "loss": 63.6277,
-      "step": 139540
-    },
-    {
-      "epoch": 0.563799658205295,
-      "grad_norm": 744.2186889648438,
-      "learning_rate": 2.4284512715692636e-05,
-      "loss": 50.3803,
-      "step": 139550
-    },
-    {
-      "epoch": 0.5638400594706626,
-      "grad_norm": 766.5017700195312,
-      "learning_rate": 2.428102349402328e-05,
-      "loss": 54.6011,
-      "step": 139560
-    },
-    {
-      "epoch": 0.5638804607360303,
-      "grad_norm": 793.3067626953125,
-      "learning_rate": 2.4277534286370752e-05,
-      "loss": 51.7648,
-      "step": 139570
-    },
-    {
-      "epoch": 0.5639208620013979,
-      "grad_norm": 1126.599609375,
-      "learning_rate": 2.4274045092803056e-05,
-      "loss": 55.7164,
-      "step": 139580
-    },
-    {
-      "epoch": 0.5639612632667655,
-      "grad_norm": 623.4227905273438,
-      "learning_rate": 2.427055591338823e-05,
-      "loss": 47.8686,
-      "step": 139590
-    },
-    {
-      "epoch": 0.5640016645321332,
-      "grad_norm": 767.97607421875,
-      "learning_rate": 2.4267066748194296e-05,
-      "loss": 36.6273,
-      "step": 139600
-    },
-    {
-      "epoch": 0.5640420657975008,
-      "grad_norm": 803.25927734375,
-      "learning_rate": 2.4263577597289267e-05,
-      "loss": 63.7557,
-      "step": 139610
-    },
-    {
-      "epoch": 0.5640824670628685,
-      "grad_norm": 576.4552001953125,
-      "learning_rate": 2.4260088460741175e-05,
-      "loss": 47.2425,
-      "step": 139620
-    },
-    {
-      "epoch": 0.564122868328236,
-      "grad_norm": 218.04110717773438,
-      "learning_rate": 2.4256599338618034e-05,
-      "loss": 56.5604,
-      "step": 139630
-    },
-    {
-      "epoch": 0.5641632695936036,
-      "grad_norm": 884.3212890625,
-      "learning_rate": 2.4253110230987878e-05,
-      "loss": 61.0225,
-      "step": 139640
-    },
-    {
-      "epoch": 0.5642036708589713,
-      "grad_norm": 1012.44189453125,
-      "learning_rate": 2.4249621137918722e-05,
-      "loss": 56.3514,
-      "step": 139650
-    },
-    {
-      "epoch": 0.5642440721243389,
-      "grad_norm": 636.1995239257812,
-      "learning_rate": 2.4246132059478578e-05,
-      "loss": 93.1031,
-      "step": 139660
-    },
-    {
-      "epoch": 0.5642844733897066,
-      "grad_norm": 1277.3994140625,
-      "learning_rate": 2.424264299573549e-05,
-      "loss": 71.3306,
-      "step": 139670
-    },
-    {
-      "epoch": 0.5643248746550742,
-      "grad_norm": 436.1991882324219,
-      "learning_rate": 2.4239153946757468e-05,
-      "loss": 52.4229,
-      "step": 139680
-    },
-    {
-      "epoch": 0.5643652759204418,
-      "grad_norm": 711.0999755859375,
-      "learning_rate": 2.4235664912612515e-05,
-      "loss": 48.8079,
-      "step": 139690
-    },
-    {
-      "epoch": 0.5644056771858095,
-      "grad_norm": 1402.82568359375,
-      "learning_rate": 2.423217589336868e-05,
-      "loss": 54.2262,
-      "step": 139700
-    },
-    {
-      "epoch": 0.5644460784511771,
-      "grad_norm": 1108.6195068359375,
-      "learning_rate": 2.422868688909398e-05,
-      "loss": 62.0169,
-      "step": 139710
-    },
-    {
-      "epoch": 0.5644864797165448,
-      "grad_norm": 677.495849609375,
-      "learning_rate": 2.4225197899856414e-05,
-      "loss": 91.8257,
-      "step": 139720
-    },
-    {
-      "epoch": 0.5645268809819124,
-      "grad_norm": 770.8883666992188,
-      "learning_rate": 2.422170892572402e-05,
-      "loss": 36.889,
-      "step": 139730
-    },
-    {
-      "epoch": 0.56456728224728,
-      "grad_norm": 1669.2225341796875,
-      "learning_rate": 2.4218219966764798e-05,
-      "loss": 69.1814,
-      "step": 139740
-    },
-    {
-      "epoch": 0.5646076835126476,
-      "grad_norm": 247.09124755859375,
-      "learning_rate": 2.4214731023046793e-05,
-      "loss": 50.9243,
-      "step": 139750
-    },
-    {
-      "epoch": 0.5646480847780152,
-      "grad_norm": 625.1920166015625,
-      "learning_rate": 2.4211242094638013e-05,
-      "loss": 85.2901,
-      "step": 139760
-    },
-    {
-      "epoch": 0.5646884860433828,
-      "grad_norm": 1005.7395629882812,
-      "learning_rate": 2.420775318160646e-05,
-      "loss": 67.7253,
-      "step": 139770
-    },
-    {
-      "epoch": 0.5647288873087505,
-      "grad_norm": 1117.547607421875,
-      "learning_rate": 2.4204264284020182e-05,
-      "loss": 71.7562,
-      "step": 139780
-    },
-    {
-      "epoch": 0.5647692885741181,
-      "grad_norm": 449.2171325683594,
-      "learning_rate": 2.4200775401947178e-05,
-      "loss": 50.0482,
-      "step": 139790
-    },
-    {
-      "epoch": 0.5648096898394858,
-      "grad_norm": 399.94256591796875,
-      "learning_rate": 2.4197286535455464e-05,
-      "loss": 49.0233,
-      "step": 139800
-    },
-    {
-      "epoch": 0.5648500911048534,
-      "grad_norm": 936.6152954101562,
-      "learning_rate": 2.419379768461307e-05,
-      "loss": 64.9566,
-      "step": 139810
-    },
-    {
-      "epoch": 0.564890492370221,
-      "grad_norm": 843.4839477539062,
-      "learning_rate": 2.419030884948799e-05,
-      "loss": 81.1626,
-      "step": 139820
-    },
-    {
-      "epoch": 0.5649308936355887,
-      "grad_norm": 725.7382202148438,
-      "learning_rate": 2.418682003014827e-05,
-      "loss": 49.8391,
-      "step": 139830
-    },
-    {
-      "epoch": 0.5649712949009563,
-      "grad_norm": 691.9060668945312,
-      "learning_rate": 2.418333122666191e-05,
-      "loss": 71.7343,
-      "step": 139840
-    },
-    {
-      "epoch": 0.565011696166324,
-      "grad_norm": 2030.63037109375,
-      "learning_rate": 2.417984243909692e-05,
-      "loss": 61.1143,
-      "step": 139850
-    },
-    {
-      "epoch": 0.5650520974316916,
-      "grad_norm": 636.2049560546875,
-      "learning_rate": 2.4176353667521335e-05,
-      "loss": 53.0477,
-      "step": 139860
-    },
-    {
-      "epoch": 0.5650924986970592,
-      "grad_norm": 443.2159118652344,
-      "learning_rate": 2.4172864912003158e-05,
-      "loss": 47.2716,
-      "step": 139870
-    },
-    {
-      "epoch": 0.5651328999624268,
-      "grad_norm": 727.209228515625,
-      "learning_rate": 2.4169376172610397e-05,
-      "loss": 65.8904,
-      "step": 139880
-    },
-    {
-      "epoch": 0.5651733012277944,
-      "grad_norm": 1317.1082763671875,
-      "learning_rate": 2.4165887449411084e-05,
-      "loss": 54.5445,
-      "step": 139890
-    },
-    {
-      "epoch": 0.565213702493162,
-      "grad_norm": 913.7978515625,
-      "learning_rate": 2.4162398742473214e-05,
-      "loss": 52.2337,
-      "step": 139900
-    },
-    {
-      "epoch": 0.5652541037585297,
-      "grad_norm": 667.3104858398438,
-      "learning_rate": 2.415891005186482e-05,
-      "loss": 47.764,
-      "step": 139910
-    },
-    {
-      "epoch": 0.5652945050238973,
-      "grad_norm": 442.9791259765625,
-      "learning_rate": 2.415542137765391e-05,
-      "loss": 53.1622,
-      "step": 139920
-    },
-    {
-      "epoch": 0.565334906289265,
-      "grad_norm": 817.6068115234375,
-      "learning_rate": 2.4151932719908483e-05,
-      "loss": 61.2308,
-      "step": 139930
-    },
-    {
-      "epoch": 0.5653753075546326,
-      "grad_norm": 594.9141845703125,
-      "learning_rate": 2.4148444078696575e-05,
-      "loss": 32.9943,
-      "step": 139940
-    },
-    {
-      "epoch": 0.5654157088200003,
-      "grad_norm": 689.52197265625,
-      "learning_rate": 2.4144955454086188e-05,
-      "loss": 54.7586,
-      "step": 139950
-    },
-    {
-      "epoch": 0.5654561100853679,
-      "grad_norm": 1552.0382080078125,
-      "learning_rate": 2.414146684614533e-05,
-      "loss": 60.1999,
-      "step": 139960
-    },
-    {
-      "epoch": 0.5654965113507355,
-      "grad_norm": 684.204833984375,
-      "learning_rate": 2.4137978254942018e-05,
-      "loss": 45.3223,
-      "step": 139970
-    },
-    {
-      "epoch": 0.5655369126161032,
-      "grad_norm": 389.4147644042969,
-      "learning_rate": 2.413448968054426e-05,
-      "loss": 40.717,
-      "step": 139980
-    },
-    {
-      "epoch": 0.5655773138814708,
-      "grad_norm": 486.74945068359375,
-      "learning_rate": 2.4131001123020083e-05,
-      "loss": 43.9063,
-      "step": 139990
-    },
-    {
-      "epoch": 0.5656177151468385,
-      "grad_norm": 2609.690673828125,
-      "learning_rate": 2.4127512582437485e-05,
-      "loss": 47.5488,
-      "step": 140000
-    },
-    {
-      "epoch": 0.565658116412206,
-      "grad_norm": 1355.923828125,
-      "learning_rate": 2.4124024058864464e-05,
-      "loss": 69.8932,
-      "step": 140010
-    },
-    {
-      "epoch": 0.5656985176775736,
-      "grad_norm": 567.3590698242188,
-      "learning_rate": 2.412053555236906e-05,
-      "loss": 63.0039,
-      "step": 140020
-    },
-    {
-      "epoch": 0.5657389189429413,
-      "grad_norm": 779.9495849609375,
-      "learning_rate": 2.4117047063019263e-05,
-      "loss": 39.6256,
-      "step": 140030
-    },
-    {
-      "epoch": 0.5657793202083089,
-      "grad_norm": 684.2831420898438,
-      "learning_rate": 2.411355859088308e-05,
-      "loss": 42.6509,
-      "step": 140040
-    },
-    {
-      "epoch": 0.5658197214736765,
-      "grad_norm": 1118.098876953125,
-      "learning_rate": 2.411007013602854e-05,
-      "loss": 68.7989,
-      "step": 140050
-    },
-    {
-      "epoch": 0.5658601227390442,
-      "grad_norm": 652.873046875,
-      "learning_rate": 2.4106581698523635e-05,
-      "loss": 61.522,
-      "step": 140060
-    },
-    {
-      "epoch": 0.5659005240044118,
-      "grad_norm": 1058.7735595703125,
-      "learning_rate": 2.4103093278436386e-05,
-      "loss": 60.4243,
-      "step": 140070
-    },
-    {
-      "epoch": 0.5659409252697795,
-      "grad_norm": 875.0758056640625,
-      "learning_rate": 2.4099604875834795e-05,
-      "loss": 60.9868,
-      "step": 140080
-    },
-    {
-      "epoch": 0.5659813265351471,
-      "grad_norm": 504.15289306640625,
-      "learning_rate": 2.4096116490786856e-05,
-      "loss": 53.1008,
-      "step": 140090
-    },
-    {
-      "epoch": 0.5660217278005147,
-      "grad_norm": 647.9260864257812,
-      "learning_rate": 2.4092628123360608e-05,
-      "loss": 44.3043,
-      "step": 140100
-    },
-    {
-      "epoch": 0.5660621290658824,
-      "grad_norm": 445.2146301269531,
-      "learning_rate": 2.4089139773624027e-05,
-      "loss": 32.4979,
-      "step": 140110
-    },
-    {
-      "epoch": 0.56610253033125,
-      "grad_norm": 583.020751953125,
-      "learning_rate": 2.408565144164515e-05,
-      "loss": 65.0225,
-      "step": 140120
-    },
-    {
-      "epoch": 0.5661429315966177,
-      "grad_norm": 782.4393310546875,
-      "learning_rate": 2.4082163127491967e-05,
-      "loss": 49.7005,
-      "step": 140130
-    },
-    {
-      "epoch": 0.5661833328619852,
-      "grad_norm": 1884.6944580078125,
-      "learning_rate": 2.407867483123248e-05,
-      "loss": 82.0714,
-      "step": 140140
-    },
-    {
-      "epoch": 0.5662237341273528,
-      "grad_norm": 710.6357421875,
-      "learning_rate": 2.4075186552934707e-05,
-      "loss": 69.646,
-      "step": 140150
-    },
-    {
-      "epoch": 0.5662641353927205,
-      "grad_norm": 165.2595672607422,
-      "learning_rate": 2.4071698292666648e-05,
-      "loss": 46.273,
-      "step": 140160
-    },
-    {
-      "epoch": 0.5663045366580881,
-      "grad_norm": 537.0858764648438,
-      "learning_rate": 2.4068210050496298e-05,
-      "loss": 76.4967,
-      "step": 140170
-    },
-    {
-      "epoch": 0.5663449379234557,
-      "grad_norm": 792.4481811523438,
-      "learning_rate": 2.406472182649168e-05,
-      "loss": 43.8985,
-      "step": 140180
-    },
-    {
-      "epoch": 0.5663853391888234,
-      "grad_norm": 611.4661865234375,
-      "learning_rate": 2.406123362072079e-05,
-      "loss": 48.2827,
-      "step": 140190
-    },
-    {
-      "epoch": 0.566425740454191,
-      "grad_norm": 1082.0413818359375,
-      "learning_rate": 2.4057745433251635e-05,
-      "loss": 72.3503,
-      "step": 140200
-    },
-    {
-      "epoch": 0.5664661417195587,
-      "grad_norm": 902.6494750976562,
-      "learning_rate": 2.4054257264152226e-05,
-      "loss": 44.0646,
-      "step": 140210
-    },
-    {
-      "epoch": 0.5665065429849263,
-      "grad_norm": 738.3258666992188,
-      "learning_rate": 2.405076911349055e-05,
-      "loss": 44.8827,
-      "step": 140220
-    },
-    {
-      "epoch": 0.566546944250294,
-      "grad_norm": 540.25732421875,
-      "learning_rate": 2.404728098133462e-05,
-      "loss": 55.6525,
-      "step": 140230
-    },
-    {
-      "epoch": 0.5665873455156616,
-      "grad_norm": 646.385498046875,
-      "learning_rate": 2.4043792867752444e-05,
-      "loss": 37.5222,
-      "step": 140240
-    },
-    {
-      "epoch": 0.5666277467810292,
-      "grad_norm": 320.0557861328125,
-      "learning_rate": 2.4040304772812002e-05,
-      "loss": 45.5283,
-      "step": 140250
-    },
-    {
-      "epoch": 0.5666681480463968,
-      "grad_norm": 398.5991516113281,
-      "learning_rate": 2.4036816696581326e-05,
-      "loss": 58.8736,
-      "step": 140260
-    },
-    {
-      "epoch": 0.5667085493117644,
-      "grad_norm": 558.6793212890625,
-      "learning_rate": 2.4033328639128394e-05,
-      "loss": 57.9081,
-      "step": 140270
-    },
-    {
-      "epoch": 0.566748950577132,
-      "grad_norm": 3701.99072265625,
-      "learning_rate": 2.4029840600521227e-05,
-      "loss": 52.849,
-      "step": 140280
-    },
-    {
-      "epoch": 0.5667893518424997,
-      "grad_norm": 725.6563110351562,
-      "learning_rate": 2.4026352580827822e-05,
-      "loss": 49.8771,
-      "step": 140290
-    },
-    {
-      "epoch": 0.5668297531078673,
-      "grad_norm": 4818.69287109375,
-      "learning_rate": 2.4022864580116164e-05,
-      "loss": 81.0483,
-      "step": 140300
-    },
-    {
-      "epoch": 0.566870154373235,
-      "grad_norm": 231.59661865234375,
-      "learning_rate": 2.4019376598454266e-05,
-      "loss": 36.7556,
-      "step": 140310
-    },
-    {
-      "epoch": 0.5669105556386026,
-      "grad_norm": 1639.919921875,
-      "learning_rate": 2.401588863591013e-05,
-      "loss": 38.1198,
-      "step": 140320
-    },
-    {
-      "epoch": 0.5669509569039702,
-      "grad_norm": 1159.2486572265625,
-      "learning_rate": 2.4012400692551746e-05,
-      "loss": 67.2019,
-      "step": 140330
-    },
-    {
-      "epoch": 0.5669913581693379,
-      "grad_norm": 452.9656982421875,
-      "learning_rate": 2.4008912768447127e-05,
-      "loss": 60.1592,
-      "step": 140340
-    },
-    {
-      "epoch": 0.5670317594347055,
-      "grad_norm": 194.69638061523438,
-      "learning_rate": 2.4005424863664254e-05,
-      "loss": 82.0656,
-      "step": 140350
-    },
-    {
-      "epoch": 0.5670721607000732,
-      "grad_norm": 792.1279907226562,
-      "learning_rate": 2.4001936978271144e-05,
-      "loss": 60.3676,
-      "step": 140360
-    },
-    {
-      "epoch": 0.5671125619654408,
-      "grad_norm": 1072.4195556640625,
-      "learning_rate": 2.3998449112335784e-05,
-      "loss": 54.9496,
-      "step": 140370
-    },
-    {
-      "epoch": 0.5671529632308084,
-      "grad_norm": 667.0706787109375,
-      "learning_rate": 2.3994961265926166e-05,
-      "loss": 74.0401,
-      "step": 140380
-    },
-    {
-      "epoch": 0.567193364496176,
-      "grad_norm": 1137.287109375,
-      "learning_rate": 2.3991473439110308e-05,
-      "loss": 47.5522,
-      "step": 140390
-    },
-    {
-      "epoch": 0.5672337657615436,
-      "grad_norm": 685.4043579101562,
-      "learning_rate": 2.398798563195619e-05,
-      "loss": 63.7304,
-      "step": 140400
-    },
-    {
-      "epoch": 0.5672741670269112,
-      "grad_norm": 426.1322937011719,
-      "learning_rate": 2.3984497844531814e-05,
-      "loss": 53.3877,
-      "step": 140410
-    },
-    {
-      "epoch": 0.5673145682922789,
-      "grad_norm": 226.16949462890625,
-      "learning_rate": 2.3981010076905174e-05,
-      "loss": 56.9861,
-      "step": 140420
-    },
-    {
-      "epoch": 0.5673549695576465,
-      "grad_norm": 736.9807739257812,
-      "learning_rate": 2.3977522329144263e-05,
-      "loss": 42.9066,
-      "step": 140430
-    },
-    {
-      "epoch": 0.5673953708230142,
-      "grad_norm": 768.5642700195312,
-      "learning_rate": 2.3974034601317086e-05,
-      "loss": 75.2792,
-      "step": 140440
-    },
-    {
-      "epoch": 0.5674357720883818,
-      "grad_norm": 561.4569702148438,
-      "learning_rate": 2.3970546893491637e-05,
-      "loss": 58.4712,
-      "step": 140450
-    },
-    {
-      "epoch": 0.5674761733537494,
-      "grad_norm": 2184.9736328125,
-      "learning_rate": 2.396705920573589e-05,
-      "loss": 69.7435,
-      "step": 140460
-    },
-    {
-      "epoch": 0.5675165746191171,
-      "grad_norm": 1226.76904296875,
-      "learning_rate": 2.3963571538117873e-05,
-      "loss": 86.3856,
-      "step": 140470
-    },
-    {
-      "epoch": 0.5675569758844847,
-      "grad_norm": 888.7282104492188,
-      "learning_rate": 2.3960083890705557e-05,
-      "loss": 46.3673,
-      "step": 140480
-    },
-    {
-      "epoch": 0.5675973771498524,
-      "grad_norm": 884.1262817382812,
-      "learning_rate": 2.395659626356694e-05,
-      "loss": 54.7737,
-      "step": 140490
-    },
-    {
-      "epoch": 0.56763777841522,
-      "grad_norm": 250.46376037597656,
-      "learning_rate": 2.3953108656770016e-05,
-      "loss": 36.1049,
-      "step": 140500
-    },
-    {
-      "epoch": 0.5676781796805876,
-      "grad_norm": 977.3258056640625,
-      "learning_rate": 2.394962107038277e-05,
-      "loss": 64.6896,
-      "step": 140510
-    },
-    {
-      "epoch": 0.5677185809459552,
-      "grad_norm": 875.897216796875,
-      "learning_rate": 2.394613350447321e-05,
-      "loss": 51.8041,
-      "step": 140520
-    },
-    {
-      "epoch": 0.5677589822113228,
-      "grad_norm": 382.20147705078125,
-      "learning_rate": 2.3942645959109324e-05,
-      "loss": 41.5387,
-      "step": 140530
-    },
-    {
-      "epoch": 0.5677993834766905,
-      "grad_norm": 0.0,
-      "learning_rate": 2.393915843435909e-05,
-      "loss": 51.2521,
-      "step": 140540
-    },
-    {
-      "epoch": 0.5678397847420581,
-      "grad_norm": 1551.5985107421875,
-      "learning_rate": 2.3935670930290512e-05,
-      "loss": 53.0409,
-      "step": 140550
-    },
-    {
-      "epoch": 0.5678801860074257,
-      "grad_norm": 665.3966674804688,
-      "learning_rate": 2.3932183446971583e-05,
-      "loss": 44.8342,
-      "step": 140560
-    },
-    {
-      "epoch": 0.5679205872727934,
-      "grad_norm": 786.4059448242188,
-      "learning_rate": 2.392869598447028e-05,
-      "loss": 54.9068,
-      "step": 140570
-    },
-    {
-      "epoch": 0.567960988538161,
-      "grad_norm": 685.6342163085938,
-      "learning_rate": 2.3925208542854608e-05,
-      "loss": 44.5049,
-      "step": 140580
-    },
-    {
-      "epoch": 0.5680013898035287,
-      "grad_norm": 780.075927734375,
-      "learning_rate": 2.3921721122192535e-05,
-      "loss": 35.4166,
-      "step": 140590
-    },
-    {
-      "epoch": 0.5680417910688963,
-      "grad_norm": 994.8262329101562,
-      "learning_rate": 2.391823372255208e-05,
-      "loss": 35.6253,
-      "step": 140600
-    },
-    {
-      "epoch": 0.5680821923342639,
-      "grad_norm": 468.07171630859375,
-      "learning_rate": 2.391474634400121e-05,
-      "loss": 56.5801,
-      "step": 140610
-    },
-    {
-      "epoch": 0.5681225935996316,
-      "grad_norm": 423.82196044921875,
-      "learning_rate": 2.3911258986607908e-05,
-      "loss": 43.0342,
-      "step": 140620
-    },
-    {
-      "epoch": 0.5681629948649992,
-      "grad_norm": 627.2518920898438,
-      "learning_rate": 2.3907771650440187e-05,
-      "loss": 41.6314,
-      "step": 140630
-    },
-    {
-      "epoch": 0.5682033961303669,
-      "grad_norm": 609.6311645507812,
-      "learning_rate": 2.390428433556601e-05,
-      "loss": 46.9864,
-      "step": 140640
-    },
-    {
-      "epoch": 0.5682437973957344,
-      "grad_norm": 608.6633911132812,
-      "learning_rate": 2.3900797042053382e-05,
-      "loss": 72.6592,
-      "step": 140650
-    },
-    {
-      "epoch": 0.568284198661102,
-      "grad_norm": 491.6672668457031,
-      "learning_rate": 2.3897309769970276e-05,
-      "loss": 37.3915,
-      "step": 140660
-    },
-    {
-      "epoch": 0.5683245999264697,
-      "grad_norm": 307.1319274902344,
-      "learning_rate": 2.3893822519384684e-05,
-      "loss": 51.833,
-      "step": 140670
-    },
-    {
-      "epoch": 0.5683650011918373,
-      "grad_norm": 899.0912475585938,
-      "learning_rate": 2.3890335290364595e-05,
-      "loss": 85.2399,
-      "step": 140680
-    },
-    {
-      "epoch": 0.568405402457205,
-      "grad_norm": 365.262451171875,
-      "learning_rate": 2.3886848082977995e-05,
-      "loss": 41.261,
-      "step": 140690
-    },
-    {
-      "epoch": 0.5684458037225726,
-      "grad_norm": 606.3248901367188,
-      "learning_rate": 2.388336089729285e-05,
-      "loss": 41.9415,
-      "step": 140700
-    },
-    {
-      "epoch": 0.5684862049879402,
-      "grad_norm": 764.3125,
-      "learning_rate": 2.387987373337717e-05,
-      "loss": 56.6378,
-      "step": 140710
-    },
-    {
-      "epoch": 0.5685266062533079,
-      "grad_norm": 835.7235107421875,
-      "learning_rate": 2.387638659129892e-05,
-      "loss": 41.2721,
-      "step": 140720
-    },
-    {
-      "epoch": 0.5685670075186755,
-      "grad_norm": 497.7888488769531,
-      "learning_rate": 2.3872899471126102e-05,
-      "loss": 51.5741,
-      "step": 140730
-    },
-    {
-      "epoch": 0.5686074087840431,
-      "grad_norm": 536.2766723632812,
-      "learning_rate": 2.3869412372926687e-05,
-      "loss": 43.3223,
-      "step": 140740
-    },
-    {
-      "epoch": 0.5686478100494108,
-      "grad_norm": 869.4346313476562,
-      "learning_rate": 2.386592529676866e-05,
-      "loss": 45.3884,
-      "step": 140750
-    },
-    {
-      "epoch": 0.5686882113147784,
-      "grad_norm": 813.7835083007812,
-      "learning_rate": 2.386243824272e-05,
-      "loss": 34.9353,
-      "step": 140760
-    },
-    {
-      "epoch": 0.5687286125801461,
-      "grad_norm": 553.7960815429688,
-      "learning_rate": 2.38589512108487e-05,
-      "loss": 55.3369,
-      "step": 140770
-    },
-    {
-      "epoch": 0.5687690138455136,
-      "grad_norm": 370.2120666503906,
-      "learning_rate": 2.3855464201222716e-05,
-      "loss": 55.3215,
-      "step": 140780
-    },
-    {
-      "epoch": 0.5688094151108812,
-      "grad_norm": 0.0,
-      "learning_rate": 2.385197721391006e-05,
-      "loss": 50.498,
-      "step": 140790
-    },
-    {
-      "epoch": 0.5688498163762489,
-      "grad_norm": 461.68670654296875,
-      "learning_rate": 2.384849024897869e-05,
-      "loss": 57.4703,
-      "step": 140800
-    },
-    {
-      "epoch": 0.5688902176416165,
-      "grad_norm": 1943.5701904296875,
-      "learning_rate": 2.3845003306496606e-05,
-      "loss": 68.1128,
-      "step": 140810
-    },
-    {
-      "epoch": 0.5689306189069842,
-      "grad_norm": 1182.188720703125,
-      "learning_rate": 2.384151638653178e-05,
-      "loss": 59.1778,
-      "step": 140820
-    },
-    {
-      "epoch": 0.5689710201723518,
-      "grad_norm": 343.8730773925781,
-      "learning_rate": 2.3838029489152176e-05,
-      "loss": 57.4862,
-      "step": 140830
-    },
-    {
-      "epoch": 0.5690114214377194,
-      "grad_norm": 609.5772094726562,
-      "learning_rate": 2.38345426144258e-05,
-      "loss": 50.4306,
-      "step": 140840
-    },
-    {
-      "epoch": 0.5690518227030871,
-      "grad_norm": 1227.27734375,
-      "learning_rate": 2.3831055762420607e-05,
-      "loss": 50.9431,
-      "step": 140850
-    },
-    {
-      "epoch": 0.5690922239684547,
-      "grad_norm": 471.8846435546875,
-      "learning_rate": 2.3827568933204575e-05,
-      "loss": 59.7956,
-      "step": 140860
-    },
-    {
-      "epoch": 0.5691326252338224,
-      "grad_norm": 601.4916381835938,
-      "learning_rate": 2.3824082126845703e-05,
-      "loss": 55.0637,
-      "step": 140870
-    },
-    {
-      "epoch": 0.56917302649919,
-      "grad_norm": 1994.75830078125,
-      "learning_rate": 2.3820595343411944e-05,
-      "loss": 69.4372,
-      "step": 140880
-    },
-    {
-      "epoch": 0.5692134277645576,
-      "grad_norm": 927.0828247070312,
-      "learning_rate": 2.3817108582971298e-05,
-      "loss": 60.4966,
-      "step": 140890
-    },
-    {
-      "epoch": 0.5692538290299252,
-      "grad_norm": 859.7433471679688,
-      "learning_rate": 2.3813621845591727e-05,
-      "loss": 42.6294,
-      "step": 140900
-    },
-    {
-      "epoch": 0.5692942302952928,
-      "grad_norm": 626.2705688476562,
-      "learning_rate": 2.3810135131341203e-05,
-      "loss": 91.8611,
-      "step": 140910
-    },
-    {
-      "epoch": 0.5693346315606604,
-      "grad_norm": 4675.134765625,
-      "learning_rate": 2.3806648440287714e-05,
-      "loss": 105.6002,
-      "step": 140920
-    },
-    {
-      "epoch": 0.5693750328260281,
-      "grad_norm": 1433.595947265625,
-      "learning_rate": 2.380316177249923e-05,
-      "loss": 60.9849,
-      "step": 140930
-    },
-    {
-      "epoch": 0.5694154340913957,
-      "grad_norm": 416.4643859863281,
-      "learning_rate": 2.3799675128043707e-05,
-      "loss": 56.4546,
-      "step": 140940
-    },
-    {
-      "epoch": 0.5694558353567634,
-      "grad_norm": 858.3992919921875,
-      "learning_rate": 2.3796188506989153e-05,
-      "loss": 51.5956,
-      "step": 140950
-    },
-    {
-      "epoch": 0.569496236622131,
-      "grad_norm": 1188.315185546875,
-      "learning_rate": 2.379270190940351e-05,
-      "loss": 63.5035,
-      "step": 140960
-    },
-    {
-      "epoch": 0.5695366378874986,
-      "grad_norm": 919.1412353515625,
-      "learning_rate": 2.3789215335354774e-05,
-      "loss": 114.3202,
-      "step": 140970
-    },
-    {
-      "epoch": 0.5695770391528663,
-      "grad_norm": 594.3109741210938,
-      "learning_rate": 2.378572878491091e-05,
-      "loss": 78.9417,
-      "step": 140980
-    },
-    {
-      "epoch": 0.5696174404182339,
-      "grad_norm": 725.9608154296875,
-      "learning_rate": 2.3782242258139883e-05,
-      "loss": 46.3523,
-      "step": 140990
-    },
-    {
-      "epoch": 0.5696578416836016,
-      "grad_norm": 424.02703857421875,
-      "learning_rate": 2.377875575510967e-05,
-      "loss": 50.4055,
-      "step": 141000
-    },
-    {
-      "epoch": 0.5696982429489692,
-      "grad_norm": 680.8141479492188,
-      "learning_rate": 2.3775269275888248e-05,
-      "loss": 51.2761,
-      "step": 141010
-    },
-    {
-      "epoch": 0.5697386442143368,
-      "grad_norm": 1021.9234008789062,
-      "learning_rate": 2.3771782820543577e-05,
-      "loss": 74.4177,
-      "step": 141020
-    },
-    {
-      "epoch": 0.5697790454797044,
-      "grad_norm": 1398.9102783203125,
-      "learning_rate": 2.3768296389143636e-05,
-      "loss": 75.1736,
-      "step": 141030
-    },
-    {
-      "epoch": 0.569819446745072,
-      "grad_norm": 403.35345458984375,
-      "learning_rate": 2.376480998175638e-05,
-      "loss": 36.8003,
-      "step": 141040
-    },
-    {
-      "epoch": 0.5698598480104397,
-      "grad_norm": 1651.0067138671875,
-      "learning_rate": 2.3761323598449803e-05,
-      "loss": 64.2051,
-      "step": 141050
-    },
-    {
-      "epoch": 0.5699002492758073,
-      "grad_norm": 802.9590454101562,
-      "learning_rate": 2.375783723929186e-05,
-      "loss": 64.763,
-      "step": 141060
-    },
-    {
-      "epoch": 0.5699406505411749,
-      "grad_norm": 294.2562255859375,
-      "learning_rate": 2.3754350904350502e-05,
-      "loss": 70.0117,
-      "step": 141070
-    },
-    {
-      "epoch": 0.5699810518065426,
-      "grad_norm": 493.0975341796875,
-      "learning_rate": 2.3750864593693732e-05,
-      "loss": 48.1589,
-      "step": 141080
-    },
-    {
-      "epoch": 0.5700214530719102,
-      "grad_norm": 789.9282836914062,
-      "learning_rate": 2.37473783073895e-05,
-      "loss": 74.4661,
-      "step": 141090
-    },
-    {
-      "epoch": 0.5700618543372779,
-      "grad_norm": 655.0347900390625,
-      "learning_rate": 2.3743892045505764e-05,
-      "loss": 55.9553,
-      "step": 141100
-    },
-    {
-      "epoch": 0.5701022556026455,
-      "grad_norm": 469.515380859375,
-      "learning_rate": 2.3740405808110504e-05,
-      "loss": 44.5552,
-      "step": 141110
-    },
-    {
-      "epoch": 0.5701426568680131,
-      "grad_norm": 2083.58251953125,
-      "learning_rate": 2.3736919595271677e-05,
-      "loss": 63.5822,
-      "step": 141120
-    },
-    {
-      "epoch": 0.5701830581333808,
-      "grad_norm": 913.0497436523438,
-      "learning_rate": 2.3733433407057257e-05,
-      "loss": 43.4769,
-      "step": 141130
-    },
-    {
-      "epoch": 0.5702234593987484,
-      "grad_norm": 1263.47119140625,
-      "learning_rate": 2.372994724353521e-05,
-      "loss": 79.1157,
-      "step": 141140
-    },
-    {
-      "epoch": 0.570263860664116,
-      "grad_norm": 968.7415771484375,
-      "learning_rate": 2.372646110477348e-05,
-      "loss": 53.7039,
-      "step": 141150
-    },
-    {
-      "epoch": 0.5703042619294836,
-      "grad_norm": 671.2269287109375,
-      "learning_rate": 2.3722974990840058e-05,
-      "loss": 48.0864,
-      "step": 141160
-    },
-    {
-      "epoch": 0.5703446631948512,
-      "grad_norm": 968.9181518554688,
-      "learning_rate": 2.3719488901802893e-05,
-      "loss": 39.3307,
-      "step": 141170
-    },
-    {
-      "epoch": 0.5703850644602189,
-      "grad_norm": 480.0361022949219,
-      "learning_rate": 2.3716002837729954e-05,
-      "loss": 68.5519,
-      "step": 141180
-    },
-    {
-      "epoch": 0.5704254657255865,
-      "grad_norm": 835.8157958984375,
-      "learning_rate": 2.3712516798689203e-05,
-      "loss": 57.1123,
-      "step": 141190
-    },
-    {
-      "epoch": 0.5704658669909541,
-      "grad_norm": 438.4929504394531,
-      "learning_rate": 2.3709030784748587e-05,
-      "loss": 38.9969,
-      "step": 141200
-    },
-    {
-      "epoch": 0.5705062682563218,
-      "grad_norm": 859.7994995117188,
-      "learning_rate": 2.370554479597609e-05,
-      "loss": 78.8213,
-      "step": 141210
-    },
-    {
-      "epoch": 0.5705466695216894,
-      "grad_norm": 1708.9825439453125,
-      "learning_rate": 2.3702058832439667e-05,
-      "loss": 62.3029,
-      "step": 141220
-    },
-    {
-      "epoch": 0.5705870707870571,
-      "grad_norm": 777.35400390625,
-      "learning_rate": 2.3698572894207262e-05,
-      "loss": 66.144,
-      "step": 141230
-    },
-    {
-      "epoch": 0.5706274720524247,
-      "grad_norm": 1250.974853515625,
-      "learning_rate": 2.369508698134686e-05,
-      "loss": 59.3208,
-      "step": 141240
-    },
-    {
-      "epoch": 0.5706678733177923,
-      "grad_norm": 541.2228393554688,
-      "learning_rate": 2.3691601093926404e-05,
-      "loss": 43.8948,
-      "step": 141250
-    },
-    {
-      "epoch": 0.57070827458316,
-      "grad_norm": 1093.102294921875,
-      "learning_rate": 2.3688115232013866e-05,
-      "loss": 59.8695,
-      "step": 141260
-    },
-    {
-      "epoch": 0.5707486758485276,
-      "grad_norm": 675.7007446289062,
-      "learning_rate": 2.3684629395677194e-05,
-      "loss": 70.8972,
-      "step": 141270
-    },
-    {
-      "epoch": 0.5707890771138953,
-      "grad_norm": 751.9618530273438,
-      "learning_rate": 2.368114358498434e-05,
-      "loss": 43.2623,
-      "step": 141280
-    },
-    {
-      "epoch": 0.5708294783792628,
-      "grad_norm": 923.1489868164062,
-      "learning_rate": 2.367765780000328e-05,
-      "loss": 68.1855,
-      "step": 141290
-    },
-    {
-      "epoch": 0.5708698796446304,
-      "grad_norm": 514.2269897460938,
-      "learning_rate": 2.367417204080196e-05,
-      "loss": 86.9622,
-      "step": 141300
-    },
-    {
-      "epoch": 0.5709102809099981,
-      "grad_norm": 423.8339538574219,
-      "learning_rate": 2.3670686307448333e-05,
-      "loss": 36.2946,
-      "step": 141310
-    },
-    {
-      "epoch": 0.5709506821753657,
-      "grad_norm": 883.6669311523438,
-      "learning_rate": 2.366720060001037e-05,
-      "loss": 67.147,
-      "step": 141320
-    },
-    {
-      "epoch": 0.5709910834407333,
-      "grad_norm": 901.2623291015625,
-      "learning_rate": 2.366371491855601e-05,
-      "loss": 44.5435,
-      "step": 141330
-    },
-    {
-      "epoch": 0.571031484706101,
-      "grad_norm": 574.0656127929688,
-      "learning_rate": 2.3660229263153215e-05,
-      "loss": 68.7471,
-      "step": 141340
-    },
-    {
-      "epoch": 0.5710718859714686,
-      "grad_norm": 710.9147338867188,
-      "learning_rate": 2.3656743633869948e-05,
-      "loss": 51.1492,
-      "step": 141350
-    },
-    {
-      "epoch": 0.5711122872368363,
-      "grad_norm": 663.8213500976562,
-      "learning_rate": 2.365325803077415e-05,
-      "loss": 38.9658,
-      "step": 141360
-    },
-    {
-      "epoch": 0.5711526885022039,
-      "grad_norm": 0.0,
-      "learning_rate": 2.3649772453933784e-05,
-      "loss": 47.6883,
-      "step": 141370
-    },
-    {
-      "epoch": 0.5711930897675715,
-      "grad_norm": 1073.9561767578125,
-      "learning_rate": 2.36462869034168e-05,
-      "loss": 55.3894,
-      "step": 141380
-    },
-    {
-      "epoch": 0.5712334910329392,
-      "grad_norm": 463.0362243652344,
-      "learning_rate": 2.364280137929114e-05,
-      "loss": 54.0964,
-      "step": 141390
-    },
-    {
-      "epoch": 0.5712738922983068,
-      "grad_norm": 1073.5634765625,
-      "learning_rate": 2.3639315881624777e-05,
-      "loss": 32.8544,
-      "step": 141400
-    },
-    {
-      "epoch": 0.5713142935636745,
-      "grad_norm": 334.5204162597656,
-      "learning_rate": 2.3635830410485638e-05,
-      "loss": 59.4493,
-      "step": 141410
-    },
-    {
-      "epoch": 0.571354694829042,
-      "grad_norm": 698.5248413085938,
-      "learning_rate": 2.36323449659417e-05,
-      "loss": 76.9526,
-      "step": 141420
-    },
-    {
-      "epoch": 0.5713950960944096,
-      "grad_norm": 510.2958068847656,
-      "learning_rate": 2.36288595480609e-05,
-      "loss": 51.604,
-      "step": 141430
-    },
-    {
-      "epoch": 0.5714354973597773,
-      "grad_norm": 393.1766052246094,
-      "learning_rate": 2.3625374156911185e-05,
-      "loss": 30.1209,
-      "step": 141440
-    },
-    {
-      "epoch": 0.5714758986251449,
-      "grad_norm": 996.3530883789062,
-      "learning_rate": 2.3621888792560515e-05,
-      "loss": 94.3675,
-      "step": 141450
-    },
-    {
-      "epoch": 0.5715162998905126,
-      "grad_norm": 583.70458984375,
-      "learning_rate": 2.361840345507683e-05,
-      "loss": 57.4963,
-      "step": 141460
-    },
-    {
-      "epoch": 0.5715567011558802,
-      "grad_norm": 797.1538696289062,
-      "learning_rate": 2.361491814452807e-05,
-      "loss": 36.3796,
-      "step": 141470
-    },
-    {
-      "epoch": 0.5715971024212478,
-      "grad_norm": 973.2113647460938,
-      "learning_rate": 2.3611432860982204e-05,
-      "loss": 51.5345,
-      "step": 141480
-    },
-    {
-      "epoch": 0.5716375036866155,
-      "grad_norm": 1032.868896484375,
-      "learning_rate": 2.360794760450716e-05,
-      "loss": 46.4819,
-      "step": 141490
-    },
-    {
-      "epoch": 0.5716779049519831,
-      "grad_norm": 393.41485595703125,
-      "learning_rate": 2.3604462375170906e-05,
-      "loss": 34.917,
-      "step": 141500
-    },
-    {
-      "epoch": 0.5717183062173508,
-      "grad_norm": 296.49212646484375,
-      "learning_rate": 2.3600977173041374e-05,
-      "loss": 51.7042,
-      "step": 141510
-    },
-    {
-      "epoch": 0.5717587074827184,
-      "grad_norm": 1662.1243896484375,
-      "learning_rate": 2.3597491998186506e-05,
-      "loss": 45.6466,
-      "step": 141520
-    },
-    {
-      "epoch": 0.571799108748086,
-      "grad_norm": 413.155029296875,
-      "learning_rate": 2.3594006850674262e-05,
-      "loss": 59.2692,
-      "step": 141530
-    },
-    {
-      "epoch": 0.5718395100134536,
-      "grad_norm": 1063.03076171875,
-      "learning_rate": 2.359052173057258e-05,
-      "loss": 57.3971,
-      "step": 141540
-    },
-    {
-      "epoch": 0.5718799112788212,
-      "grad_norm": 440.7333679199219,
-      "learning_rate": 2.3587036637949388e-05,
-      "loss": 41.32,
-      "step": 141550
-    },
-    {
-      "epoch": 0.5719203125441888,
-      "grad_norm": 868.6510620117188,
-      "learning_rate": 2.3583551572872656e-05,
-      "loss": 56.1956,
-      "step": 141560
-    },
-    {
-      "epoch": 0.5719607138095565,
-      "grad_norm": 1081.4976806640625,
-      "learning_rate": 2.3580066535410307e-05,
-      "loss": 45.4585,
-      "step": 141570
-    },
-    {
-      "epoch": 0.5720011150749241,
-      "grad_norm": 320.47918701171875,
-      "learning_rate": 2.35765815256303e-05,
-      "loss": 53.1529,
-      "step": 141580
-    },
-    {
-      "epoch": 0.5720415163402918,
-      "grad_norm": 540.5079956054688,
-      "learning_rate": 2.3573096543600567e-05,
-      "loss": 33.4318,
-      "step": 141590
-    },
-    {
-      "epoch": 0.5720819176056594,
-      "grad_norm": 695.275146484375,
-      "learning_rate": 2.3569611589389047e-05,
-      "loss": 55.8058,
-      "step": 141600
-    },
-    {
-      "epoch": 0.572122318871027,
-      "grad_norm": 744.6594848632812,
-      "learning_rate": 2.3566126663063695e-05,
-      "loss": 69.8493,
-      "step": 141610
-    },
-    {
-      "epoch": 0.5721627201363947,
-      "grad_norm": 1078.035400390625,
-      "learning_rate": 2.3562641764692433e-05,
-      "loss": 42.8643,
-      "step": 141620
-    },
-    {
-      "epoch": 0.5722031214017623,
-      "grad_norm": 1568.65625,
-      "learning_rate": 2.3559156894343213e-05,
-      "loss": 76.2597,
-      "step": 141630
-    },
-    {
-      "epoch": 0.57224352266713,
-      "grad_norm": 221.61920166015625,
-      "learning_rate": 2.355567205208397e-05,
-      "loss": 50.2042,
-      "step": 141640
-    },
-    {
-      "epoch": 0.5722839239324976,
-      "grad_norm": 523.4913940429688,
-      "learning_rate": 2.355218723798264e-05,
-      "loss": 62.5304,
-      "step": 141650
-    },
-    {
-      "epoch": 0.5723243251978652,
-      "grad_norm": 475.74395751953125,
-      "learning_rate": 2.3548702452107175e-05,
-      "loss": 43.9872,
-      "step": 141660
-    },
-    {
-      "epoch": 0.5723647264632328,
-      "grad_norm": 1220.044921875,
-      "learning_rate": 2.3545217694525498e-05,
-      "loss": 67.9373,
-      "step": 141670
-    },
-    {
-      "epoch": 0.5724051277286004,
-      "grad_norm": 1851.209228515625,
-      "learning_rate": 2.3541732965305543e-05,
-      "loss": 57.7291,
-      "step": 141680
-    },
-    {
-      "epoch": 0.5724455289939681,
-      "grad_norm": 757.2409057617188,
-      "learning_rate": 2.3538248264515266e-05,
-      "loss": 49.6187,
-      "step": 141690
-    },
-    {
-      "epoch": 0.5724859302593357,
-      "grad_norm": 5554.1015625,
-      "learning_rate": 2.3534763592222586e-05,
-      "loss": 68.57,
-      "step": 141700
-    },
-    {
-      "epoch": 0.5725263315247033,
-      "grad_norm": 962.51611328125,
-      "learning_rate": 2.353127894849545e-05,
-      "loss": 64.5327,
-      "step": 141710
-    },
-    {
-      "epoch": 0.572566732790071,
-      "grad_norm": 1030.315673828125,
-      "learning_rate": 2.3527794333401786e-05,
-      "loss": 101.8608,
-      "step": 141720
-    },
-    {
-      "epoch": 0.5726071340554386,
-      "grad_norm": 1969.6497802734375,
-      "learning_rate": 2.3524309747009522e-05,
-      "loss": 75.3899,
-      "step": 141730
-    },
-    {
-      "epoch": 0.5726475353208063,
-      "grad_norm": 591.5418701171875,
-      "learning_rate": 2.3520825189386604e-05,
-      "loss": 48.2578,
-      "step": 141740
-    },
-    {
-      "epoch": 0.5726879365861739,
-      "grad_norm": 844.501953125,
-      "learning_rate": 2.3517340660600964e-05,
-      "loss": 56.6854,
-      "step": 141750
-    },
-    {
-      "epoch": 0.5727283378515415,
-      "grad_norm": 694.3187866210938,
-      "learning_rate": 2.3513856160720522e-05,
-      "loss": 81.1811,
-      "step": 141760
-    },
-    {
-      "epoch": 0.5727687391169092,
-      "grad_norm": 1127.83544921875,
-      "learning_rate": 2.3510371689813226e-05,
-      "loss": 62.8335,
-      "step": 141770
-    },
-    {
-      "epoch": 0.5728091403822768,
-      "grad_norm": 544.0736694335938,
-      "learning_rate": 2.3506887247947e-05,
-      "loss": 46.8202,
-      "step": 141780
-    },
-    {
-      "epoch": 0.5728495416476445,
-      "grad_norm": 1501.36083984375,
-      "learning_rate": 2.3503402835189778e-05,
-      "loss": 52.4827,
-      "step": 141790
-    },
-    {
-      "epoch": 0.572889942913012,
-      "grad_norm": 1339.47705078125,
-      "learning_rate": 2.349991845160949e-05,
-      "loss": 54.5597,
-      "step": 141800
-    },
-    {
-      "epoch": 0.5729303441783796,
-      "grad_norm": 1593.72705078125,
-      "learning_rate": 2.349643409727405e-05,
-      "loss": 53.1381,
-      "step": 141810
-    },
-    {
-      "epoch": 0.5729707454437473,
-      "grad_norm": 657.8176879882812,
-      "learning_rate": 2.3492949772251414e-05,
-      "loss": 51.3417,
-      "step": 141820
-    },
-    {
-      "epoch": 0.5730111467091149,
-      "grad_norm": 1007.0821533203125,
-      "learning_rate": 2.3489465476609502e-05,
-      "loss": 77.4722,
-      "step": 141830
-    },
-    {
-      "epoch": 0.5730515479744825,
-      "grad_norm": 613.6268920898438,
-      "learning_rate": 2.348598121041622e-05,
-      "loss": 64.9442,
-      "step": 141840
-    },
-    {
-      "epoch": 0.5730919492398502,
-      "grad_norm": 847.4960327148438,
-      "learning_rate": 2.348249697373953e-05,
-      "loss": 36.6914,
-      "step": 141850
-    },
-    {
-      "epoch": 0.5731323505052178,
-      "grad_norm": 692.1813354492188,
-      "learning_rate": 2.3479012766647335e-05,
-      "loss": 68.5336,
-      "step": 141860
-    },
-    {
-      "epoch": 0.5731727517705855,
-      "grad_norm": 813.51123046875,
-      "learning_rate": 2.3475528589207576e-05,
-      "loss": 48.4441,
-      "step": 141870
-    },
-    {
-      "epoch": 0.5732131530359531,
-      "grad_norm": 572.866943359375,
-      "learning_rate": 2.3472044441488174e-05,
-      "loss": 45.9235,
-      "step": 141880
-    },
-    {
-      "epoch": 0.5732535543013207,
-      "grad_norm": 273.5608215332031,
-      "learning_rate": 2.3468560323557036e-05,
-      "loss": 57.0392,
-      "step": 141890
-    },
-    {
-      "epoch": 0.5732939555666884,
-      "grad_norm": 889.541015625,
-      "learning_rate": 2.3465076235482116e-05,
-      "loss": 33.2406,
-      "step": 141900
-    },
-    {
-      "epoch": 0.573334356832056,
-      "grad_norm": 0.0,
-      "learning_rate": 2.3461592177331325e-05,
-      "loss": 46.7832,
-      "step": 141910
-    },
-    {
-      "epoch": 0.5733747580974237,
-      "grad_norm": 558.1715698242188,
-      "learning_rate": 2.345810814917258e-05,
-      "loss": 43.9247,
-      "step": 141920
-    },
-    {
-      "epoch": 0.5734151593627912,
-      "grad_norm": 800.1122436523438,
-      "learning_rate": 2.3454624151073815e-05,
-      "loss": 51.9048,
-      "step": 141930
-    },
-    {
-      "epoch": 0.5734555606281588,
-      "grad_norm": 1590.205078125,
-      "learning_rate": 2.3451140183102945e-05,
-      "loss": 58.2996,
-      "step": 141940
-    },
-    {
-      "epoch": 0.5734959618935265,
-      "grad_norm": 594.8118896484375,
-      "learning_rate": 2.34476562453279e-05,
-      "loss": 56.2581,
-      "step": 141950
-    },
-    {
-      "epoch": 0.5735363631588941,
-      "grad_norm": 734.5135498046875,
-      "learning_rate": 2.3444172337816592e-05,
-      "loss": 51.2526,
-      "step": 141960
-    },
-    {
-      "epoch": 0.5735767644242618,
-      "grad_norm": 316.3956298828125,
-      "learning_rate": 2.344068846063694e-05,
-      "loss": 46.3092,
-      "step": 141970
-    },
-    {
-      "epoch": 0.5736171656896294,
-      "grad_norm": 703.2752075195312,
-      "learning_rate": 2.343720461385688e-05,
-      "loss": 38.9623,
-      "step": 141980
-    },
-    {
-      "epoch": 0.573657566954997,
-      "grad_norm": 471.1736755371094,
-      "learning_rate": 2.3433720797544316e-05,
-      "loss": 47.8452,
-      "step": 141990
-    },
-    {
-      "epoch": 0.5736979682203647,
-      "grad_norm": 1363.9534912109375,
-      "learning_rate": 2.3430237011767167e-05,
-      "loss": 46.777,
-      "step": 142000
-    },
-    {
-      "epoch": 0.5737383694857323,
-      "grad_norm": 956.2443237304688,
-      "learning_rate": 2.3426753256593363e-05,
-      "loss": 52.6443,
-      "step": 142010
-    },
-    {
-      "epoch": 0.5737787707511,
-      "grad_norm": 627.6842041015625,
-      "learning_rate": 2.3423269532090802e-05,
-      "loss": 55.9855,
-      "step": 142020
-    },
-    {
-      "epoch": 0.5738191720164676,
-      "grad_norm": 1048.773681640625,
-      "learning_rate": 2.3419785838327425e-05,
-      "loss": 65.7966,
-      "step": 142030
-    },
-    {
-      "epoch": 0.5738595732818352,
-      "grad_norm": 985.0260620117188,
-      "learning_rate": 2.3416302175371138e-05,
-      "loss": 53.6217,
-      "step": 142040
-    },
-    {
-      "epoch": 0.5738999745472029,
-      "grad_norm": 530.5432739257812,
-      "learning_rate": 2.341281854328985e-05,
-      "loss": 77.8192,
-      "step": 142050
-    },
-    {
-      "epoch": 0.5739403758125704,
-      "grad_norm": 1114.1192626953125,
-      "learning_rate": 2.3409334942151484e-05,
-      "loss": 64.7953,
-      "step": 142060
-    },
-    {
-      "epoch": 0.573980777077938,
-      "grad_norm": 861.7225952148438,
-      "learning_rate": 2.3405851372023953e-05,
-      "loss": 51.9311,
-      "step": 142070
-    },
-    {
-      "epoch": 0.5740211783433057,
-      "grad_norm": 636.5674438476562,
-      "learning_rate": 2.3402367832975163e-05,
-      "loss": 49.7294,
-      "step": 142080
-    },
-    {
-      "epoch": 0.5740615796086733,
-      "grad_norm": 892.6287841796875,
-      "learning_rate": 2.339888432507304e-05,
-      "loss": 67.6866,
-      "step": 142090
-    },
-    {
-      "epoch": 0.574101980874041,
-      "grad_norm": 596.37646484375,
-      "learning_rate": 2.3395400848385486e-05,
-      "loss": 55.9081,
-      "step": 142100
-    },
-    {
-      "epoch": 0.5741423821394086,
-      "grad_norm": 514.5144653320312,
-      "learning_rate": 2.3391917402980426e-05,
-      "loss": 54.0444,
-      "step": 142110
-    },
-    {
-      "epoch": 0.5741827834047762,
-      "grad_norm": 2546.439208984375,
-      "learning_rate": 2.3388433988925763e-05,
-      "loss": 93.0408,
-      "step": 142120
-    },
-    {
-      "epoch": 0.5742231846701439,
-      "grad_norm": 704.0580444335938,
-      "learning_rate": 2.3384950606289406e-05,
-      "loss": 76.4964,
-      "step": 142130
-    },
-    {
-      "epoch": 0.5742635859355115,
-      "grad_norm": 818.4609985351562,
-      "learning_rate": 2.3381467255139276e-05,
-      "loss": 37.8892,
-      "step": 142140
-    },
-    {
-      "epoch": 0.5743039872008792,
-      "grad_norm": 854.7376708984375,
-      "learning_rate": 2.3377983935543264e-05,
-      "loss": 50.8855,
-      "step": 142150
-    },
-    {
-      "epoch": 0.5743443884662468,
-      "grad_norm": 517.0084838867188,
-      "learning_rate": 2.3374500647569297e-05,
-      "loss": 60.2467,
-      "step": 142160
-    },
-    {
-      "epoch": 0.5743847897316144,
-      "grad_norm": 550.582275390625,
-      "learning_rate": 2.3371017391285278e-05,
-      "loss": 59.1301,
-      "step": 142170
-    },
-    {
-      "epoch": 0.574425190996982,
-      "grad_norm": 652.0669555664062,
-      "learning_rate": 2.3367534166759102e-05,
-      "loss": 57.5477,
-      "step": 142180
-    },
-    {
-      "epoch": 0.5744655922623496,
-      "grad_norm": 714.7897338867188,
-      "learning_rate": 2.3364050974058703e-05,
-      "loss": 71.8189,
-      "step": 142190
-    },
-    {
-      "epoch": 0.5745059935277173,
-      "grad_norm": 288.60345458984375,
-      "learning_rate": 2.336056781325197e-05,
-      "loss": 40.1994,
-      "step": 142200
-    },
-    {
-      "epoch": 0.5745463947930849,
-      "grad_norm": 4422.25830078125,
-      "learning_rate": 2.3357084684406806e-05,
-      "loss": 63.4722,
-      "step": 142210
-    },
-    {
-      "epoch": 0.5745867960584525,
-      "grad_norm": 474.19500732421875,
-      "learning_rate": 2.3353601587591125e-05,
-      "loss": 53.761,
-      "step": 142220
-    },
-    {
-      "epoch": 0.5746271973238202,
-      "grad_norm": 0.0,
-      "learning_rate": 2.3350118522872823e-05,
-      "loss": 50.8316,
-      "step": 142230
-    },
-    {
-      "epoch": 0.5746675985891878,
-      "grad_norm": 610.2474975585938,
-      "learning_rate": 2.3346635490319814e-05,
-      "loss": 58.9436,
-      "step": 142240
-    },
-    {
-      "epoch": 0.5747079998545555,
-      "grad_norm": 1580.07421875,
-      "learning_rate": 2.3343152490000004e-05,
-      "loss": 88.0132,
-      "step": 142250
-    },
-    {
-      "epoch": 0.5747484011199231,
-      "grad_norm": 365.8472900390625,
-      "learning_rate": 2.3339669521981273e-05,
-      "loss": 51.9967,
-      "step": 142260
-    },
-    {
-      "epoch": 0.5747888023852907,
-      "grad_norm": 1166.15087890625,
-      "learning_rate": 2.3336186586331555e-05,
-      "loss": 51.4411,
-      "step": 142270
-    },
-    {
-      "epoch": 0.5748292036506584,
-      "grad_norm": 1141.8216552734375,
-      "learning_rate": 2.3332703683118732e-05,
-      "loss": 78.2708,
-      "step": 142280
-    },
-    {
-      "epoch": 0.574869604916026,
-      "grad_norm": 956.0130004882812,
-      "learning_rate": 2.3329220812410704e-05,
-      "loss": 40.9031,
-      "step": 142290
-    },
-    {
-      "epoch": 0.5749100061813937,
-      "grad_norm": 1446.7919921875,
-      "learning_rate": 2.332573797427538e-05,
-      "loss": 78.6235,
-      "step": 142300
-    },
-    {
-      "epoch": 0.5749504074467612,
-      "grad_norm": 834.1378784179688,
-      "learning_rate": 2.3322255168780655e-05,
-      "loss": 33.3753,
-      "step": 142310
-    },
-    {
-      "epoch": 0.5749908087121288,
-      "grad_norm": 970.3228759765625,
-      "learning_rate": 2.3318772395994433e-05,
-      "loss": 57.855,
-      "step": 142320
-    },
-    {
-      "epoch": 0.5750312099774965,
-      "grad_norm": 1489.4097900390625,
-      "learning_rate": 2.3315289655984605e-05,
-      "loss": 63.0613,
-      "step": 142330
-    },
-    {
-      "epoch": 0.5750716112428641,
-      "grad_norm": 610.0813598632812,
-      "learning_rate": 2.331180694881907e-05,
-      "loss": 73.1477,
-      "step": 142340
-    },
-    {
-      "epoch": 0.5751120125082317,
-      "grad_norm": 694.2882080078125,
-      "learning_rate": 2.3308324274565734e-05,
-      "loss": 38.6415,
-      "step": 142350
-    },
-    {
-      "epoch": 0.5751524137735994,
-      "grad_norm": 648.3965454101562,
-      "learning_rate": 2.3304841633292487e-05,
-      "loss": 41.1692,
-      "step": 142360
-    },
-    {
-      "epoch": 0.575192815038967,
-      "grad_norm": 966.6323852539062,
-      "learning_rate": 2.3301359025067216e-05,
-      "loss": 56.3727,
-      "step": 142370
-    },
-    {
-      "epoch": 0.5752332163043347,
-      "grad_norm": 769.1686401367188,
-      "learning_rate": 2.3297876449957834e-05,
-      "loss": 78.7668,
-      "step": 142380
-    },
-    {
-      "epoch": 0.5752736175697023,
-      "grad_norm": 397.4100341796875,
-      "learning_rate": 2.329439390803222e-05,
-      "loss": 61.6241,
-      "step": 142390
-    },
-    {
-      "epoch": 0.5753140188350699,
-      "grad_norm": 809.5210571289062,
-      "learning_rate": 2.3290911399358285e-05,
-      "loss": 49.5255,
-      "step": 142400
-    },
-    {
-      "epoch": 0.5753544201004376,
-      "grad_norm": 293.4340515136719,
-      "learning_rate": 2.3287428924003912e-05,
-      "loss": 43.1965,
-      "step": 142410
-    },
-    {
-      "epoch": 0.5753948213658052,
-      "grad_norm": 966.3424682617188,
-      "learning_rate": 2.3283946482036982e-05,
-      "loss": 53.9423,
-      "step": 142420
-    },
-    {
-      "epoch": 0.5754352226311729,
-      "grad_norm": 483.5935974121094,
-      "learning_rate": 2.3280464073525407e-05,
-      "loss": 47.5288,
-      "step": 142430
-    },
-    {
-      "epoch": 0.5754756238965404,
-      "grad_norm": 422.25970458984375,
-      "learning_rate": 2.327698169853707e-05,
-      "loss": 26.3258,
-      "step": 142440
-    },
-    {
-      "epoch": 0.575516025161908,
-      "grad_norm": 942.5641479492188,
-      "learning_rate": 2.3273499357139855e-05,
-      "loss": 57.1318,
-      "step": 142450
-    },
-    {
-      "epoch": 0.5755564264272757,
-      "grad_norm": 736.4271240234375,
-      "learning_rate": 2.3270017049401664e-05,
-      "loss": 56.0266,
-      "step": 142460
-    },
-    {
-      "epoch": 0.5755968276926433,
-      "grad_norm": 952.470947265625,
-      "learning_rate": 2.3266534775390383e-05,
-      "loss": 61.0546,
-      "step": 142470
-    },
-    {
-      "epoch": 0.575637228958011,
-      "grad_norm": 410.7485656738281,
-      "learning_rate": 2.32630525351739e-05,
-      "loss": 32.2695,
-      "step": 142480
-    },
-    {
-      "epoch": 0.5756776302233786,
-      "grad_norm": 528.5092163085938,
-      "learning_rate": 2.3259570328820106e-05,
-      "loss": 46.2177,
-      "step": 142490
-    },
-    {
-      "epoch": 0.5757180314887462,
-      "grad_norm": 528.0546264648438,
-      "learning_rate": 2.3256088156396868e-05,
-      "loss": 44.6508,
-      "step": 142500
-    },
-    {
-      "epoch": 0.5757584327541139,
-      "grad_norm": 658.5557861328125,
-      "learning_rate": 2.3252606017972103e-05,
-      "loss": 34.0906,
-      "step": 142510
-    },
-    {
-      "epoch": 0.5757988340194815,
-      "grad_norm": 507.3339538574219,
-      "learning_rate": 2.324912391361368e-05,
-      "loss": 49.9197,
-      "step": 142520
-    },
-    {
-      "epoch": 0.5758392352848491,
-      "grad_norm": 940.519775390625,
-      "learning_rate": 2.324564184338948e-05,
-      "loss": 64.702,
-      "step": 142530
-    },
-    {
-      "epoch": 0.5758796365502168,
-      "grad_norm": 806.0038452148438,
-      "learning_rate": 2.3242159807367408e-05,
-      "loss": 70.4329,
-      "step": 142540
-    },
-    {
-      "epoch": 0.5759200378155844,
-      "grad_norm": 641.6761474609375,
-      "learning_rate": 2.323867780561533e-05,
-      "loss": 43.513,
-      "step": 142550
-    },
-    {
-      "epoch": 0.5759604390809521,
-      "grad_norm": 577.0625610351562,
-      "learning_rate": 2.323519583820114e-05,
-      "loss": 72.0487,
-      "step": 142560
-    },
-    {
-      "epoch": 0.5760008403463196,
-      "grad_norm": 625.9312744140625,
-      "learning_rate": 2.323171390519271e-05,
-      "loss": 43.7065,
-      "step": 142570
-    },
-    {
-      "epoch": 0.5760412416116872,
-      "grad_norm": 567.3580322265625,
-      "learning_rate": 2.3228232006657923e-05,
-      "loss": 54.7438,
-      "step": 142580
-    },
-    {
-      "epoch": 0.5760816428770549,
-      "grad_norm": 770.6826782226562,
-      "learning_rate": 2.3224750142664677e-05,
-      "loss": 59.1607,
-      "step": 142590
-    },
-    {
-      "epoch": 0.5761220441424225,
-      "grad_norm": 758.7730712890625,
-      "learning_rate": 2.3221268313280838e-05,
-      "loss": 63.5247,
-      "step": 142600
-    },
-    {
-      "epoch": 0.5761624454077902,
-      "grad_norm": 665.5321044921875,
-      "learning_rate": 2.3217786518574273e-05,
-      "loss": 41.7696,
-      "step": 142610
-    },
-    {
-      "epoch": 0.5762028466731578,
-      "grad_norm": 671.5497436523438,
-      "learning_rate": 2.32143047586129e-05,
-      "loss": 58.2926,
-      "step": 142620
-    },
-    {
-      "epoch": 0.5762432479385254,
-      "grad_norm": 843.5665283203125,
-      "learning_rate": 2.321082303346456e-05,
-      "loss": 52.1115,
-      "step": 142630
-    },
-    {
-      "epoch": 0.5762836492038931,
-      "grad_norm": 1791.599853515625,
-      "learning_rate": 2.320734134319715e-05,
-      "loss": 52.7765,
-      "step": 142640
-    },
-    {
-      "epoch": 0.5763240504692607,
-      "grad_norm": 618.1298217773438,
-      "learning_rate": 2.3203859687878548e-05,
-      "loss": 50.9905,
-      "step": 142650
-    },
-    {
-      "epoch": 0.5763644517346284,
-      "grad_norm": 620.7821655273438,
-      "learning_rate": 2.320037806757662e-05,
-      "loss": 45.4659,
-      "step": 142660
-    },
-    {
-      "epoch": 0.576404852999996,
-      "grad_norm": 1306.3067626953125,
-      "learning_rate": 2.3196896482359253e-05,
-      "loss": 58.8106,
-      "step": 142670
-    },
-    {
-      "epoch": 0.5764452542653636,
-      "grad_norm": 582.6734619140625,
-      "learning_rate": 2.319341493229431e-05,
-      "loss": 48.8326,
-      "step": 142680
-    },
-    {
-      "epoch": 0.5764856555307313,
-      "grad_norm": 733.7221069335938,
-      "learning_rate": 2.318993341744968e-05,
-      "loss": 61.6943,
-      "step": 142690
-    },
-    {
-      "epoch": 0.5765260567960988,
-      "grad_norm": 670.154296875,
-      "learning_rate": 2.3186451937893235e-05,
-      "loss": 67.0611,
-      "step": 142700
-    },
-    {
-      "epoch": 0.5765664580614664,
-      "grad_norm": 633.4531860351562,
-      "learning_rate": 2.3182970493692827e-05,
-      "loss": 60.647,
-      "step": 142710
-    },
-    {
-      "epoch": 0.5766068593268341,
-      "grad_norm": 878.5447998046875,
-      "learning_rate": 2.3179489084916358e-05,
-      "loss": 43.6026,
-      "step": 142720
-    },
-    {
-      "epoch": 0.5766472605922017,
-      "grad_norm": 410.3040466308594,
-      "learning_rate": 2.3176007711631687e-05,
-      "loss": 25.5606,
-      "step": 142730
-    },
-    {
-      "epoch": 0.5766876618575694,
-      "grad_norm": 833.0294189453125,
-      "learning_rate": 2.317252637390668e-05,
-      "loss": 48.7416,
-      "step": 142740
-    },
-    {
-      "epoch": 0.576728063122937,
-      "grad_norm": 1507.3426513671875,
-      "learning_rate": 2.3169045071809215e-05,
-      "loss": 62.4276,
-      "step": 142750
-    },
-    {
-      "epoch": 0.5767684643883046,
-      "grad_norm": 1039.6998291015625,
-      "learning_rate": 2.316556380540715e-05,
-      "loss": 44.9483,
-      "step": 142760
-    },
-    {
-      "epoch": 0.5768088656536723,
-      "grad_norm": 594.3389892578125,
-      "learning_rate": 2.316208257476837e-05,
-      "loss": 56.9623,
-      "step": 142770
-    },
-    {
-      "epoch": 0.5768492669190399,
-      "grad_norm": 1160.0947265625,
-      "learning_rate": 2.315860137996074e-05,
-      "loss": 62.822,
-      "step": 142780
-    },
-    {
-      "epoch": 0.5768896681844076,
-      "grad_norm": 651.6566162109375,
-      "learning_rate": 2.3155120221052112e-05,
-      "loss": 54.3397,
-      "step": 142790
-    },
-    {
-      "epoch": 0.5769300694497752,
-      "grad_norm": 0.0,
-      "learning_rate": 2.3151639098110377e-05,
-      "loss": 64.4152,
-      "step": 142800
-    },
-    {
-      "epoch": 0.5769704707151428,
-      "grad_norm": 653.4903564453125,
-      "learning_rate": 2.3148158011203388e-05,
-      "loss": 77.8834,
-      "step": 142810
-    },
-    {
-      "epoch": 0.5770108719805104,
-      "grad_norm": 596.7733154296875,
-      "learning_rate": 2.314467696039901e-05,
-      "loss": 65.638,
-      "step": 142820
-    },
-    {
-      "epoch": 0.577051273245878,
-      "grad_norm": 552.674072265625,
-      "learning_rate": 2.314119594576511e-05,
-      "loss": 54.3161,
-      "step": 142830
-    },
-    {
-      "epoch": 0.5770916745112457,
-      "grad_norm": 296.8179931640625,
-      "learning_rate": 2.3137714967369545e-05,
-      "loss": 38.5794,
-      "step": 142840
-    },
-    {
-      "epoch": 0.5771320757766133,
-      "grad_norm": 487.05706787109375,
-      "learning_rate": 2.3134234025280196e-05,
-      "loss": 60.3549,
-      "step": 142850
-    },
-    {
-      "epoch": 0.5771724770419809,
-      "grad_norm": 747.7566528320312,
-      "learning_rate": 2.3130753119564915e-05,
-      "loss": 53.9551,
-      "step": 142860
-    },
-    {
-      "epoch": 0.5772128783073486,
-      "grad_norm": 469.7563171386719,
-      "learning_rate": 2.3127272250291553e-05,
-      "loss": 56.6801,
-      "step": 142870
-    },
-    {
-      "epoch": 0.5772532795727162,
-      "grad_norm": 433.9674377441406,
-      "learning_rate": 2.3123791417527994e-05,
-      "loss": 62.3665,
-      "step": 142880
-    },
-    {
-      "epoch": 0.5772936808380839,
-      "grad_norm": 355.9599304199219,
-      "learning_rate": 2.3120310621342086e-05,
-      "loss": 49.0493,
-      "step": 142890
-    },
-    {
-      "epoch": 0.5773340821034515,
-      "grad_norm": 1589.1834716796875,
-      "learning_rate": 2.3116829861801686e-05,
-      "loss": 76.2406,
-      "step": 142900
-    },
-    {
-      "epoch": 0.5773744833688191,
-      "grad_norm": 1414.6934814453125,
-      "learning_rate": 2.311334913897466e-05,
-      "loss": 66.9008,
-      "step": 142910
-    },
-    {
-      "epoch": 0.5774148846341868,
-      "grad_norm": 681.8200073242188,
-      "learning_rate": 2.3109868452928855e-05,
-      "loss": 50.0082,
-      "step": 142920
-    },
-    {
-      "epoch": 0.5774552858995544,
-      "grad_norm": 1097.9447021484375,
-      "learning_rate": 2.310638780373215e-05,
-      "loss": 59.7922,
-      "step": 142930
-    },
-    {
-      "epoch": 0.5774956871649221,
-      "grad_norm": 532.4230346679688,
-      "learning_rate": 2.3102907191452388e-05,
-      "loss": 41.8895,
-      "step": 142940
-    },
-    {
-      "epoch": 0.5775360884302896,
-      "grad_norm": 990.6885986328125,
-      "learning_rate": 2.309942661615742e-05,
-      "loss": 60.9387,
-      "step": 142950
-    },
-    {
-      "epoch": 0.5775764896956572,
-      "grad_norm": 980.983154296875,
-      "learning_rate": 2.3095946077915114e-05,
-      "loss": 61.0354,
-      "step": 142960
-    },
-    {
-      "epoch": 0.5776168909610249,
-      "grad_norm": 697.8203735351562,
-      "learning_rate": 2.309246557679332e-05,
-      "loss": 44.8736,
-      "step": 142970
-    },
-    {
-      "epoch": 0.5776572922263925,
-      "grad_norm": 645.3919067382812,
-      "learning_rate": 2.3088985112859884e-05,
-      "loss": 37.8088,
-      "step": 142980
-    },
-    {
-      "epoch": 0.5776976934917601,
-      "grad_norm": 1883.624755859375,
-      "learning_rate": 2.3085504686182673e-05,
-      "loss": 40.0972,
-      "step": 142990
-    },
-    {
-      "epoch": 0.5777380947571278,
-      "grad_norm": 841.05029296875,
-      "learning_rate": 2.3082024296829536e-05,
-      "loss": 60.6423,
-      "step": 143000
-    },
-    {
-      "epoch": 0.5777784960224954,
-      "grad_norm": 396.6905212402344,
-      "learning_rate": 2.3078543944868323e-05,
-      "loss": 45.2093,
-      "step": 143010
-    },
-    {
-      "epoch": 0.5778188972878631,
-      "grad_norm": 663.75,
-      "learning_rate": 2.3075063630366884e-05,
-      "loss": 70.7723,
-      "step": 143020
-    },
-    {
-      "epoch": 0.5778592985532307,
-      "grad_norm": 396.67694091796875,
-      "learning_rate": 2.307158335339306e-05,
-      "loss": 47.533,
-      "step": 143030
-    },
-    {
-      "epoch": 0.5778996998185983,
-      "grad_norm": 704.8739624023438,
-      "learning_rate": 2.3068103114014726e-05,
-      "loss": 41.4563,
-      "step": 143040
-    },
-    {
-      "epoch": 0.577940101083966,
-      "grad_norm": 1157.4110107421875,
-      "learning_rate": 2.3064622912299712e-05,
-      "loss": 49.6344,
-      "step": 143050
-    },
-    {
-      "epoch": 0.5779805023493336,
-      "grad_norm": 665.005615234375,
-      "learning_rate": 2.3061142748315865e-05,
-      "loss": 58.9296,
-      "step": 143060
-    },
-    {
-      "epoch": 0.5780209036147013,
-      "grad_norm": 542.858154296875,
-      "learning_rate": 2.3057662622131047e-05,
-      "loss": 49.8475,
-      "step": 143070
-    },
-    {
-      "epoch": 0.5780613048800688,
-      "grad_norm": 755.5459594726562,
-      "learning_rate": 2.3054182533813087e-05,
-      "loss": 47.4976,
-      "step": 143080
-    },
-    {
-      "epoch": 0.5781017061454364,
-      "grad_norm": 445.7283935546875,
-      "learning_rate": 2.3050702483429846e-05,
-      "loss": 48.5714,
-      "step": 143090
-    },
-    {
-      "epoch": 0.5781421074108041,
-      "grad_norm": 512.7163696289062,
-      "learning_rate": 2.304722247104917e-05,
-      "loss": 73.921,
-      "step": 143100
-    },
-    {
-      "epoch": 0.5781825086761717,
-      "grad_norm": 332.2102355957031,
-      "learning_rate": 2.3043742496738883e-05,
-      "loss": 48.0645,
-      "step": 143110
-    },
-    {
-      "epoch": 0.5782229099415394,
-      "grad_norm": 330.4541015625,
-      "learning_rate": 2.304026256056685e-05,
-      "loss": 43.9408,
-      "step": 143120
-    },
-    {
-      "epoch": 0.578263311206907,
-      "grad_norm": 639.76416015625,
-      "learning_rate": 2.3036782662600915e-05,
-      "loss": 55.066,
-      "step": 143130
-    },
-    {
-      "epoch": 0.5783037124722746,
-      "grad_norm": 491.55560302734375,
-      "learning_rate": 2.3033302802908894e-05,
-      "loss": 52.4497,
-      "step": 143140
-    },
-    {
-      "epoch": 0.5783441137376423,
-      "grad_norm": 1402.951416015625,
-      "learning_rate": 2.3029822981558667e-05,
-      "loss": 55.6616,
-      "step": 143150
-    },
-    {
-      "epoch": 0.5783845150030099,
-      "grad_norm": 418.091552734375,
-      "learning_rate": 2.3026343198618043e-05,
-      "loss": 36.3308,
-      "step": 143160
-    },
-    {
-      "epoch": 0.5784249162683776,
-      "grad_norm": 1122.232177734375,
-      "learning_rate": 2.302286345415488e-05,
-      "loss": 37.5329,
-      "step": 143170
-    },
-    {
-      "epoch": 0.5784653175337452,
-      "grad_norm": 1635.8270263671875,
-      "learning_rate": 2.3019383748237015e-05,
-      "loss": 118.2091,
-      "step": 143180
-    },
-    {
-      "epoch": 0.5785057187991128,
-      "grad_norm": 747.5432739257812,
-      "learning_rate": 2.3015904080932272e-05,
-      "loss": 49.2059,
-      "step": 143190
-    },
-    {
-      "epoch": 0.5785461200644805,
-      "grad_norm": 894.6597900390625,
-      "learning_rate": 2.301242445230851e-05,
-      "loss": 51.8367,
-      "step": 143200
-    },
-    {
-      "epoch": 0.578586521329848,
-      "grad_norm": 758.6695556640625,
-      "learning_rate": 2.3008944862433547e-05,
-      "loss": 64.8202,
-      "step": 143210
-    },
-    {
-      "epoch": 0.5786269225952156,
-      "grad_norm": 359.57843017578125,
-      "learning_rate": 2.3005465311375237e-05,
-      "loss": 30.1586,
-      "step": 143220
-    },
-    {
-      "epoch": 0.5786673238605833,
-      "grad_norm": 484.9412536621094,
-      "learning_rate": 2.300198579920141e-05,
-      "loss": 49.8715,
-      "step": 143230
-    },
-    {
-      "epoch": 0.5787077251259509,
-      "grad_norm": 1004.665283203125,
-      "learning_rate": 2.2998506325979894e-05,
-      "loss": 66.9634,
-      "step": 143240
-    },
-    {
-      "epoch": 0.5787481263913186,
-      "grad_norm": 980.5902099609375,
-      "learning_rate": 2.299502689177853e-05,
-      "loss": 69.9582,
-      "step": 143250
-    },
-    {
-      "epoch": 0.5787885276566862,
-      "grad_norm": 621.208984375,
-      "learning_rate": 2.2991547496665148e-05,
-      "loss": 52.7556,
-      "step": 143260
-    },
-    {
-      "epoch": 0.5788289289220538,
-      "grad_norm": 673.4203491210938,
-      "learning_rate": 2.298806814070758e-05,
-      "loss": 60.5647,
-      "step": 143270
-    },
-    {
-      "epoch": 0.5788693301874215,
-      "grad_norm": 748.7210693359375,
-      "learning_rate": 2.2984588823973662e-05,
-      "loss": 61.0294,
-      "step": 143280
-    },
-    {
-      "epoch": 0.5789097314527891,
-      "grad_norm": 510.8747863769531,
-      "learning_rate": 2.2981109546531215e-05,
-      "loss": 54.1331,
-      "step": 143290
-    },
-    {
-      "epoch": 0.5789501327181568,
-      "grad_norm": 573.7908935546875,
-      "learning_rate": 2.2977630308448085e-05,
-      "loss": 50.7809,
-      "step": 143300
-    },
-    {
-      "epoch": 0.5789905339835244,
-      "grad_norm": 1115.3363037109375,
-      "learning_rate": 2.2974151109792096e-05,
-      "loss": 77.0252,
-      "step": 143310
-    },
-    {
-      "epoch": 0.579030935248892,
-      "grad_norm": 732.9327392578125,
-      "learning_rate": 2.2970671950631064e-05,
-      "loss": 41.6058,
-      "step": 143320
-    },
-    {
-      "epoch": 0.5790713365142596,
-      "grad_norm": 749.0936889648438,
-      "learning_rate": 2.296719283103283e-05,
-      "loss": 46.5037,
-      "step": 143330
-    },
-    {
-      "epoch": 0.5791117377796272,
-      "grad_norm": 416.95855712890625,
-      "learning_rate": 2.2963713751065223e-05,
-      "loss": 68.9422,
-      "step": 143340
-    },
-    {
-      "epoch": 0.5791521390449949,
-      "grad_norm": 234.63787841796875,
-      "learning_rate": 2.2960234710796063e-05,
-      "loss": 44.6314,
-      "step": 143350
-    },
-    {
-      "epoch": 0.5791925403103625,
-      "grad_norm": 1270.1083984375,
-      "learning_rate": 2.2956755710293183e-05,
-      "loss": 74.6942,
-      "step": 143360
-    },
-    {
-      "epoch": 0.5792329415757301,
-      "grad_norm": 1411.5767822265625,
-      "learning_rate": 2.2953276749624386e-05,
-      "loss": 63.6014,
-      "step": 143370
-    },
-    {
-      "epoch": 0.5792733428410978,
-      "grad_norm": 371.9848327636719,
-      "learning_rate": 2.2949797828857525e-05,
-      "loss": 38.8278,
-      "step": 143380
-    },
-    {
-      "epoch": 0.5793137441064654,
-      "grad_norm": 1254.4080810546875,
-      "learning_rate": 2.2946318948060413e-05,
-      "loss": 59.7256,
-      "step": 143390
-    },
-    {
-      "epoch": 0.579354145371833,
-      "grad_norm": 1216.3907470703125,
-      "learning_rate": 2.294284010730086e-05,
-      "loss": 38.042,
-      "step": 143400
-    },
-    {
-      "epoch": 0.5793945466372007,
-      "grad_norm": 910.777099609375,
-      "learning_rate": 2.2939361306646706e-05,
-      "loss": 69.6151,
-      "step": 143410
-    },
-    {
-      "epoch": 0.5794349479025683,
-      "grad_norm": 640.028564453125,
-      "learning_rate": 2.2935882546165766e-05,
-      "loss": 58.462,
-      "step": 143420
-    },
-    {
-      "epoch": 0.579475349167936,
-      "grad_norm": 915.5157470703125,
-      "learning_rate": 2.293240382592585e-05,
-      "loss": 55.9273,
-      "step": 143430
-    },
-    {
-      "epoch": 0.5795157504333036,
-      "grad_norm": 819.3345336914062,
-      "learning_rate": 2.2928925145994794e-05,
-      "loss": 47.2641,
-      "step": 143440
-    },
-    {
-      "epoch": 0.5795561516986713,
-      "grad_norm": 288.2989196777344,
-      "learning_rate": 2.29254465064404e-05,
-      "loss": 61.8187,
-      "step": 143450
-    },
-    {
-      "epoch": 0.5795965529640388,
-      "grad_norm": 467.6305847167969,
-      "learning_rate": 2.2921967907330504e-05,
-      "loss": 59.036,
-      "step": 143460
-    },
-    {
-      "epoch": 0.5796369542294064,
-      "grad_norm": 1217.71142578125,
-      "learning_rate": 2.291848934873291e-05,
-      "loss": 53.756,
-      "step": 143470
-    },
-    {
-      "epoch": 0.5796773554947741,
-      "grad_norm": 964.2899780273438,
-      "learning_rate": 2.291501083071543e-05,
-      "loss": 42.6057,
-      "step": 143480
-    },
-    {
-      "epoch": 0.5797177567601417,
-      "grad_norm": 787.3900146484375,
-      "learning_rate": 2.29115323533459e-05,
-      "loss": 45.8454,
-      "step": 143490
-    },
-    {
-      "epoch": 0.5797581580255093,
-      "grad_norm": 519.1593627929688,
-      "learning_rate": 2.2908053916692117e-05,
-      "loss": 56.6687,
-      "step": 143500
-    },
-    {
-      "epoch": 0.579798559290877,
-      "grad_norm": 649.1287841796875,
-      "learning_rate": 2.2904575520821897e-05,
-      "loss": 38.6635,
-      "step": 143510
-    },
-    {
-      "epoch": 0.5798389605562446,
-      "grad_norm": 696.3875122070312,
-      "learning_rate": 2.2901097165803062e-05,
-      "loss": 93.9158,
-      "step": 143520
-    },
-    {
-      "epoch": 0.5798793618216123,
-      "grad_norm": 826.119384765625,
-      "learning_rate": 2.2897618851703407e-05,
-      "loss": 55.8066,
-      "step": 143530
-    },
-    {
-      "epoch": 0.5799197630869799,
-      "grad_norm": 742.7290649414062,
-      "learning_rate": 2.289414057859077e-05,
-      "loss": 31.2436,
-      "step": 143540
-    },
-    {
-      "epoch": 0.5799601643523475,
-      "grad_norm": 682.001220703125,
-      "learning_rate": 2.2890662346532936e-05,
-      "loss": 60.7501,
-      "step": 143550
-    },
-    {
-      "epoch": 0.5800005656177152,
-      "grad_norm": 1402.470703125,
-      "learning_rate": 2.2887184155597723e-05,
-      "loss": 54.5233,
-      "step": 143560
-    },
-    {
-      "epoch": 0.5800409668830828,
-      "grad_norm": 1101.815673828125,
-      "learning_rate": 2.288370600585295e-05,
-      "loss": 46.7544,
-      "step": 143570
-    },
-    {
-      "epoch": 0.5800813681484505,
-      "grad_norm": 836.9398803710938,
-      "learning_rate": 2.2880227897366422e-05,
-      "loss": 45.7828,
-      "step": 143580
-    },
-    {
-      "epoch": 0.580121769413818,
-      "grad_norm": 322.18975830078125,
-      "learning_rate": 2.2876749830205934e-05,
-      "loss": 29.6777,
-      "step": 143590
-    },
-    {
-      "epoch": 0.5801621706791856,
-      "grad_norm": 852.6254272460938,
-      "learning_rate": 2.28732718044393e-05,
-      "loss": 44.7059,
-      "step": 143600
-    },
-    {
-      "epoch": 0.5802025719445533,
-      "grad_norm": 1030.384033203125,
-      "learning_rate": 2.2869793820134332e-05,
-      "loss": 41.8577,
-      "step": 143610
-    },
-    {
-      "epoch": 0.5802429732099209,
-      "grad_norm": 798.349609375,
-      "learning_rate": 2.286631587735883e-05,
-      "loss": 53.5231,
-      "step": 143620
-    },
-    {
-      "epoch": 0.5802833744752885,
-      "grad_norm": 440.0967712402344,
-      "learning_rate": 2.28628379761806e-05,
-      "loss": 43.9023,
-      "step": 143630
-    },
-    {
-      "epoch": 0.5803237757406562,
-      "grad_norm": 1532.7510986328125,
-      "learning_rate": 2.2859360116667432e-05,
-      "loss": 55.8753,
-      "step": 143640
-    },
-    {
-      "epoch": 0.5803641770060238,
-      "grad_norm": 280.5191345214844,
-      "learning_rate": 2.2855882298887156e-05,
-      "loss": 76.918,
-      "step": 143650
-    },
-    {
-      "epoch": 0.5804045782713915,
-      "grad_norm": 721.1527709960938,
-      "learning_rate": 2.2852404522907556e-05,
-      "loss": 51.7469,
-      "step": 143660
-    },
-    {
-      "epoch": 0.5804449795367591,
-      "grad_norm": 0.0,
-      "learning_rate": 2.2848926788796425e-05,
-      "loss": 49.8166,
-      "step": 143670
-    },
-    {
-      "epoch": 0.5804853808021267,
-      "grad_norm": 638.0519409179688,
-      "learning_rate": 2.284544909662158e-05,
-      "loss": 34.2589,
-      "step": 143680
-    },
-    {
-      "epoch": 0.5805257820674944,
-      "grad_norm": 576.2401733398438,
-      "learning_rate": 2.2841971446450815e-05,
-      "loss": 54.998,
-      "step": 143690
-    },
-    {
-      "epoch": 0.580566183332862,
-      "grad_norm": 785.57568359375,
-      "learning_rate": 2.2838493838351933e-05,
-      "loss": 65.9362,
-      "step": 143700
-    },
-    {
-      "epoch": 0.5806065845982297,
-      "grad_norm": 753.3636474609375,
-      "learning_rate": 2.2835016272392722e-05,
-      "loss": 45.9941,
-      "step": 143710
-    },
-    {
-      "epoch": 0.5806469858635972,
-      "grad_norm": 822.4500122070312,
-      "learning_rate": 2.2831538748640974e-05,
-      "loss": 50.9894,
-      "step": 143720
-    },
-    {
-      "epoch": 0.5806873871289648,
-      "grad_norm": 608.1663818359375,
-      "learning_rate": 2.2828061267164508e-05,
-      "loss": 38.5246,
-      "step": 143730
-    },
-    {
-      "epoch": 0.5807277883943325,
-      "grad_norm": 460.6668701171875,
-      "learning_rate": 2.282458382803109e-05,
-      "loss": 36.4291,
-      "step": 143740
-    },
-    {
-      "epoch": 0.5807681896597001,
-      "grad_norm": 427.8645935058594,
-      "learning_rate": 2.2821106431308544e-05,
-      "loss": 34.1155,
-      "step": 143750
-    },
-    {
-      "epoch": 0.5808085909250678,
-      "grad_norm": 926.2526245117188,
-      "learning_rate": 2.281762907706465e-05,
-      "loss": 67.177,
-      "step": 143760
-    },
-    {
-      "epoch": 0.5808489921904354,
-      "grad_norm": 851.1618041992188,
-      "learning_rate": 2.28141517653672e-05,
-      "loss": 58.3684,
-      "step": 143770
-    },
-    {
-      "epoch": 0.580889393455803,
-      "grad_norm": 392.8139953613281,
-      "learning_rate": 2.2810674496283984e-05,
-      "loss": 61.7282,
-      "step": 143780
-    },
-    {
-      "epoch": 0.5809297947211707,
-      "grad_norm": 1116.66357421875,
-      "learning_rate": 2.28071972698828e-05,
-      "loss": 39.3322,
-      "step": 143790
-    },
-    {
-      "epoch": 0.5809701959865383,
-      "grad_norm": 493.0304870605469,
-      "learning_rate": 2.280372008623142e-05,
-      "loss": 44.2897,
-      "step": 143800
-    },
-    {
-      "epoch": 0.581010597251906,
-      "grad_norm": 582.5764770507812,
-      "learning_rate": 2.280024294539766e-05,
-      "loss": 44.4858,
-      "step": 143810
-    },
-    {
-      "epoch": 0.5810509985172736,
-      "grad_norm": 3156.2939453125,
-      "learning_rate": 2.279676584744929e-05,
-      "loss": 63.7611,
-      "step": 143820
-    },
-    {
-      "epoch": 0.5810913997826412,
-      "grad_norm": 694.3815307617188,
-      "learning_rate": 2.279328879245411e-05,
-      "loss": 50.649,
-      "step": 143830
-    },
-    {
-      "epoch": 0.5811318010480089,
-      "grad_norm": 732.0067749023438,
-      "learning_rate": 2.27898117804799e-05,
-      "loss": 33.2757,
-      "step": 143840
-    },
-    {
-      "epoch": 0.5811722023133764,
-      "grad_norm": 2266.329345703125,
-      "learning_rate": 2.2786334811594446e-05,
-      "loss": 49.3162,
-      "step": 143850
-    },
-    {
-      "epoch": 0.581212603578744,
-      "grad_norm": 1178.98193359375,
-      "learning_rate": 2.278285788586554e-05,
-      "loss": 66.4311,
-      "step": 143860
-    },
-    {
-      "epoch": 0.5812530048441117,
-      "grad_norm": 523.5665283203125,
-      "learning_rate": 2.2779381003360958e-05,
-      "loss": 56.1309,
-      "step": 143870
-    },
-    {
-      "epoch": 0.5812934061094793,
-      "grad_norm": 926.5538940429688,
-      "learning_rate": 2.2775904164148477e-05,
-      "loss": 44.5801,
-      "step": 143880
-    },
-    {
-      "epoch": 0.581333807374847,
-      "grad_norm": 311.15753173828125,
-      "learning_rate": 2.27724273682959e-05,
-      "loss": 35.8682,
-      "step": 143890
-    },
-    {
-      "epoch": 0.5813742086402146,
-      "grad_norm": 473.8144226074219,
-      "learning_rate": 2.2768950615870986e-05,
-      "loss": 40.8588,
-      "step": 143900
-    },
-    {
-      "epoch": 0.5814146099055822,
-      "grad_norm": 1385.28125,
-      "learning_rate": 2.276547390694154e-05,
-      "loss": 50.3639,
-      "step": 143910
-    },
-    {
-      "epoch": 0.5814550111709499,
-      "grad_norm": 824.149169921875,
-      "learning_rate": 2.2761997241575333e-05,
-      "loss": 49.5366,
-      "step": 143920
-    },
-    {
-      "epoch": 0.5814954124363175,
-      "grad_norm": 310.1435546875,
-      "learning_rate": 2.2758520619840134e-05,
-      "loss": 45.1334,
-      "step": 143930
-    },
-    {
-      "epoch": 0.5815358137016852,
-      "grad_norm": 384.5504455566406,
-      "learning_rate": 2.2755044041803738e-05,
-      "loss": 64.554,
-      "step": 143940
-    },
-    {
-      "epoch": 0.5815762149670528,
-      "grad_norm": 1101.164306640625,
-      "learning_rate": 2.2751567507533906e-05,
-      "loss": 55.4663,
-      "step": 143950
-    },
-    {
-      "epoch": 0.5816166162324204,
-      "grad_norm": 806.0096435546875,
-      "learning_rate": 2.2748091017098423e-05,
-      "loss": 58.164,
-      "step": 143960
-    },
-    {
-      "epoch": 0.581657017497788,
-      "grad_norm": 1249.18212890625,
-      "learning_rate": 2.274461457056507e-05,
-      "loss": 78.7904,
-      "step": 143970
-    },
-    {
-      "epoch": 0.5816974187631556,
-      "grad_norm": 484.7062072753906,
-      "learning_rate": 2.2741138168001608e-05,
-      "loss": 44.8782,
-      "step": 143980
-    },
-    {
-      "epoch": 0.5817378200285233,
-      "grad_norm": 744.9406127929688,
-      "learning_rate": 2.273766180947583e-05,
-      "loss": 62.8914,
-      "step": 143990
-    },
-    {
-      "epoch": 0.5817782212938909,
-      "grad_norm": 422.4091491699219,
-      "learning_rate": 2.2734185495055503e-05,
-      "loss": 50.1495,
-      "step": 144000
-    },
-    {
-      "epoch": 0.5818186225592585,
-      "grad_norm": 967.376708984375,
-      "learning_rate": 2.273070922480838e-05,
-      "loss": 48.5318,
-      "step": 144010
-    },
-    {
-      "epoch": 0.5818590238246262,
-      "grad_norm": 876.5336303710938,
-      "learning_rate": 2.2727232998802263e-05,
-      "loss": 63.444,
-      "step": 144020
-    },
-    {
-      "epoch": 0.5818994250899938,
-      "grad_norm": 1013.9330444335938,
-      "learning_rate": 2.272375681710491e-05,
-      "loss": 63.0844,
-      "step": 144030
-    },
-    {
-      "epoch": 0.5819398263553615,
-      "grad_norm": 759.7598876953125,
-      "learning_rate": 2.272028067978408e-05,
-      "loss": 56.8256,
-      "step": 144040
-    },
-    {
-      "epoch": 0.5819802276207291,
-      "grad_norm": 357.03948974609375,
-      "learning_rate": 2.271680458690756e-05,
-      "loss": 76.433,
-      "step": 144050
-    },
-    {
-      "epoch": 0.5820206288860967,
-      "grad_norm": 662.3284912109375,
-      "learning_rate": 2.27133285385431e-05,
-      "loss": 38.5279,
-      "step": 144060
-    },
-    {
-      "epoch": 0.5820610301514644,
-      "grad_norm": 1271.0579833984375,
-      "learning_rate": 2.270985253475849e-05,
-      "loss": 65.1111,
-      "step": 144070
-    },
-    {
-      "epoch": 0.582101431416832,
-      "grad_norm": 302.46990966796875,
-      "learning_rate": 2.270637657562148e-05,
-      "loss": 31.1196,
-      "step": 144080
-    },
-    {
-      "epoch": 0.5821418326821997,
-      "grad_norm": 683.177490234375,
-      "learning_rate": 2.2702900661199827e-05,
-      "loss": 48.2238,
-      "step": 144090
-    },
-    {
-      "epoch": 0.5821822339475672,
-      "grad_norm": 970.8692016601562,
-      "learning_rate": 2.2699424791561326e-05,
-      "loss": 65.3873,
-      "step": 144100
-    },
-    {
-      "epoch": 0.5822226352129348,
-      "grad_norm": 547.6918334960938,
-      "learning_rate": 2.2695948966773718e-05,
-      "loss": 62.7877,
-      "step": 144110
-    },
-    {
-      "epoch": 0.5822630364783025,
-      "grad_norm": 952.3994140625,
-      "learning_rate": 2.2692473186904765e-05,
-      "loss": 58.3059,
-      "step": 144120
-    },
-    {
-      "epoch": 0.5823034377436701,
-      "grad_norm": 368.6611633300781,
-      "learning_rate": 2.2688997452022244e-05,
-      "loss": 63.0227,
-      "step": 144130
-    },
-    {
-      "epoch": 0.5823438390090377,
-      "grad_norm": 1159.48876953125,
-      "learning_rate": 2.2685521762193892e-05,
-      "loss": 58.016,
-      "step": 144140
-    },
-    {
-      "epoch": 0.5823842402744054,
-      "grad_norm": 823.868408203125,
-      "learning_rate": 2.2682046117487498e-05,
-      "loss": 40.8645,
-      "step": 144150
-    },
-    {
-      "epoch": 0.582424641539773,
-      "grad_norm": 655.6223754882812,
-      "learning_rate": 2.267857051797081e-05,
-      "loss": 88.7146,
-      "step": 144160
-    },
-    {
-      "epoch": 0.5824650428051407,
-      "grad_norm": 1600.1575927734375,
-      "learning_rate": 2.2675094963711572e-05,
-      "loss": 49.5671,
-      "step": 144170
-    },
-    {
-      "epoch": 0.5825054440705083,
-      "grad_norm": 1083.625,
-      "learning_rate": 2.2671619454777566e-05,
-      "loss": 68.5955,
-      "step": 144180
-    },
-    {
-      "epoch": 0.582545845335876,
-      "grad_norm": 686.4727172851562,
-      "learning_rate": 2.266814399123654e-05,
-      "loss": 51.8318,
-      "step": 144190
-    },
-    {
-      "epoch": 0.5825862466012436,
-      "grad_norm": 745.5236206054688,
-      "learning_rate": 2.266466857315624e-05,
-      "loss": 50.2613,
-      "step": 144200
-    },
-    {
-      "epoch": 0.5826266478666112,
-      "grad_norm": 1290.810302734375,
-      "learning_rate": 2.2661193200604432e-05,
-      "loss": 53.2357,
-      "step": 144210
-    },
-    {
-      "epoch": 0.5826670491319789,
-      "grad_norm": 589.5169067382812,
-      "learning_rate": 2.265771787364886e-05,
-      "loss": 62.4375,
-      "step": 144220
-    },
-    {
-      "epoch": 0.5827074503973464,
-      "grad_norm": 1309.25927734375,
-      "learning_rate": 2.265424259235729e-05,
-      "loss": 48.7557,
-      "step": 144230
-    },
-    {
-      "epoch": 0.582747851662714,
-      "grad_norm": 578.2619018554688,
-      "learning_rate": 2.2650767356797474e-05,
-      "loss": 46.7097,
-      "step": 144240
-    },
-    {
-      "epoch": 0.5827882529280817,
-      "grad_norm": 348.2593994140625,
-      "learning_rate": 2.2647292167037144e-05,
-      "loss": 38.4774,
-      "step": 144250
-    },
-    {
-      "epoch": 0.5828286541934493,
-      "grad_norm": 1188.864990234375,
-      "learning_rate": 2.2643817023144072e-05,
-      "loss": 83.3154,
-      "step": 144260
-    },
-    {
-      "epoch": 0.582869055458817,
-      "grad_norm": 400.77227783203125,
-      "learning_rate": 2.2640341925185997e-05,
-      "loss": 34.9298,
-      "step": 144270
-    },
-    {
-      "epoch": 0.5829094567241846,
-      "grad_norm": 786.1428833007812,
-      "learning_rate": 2.2636866873230677e-05,
-      "loss": 68.8352,
-      "step": 144280
-    },
-    {
-      "epoch": 0.5829498579895522,
-      "grad_norm": 822.0831298828125,
-      "learning_rate": 2.263339186734585e-05,
-      "loss": 52.5729,
-      "step": 144290
-    },
-    {
-      "epoch": 0.5829902592549199,
-      "grad_norm": 477.41424560546875,
-      "learning_rate": 2.2629916907599268e-05,
-      "loss": 48.0709,
-      "step": 144300
-    },
-    {
-      "epoch": 0.5830306605202875,
-      "grad_norm": 962.1996459960938,
-      "learning_rate": 2.262644199405868e-05,
-      "loss": 64.8363,
-      "step": 144310
-    },
-    {
-      "epoch": 0.5830710617856552,
-      "grad_norm": 490.6531066894531,
-      "learning_rate": 2.2622967126791823e-05,
-      "loss": 51.404,
-      "step": 144320
-    },
-    {
-      "epoch": 0.5831114630510228,
-      "grad_norm": 430.4482727050781,
-      "learning_rate": 2.2619492305866437e-05,
-      "loss": 35.495,
-      "step": 144330
-    },
-    {
-      "epoch": 0.5831518643163904,
-      "grad_norm": 1517.2744140625,
-      "learning_rate": 2.2616017531350288e-05,
-      "loss": 64.5694,
-      "step": 144340
-    },
-    {
-      "epoch": 0.5831922655817581,
-      "grad_norm": 534.8115234375,
-      "learning_rate": 2.2612542803311094e-05,
-      "loss": 50.1855,
-      "step": 144350
-    },
-    {
-      "epoch": 0.5832326668471256,
-      "grad_norm": 859.8734741210938,
-      "learning_rate": 2.2609068121816612e-05,
-      "loss": 54.9618,
-      "step": 144360
-    },
-    {
-      "epoch": 0.5832730681124932,
-      "grad_norm": 407.89892578125,
-      "learning_rate": 2.2605593486934583e-05,
-      "loss": 45.6449,
-      "step": 144370
-    },
-    {
-      "epoch": 0.5833134693778609,
-      "grad_norm": 837.1588745117188,
-      "learning_rate": 2.2602118898732736e-05,
-      "loss": 53.8316,
-      "step": 144380
-    },
-    {
-      "epoch": 0.5833538706432285,
-      "grad_norm": 351.80755615234375,
-      "learning_rate": 2.259864435727882e-05,
-      "loss": 61.8373,
-      "step": 144390
-    },
-    {
-      "epoch": 0.5833942719085962,
-      "grad_norm": 1027.636962890625,
-      "learning_rate": 2.2595169862640568e-05,
-      "loss": 73.4547,
-      "step": 144400
-    },
-    {
-      "epoch": 0.5834346731739638,
-      "grad_norm": 592.4967651367188,
-      "learning_rate": 2.2591695414885708e-05,
-      "loss": 54.6856,
-      "step": 144410
-    },
-    {
-      "epoch": 0.5834750744393314,
-      "grad_norm": 711.134521484375,
-      "learning_rate": 2.2588221014081996e-05,
-      "loss": 52.1077,
-      "step": 144420
-    },
-    {
-      "epoch": 0.5835154757046991,
-      "grad_norm": 541.2207641601562,
-      "learning_rate": 2.2584746660297144e-05,
-      "loss": 30.0791,
-      "step": 144430
-    },
-    {
-      "epoch": 0.5835558769700667,
-      "grad_norm": 475.95074462890625,
-      "learning_rate": 2.2581272353598915e-05,
-      "loss": 59.0276,
-      "step": 144440
-    },
-    {
-      "epoch": 0.5835962782354344,
-      "grad_norm": 687.62548828125,
-      "learning_rate": 2.2577798094055023e-05,
-      "loss": 48.3016,
-      "step": 144450
-    },
-    {
-      "epoch": 0.583636679500802,
-      "grad_norm": 595.2200927734375,
-      "learning_rate": 2.25743238817332e-05,
-      "loss": 49.3811,
-      "step": 144460
-    },
-    {
-      "epoch": 0.5836770807661696,
-      "grad_norm": 812.7335205078125,
-      "learning_rate": 2.2570849716701185e-05,
-      "loss": 52.779,
-      "step": 144470
-    },
-    {
-      "epoch": 0.5837174820315373,
-      "grad_norm": 784.1361694335938,
-      "learning_rate": 2.256737559902671e-05,
-      "loss": 56.5076,
-      "step": 144480
-    },
-    {
-      "epoch": 0.5837578832969048,
-      "grad_norm": 862.160888671875,
-      "learning_rate": 2.2563901528777487e-05,
-      "loss": 36.3721,
-      "step": 144490
-    },
-    {
-      "epoch": 0.5837982845622725,
-      "grad_norm": 959.0160522460938,
-      "learning_rate": 2.2560427506021266e-05,
-      "loss": 62.3528,
-      "step": 144500
-    },
-    {
-      "epoch": 0.5838386858276401,
-      "grad_norm": 665.8561401367188,
-      "learning_rate": 2.2556953530825762e-05,
-      "loss": 49.8962,
-      "step": 144510
-    },
-    {
-      "epoch": 0.5838790870930077,
-      "grad_norm": 338.06787109375,
-      "learning_rate": 2.255347960325871e-05,
-      "loss": 69.7289,
-      "step": 144520
-    },
-    {
-      "epoch": 0.5839194883583754,
-      "grad_norm": 543.3985595703125,
-      "learning_rate": 2.2550005723387838e-05,
-      "loss": 37.3539,
-      "step": 144530
-    },
-    {
-      "epoch": 0.583959889623743,
-      "grad_norm": 1258.537353515625,
-      "learning_rate": 2.2546531891280857e-05,
-      "loss": 37.1479,
-      "step": 144540
-    },
-    {
-      "epoch": 0.5840002908891107,
-      "grad_norm": 1533.1956787109375,
-      "learning_rate": 2.254305810700551e-05,
-      "loss": 56.9214,
-      "step": 144550
-    },
-    {
-      "epoch": 0.5840406921544783,
-      "grad_norm": 497.5652770996094,
-      "learning_rate": 2.2539584370629508e-05,
-      "loss": 48.9781,
-      "step": 144560
-    },
-    {
-      "epoch": 0.5840810934198459,
-      "grad_norm": 1660.9959716796875,
-      "learning_rate": 2.2536110682220563e-05,
-      "loss": 51.9823,
-      "step": 144570
-    },
-    {
-      "epoch": 0.5841214946852136,
-      "grad_norm": 882.5232543945312,
-      "learning_rate": 2.2532637041846422e-05,
-      "loss": 47.9464,
-      "step": 144580
-    },
-    {
-      "epoch": 0.5841618959505812,
-      "grad_norm": 1142.7230224609375,
-      "learning_rate": 2.2529163449574778e-05,
-      "loss": 74.0414,
-      "step": 144590
-    },
-    {
-      "epoch": 0.5842022972159489,
-      "grad_norm": 899.859375,
-      "learning_rate": 2.2525689905473376e-05,
-      "loss": 41.7679,
-      "step": 144600
-    },
-    {
-      "epoch": 0.5842426984813164,
-      "grad_norm": 1208.34765625,
-      "learning_rate": 2.2522216409609924e-05,
-      "loss": 46.6088,
-      "step": 144610
-    },
-    {
-      "epoch": 0.584283099746684,
-      "grad_norm": 609.228515625,
-      "learning_rate": 2.2518742962052137e-05,
-      "loss": 49.8582,
-      "step": 144620
-    },
-    {
-      "epoch": 0.5843235010120517,
-      "grad_norm": 791.5541381835938,
-      "learning_rate": 2.2515269562867728e-05,
-      "loss": 46.628,
-      "step": 144630
-    },
-    {
-      "epoch": 0.5843639022774193,
-      "grad_norm": 550.5863037109375,
-      "learning_rate": 2.2511796212124425e-05,
-      "loss": 57.1984,
-      "step": 144640
-    },
-    {
-      "epoch": 0.5844043035427869,
-      "grad_norm": 1214.3863525390625,
-      "learning_rate": 2.250832290988993e-05,
-      "loss": 59.4914,
-      "step": 144650
-    },
-    {
-      "epoch": 0.5844447048081546,
-      "grad_norm": 2203.878173828125,
-      "learning_rate": 2.250484965623197e-05,
-      "loss": 51.4256,
-      "step": 144660
-    },
-    {
-      "epoch": 0.5844851060735222,
-      "grad_norm": 786.3572998046875,
-      "learning_rate": 2.250137645121824e-05,
-      "loss": 42.0186,
-      "step": 144670
-    },
-    {
-      "epoch": 0.5845255073388899,
-      "grad_norm": 846.0687255859375,
-      "learning_rate": 2.2497903294916474e-05,
-      "loss": 49.2691,
-      "step": 144680
-    },
-    {
-      "epoch": 0.5845659086042575,
-      "grad_norm": 1059.890380859375,
-      "learning_rate": 2.2494430187394365e-05,
-      "loss": 58.0176,
-      "step": 144690
-    },
-    {
-      "epoch": 0.5846063098696251,
-      "grad_norm": 653.216552734375,
-      "learning_rate": 2.2490957128719624e-05,
-      "loss": 50.118,
-      "step": 144700
-    },
-    {
-      "epoch": 0.5846467111349928,
-      "grad_norm": 1580.69873046875,
-      "learning_rate": 2.2487484118959978e-05,
-      "loss": 65.8694,
-      "step": 144710
-    },
-    {
-      "epoch": 0.5846871124003604,
-      "grad_norm": 964.9862670898438,
-      "learning_rate": 2.248401115818312e-05,
-      "loss": 44.7298,
-      "step": 144720
-    },
-    {
-      "epoch": 0.5847275136657281,
-      "grad_norm": 1182.8531494140625,
-      "learning_rate": 2.2480538246456752e-05,
-      "loss": 48.8191,
-      "step": 144730
-    },
-    {
-      "epoch": 0.5847679149310956,
-      "grad_norm": 617.6116943359375,
-      "learning_rate": 2.24770653838486e-05,
-      "loss": 52.9255,
-      "step": 144740
-    },
-    {
-      "epoch": 0.5848083161964632,
-      "grad_norm": 639.6339721679688,
-      "learning_rate": 2.247359257042634e-05,
-      "loss": 55.6195,
-      "step": 144750
-    },
-    {
-      "epoch": 0.5848487174618309,
-      "grad_norm": 574.1550903320312,
-      "learning_rate": 2.247011980625771e-05,
-      "loss": 50.8011,
-      "step": 144760
-    },
-    {
-      "epoch": 0.5848891187271985,
-      "grad_norm": 733.3055419921875,
-      "learning_rate": 2.246664709141039e-05,
-      "loss": 62.5022,
-      "step": 144770
-    },
-    {
-      "epoch": 0.5849295199925661,
-      "grad_norm": 1303.0869140625,
-      "learning_rate": 2.2463174425952084e-05,
-      "loss": 50.9016,
-      "step": 144780
-    },
-    {
-      "epoch": 0.5849699212579338,
-      "grad_norm": 424.1533203125,
-      "learning_rate": 2.2459701809950506e-05,
-      "loss": 31.3942,
-      "step": 144790
-    },
-    {
-      "epoch": 0.5850103225233014,
-      "grad_norm": 719.8305053710938,
-      "learning_rate": 2.2456229243473345e-05,
-      "loss": 59.4844,
-      "step": 144800
-    },
-    {
-      "epoch": 0.5850507237886691,
-      "grad_norm": 1231.0697021484375,
-      "learning_rate": 2.2452756726588307e-05,
-      "loss": 61.0486,
-      "step": 144810
-    },
-    {
-      "epoch": 0.5850911250540367,
-      "grad_norm": 0.0,
-      "learning_rate": 2.2449284259363093e-05,
-      "loss": 59.8069,
-      "step": 144820
-    },
-    {
-      "epoch": 0.5851315263194043,
-      "grad_norm": 1179.6441650390625,
-      "learning_rate": 2.2445811841865383e-05,
-      "loss": 48.1206,
-      "step": 144830
-    },
-    {
-      "epoch": 0.585171927584772,
-      "grad_norm": 588.29638671875,
-      "learning_rate": 2.2442339474162898e-05,
-      "loss": 79.9188,
-      "step": 144840
-    },
-    {
-      "epoch": 0.5852123288501396,
-      "grad_norm": 1237.6119384765625,
-      "learning_rate": 2.243886715632332e-05,
-      "loss": 48.1361,
-      "step": 144850
-    },
-    {
-      "epoch": 0.5852527301155073,
-      "grad_norm": 945.63232421875,
-      "learning_rate": 2.2435394888414334e-05,
-      "loss": 44.111,
-      "step": 144860
-    },
-    {
-      "epoch": 0.5852931313808748,
-      "grad_norm": 789.8206176757812,
-      "learning_rate": 2.243192267050366e-05,
-      "loss": 62.4617,
-      "step": 144870
-    },
-    {
-      "epoch": 0.5853335326462424,
-      "grad_norm": 2056.07470703125,
-      "learning_rate": 2.2428450502658967e-05,
-      "loss": 79.2203,
-      "step": 144880
-    },
-    {
-      "epoch": 0.5853739339116101,
-      "grad_norm": 576.0573120117188,
-      "learning_rate": 2.242497838494796e-05,
-      "loss": 45.161,
-      "step": 144890
-    },
-    {
-      "epoch": 0.5854143351769777,
-      "grad_norm": 596.4187622070312,
-      "learning_rate": 2.2421506317438325e-05,
-      "loss": 75.1115,
-      "step": 144900
-    },
-    {
-      "epoch": 0.5854547364423454,
-      "grad_norm": 652.2193603515625,
-      "learning_rate": 2.241803430019774e-05,
-      "loss": 51.4237,
-      "step": 144910
-    },
-    {
-      "epoch": 0.585495137707713,
-      "grad_norm": 946.1011352539062,
-      "learning_rate": 2.241456233329392e-05,
-      "loss": 47.0027,
-      "step": 144920
-    },
-    {
-      "epoch": 0.5855355389730806,
-      "grad_norm": 711.0994873046875,
-      "learning_rate": 2.2411090416794538e-05,
-      "loss": 62.8833,
-      "step": 144930
-    },
-    {
-      "epoch": 0.5855759402384483,
-      "grad_norm": 1949.346923828125,
-      "learning_rate": 2.2407618550767268e-05,
-      "loss": 68.7776,
-      "step": 144940
-    },
-    {
-      "epoch": 0.5856163415038159,
-      "grad_norm": 676.929931640625,
-      "learning_rate": 2.2404146735279822e-05,
-      "loss": 46.8256,
-      "step": 144950
-    },
-    {
-      "epoch": 0.5856567427691836,
-      "grad_norm": 271.9763488769531,
-      "learning_rate": 2.2400674970399863e-05,
-      "loss": 46.4138,
-      "step": 144960
-    },
-    {
-      "epoch": 0.5856971440345512,
-      "grad_norm": 278.78790283203125,
-      "learning_rate": 2.2397203256195087e-05,
-      "loss": 46.952,
-      "step": 144970
-    },
-    {
-      "epoch": 0.5857375452999188,
-      "grad_norm": 603.0307006835938,
-      "learning_rate": 2.239373159273318e-05,
-      "loss": 42.9002,
-      "step": 144980
-    },
-    {
-      "epoch": 0.5857779465652865,
-      "grad_norm": 573.30322265625,
-      "learning_rate": 2.2390259980081805e-05,
-      "loss": 56.5838,
-      "step": 144990
-    },
-    {
-      "epoch": 0.585818347830654,
-      "grad_norm": 972.4603271484375,
-      "learning_rate": 2.238678841830867e-05,
-      "loss": 52.5977,
-      "step": 145000
-    },
-    {
-      "epoch": 0.5858587490960216,
-      "grad_norm": 383.62628173828125,
-      "learning_rate": 2.2383316907481433e-05,
-      "loss": 37.6184,
-      "step": 145010
-    },
-    {
-      "epoch": 0.5858991503613893,
-      "grad_norm": 496.90045166015625,
-      "learning_rate": 2.237984544766777e-05,
-      "loss": 44.9758,
-      "step": 145020
-    },
-    {
-      "epoch": 0.5859395516267569,
-      "grad_norm": 528.111572265625,
-      "learning_rate": 2.2376374038935384e-05,
-      "loss": 51.424,
-      "step": 145030
-    },
-    {
-      "epoch": 0.5859799528921246,
-      "grad_norm": 857.6492919921875,
-      "learning_rate": 2.2372902681351923e-05,
-      "loss": 49.9991,
-      "step": 145040
-    },
-    {
-      "epoch": 0.5860203541574922,
-      "grad_norm": 642.615966796875,
-      "learning_rate": 2.236943137498509e-05,
-      "loss": 42.532,
-      "step": 145050
-    },
-    {
-      "epoch": 0.5860607554228598,
-      "grad_norm": 1281.65625,
-      "learning_rate": 2.2365960119902545e-05,
-      "loss": 71.6778,
-      "step": 145060
-    },
-    {
-      "epoch": 0.5861011566882275,
-      "grad_norm": 406.2398986816406,
-      "learning_rate": 2.236248891617196e-05,
-      "loss": 41.7716,
-      "step": 145070
-    },
-    {
-      "epoch": 0.5861415579535951,
-      "grad_norm": 514.5000610351562,
-      "learning_rate": 2.235901776386101e-05,
-      "loss": 93.0862,
-      "step": 145080
-    },
-    {
-      "epoch": 0.5861819592189628,
-      "grad_norm": 429.3301696777344,
-      "learning_rate": 2.2355546663037373e-05,
-      "loss": 72.3577,
-      "step": 145090
-    },
-    {
-      "epoch": 0.5862223604843304,
-      "grad_norm": 928.6071166992188,
-      "learning_rate": 2.2352075613768707e-05,
-      "loss": 49.8982,
-      "step": 145100
-    },
-    {
-      "epoch": 0.586262761749698,
-      "grad_norm": 854.5071411132812,
-      "learning_rate": 2.2348604616122698e-05,
-      "loss": 55.09,
-      "step": 145110
-    },
-    {
-      "epoch": 0.5863031630150657,
-      "grad_norm": 980.25390625,
-      "learning_rate": 2.2345133670166997e-05,
-      "loss": 63.9885,
-      "step": 145120
-    },
-    {
-      "epoch": 0.5863435642804332,
-      "grad_norm": 1633.0760498046875,
-      "learning_rate": 2.2341662775969295e-05,
-      "loss": 69.4659,
-      "step": 145130
-    },
-    {
-      "epoch": 0.5863839655458009,
-      "grad_norm": 855.7648315429688,
-      "learning_rate": 2.2338191933597242e-05,
-      "loss": 62.3209,
-      "step": 145140
-    },
-    {
-      "epoch": 0.5864243668111685,
-      "grad_norm": 747.5701293945312,
-      "learning_rate": 2.2334721143118504e-05,
-      "loss": 69.4531,
-      "step": 145150
-    },
-    {
-      "epoch": 0.5864647680765361,
-      "grad_norm": 923.6497802734375,
-      "learning_rate": 2.2331250404600755e-05,
-      "loss": 56.0059,
-      "step": 145160
-    },
-    {
-      "epoch": 0.5865051693419038,
-      "grad_norm": 832.5083618164062,
-      "learning_rate": 2.2327779718111652e-05,
-      "loss": 71.0052,
-      "step": 145170
-    },
-    {
-      "epoch": 0.5865455706072714,
-      "grad_norm": 700.8128051757812,
-      "learning_rate": 2.232430908371885e-05,
-      "loss": 56.0096,
-      "step": 145180
-    },
-    {
-      "epoch": 0.5865859718726391,
-      "grad_norm": 651.7989501953125,
-      "learning_rate": 2.232083850149003e-05,
-      "loss": 44.317,
-      "step": 145190
-    },
-    {
-      "epoch": 0.5866263731380067,
-      "grad_norm": 846.4608764648438,
-      "learning_rate": 2.2317367971492835e-05,
-      "loss": 68.4186,
-      "step": 145200
-    },
-    {
-      "epoch": 0.5866667744033743,
-      "grad_norm": 601.2640380859375,
-      "learning_rate": 2.2313897493794938e-05,
-      "loss": 64.6769,
-      "step": 145210
-    },
-    {
-      "epoch": 0.586707175668742,
-      "grad_norm": 393.1311950683594,
-      "learning_rate": 2.2310427068463996e-05,
-      "loss": 35.9002,
-      "step": 145220
-    },
-    {
-      "epoch": 0.5867475769341096,
-      "grad_norm": 591.4609375,
-      "learning_rate": 2.2306956695567655e-05,
-      "loss": 51.3608,
-      "step": 145230
-    },
-    {
-      "epoch": 0.5867879781994773,
-      "grad_norm": 760.8458862304688,
-      "learning_rate": 2.2303486375173585e-05,
-      "loss": 77.252,
-      "step": 145240
-    },
-    {
-      "epoch": 0.5868283794648448,
-      "grad_norm": 1265.822021484375,
-      "learning_rate": 2.230001610734943e-05,
-      "loss": 52.4753,
-      "step": 145250
-    },
-    {
-      "epoch": 0.5868687807302124,
-      "grad_norm": 751.3007202148438,
-      "learning_rate": 2.2296545892162858e-05,
-      "loss": 43.7793,
-      "step": 145260
-    },
-    {
-      "epoch": 0.5869091819955801,
-      "grad_norm": 706.1170043945312,
-      "learning_rate": 2.2293075729681513e-05,
-      "loss": 41.4295,
-      "step": 145270
-    },
-    {
-      "epoch": 0.5869495832609477,
-      "grad_norm": 778.751708984375,
-      "learning_rate": 2.2289605619973045e-05,
-      "loss": 36.7155,
-      "step": 145280
-    },
-    {
-      "epoch": 0.5869899845263153,
-      "grad_norm": 2680.686767578125,
-      "learning_rate": 2.2286135563105117e-05,
-      "loss": 60.0938,
-      "step": 145290
-    },
-    {
-      "epoch": 0.587030385791683,
-      "grad_norm": 557.2347412109375,
-      "learning_rate": 2.2282665559145378e-05,
-      "loss": 51.4661,
-      "step": 145300
-    },
-    {
-      "epoch": 0.5870707870570506,
-      "grad_norm": 905.8478393554688,
-      "learning_rate": 2.227919560816146e-05,
-      "loss": 65.2185,
-      "step": 145310
-    },
-    {
-      "epoch": 0.5871111883224183,
-      "grad_norm": 652.237548828125,
-      "learning_rate": 2.2275725710221035e-05,
-      "loss": 43.8062,
-      "step": 145320
-    },
-    {
-      "epoch": 0.5871515895877859,
-      "grad_norm": 666.0792846679688,
-      "learning_rate": 2.2272255865391733e-05,
-      "loss": 47.7678,
-      "step": 145330
-    },
-    {
-      "epoch": 0.5871919908531535,
-      "grad_norm": 1246.526611328125,
-      "learning_rate": 2.2268786073741216e-05,
-      "loss": 68.3254,
-      "step": 145340
-    },
-    {
-      "epoch": 0.5872323921185212,
-      "grad_norm": 414.19940185546875,
-      "learning_rate": 2.226531633533712e-05,
-      "loss": 52.0656,
-      "step": 145350
-    },
-    {
-      "epoch": 0.5872727933838888,
-      "grad_norm": 870.029541015625,
-      "learning_rate": 2.2261846650247075e-05,
-      "loss": 54.3534,
-      "step": 145360
-    },
-    {
-      "epoch": 0.5873131946492565,
-      "grad_norm": 1294.477294921875,
-      "learning_rate": 2.2258377018538753e-05,
-      "loss": 47.3047,
-      "step": 145370
-    },
-    {
-      "epoch": 0.587353595914624,
-      "grad_norm": 661.3151245117188,
-      "learning_rate": 2.2254907440279786e-05,
-      "loss": 48.7553,
-      "step": 145380
-    },
-    {
-      "epoch": 0.5873939971799916,
-      "grad_norm": 624.5767822265625,
-      "learning_rate": 2.2251437915537797e-05,
-      "loss": 41.0314,
-      "step": 145390
-    },
-    {
-      "epoch": 0.5874343984453593,
-      "grad_norm": 1039.283935546875,
-      "learning_rate": 2.224796844438045e-05,
-      "loss": 59.0014,
-      "step": 145400
-    },
-    {
-      "epoch": 0.5874747997107269,
-      "grad_norm": 1308.9102783203125,
-      "learning_rate": 2.2244499026875374e-05,
-      "loss": 70.056,
-      "step": 145410
-    },
-    {
-      "epoch": 0.5875152009760946,
-      "grad_norm": 752.0531005859375,
-      "learning_rate": 2.224102966309021e-05,
-      "loss": 37.8407,
-      "step": 145420
-    },
-    {
-      "epoch": 0.5875556022414622,
-      "grad_norm": 584.2730712890625,
-      "learning_rate": 2.2237560353092592e-05,
-      "loss": 53.7454,
-      "step": 145430
-    },
-    {
-      "epoch": 0.5875960035068298,
-      "grad_norm": 825.1495971679688,
-      "learning_rate": 2.223409109695015e-05,
-      "loss": 53.0734,
-      "step": 145440
-    },
-    {
-      "epoch": 0.5876364047721975,
-      "grad_norm": 853.8151245117188,
-      "learning_rate": 2.2230621894730536e-05,
-      "loss": 39.1982,
-      "step": 145450
-    },
-    {
-      "epoch": 0.5876768060375651,
-      "grad_norm": 0.0,
-      "learning_rate": 2.2227152746501373e-05,
-      "loss": 28.6321,
-      "step": 145460
-    },
-    {
-      "epoch": 0.5877172073029328,
-      "grad_norm": 499.9764404296875,
-      "learning_rate": 2.2223683652330283e-05,
-      "loss": 39.2175,
-      "step": 145470
-    },
-    {
-      "epoch": 0.5877576085683004,
-      "grad_norm": 1175.066650390625,
-      "learning_rate": 2.2220214612284924e-05,
-      "loss": 57.1001,
-      "step": 145480
-    },
-    {
-      "epoch": 0.587798009833668,
-      "grad_norm": 566.9341430664062,
-      "learning_rate": 2.2216745626432906e-05,
-      "loss": 70.2406,
-      "step": 145490
-    },
-    {
-      "epoch": 0.5878384110990357,
-      "grad_norm": 664.6567993164062,
-      "learning_rate": 2.2213276694841866e-05,
-      "loss": 46.9813,
-      "step": 145500
-    },
-    {
-      "epoch": 0.5878788123644032,
-      "grad_norm": 540.4669799804688,
-      "learning_rate": 2.2209807817579438e-05,
-      "loss": 72.8235,
-      "step": 145510
-    },
-    {
-      "epoch": 0.5879192136297708,
-      "grad_norm": 761.8043823242188,
-      "learning_rate": 2.2206338994713228e-05,
-      "loss": 45.2406,
-      "step": 145520
-    },
-    {
-      "epoch": 0.5879596148951385,
-      "grad_norm": 331.9342346191406,
-      "learning_rate": 2.2202870226310888e-05,
-      "loss": 35.2402,
-      "step": 145530
-    },
-    {
-      "epoch": 0.5880000161605061,
-      "grad_norm": 220.01025390625,
-      "learning_rate": 2.2199401512440034e-05,
-      "loss": 38.3728,
-      "step": 145540
-    },
-    {
-      "epoch": 0.5880404174258738,
-      "grad_norm": 498.82073974609375,
-      "learning_rate": 2.2195932853168278e-05,
-      "loss": 29.1557,
-      "step": 145550
-    },
-    {
-      "epoch": 0.5880808186912414,
-      "grad_norm": 2570.12255859375,
-      "learning_rate": 2.2192464248563265e-05,
-      "loss": 71.8326,
-      "step": 145560
-    },
-    {
-      "epoch": 0.588121219956609,
-      "grad_norm": 857.0679931640625,
-      "learning_rate": 2.2188995698692606e-05,
-      "loss": 53.2321,
-      "step": 145570
-    },
-    {
-      "epoch": 0.5881616212219767,
-      "grad_norm": 427.507568359375,
-      "learning_rate": 2.2185527203623922e-05,
-      "loss": 40.6259,
-      "step": 145580
-    },
-    {
-      "epoch": 0.5882020224873443,
-      "grad_norm": 223.01893615722656,
-      "learning_rate": 2.2182058763424833e-05,
-      "loss": 37.5061,
-      "step": 145590
-    },
-    {
-      "epoch": 0.588242423752712,
-      "grad_norm": 557.50390625,
-      "learning_rate": 2.217859037816296e-05,
-      "loss": 82.0438,
-      "step": 145600
-    },
-    {
-      "epoch": 0.5882828250180796,
-      "grad_norm": 779.0130615234375,
-      "learning_rate": 2.217512204790592e-05,
-      "loss": 44.8452,
-      "step": 145610
-    },
-    {
-      "epoch": 0.5883232262834472,
-      "grad_norm": 323.9913024902344,
-      "learning_rate": 2.2171653772721335e-05,
-      "loss": 51.8958,
-      "step": 145620
-    },
-    {
-      "epoch": 0.5883636275488149,
-      "grad_norm": 635.4793090820312,
-      "learning_rate": 2.2168185552676805e-05,
-      "loss": 64.8756,
-      "step": 145630
-    },
-    {
-      "epoch": 0.5884040288141824,
-      "grad_norm": 477.123779296875,
-      "learning_rate": 2.2164717387839966e-05,
-      "loss": 37.9061,
-      "step": 145640
-    },
-    {
-      "epoch": 0.58844443007955,
-      "grad_norm": 990.1459350585938,
-      "learning_rate": 2.2161249278278405e-05,
-      "loss": 57.033,
-      "step": 145650
-    },
-    {
-      "epoch": 0.5884848313449177,
-      "grad_norm": 519.134765625,
-      "learning_rate": 2.2157781224059772e-05,
-      "loss": 63.8006,
-      "step": 145660
-    },
-    {
-      "epoch": 0.5885252326102853,
-      "grad_norm": 540.0624389648438,
-      "learning_rate": 2.215431322525165e-05,
-      "loss": 66.9346,
-      "step": 145670
-    },
-    {
-      "epoch": 0.588565633875653,
-      "grad_norm": 429.2971496582031,
-      "learning_rate": 2.215084528192165e-05,
-      "loss": 41.9368,
-      "step": 145680
-    },
-    {
-      "epoch": 0.5886060351410206,
-      "grad_norm": 287.19329833984375,
-      "learning_rate": 2.2147377394137398e-05,
-      "loss": 53.8232,
-      "step": 145690
-    },
-    {
-      "epoch": 0.5886464364063883,
-      "grad_norm": 2038.7117919921875,
-      "learning_rate": 2.214390956196649e-05,
-      "loss": 53.2171,
-      "step": 145700
-    },
-    {
-      "epoch": 0.5886868376717559,
-      "grad_norm": 342.96429443359375,
-      "learning_rate": 2.2140441785476526e-05,
-      "loss": 70.9862,
-      "step": 145710
-    },
-    {
-      "epoch": 0.5887272389371235,
-      "grad_norm": 1174.8426513671875,
-      "learning_rate": 2.213697406473513e-05,
-      "loss": 59.1554,
-      "step": 145720
-    },
-    {
-      "epoch": 0.5887676402024912,
-      "grad_norm": 446.2856750488281,
-      "learning_rate": 2.213350639980989e-05,
-      "loss": 84.1959,
-      "step": 145730
-    },
-    {
-      "epoch": 0.5888080414678588,
-      "grad_norm": 910.0787353515625,
-      "learning_rate": 2.213003879076843e-05,
-      "loss": 59.0944,
-      "step": 145740
-    },
-    {
-      "epoch": 0.5888484427332265,
-      "grad_norm": 379.2336120605469,
-      "learning_rate": 2.212657123767834e-05,
-      "loss": 47.8588,
-      "step": 145750
-    },
-    {
-      "epoch": 0.5888888439985941,
-      "grad_norm": 545.8311767578125,
-      "learning_rate": 2.2123103740607215e-05,
-      "loss": 41.2895,
-      "step": 145760
-    },
-    {
-      "epoch": 0.5889292452639616,
-      "grad_norm": 451.2540283203125,
-      "learning_rate": 2.2119636299622675e-05,
-      "loss": 49.8382,
-      "step": 145770
-    },
-    {
-      "epoch": 0.5889696465293293,
-      "grad_norm": 2494.452880859375,
-      "learning_rate": 2.2116168914792292e-05,
-      "loss": 51.3358,
-      "step": 145780
-    },
-    {
-      "epoch": 0.5890100477946969,
-      "grad_norm": 1018.56103515625,
-      "learning_rate": 2.211270158618369e-05,
-      "loss": 80.1487,
-      "step": 145790
-    },
-    {
-      "epoch": 0.5890504490600645,
-      "grad_norm": 700.343994140625,
-      "learning_rate": 2.2109234313864465e-05,
-      "loss": 52.8399,
-      "step": 145800
-    },
-    {
-      "epoch": 0.5890908503254322,
-      "grad_norm": 401.8301696777344,
-      "learning_rate": 2.210576709790219e-05,
-      "loss": 38.5016,
-      "step": 145810
-    },
-    {
-      "epoch": 0.5891312515907998,
-      "grad_norm": 379.638427734375,
-      "learning_rate": 2.210229993836449e-05,
-      "loss": 67.7898,
-      "step": 145820
-    },
-    {
-      "epoch": 0.5891716528561675,
-      "grad_norm": 1105.470458984375,
-      "learning_rate": 2.209883283531894e-05,
-      "loss": 62.1052,
-      "step": 145830
-    },
-    {
-      "epoch": 0.5892120541215351,
-      "grad_norm": 409.69854736328125,
-      "learning_rate": 2.209536578883313e-05,
-      "loss": 52.7116,
-      "step": 145840
-    },
-    {
-      "epoch": 0.5892524553869027,
-      "grad_norm": 442.1845397949219,
-      "learning_rate": 2.209189879897467e-05,
-      "loss": 45.8233,
-      "step": 145850
-    },
-    {
-      "epoch": 0.5892928566522704,
-      "grad_norm": 284.5019836425781,
-      "learning_rate": 2.2088431865811127e-05,
-      "loss": 58.4883,
-      "step": 145860
-    },
-    {
-      "epoch": 0.589333257917638,
-      "grad_norm": 762.030029296875,
-      "learning_rate": 2.2084964989410113e-05,
-      "loss": 61.4194,
-      "step": 145870
-    },
-    {
-      "epoch": 0.5893736591830057,
-      "grad_norm": 770.3872680664062,
-      "learning_rate": 2.208149816983921e-05,
-      "loss": 69.5855,
-      "step": 145880
-    },
-    {
-      "epoch": 0.5894140604483732,
-      "grad_norm": 322.16644287109375,
-      "learning_rate": 2.2078031407165993e-05,
-      "loss": 64.2785,
-      "step": 145890
-    },
-    {
-      "epoch": 0.5894544617137408,
-      "grad_norm": 483.3517150878906,
-      "learning_rate": 2.2074564701458065e-05,
-      "loss": 39.8775,
-      "step": 145900
-    },
-    {
-      "epoch": 0.5894948629791085,
-      "grad_norm": 506.798828125,
-      "learning_rate": 2.2071098052783008e-05,
-      "loss": 52.2202,
-      "step": 145910
-    },
-    {
-      "epoch": 0.5895352642444761,
-      "grad_norm": 342.7799072265625,
-      "learning_rate": 2.2067631461208393e-05,
-      "loss": 38.5798,
-      "step": 145920
-    },
-    {
-      "epoch": 0.5895756655098437,
-      "grad_norm": 892.741943359375,
-      "learning_rate": 2.2064164926801817e-05,
-      "loss": 56.9719,
-      "step": 145930
-    },
-    {
-      "epoch": 0.5896160667752114,
-      "grad_norm": 1077.632080078125,
-      "learning_rate": 2.2060698449630853e-05,
-      "loss": 61.5984,
-      "step": 145940
-    },
-    {
-      "epoch": 0.589656468040579,
-      "grad_norm": 469.6213684082031,
-      "learning_rate": 2.205723202976309e-05,
-      "loss": 33.4894,
-      "step": 145950
-    },
-    {
-      "epoch": 0.5896968693059467,
-      "grad_norm": 283.9982604980469,
-      "learning_rate": 2.205376566726611e-05,
-      "loss": 30.2017,
-      "step": 145960
-    },
-    {
-      "epoch": 0.5897372705713143,
-      "grad_norm": 1458.912841796875,
-      "learning_rate": 2.2050299362207472e-05,
-      "loss": 77.3268,
-      "step": 145970
-    },
-    {
-      "epoch": 0.589777671836682,
-      "grad_norm": 503.4773254394531,
-      "learning_rate": 2.2046833114654773e-05,
-      "loss": 56.1382,
-      "step": 145980
-    },
-    {
-      "epoch": 0.5898180731020496,
-      "grad_norm": 2263.208740234375,
-      "learning_rate": 2.204336692467559e-05,
-      "loss": 53.7279,
-      "step": 145990
-    },
-    {
-      "epoch": 0.5898584743674172,
-      "grad_norm": 752.0081176757812,
-      "learning_rate": 2.2039900792337474e-05,
-      "loss": 52.1207,
-      "step": 146000
-    },
-    {
-      "epoch": 0.5898988756327849,
-      "grad_norm": 748.5038452148438,
-      "learning_rate": 2.203643471770803e-05,
-      "loss": 49.4564,
-      "step": 146010
-    },
-    {
-      "epoch": 0.5899392768981524,
-      "grad_norm": 529.9098510742188,
-      "learning_rate": 2.203296870085481e-05,
-      "loss": 60.0309,
-      "step": 146020
-    },
-    {
-      "epoch": 0.58997967816352,
-      "grad_norm": 878.1099853515625,
-      "learning_rate": 2.20295027418454e-05,
-      "loss": 63.0526,
-      "step": 146030
-    },
-    {
-      "epoch": 0.5900200794288877,
-      "grad_norm": 1283.2969970703125,
-      "learning_rate": 2.202603684074736e-05,
-      "loss": 59.6128,
-      "step": 146040
-    },
-    {
-      "epoch": 0.5900604806942553,
-      "grad_norm": 2465.196533203125,
-      "learning_rate": 2.2022570997628256e-05,
-      "loss": 76.0184,
-      "step": 146050
-    },
-    {
-      "epoch": 0.590100881959623,
-      "grad_norm": 223.83999633789062,
-      "learning_rate": 2.2019105212555675e-05,
-      "loss": 44.1915,
-      "step": 146060
-    },
-    {
-      "epoch": 0.5901412832249906,
-      "grad_norm": 1009.5184326171875,
-      "learning_rate": 2.2015639485597168e-05,
-      "loss": 47.8159,
-      "step": 146070
-    },
-    {
-      "epoch": 0.5901816844903582,
-      "grad_norm": 774.2605590820312,
-      "learning_rate": 2.2012173816820297e-05,
-      "loss": 42.978,
-      "step": 146080
-    },
-    {
-      "epoch": 0.5902220857557259,
-      "grad_norm": 537.9838256835938,
-      "learning_rate": 2.2008708206292645e-05,
-      "loss": 58.8493,
-      "step": 146090
-    },
-    {
-      "epoch": 0.5902624870210935,
-      "grad_norm": 209.31063842773438,
-      "learning_rate": 2.200524265408176e-05,
-      "loss": 48.89,
-      "step": 146100
-    },
-    {
-      "epoch": 0.5903028882864612,
-      "grad_norm": 923.2644653320312,
-      "learning_rate": 2.2001777160255222e-05,
-      "loss": 57.573,
-      "step": 146110
-    },
-    {
-      "epoch": 0.5903432895518288,
-      "grad_norm": 531.9500122070312,
-      "learning_rate": 2.199831172488058e-05,
-      "loss": 63.6238,
-      "step": 146120
-    },
-    {
-      "epoch": 0.5903836908171964,
-      "grad_norm": 1522.965576171875,
-      "learning_rate": 2.1994846348025385e-05,
-      "loss": 35.7507,
-      "step": 146130
-    },
-    {
-      "epoch": 0.5904240920825641,
-      "grad_norm": 1366.9095458984375,
-      "learning_rate": 2.1991381029757215e-05,
-      "loss": 90.262,
-      "step": 146140
-    },
-    {
-      "epoch": 0.5904644933479316,
-      "grad_norm": 804.5481567382812,
-      "learning_rate": 2.1987915770143624e-05,
-      "loss": 53.6699,
-      "step": 146150
-    },
-    {
-      "epoch": 0.5905048946132992,
-      "grad_norm": 509.3193054199219,
-      "learning_rate": 2.1984450569252154e-05,
-      "loss": 51.2727,
-      "step": 146160
-    },
-    {
-      "epoch": 0.5905452958786669,
-      "grad_norm": 789.99755859375,
-      "learning_rate": 2.1980985427150385e-05,
-      "loss": 44.7793,
-      "step": 146170
-    },
-    {
-      "epoch": 0.5905856971440345,
-      "grad_norm": 731.342041015625,
-      "learning_rate": 2.197752034390585e-05,
-      "loss": 62.3214,
-      "step": 146180
-    },
-    {
-      "epoch": 0.5906260984094022,
-      "grad_norm": 371.409423828125,
-      "learning_rate": 2.1974055319586124e-05,
-      "loss": 78.2004,
-      "step": 146190
-    },
-    {
-      "epoch": 0.5906664996747698,
-      "grad_norm": 3826.60009765625,
-      "learning_rate": 2.1970590354258745e-05,
-      "loss": 73.2504,
-      "step": 146200
-    },
-    {
-      "epoch": 0.5907069009401374,
-      "grad_norm": 564.4078979492188,
-      "learning_rate": 2.1967125447991254e-05,
-      "loss": 42.8259,
-      "step": 146210
-    },
-    {
-      "epoch": 0.5907473022055051,
-      "grad_norm": 309.908203125,
-      "learning_rate": 2.1963660600851225e-05,
-      "loss": 58.1054,
-      "step": 146220
-    },
-    {
-      "epoch": 0.5907877034708727,
-      "grad_norm": 501.643310546875,
-      "learning_rate": 2.1960195812906196e-05,
-      "loss": 41.4755,
-      "step": 146230
-    },
-    {
-      "epoch": 0.5908281047362404,
-      "grad_norm": 546.9186401367188,
-      "learning_rate": 2.1956731084223702e-05,
-      "loss": 48.2207,
-      "step": 146240
-    },
-    {
-      "epoch": 0.590868506001608,
-      "grad_norm": 934.6657104492188,
-      "learning_rate": 2.195326641487132e-05,
-      "loss": 44.1907,
-      "step": 146250
-    },
-    {
-      "epoch": 0.5909089072669756,
-      "grad_norm": 864.5132446289062,
-      "learning_rate": 2.1949801804916566e-05,
-      "loss": 76.9655,
-      "step": 146260
-    },
-    {
-      "epoch": 0.5909493085323433,
-      "grad_norm": 763.5232543945312,
-      "learning_rate": 2.1946337254426998e-05,
-      "loss": 34.84,
-      "step": 146270
-    },
-    {
-      "epoch": 0.5909897097977108,
-      "grad_norm": 1223.5108642578125,
-      "learning_rate": 2.194287276347016e-05,
-      "loss": 55.184,
-      "step": 146280
-    },
-    {
-      "epoch": 0.5910301110630785,
-      "grad_norm": 392.8327331542969,
-      "learning_rate": 2.193940833211359e-05,
-      "loss": 36.6568,
-      "step": 146290
-    },
-    {
-      "epoch": 0.5910705123284461,
-      "grad_norm": 873.4593505859375,
-      "learning_rate": 2.1935943960424833e-05,
-      "loss": 68.934,
-      "step": 146300
-    },
-    {
-      "epoch": 0.5911109135938137,
-      "grad_norm": 858.9896240234375,
-      "learning_rate": 2.1932479648471416e-05,
-      "loss": 52.654,
-      "step": 146310
-    },
-    {
-      "epoch": 0.5911513148591814,
-      "grad_norm": 2630.4091796875,
-      "learning_rate": 2.19290153963209e-05,
-      "loss": 80.9151,
-      "step": 146320
-    },
-    {
-      "epoch": 0.591191716124549,
-      "grad_norm": 438.961669921875,
-      "learning_rate": 2.1925551204040806e-05,
-      "loss": 76.9184,
-      "step": 146330
-    },
-    {
-      "epoch": 0.5912321173899167,
-      "grad_norm": 916.41357421875,
-      "learning_rate": 2.1922087071698665e-05,
-      "loss": 42.8016,
-      "step": 146340
-    },
-    {
-      "epoch": 0.5912725186552843,
-      "grad_norm": 1062.0850830078125,
-      "learning_rate": 2.1918622999362035e-05,
-      "loss": 52.5087,
-      "step": 146350
-    },
-    {
-      "epoch": 0.5913129199206519,
-      "grad_norm": 918.3807983398438,
-      "learning_rate": 2.1915158987098432e-05,
-      "loss": 52.7011,
-      "step": 146360
-    },
-    {
-      "epoch": 0.5913533211860196,
-      "grad_norm": 549.4228515625,
-      "learning_rate": 2.191169503497539e-05,
-      "loss": 57.5452,
-      "step": 146370
-    },
-    {
-      "epoch": 0.5913937224513872,
-      "grad_norm": 684.5689697265625,
-      "learning_rate": 2.190823114306045e-05,
-      "loss": 58.4194,
-      "step": 146380
-    },
-    {
-      "epoch": 0.5914341237167549,
-      "grad_norm": 633.138427734375,
-      "learning_rate": 2.190476731142112e-05,
-      "loss": 70.747,
-      "step": 146390
-    },
-    {
-      "epoch": 0.5914745249821224,
-      "grad_norm": 353.6347351074219,
-      "learning_rate": 2.1901303540124956e-05,
-      "loss": 47.8663,
-      "step": 146400
-    },
-    {
-      "epoch": 0.59151492624749,
-      "grad_norm": 1246.96826171875,
-      "learning_rate": 2.189783982923948e-05,
-      "loss": 49.267,
-      "step": 146410
-    },
-    {
-      "epoch": 0.5915553275128577,
-      "grad_norm": 963.0771484375,
-      "learning_rate": 2.18943761788322e-05,
-      "loss": 74.9258,
-      "step": 146420
-    },
-    {
-      "epoch": 0.5915957287782253,
-      "grad_norm": 543.4385986328125,
-      "learning_rate": 2.1890912588970662e-05,
-      "loss": 48.036,
-      "step": 146430
-    },
-    {
-      "epoch": 0.591636130043593,
-      "grad_norm": 1466.5035400390625,
-      "learning_rate": 2.188744905972239e-05,
-      "loss": 56.9263,
-      "step": 146440
-    },
-    {
-      "epoch": 0.5916765313089606,
-      "grad_norm": 893.92333984375,
-      "learning_rate": 2.1883985591154893e-05,
-      "loss": 65.8888,
-      "step": 146450
-    },
-    {
-      "epoch": 0.5917169325743282,
-      "grad_norm": 614.3671264648438,
-      "learning_rate": 2.188052218333571e-05,
-      "loss": 35.0252,
-      "step": 146460
-    },
-    {
-      "epoch": 0.5917573338396959,
-      "grad_norm": 581.8492431640625,
-      "learning_rate": 2.1877058836332335e-05,
-      "loss": 46.298,
-      "step": 146470
-    },
-    {
-      "epoch": 0.5917977351050635,
-      "grad_norm": 1633.851318359375,
-      "learning_rate": 2.187359555021232e-05,
-      "loss": 63.4169,
-      "step": 146480
-    },
-    {
-      "epoch": 0.5918381363704311,
-      "grad_norm": 484.6391296386719,
-      "learning_rate": 2.187013232504317e-05,
-      "loss": 114.912,
-      "step": 146490
-    },
-    {
-      "epoch": 0.5918785376357988,
-      "grad_norm": 1368.1236572265625,
-      "learning_rate": 2.186666916089239e-05,
-      "loss": 54.1732,
-      "step": 146500
-    },
-    {
-      "epoch": 0.5919189389011664,
-      "grad_norm": 131.18450927734375,
-      "learning_rate": 2.1863206057827523e-05,
-      "loss": 41.559,
-      "step": 146510
-    },
-    {
-      "epoch": 0.5919593401665341,
-      "grad_norm": 682.2334594726562,
-      "learning_rate": 2.1859743015916065e-05,
-      "loss": 79.2423,
-      "step": 146520
-    },
-    {
-      "epoch": 0.5919997414319016,
-      "grad_norm": 236.24301147460938,
-      "learning_rate": 2.1856280035225527e-05,
-      "loss": 37.946,
-      "step": 146530
-    },
-    {
-      "epoch": 0.5920401426972692,
-      "grad_norm": 541.71728515625,
-      "learning_rate": 2.1852817115823436e-05,
-      "loss": 45.4416,
-      "step": 146540
-    },
-    {
-      "epoch": 0.5920805439626369,
-      "grad_norm": 1449.23046875,
-      "learning_rate": 2.184935425777728e-05,
-      "loss": 76.4084,
-      "step": 146550
-    },
-    {
-      "epoch": 0.5921209452280045,
-      "grad_norm": 752.7134399414062,
-      "learning_rate": 2.1845891461154602e-05,
-      "loss": 50.192,
-      "step": 146560
-    },
-    {
-      "epoch": 0.5921613464933722,
-      "grad_norm": 1388.42626953125,
-      "learning_rate": 2.1842428726022892e-05,
-      "loss": 48.8798,
-      "step": 146570
-    },
-    {
-      "epoch": 0.5922017477587398,
-      "grad_norm": 1135.522216796875,
-      "learning_rate": 2.183896605244965e-05,
-      "loss": 66.7696,
-      "step": 146580
-    },
-    {
-      "epoch": 0.5922421490241074,
-      "grad_norm": 1091.169921875,
-      "learning_rate": 2.18355034405024e-05,
-      "loss": 52.7894,
-      "step": 146590
-    },
-    {
-      "epoch": 0.5922825502894751,
-      "grad_norm": 315.2186279296875,
-      "learning_rate": 2.183204089024864e-05,
-      "loss": 31.6009,
-      "step": 146600
-    },
-    {
-      "epoch": 0.5923229515548427,
-      "grad_norm": 1093.932861328125,
-      "learning_rate": 2.182857840175587e-05,
-      "loss": 56.5924,
-      "step": 146610
-    },
-    {
-      "epoch": 0.5923633528202104,
-      "grad_norm": 504.3761291503906,
-      "learning_rate": 2.1825115975091596e-05,
-      "loss": 51.7465,
-      "step": 146620
-    },
-    {
-      "epoch": 0.592403754085578,
-      "grad_norm": 1324.4403076171875,
-      "learning_rate": 2.1821653610323318e-05,
-      "loss": 62.9345,
-      "step": 146630
-    },
-    {
-      "epoch": 0.5924441553509456,
-      "grad_norm": 867.4586791992188,
-      "learning_rate": 2.181819130751855e-05,
-      "loss": 51.7099,
-      "step": 146640
-    },
-    {
-      "epoch": 0.5924845566163133,
-      "grad_norm": 476.7613525390625,
-      "learning_rate": 2.1814729066744776e-05,
-      "loss": 50.1964,
-      "step": 146650
-    },
-    {
-      "epoch": 0.5925249578816808,
-      "grad_norm": 654.5791015625,
-      "learning_rate": 2.181126688806949e-05,
-      "loss": 56.1941,
-      "step": 146660
-    },
-    {
-      "epoch": 0.5925653591470484,
-      "grad_norm": 834.60498046875,
-      "learning_rate": 2.180780477156021e-05,
-      "loss": 52.9529,
-      "step": 146670
-    },
-    {
-      "epoch": 0.5926057604124161,
-      "grad_norm": 4547.328125,
-      "learning_rate": 2.1804342717284415e-05,
-      "loss": 59.4421,
-      "step": 146680
-    },
-    {
-      "epoch": 0.5926461616777837,
-      "grad_norm": 492.46307373046875,
-      "learning_rate": 2.18008807253096e-05,
-      "loss": 50.6839,
-      "step": 146690
-    },
-    {
-      "epoch": 0.5926865629431514,
-      "grad_norm": 3489.217041015625,
-      "learning_rate": 2.1797418795703267e-05,
-      "loss": 61.9006,
-      "step": 146700
-    },
-    {
-      "epoch": 0.592726964208519,
-      "grad_norm": 392.3635559082031,
-      "learning_rate": 2.17939569285329e-05,
-      "loss": 35.6025,
-      "step": 146710
-    },
-    {
-      "epoch": 0.5927673654738866,
-      "grad_norm": 1170.8079833984375,
-      "learning_rate": 2.1790495123866e-05,
-      "loss": 68.8476,
-      "step": 146720
-    },
-    {
-      "epoch": 0.5928077667392543,
-      "grad_norm": 654.0852661132812,
-      "learning_rate": 2.1787033381770045e-05,
-      "loss": 49.868,
-      "step": 146730
-    },
-    {
-      "epoch": 0.5928481680046219,
-      "grad_norm": 387.08111572265625,
-      "learning_rate": 2.1783571702312523e-05,
-      "loss": 39.509,
-      "step": 146740
-    },
-    {
-      "epoch": 0.5928885692699896,
-      "grad_norm": 876.9544067382812,
-      "learning_rate": 2.1780110085560935e-05,
-      "loss": 54.9167,
-      "step": 146750
-    },
-    {
-      "epoch": 0.5929289705353572,
-      "grad_norm": 867.4058227539062,
-      "learning_rate": 2.177664853158276e-05,
-      "loss": 58.304,
-      "step": 146760
-    },
-    {
-      "epoch": 0.5929693718007248,
-      "grad_norm": 269.5369873046875,
-      "learning_rate": 2.1773187040445465e-05,
-      "loss": 44.2142,
-      "step": 146770
-    },
-    {
-      "epoch": 0.5930097730660925,
-      "grad_norm": 1281.8651123046875,
-      "learning_rate": 2.1769725612216567e-05,
-      "loss": 52.3775,
-      "step": 146780
-    },
-    {
-      "epoch": 0.59305017433146,
-      "grad_norm": 701.8643188476562,
-      "learning_rate": 2.1766264246963525e-05,
-      "loss": 54.0386,
-      "step": 146790
-    },
-    {
-      "epoch": 0.5930905755968277,
-      "grad_norm": 791.13720703125,
-      "learning_rate": 2.176280294475383e-05,
-      "loss": 49.9338,
-      "step": 146800
-    },
-    {
-      "epoch": 0.5931309768621953,
-      "grad_norm": 181.1816864013672,
-      "learning_rate": 2.1759341705654958e-05,
-      "loss": 45.6333,
-      "step": 146810
-    },
-    {
-      "epoch": 0.5931713781275629,
-      "grad_norm": 940.6576538085938,
-      "learning_rate": 2.175588052973438e-05,
-      "loss": 68.6767,
-      "step": 146820
-    },
-    {
-      "epoch": 0.5932117793929306,
-      "grad_norm": 583.0791625976562,
-      "learning_rate": 2.175241941705959e-05,
-      "loss": 59.7825,
-      "step": 146830
-    },
-    {
-      "epoch": 0.5932521806582982,
-      "grad_norm": 924.9185791015625,
-      "learning_rate": 2.1748958367698046e-05,
-      "loss": 42.2929,
-      "step": 146840
-    },
-    {
-      "epoch": 0.5932925819236659,
-      "grad_norm": 464.9294128417969,
-      "learning_rate": 2.174549738171724e-05,
-      "loss": 40.7214,
-      "step": 146850
-    },
-    {
-      "epoch": 0.5933329831890335,
-      "grad_norm": 495.89080810546875,
-      "learning_rate": 2.174203645918464e-05,
-      "loss": 90.9891,
-      "step": 146860
-    },
-    {
-      "epoch": 0.5933733844544011,
-      "grad_norm": 715.9097290039062,
-      "learning_rate": 2.1738575600167713e-05,
-      "loss": 46.5196,
-      "step": 146870
-    },
-    {
-      "epoch": 0.5934137857197688,
-      "grad_norm": 630.5188598632812,
-      "learning_rate": 2.1735114804733938e-05,
-      "loss": 34.2869,
-      "step": 146880
-    },
-    {
-      "epoch": 0.5934541869851364,
-      "grad_norm": 923.3815307617188,
-      "learning_rate": 2.173165407295078e-05,
-      "loss": 80.3715,
-      "step": 146890
-    },
-    {
-      "epoch": 0.593494588250504,
-      "grad_norm": 985.5263671875,
-      "learning_rate": 2.1728193404885697e-05,
-      "loss": 68.0232,
-      "step": 146900
-    },
-    {
-      "epoch": 0.5935349895158717,
-      "grad_norm": 1455.7779541015625,
-      "learning_rate": 2.172473280060618e-05,
-      "loss": 55.3433,
-      "step": 146910
-    },
-    {
-      "epoch": 0.5935753907812392,
-      "grad_norm": 679.5105590820312,
-      "learning_rate": 2.172127226017967e-05,
-      "loss": 48.74,
-      "step": 146920
-    },
-    {
-      "epoch": 0.5936157920466069,
-      "grad_norm": 484.81591796875,
-      "learning_rate": 2.1717811783673657e-05,
-      "loss": 29.9719,
-      "step": 146930
-    },
-    {
-      "epoch": 0.5936561933119745,
-      "grad_norm": 623.7422485351562,
-      "learning_rate": 2.1714351371155596e-05,
-      "loss": 70.4198,
-      "step": 146940
-    },
-    {
-      "epoch": 0.5936965945773421,
-      "grad_norm": 789.708740234375,
-      "learning_rate": 2.171089102269294e-05,
-      "loss": 34.6678,
-      "step": 146950
-    },
-    {
-      "epoch": 0.5937369958427098,
-      "grad_norm": 1196.701904296875,
-      "learning_rate": 2.170743073835316e-05,
-      "loss": 52.6208,
-      "step": 146960
-    },
-    {
-      "epoch": 0.5937773971080774,
-      "grad_norm": 540.9883422851562,
-      "learning_rate": 2.170397051820371e-05,
-      "loss": 57.1448,
-      "step": 146970
-    },
-    {
-      "epoch": 0.5938177983734451,
-      "grad_norm": 4241.60205078125,
-      "learning_rate": 2.1700510362312052e-05,
-      "loss": 70.6284,
-      "step": 146980
-    },
-    {
-      "epoch": 0.5938581996388127,
-      "grad_norm": 664.571044921875,
-      "learning_rate": 2.1697050270745648e-05,
-      "loss": 86.9309,
-      "step": 146990
-    },
-    {
-      "epoch": 0.5938986009041803,
-      "grad_norm": 1107.1243896484375,
-      "learning_rate": 2.1693590243571938e-05,
-      "loss": 43.3203,
-      "step": 147000
-    },
-    {
-      "epoch": 0.593939002169548,
-      "grad_norm": 1379.3629150390625,
-      "learning_rate": 2.1690130280858398e-05,
-      "loss": 63.4872,
-      "step": 147010
-    },
-    {
-      "epoch": 0.5939794034349156,
-      "grad_norm": 534.1649780273438,
-      "learning_rate": 2.1686670382672475e-05,
-      "loss": 56.6214,
-      "step": 147020
-    },
-    {
-      "epoch": 0.5940198047002833,
-      "grad_norm": 675.9949951171875,
-      "learning_rate": 2.1683210549081607e-05,
-      "loss": 53.4801,
-      "step": 147030
-    },
-    {
-      "epoch": 0.5940602059656508,
-      "grad_norm": 450.9206848144531,
-      "learning_rate": 2.1679750780153267e-05,
-      "loss": 45.3882,
-      "step": 147040
-    },
-    {
-      "epoch": 0.5941006072310184,
-      "grad_norm": 513.6524047851562,
-      "learning_rate": 2.1676291075954894e-05,
-      "loss": 56.7767,
-      "step": 147050
-    },
-    {
-      "epoch": 0.5941410084963861,
-      "grad_norm": 938.4927368164062,
-      "learning_rate": 2.1672831436553935e-05,
-      "loss": 38.4072,
-      "step": 147060
-    },
-    {
-      "epoch": 0.5941814097617537,
-      "grad_norm": 520.1245727539062,
-      "learning_rate": 2.166937186201784e-05,
-      "loss": 49.0356,
-      "step": 147070
-    },
-    {
-      "epoch": 0.5942218110271213,
-      "grad_norm": 508.855224609375,
-      "learning_rate": 2.166591235241405e-05,
-      "loss": 45.4661,
-      "step": 147080
-    },
-    {
-      "epoch": 0.594262212292489,
-      "grad_norm": 1004.2674560546875,
-      "learning_rate": 2.1662452907810024e-05,
-      "loss": 49.2036,
-      "step": 147090
-    },
-    {
-      "epoch": 0.5943026135578566,
-      "grad_norm": 769.294189453125,
-      "learning_rate": 2.1658993528273197e-05,
-      "loss": 65.9896,
-      "step": 147100
-    },
-    {
-      "epoch": 0.5943430148232243,
-      "grad_norm": 256.8412780761719,
-      "learning_rate": 2.1655534213871e-05,
-      "loss": 55.592,
-      "step": 147110
-    },
-    {
-      "epoch": 0.5943834160885919,
-      "grad_norm": 1590.532958984375,
-      "learning_rate": 2.16520749646709e-05,
-      "loss": 56.2808,
-      "step": 147120
-    },
-    {
-      "epoch": 0.5944238173539595,
-      "grad_norm": 846.234619140625,
-      "learning_rate": 2.1648615780740316e-05,
-      "loss": 50.1012,
-      "step": 147130
-    },
-    {
-      "epoch": 0.5944642186193272,
-      "grad_norm": 884.344482421875,
-      "learning_rate": 2.164515666214669e-05,
-      "loss": 40.8878,
-      "step": 147140
-    },
-    {
-      "epoch": 0.5945046198846948,
-      "grad_norm": 409.51025390625,
-      "learning_rate": 2.1641697608957466e-05,
-      "loss": 50.304,
-      "step": 147150
-    },
-    {
-      "epoch": 0.5945450211500625,
-      "grad_norm": 934.0576782226562,
-      "learning_rate": 2.163823862124007e-05,
-      "loss": 46.8148,
-      "step": 147160
-    },
-    {
-      "epoch": 0.59458542241543,
-      "grad_norm": 1109.9046630859375,
-      "learning_rate": 2.163477969906195e-05,
-      "loss": 47.4143,
-      "step": 147170
-    },
-    {
-      "epoch": 0.5946258236807976,
-      "grad_norm": 733.655517578125,
-      "learning_rate": 2.1631320842490532e-05,
-      "loss": 42.5417,
-      "step": 147180
-    },
-    {
-      "epoch": 0.5946662249461653,
-      "grad_norm": 121.8224868774414,
-      "learning_rate": 2.162786205159324e-05,
-      "loss": 52.3802,
-      "step": 147190
-    },
-    {
-      "epoch": 0.5947066262115329,
-      "grad_norm": 1324.3228759765625,
-      "learning_rate": 2.1624403326437523e-05,
-      "loss": 67.4435,
-      "step": 147200
-    },
-    {
-      "epoch": 0.5947470274769006,
-      "grad_norm": 911.5564575195312,
-      "learning_rate": 2.16209446670908e-05,
-      "loss": 43.5101,
-      "step": 147210
-    },
-    {
-      "epoch": 0.5947874287422682,
-      "grad_norm": 460.3110656738281,
-      "learning_rate": 2.1617486073620498e-05,
-      "loss": 59.1632,
-      "step": 147220
-    },
-    {
-      "epoch": 0.5948278300076358,
-      "grad_norm": 569.986572265625,
-      "learning_rate": 2.1614027546094048e-05,
-      "loss": 35.5932,
-      "step": 147230
-    },
-    {
-      "epoch": 0.5948682312730035,
-      "grad_norm": 1580.7315673828125,
-      "learning_rate": 2.1610569084578867e-05,
-      "loss": 39.1384,
-      "step": 147240
-    },
-    {
-      "epoch": 0.5949086325383711,
-      "grad_norm": 675.512939453125,
-      "learning_rate": 2.1607110689142393e-05,
-      "loss": 55.1546,
-      "step": 147250
-    },
-    {
-      "epoch": 0.5949490338037388,
-      "grad_norm": 978.5493774414062,
-      "learning_rate": 2.1603652359852044e-05,
-      "loss": 38.9832,
-      "step": 147260
-    },
-    {
-      "epoch": 0.5949894350691064,
-      "grad_norm": 1820.423583984375,
-      "learning_rate": 2.1600194096775233e-05,
-      "loss": 65.3882,
-      "step": 147270
-    },
-    {
-      "epoch": 0.595029836334474,
-      "grad_norm": 451.9998474121094,
-      "learning_rate": 2.1596735899979396e-05,
-      "loss": 40.6909,
-      "step": 147280
-    },
-    {
-      "epoch": 0.5950702375998417,
-      "grad_norm": 1065.4068603515625,
-      "learning_rate": 2.159327776953195e-05,
-      "loss": 66.7301,
-      "step": 147290
-    },
-    {
-      "epoch": 0.5951106388652092,
-      "grad_norm": 672.516845703125,
-      "learning_rate": 2.158981970550029e-05,
-      "loss": 50.5191,
-      "step": 147300
-    },
-    {
-      "epoch": 0.5951510401305768,
-      "grad_norm": 2511.90087890625,
-      "learning_rate": 2.1586361707951866e-05,
-      "loss": 59.1653,
-      "step": 147310
-    },
-    {
-      "epoch": 0.5951914413959445,
-      "grad_norm": 489.2138366699219,
-      "learning_rate": 2.158290377695407e-05,
-      "loss": 50.135,
-      "step": 147320
-    },
-    {
-      "epoch": 0.5952318426613121,
-      "grad_norm": 790.0176391601562,
-      "learning_rate": 2.1579445912574333e-05,
-      "loss": 47.7796,
-      "step": 147330
-    },
-    {
-      "epoch": 0.5952722439266798,
-      "grad_norm": 1207.177734375,
-      "learning_rate": 2.157598811488006e-05,
-      "loss": 53.8064,
-      "step": 147340
-    },
-    {
-      "epoch": 0.5953126451920474,
-      "grad_norm": 1556.337158203125,
-      "learning_rate": 2.1572530383938645e-05,
-      "loss": 74.1998,
-      "step": 147350
-    },
-    {
-      "epoch": 0.595353046457415,
-      "grad_norm": 289.1591796875,
-      "learning_rate": 2.1569072719817526e-05,
-      "loss": 47.7599,
-      "step": 147360
-    },
-    {
-      "epoch": 0.5953934477227827,
-      "grad_norm": 793.8142700195312,
-      "learning_rate": 2.1565615122584092e-05,
-      "loss": 50.358,
-      "step": 147370
-    },
-    {
-      "epoch": 0.5954338489881503,
-      "grad_norm": 424.2123718261719,
-      "learning_rate": 2.156215759230577e-05,
-      "loss": 50.9866,
-      "step": 147380
-    },
-    {
-      "epoch": 0.595474250253518,
-      "grad_norm": 680.01318359375,
-      "learning_rate": 2.155870012904996e-05,
-      "loss": 34.8699,
-      "step": 147390
-    },
-    {
-      "epoch": 0.5955146515188856,
-      "grad_norm": 1084.6480712890625,
-      "learning_rate": 2.155524273288405e-05,
-      "loss": 54.3668,
-      "step": 147400
-    },
-    {
-      "epoch": 0.5955550527842532,
-      "grad_norm": 1713.6904296875,
-      "learning_rate": 2.155178540387546e-05,
-      "loss": 55.3956,
-      "step": 147410
-    },
-    {
-      "epoch": 0.5955954540496209,
-      "grad_norm": 677.2518310546875,
-      "learning_rate": 2.1548328142091598e-05,
-      "loss": 48.0534,
-      "step": 147420
-    },
-    {
-      "epoch": 0.5956358553149884,
-      "grad_norm": 976.79443359375,
-      "learning_rate": 2.154487094759984e-05,
-      "loss": 50.823,
-      "step": 147430
-    },
-    {
-      "epoch": 0.5956762565803561,
-      "grad_norm": 306.4274597167969,
-      "learning_rate": 2.1541413820467615e-05,
-      "loss": 70.2195,
-      "step": 147440
-    },
-    {
-      "epoch": 0.5957166578457237,
-      "grad_norm": 870.5831298828125,
-      "learning_rate": 2.1537956760762295e-05,
-      "loss": 44.3455,
-      "step": 147450
-    },
-    {
-      "epoch": 0.5957570591110913,
-      "grad_norm": 2688.650146484375,
-      "learning_rate": 2.15344997685513e-05,
-      "loss": 55.4343,
-      "step": 147460
-    },
-    {
-      "epoch": 0.595797460376459,
-      "grad_norm": 1122.9935302734375,
-      "learning_rate": 2.1531042843902018e-05,
-      "loss": 54.8769,
-      "step": 147470
-    },
-    {
-      "epoch": 0.5958378616418266,
-      "grad_norm": 810.6016845703125,
-      "learning_rate": 2.1527585986881837e-05,
-      "loss": 33.0876,
-      "step": 147480
-    },
-    {
-      "epoch": 0.5958782629071943,
-      "grad_norm": 579.2329711914062,
-      "learning_rate": 2.152412919755816e-05,
-      "loss": 56.4312,
-      "step": 147490
-    },
-    {
-      "epoch": 0.5959186641725619,
-      "grad_norm": 578.1658325195312,
-      "learning_rate": 2.1520672475998373e-05,
-      "loss": 60.0798,
-      "step": 147500
-    },
-    {
-      "epoch": 0.5959590654379295,
-      "grad_norm": 828.1384887695312,
-      "learning_rate": 2.1517215822269857e-05,
-      "loss": 53.2631,
-      "step": 147510
-    },
-    {
-      "epoch": 0.5959994667032972,
-      "grad_norm": 278.0408935546875,
-      "learning_rate": 2.1513759236440023e-05,
-      "loss": 46.6886,
-      "step": 147520
-    },
-    {
-      "epoch": 0.5960398679686648,
-      "grad_norm": 1062.5751953125,
-      "learning_rate": 2.1510302718576232e-05,
-      "loss": 67.722,
-      "step": 147530
-    },
-    {
-      "epoch": 0.5960802692340325,
-      "grad_norm": 1039.3131103515625,
-      "learning_rate": 2.1506846268745903e-05,
-      "loss": 37.7568,
-      "step": 147540
-    },
-    {
-      "epoch": 0.5961206704994001,
-      "grad_norm": 186.1059112548828,
-      "learning_rate": 2.1503389887016404e-05,
-      "loss": 49.6921,
-      "step": 147550
-    },
-    {
-      "epoch": 0.5961610717647676,
-      "grad_norm": 664.5225219726562,
-      "learning_rate": 2.149993357345511e-05,
-      "loss": 50.4448,
-      "step": 147560
-    },
-    {
-      "epoch": 0.5962014730301353,
-      "grad_norm": 636.2140502929688,
-      "learning_rate": 2.149647732812942e-05,
-      "loss": 46.299,
-      "step": 147570
-    },
-    {
-      "epoch": 0.5962418742955029,
-      "grad_norm": 631.2529907226562,
-      "learning_rate": 2.1493021151106703e-05,
-      "loss": 51.5016,
-      "step": 147580
-    },
-    {
-      "epoch": 0.5962822755608705,
-      "grad_norm": 579.6639404296875,
-      "learning_rate": 2.1489565042454344e-05,
-      "loss": 56.0493,
-      "step": 147590
-    },
-    {
-      "epoch": 0.5963226768262382,
-      "grad_norm": 571.759033203125,
-      "learning_rate": 2.148610900223973e-05,
-      "loss": 59.5269,
-      "step": 147600
-    },
-    {
-      "epoch": 0.5963630780916058,
-      "grad_norm": 580.6973876953125,
-      "learning_rate": 2.1482653030530217e-05,
-      "loss": 55.8021,
-      "step": 147610
-    },
-    {
-      "epoch": 0.5964034793569735,
-      "grad_norm": 921.6025390625,
-      "learning_rate": 2.1479197127393204e-05,
-      "loss": 43.2684,
-      "step": 147620
-    },
-    {
-      "epoch": 0.5964438806223411,
-      "grad_norm": 646.3934326171875,
-      "learning_rate": 2.1475741292896055e-05,
-      "loss": 64.2098,
-      "step": 147630
-    },
-    {
-      "epoch": 0.5964842818877087,
-      "grad_norm": 453.28741455078125,
-      "learning_rate": 2.1472285527106137e-05,
-      "loss": 40.0422,
-      "step": 147640
-    },
-    {
-      "epoch": 0.5965246831530764,
-      "grad_norm": 952.0294799804688,
-      "learning_rate": 2.1468829830090838e-05,
-      "loss": 54.1033,
-      "step": 147650
-    },
-    {
-      "epoch": 0.596565084418444,
-      "grad_norm": 332.8349609375,
-      "learning_rate": 2.1465374201917518e-05,
-      "loss": 39.744,
-      "step": 147660
-    },
-    {
-      "epoch": 0.5966054856838117,
-      "grad_norm": 783.0602416992188,
-      "learning_rate": 2.146191864265354e-05,
-      "loss": 43.308,
-      "step": 147670
-    },
-    {
-      "epoch": 0.5966458869491792,
-      "grad_norm": 1387.0821533203125,
-      "learning_rate": 2.145846315236629e-05,
-      "loss": 70.0475,
-      "step": 147680
-    },
-    {
-      "epoch": 0.5966862882145468,
-      "grad_norm": 699.0653076171875,
-      "learning_rate": 2.1455007731123112e-05,
-      "loss": 33.2471,
-      "step": 147690
-    },
-    {
-      "epoch": 0.5967266894799145,
-      "grad_norm": 684.11572265625,
-      "learning_rate": 2.1451552378991392e-05,
-      "loss": 51.8979,
-      "step": 147700
-    },
-    {
-      "epoch": 0.5967670907452821,
-      "grad_norm": 799.9059448242188,
-      "learning_rate": 2.144809709603849e-05,
-      "loss": 90.1518,
-      "step": 147710
-    },
-    {
-      "epoch": 0.5968074920106498,
-      "grad_norm": 564.4153442382812,
-      "learning_rate": 2.1444641882331744e-05,
-      "loss": 33.3515,
-      "step": 147720
-    },
-    {
-      "epoch": 0.5968478932760174,
-      "grad_norm": 900.5286865234375,
-      "learning_rate": 2.1441186737938555e-05,
-      "loss": 50.0462,
-      "step": 147730
-    },
-    {
-      "epoch": 0.596888294541385,
-      "grad_norm": 479.7302551269531,
-      "learning_rate": 2.1437731662926258e-05,
-      "loss": 41.8366,
-      "step": 147740
-    },
-    {
-      "epoch": 0.5969286958067527,
-      "grad_norm": 1118.731201171875,
-      "learning_rate": 2.1434276657362213e-05,
-      "loss": 68.8751,
-      "step": 147750
-    },
-    {
-      "epoch": 0.5969690970721203,
-      "grad_norm": 297.614013671875,
-      "learning_rate": 2.1430821721313782e-05,
-      "loss": 66.7755,
-      "step": 147760
-    },
-    {
-      "epoch": 0.597009498337488,
-      "grad_norm": 3487.298828125,
-      "learning_rate": 2.142736685484831e-05,
-      "loss": 63.8574,
-      "step": 147770
-    },
-    {
-      "epoch": 0.5970498996028556,
-      "grad_norm": 596.0402221679688,
-      "learning_rate": 2.1423912058033174e-05,
-      "loss": 51.5885,
-      "step": 147780
-    },
-    {
-      "epoch": 0.5970903008682232,
-      "grad_norm": 1751.547119140625,
-      "learning_rate": 2.142045733093571e-05,
-      "loss": 62.8217,
-      "step": 147790
-    },
-    {
-      "epoch": 0.5971307021335909,
-      "grad_norm": 951.2716674804688,
-      "learning_rate": 2.1417002673623264e-05,
-      "loss": 44.525,
-      "step": 147800
-    },
-    {
-      "epoch": 0.5971711033989584,
-      "grad_norm": 464.1297912597656,
-      "learning_rate": 2.141354808616321e-05,
-      "loss": 52.8699,
-      "step": 147810
-    },
-    {
-      "epoch": 0.597211504664326,
-      "grad_norm": 581.2355346679688,
-      "learning_rate": 2.1410093568622878e-05,
-      "loss": 50.8966,
-      "step": 147820
-    },
-    {
-      "epoch": 0.5972519059296937,
-      "grad_norm": 861.3636474609375,
-      "learning_rate": 2.1406639121069617e-05,
-      "loss": 56.8085,
-      "step": 147830
-    },
-    {
-      "epoch": 0.5972923071950613,
-      "grad_norm": 1141.43505859375,
-      "learning_rate": 2.1403184743570778e-05,
-      "loss": 49.3954,
-      "step": 147840
-    },
-    {
-      "epoch": 0.597332708460429,
-      "grad_norm": 708.7974853515625,
-      "learning_rate": 2.1399730436193697e-05,
-      "loss": 37.2018,
-      "step": 147850
-    },
-    {
-      "epoch": 0.5973731097257966,
-      "grad_norm": 584.9114990234375,
-      "learning_rate": 2.1396276199005734e-05,
-      "loss": 39.1667,
-      "step": 147860
-    },
-    {
-      "epoch": 0.5974135109911642,
-      "grad_norm": 890.6394653320312,
-      "learning_rate": 2.1392822032074224e-05,
-      "loss": 41.9098,
-      "step": 147870
-    },
-    {
-      "epoch": 0.5974539122565319,
-      "grad_norm": 1730.406494140625,
-      "learning_rate": 2.138936793546649e-05,
-      "loss": 51.9415,
-      "step": 147880
-    },
-    {
-      "epoch": 0.5974943135218995,
-      "grad_norm": 970.1226806640625,
-      "learning_rate": 2.13859139092499e-05,
-      "loss": 54.9884,
-      "step": 147890
-    },
-    {
-      "epoch": 0.5975347147872672,
-      "grad_norm": 559.8754272460938,
-      "learning_rate": 2.1382459953491774e-05,
-      "loss": 48.8598,
-      "step": 147900
-    },
-    {
-      "epoch": 0.5975751160526348,
-      "grad_norm": 5372.072265625,
-      "learning_rate": 2.137900606825946e-05,
-      "loss": 54.8929,
-      "step": 147910
-    },
-    {
-      "epoch": 0.5976155173180024,
-      "grad_norm": 871.8217163085938,
-      "learning_rate": 2.137555225362028e-05,
-      "loss": 46.6193,
-      "step": 147920
-    },
-    {
-      "epoch": 0.5976559185833701,
-      "grad_norm": 585.6782836914062,
-      "learning_rate": 2.1372098509641574e-05,
-      "loss": 47.909,
-      "step": 147930
-    },
-    {
-      "epoch": 0.5976963198487376,
-      "grad_norm": 884.8446655273438,
-      "learning_rate": 2.1368644836390684e-05,
-      "loss": 41.8532,
-      "step": 147940
-    },
-    {
-      "epoch": 0.5977367211141053,
-      "grad_norm": 589.468017578125,
-      "learning_rate": 2.136519123393493e-05,
-      "loss": 57.2476,
-      "step": 147950
-    },
-    {
-      "epoch": 0.5977771223794729,
-      "grad_norm": 1855.3509521484375,
-      "learning_rate": 2.1361737702341634e-05,
-      "loss": 71.2147,
-      "step": 147960
-    },
-    {
-      "epoch": 0.5978175236448405,
-      "grad_norm": 304.5381774902344,
-      "learning_rate": 2.1358284241678146e-05,
-      "loss": 42.8469,
-      "step": 147970
-    },
-    {
-      "epoch": 0.5978579249102082,
-      "grad_norm": 451.1543273925781,
-      "learning_rate": 2.135483085201177e-05,
-      "loss": 64.4844,
-      "step": 147980
-    },
-    {
-      "epoch": 0.5978983261755758,
-      "grad_norm": 1317.0662841796875,
-      "learning_rate": 2.1351377533409855e-05,
-      "loss": 83.4466,
-      "step": 147990
-    },
-    {
-      "epoch": 0.5979387274409435,
-      "grad_norm": 754.2716064453125,
-      "learning_rate": 2.1347924285939714e-05,
-      "loss": 61.419,
-      "step": 148000
-    },
-    {
-      "epoch": 0.5979791287063111,
-      "grad_norm": 934.2620239257812,
-      "learning_rate": 2.1344471109668662e-05,
-      "loss": 56.7021,
-      "step": 148010
-    },
-    {
-      "epoch": 0.5980195299716787,
-      "grad_norm": 708.87451171875,
-      "learning_rate": 2.134101800466403e-05,
-      "loss": 59.7696,
-      "step": 148020
-    },
-    {
-      "epoch": 0.5980599312370464,
-      "grad_norm": 717.4935302734375,
-      "learning_rate": 2.1337564970993145e-05,
-      "loss": 41.5032,
-      "step": 148030
-    },
-    {
-      "epoch": 0.598100332502414,
-      "grad_norm": 664.8682861328125,
-      "learning_rate": 2.1334112008723297e-05,
-      "loss": 44.9038,
-      "step": 148040
-    },
-    {
-      "epoch": 0.5981407337677817,
-      "grad_norm": 607.526611328125,
-      "learning_rate": 2.1330659117921837e-05,
-      "loss": 72.0059,
-      "step": 148050
-    },
-    {
-      "epoch": 0.5981811350331493,
-      "grad_norm": 848.9324340820312,
-      "learning_rate": 2.1327206298656056e-05,
-      "loss": 56.5129,
-      "step": 148060
-    },
-    {
-      "epoch": 0.5982215362985168,
-      "grad_norm": 790.6737670898438,
-      "learning_rate": 2.1323753550993288e-05,
-      "loss": 40.4335,
-      "step": 148070
-    },
-    {
-      "epoch": 0.5982619375638845,
-      "grad_norm": 196.77239990234375,
-      "learning_rate": 2.132030087500084e-05,
-      "loss": 32.2374,
-      "step": 148080
-    },
-    {
-      "epoch": 0.5983023388292521,
-      "grad_norm": 645.4754638671875,
-      "learning_rate": 2.1316848270746015e-05,
-      "loss": 67.4574,
-      "step": 148090
-    },
-    {
-      "epoch": 0.5983427400946197,
-      "grad_norm": 1457.8206787109375,
-      "learning_rate": 2.1313395738296134e-05,
-      "loss": 56.8633,
-      "step": 148100
-    },
-    {
-      "epoch": 0.5983831413599874,
-      "grad_norm": 773.4266357421875,
-      "learning_rate": 2.1309943277718497e-05,
-      "loss": 50.3852,
-      "step": 148110
-    },
-    {
-      "epoch": 0.598423542625355,
-      "grad_norm": 988.7295532226562,
-      "learning_rate": 2.130649088908041e-05,
-      "loss": 49.5678,
-      "step": 148120
-    },
-    {
-      "epoch": 0.5984639438907227,
-      "grad_norm": 0.0,
-      "learning_rate": 2.1303038572449192e-05,
-      "loss": 40.0559,
-      "step": 148130
-    },
-    {
-      "epoch": 0.5985043451560903,
-      "grad_norm": 1050.348876953125,
-      "learning_rate": 2.1299586327892133e-05,
-      "loss": 58.7035,
-      "step": 148140
-    },
-    {
-      "epoch": 0.5985447464214579,
-      "grad_norm": 1048.6171875,
-      "learning_rate": 2.129613415547655e-05,
-      "loss": 62.4022,
-      "step": 148150
-    },
-    {
-      "epoch": 0.5985851476868256,
-      "grad_norm": 391.7165832519531,
-      "learning_rate": 2.1292682055269745e-05,
-      "loss": 64.7473,
-      "step": 148160
-    },
-    {
-      "epoch": 0.5986255489521932,
-      "grad_norm": 542.6089477539062,
-      "learning_rate": 2.1289230027339003e-05,
-      "loss": 72.982,
-      "step": 148170
-    },
-    {
-      "epoch": 0.5986659502175609,
-      "grad_norm": 1377.5733642578125,
-      "learning_rate": 2.1285778071751634e-05,
-      "loss": 50.1366,
-      "step": 148180
-    },
-    {
-      "epoch": 0.5987063514829285,
-      "grad_norm": 574.2361450195312,
-      "learning_rate": 2.128232618857494e-05,
-      "loss": 64.3552,
-      "step": 148190
-    },
-    {
-      "epoch": 0.598746752748296,
-      "grad_norm": 1206.60888671875,
-      "learning_rate": 2.1278874377876197e-05,
-      "loss": 71.0976,
-      "step": 148200
-    },
-    {
-      "epoch": 0.5987871540136637,
-      "grad_norm": 402.2405090332031,
-      "learning_rate": 2.1275422639722724e-05,
-      "loss": 42.0494,
-      "step": 148210
-    },
-    {
-      "epoch": 0.5988275552790313,
-      "grad_norm": 777.2664184570312,
-      "learning_rate": 2.1271970974181795e-05,
-      "loss": 48.6818,
-      "step": 148220
-    },
-    {
-      "epoch": 0.598867956544399,
-      "grad_norm": 840.8900146484375,
-      "learning_rate": 2.1268519381320724e-05,
-      "loss": 63.5646,
-      "step": 148230
-    },
-    {
-      "epoch": 0.5989083578097666,
-      "grad_norm": 532.3710327148438,
-      "learning_rate": 2.1265067861206784e-05,
-      "loss": 55.8238,
-      "step": 148240
-    },
-    {
-      "epoch": 0.5989487590751342,
-      "grad_norm": 466.35003662109375,
-      "learning_rate": 2.1261616413907265e-05,
-      "loss": 58.5717,
-      "step": 148250
-    },
-    {
-      "epoch": 0.5989891603405019,
-      "grad_norm": 329.7406311035156,
-      "learning_rate": 2.125816503948946e-05,
-      "loss": 54.9311,
-      "step": 148260
-    },
-    {
-      "epoch": 0.5990295616058695,
-      "grad_norm": 347.29327392578125,
-      "learning_rate": 2.1254713738020658e-05,
-      "loss": 55.2146,
-      "step": 148270
-    },
-    {
-      "epoch": 0.5990699628712371,
-      "grad_norm": 513.181640625,
-      "learning_rate": 2.1251262509568133e-05,
-      "loss": 45.6338,
-      "step": 148280
-    },
-    {
-      "epoch": 0.5991103641366048,
-      "grad_norm": 308.05279541015625,
-      "learning_rate": 2.1247811354199184e-05,
-      "loss": 51.3038,
-      "step": 148290
-    },
-    {
-      "epoch": 0.5991507654019724,
-      "grad_norm": 831.5238037109375,
-      "learning_rate": 2.1244360271981073e-05,
-      "loss": 60.9007,
-      "step": 148300
-    },
-    {
-      "epoch": 0.5991911666673401,
-      "grad_norm": 539.5789184570312,
-      "learning_rate": 2.12409092629811e-05,
-      "loss": 41.7726,
-      "step": 148310
-    },
-    {
-      "epoch": 0.5992315679327076,
-      "grad_norm": 257.03363037109375,
-      "learning_rate": 2.123745832726654e-05,
-      "loss": 55.673,
-      "step": 148320
-    },
-    {
-      "epoch": 0.5992719691980752,
-      "grad_norm": 380.1696472167969,
-      "learning_rate": 2.1234007464904654e-05,
-      "loss": 42.6603,
-      "step": 148330
-    },
-    {
-      "epoch": 0.5993123704634429,
-      "grad_norm": 809.4993286132812,
-      "learning_rate": 2.1230556675962744e-05,
-      "loss": 30.5962,
-      "step": 148340
-    },
-    {
-      "epoch": 0.5993527717288105,
-      "grad_norm": 761.6078491210938,
-      "learning_rate": 2.1227105960508063e-05,
-      "loss": 53.1363,
-      "step": 148350
-    },
-    {
-      "epoch": 0.5993931729941782,
-      "grad_norm": 471.2385559082031,
-      "learning_rate": 2.1223655318607904e-05,
-      "loss": 54.6683,
-      "step": 148360
-    },
-    {
-      "epoch": 0.5994335742595458,
-      "grad_norm": 379.0174865722656,
-      "learning_rate": 2.1220204750329525e-05,
-      "loss": 41.1171,
-      "step": 148370
-    },
-    {
-      "epoch": 0.5994739755249134,
-      "grad_norm": 635.4507446289062,
-      "learning_rate": 2.1216754255740193e-05,
-      "loss": 43.5764,
-      "step": 148380
-    },
-    {
-      "epoch": 0.5995143767902811,
-      "grad_norm": 243.54612731933594,
-      "learning_rate": 2.121330383490719e-05,
-      "loss": 54.9277,
-      "step": 148390
-    },
-    {
-      "epoch": 0.5995547780556487,
-      "grad_norm": 855.86181640625,
-      "learning_rate": 2.1209853487897784e-05,
-      "loss": 58.4136,
-      "step": 148400
-    },
-    {
-      "epoch": 0.5995951793210164,
-      "grad_norm": 666.0252685546875,
-      "learning_rate": 2.1206403214779223e-05,
-      "loss": 63.9464,
-      "step": 148410
-    },
-    {
-      "epoch": 0.599635580586384,
-      "grad_norm": 909.7586059570312,
-      "learning_rate": 2.1202953015618794e-05,
-      "loss": 57.6522,
-      "step": 148420
-    },
-    {
-      "epoch": 0.5996759818517516,
-      "grad_norm": 1767.159912109375,
-      "learning_rate": 2.1199502890483747e-05,
-      "loss": 58.3725,
-      "step": 148430
-    },
-    {
-      "epoch": 0.5997163831171193,
-      "grad_norm": 640.2388916015625,
-      "learning_rate": 2.119605283944135e-05,
-      "loss": 40.1175,
-      "step": 148440
-    },
-    {
-      "epoch": 0.5997567843824868,
-      "grad_norm": 512.8927001953125,
-      "learning_rate": 2.1192602862558864e-05,
-      "loss": 59.5971,
-      "step": 148450
-    },
-    {
-      "epoch": 0.5997971856478544,
-      "grad_norm": 613.088134765625,
-      "learning_rate": 2.1189152959903536e-05,
-      "loss": 59.3687,
-      "step": 148460
-    },
-    {
-      "epoch": 0.5998375869132221,
-      "grad_norm": 489.1797180175781,
-      "learning_rate": 2.118570313154264e-05,
-      "loss": 53.7565,
-      "step": 148470
-    },
-    {
-      "epoch": 0.5998779881785897,
-      "grad_norm": 593.52734375,
-      "learning_rate": 2.1182253377543425e-05,
-      "loss": 53.75,
-      "step": 148480
-    },
-    {
-      "epoch": 0.5999183894439574,
-      "grad_norm": 644.7229614257812,
-      "learning_rate": 2.117880369797314e-05,
-      "loss": 67.9577,
-      "step": 148490
-    },
-    {
-      "epoch": 0.599958790709325,
-      "grad_norm": 755.2145385742188,
-      "learning_rate": 2.117535409289905e-05,
-      "loss": 43.5806,
-      "step": 148500
-    },
-    {
-      "epoch": 0.5999991919746926,
-      "grad_norm": 499.0184326171875,
-      "learning_rate": 2.11719045623884e-05,
-      "loss": 47.2123,
-      "step": 148510
-    },
-    {
-      "epoch": 0.6000395932400603,
-      "grad_norm": 620.1091918945312,
-      "learning_rate": 2.1168455106508446e-05,
-      "loss": 105.0849,
-      "step": 148520
-    },
-    {
-      "epoch": 0.6000799945054279,
-      "grad_norm": 655.6963500976562,
-      "learning_rate": 2.1165005725326427e-05,
-      "loss": 65.4218,
-      "step": 148530
-    },
-    {
-      "epoch": 0.6001203957707956,
-      "grad_norm": 615.6967163085938,
-      "learning_rate": 2.116155641890959e-05,
-      "loss": 38.5551,
-      "step": 148540
-    },
-    {
-      "epoch": 0.6001607970361632,
-      "grad_norm": 774.8275756835938,
-      "learning_rate": 2.1158107187325198e-05,
-      "loss": 57.5716,
-      "step": 148550
-    },
-    {
-      "epoch": 0.6002011983015308,
-      "grad_norm": 1074.4395751953125,
-      "learning_rate": 2.1154658030640483e-05,
-      "loss": 75.7318,
-      "step": 148560
-    },
-    {
-      "epoch": 0.6002415995668985,
-      "grad_norm": 223.1709747314453,
-      "learning_rate": 2.1151208948922676e-05,
-      "loss": 35.348,
-      "step": 148570
-    },
-    {
-      "epoch": 0.600282000832266,
-      "grad_norm": 767.1890869140625,
-      "learning_rate": 2.1147759942239046e-05,
-      "loss": 54.7006,
-      "step": 148580
-    },
-    {
-      "epoch": 0.6003224020976337,
-      "grad_norm": 979.3482666015625,
-      "learning_rate": 2.1144311010656813e-05,
-      "loss": 57.0623,
-      "step": 148590
-    },
-    {
-      "epoch": 0.6003628033630013,
-      "grad_norm": 669.6612548828125,
-      "learning_rate": 2.114086215424322e-05,
-      "loss": 51.9669,
-      "step": 148600
-    },
-    {
-      "epoch": 0.6004032046283689,
-      "grad_norm": 1702.1258544921875,
-      "learning_rate": 2.1137413373065515e-05,
-      "loss": 69.3868,
-      "step": 148610
-    },
-    {
-      "epoch": 0.6004436058937366,
-      "grad_norm": 1155.5137939453125,
-      "learning_rate": 2.1133964667190918e-05,
-      "loss": 63.2453,
-      "step": 148620
-    },
-    {
-      "epoch": 0.6004840071591042,
-      "grad_norm": 1104.260986328125,
-      "learning_rate": 2.1130516036686675e-05,
-      "loss": 43.4092,
-      "step": 148630
-    },
-    {
-      "epoch": 0.6005244084244719,
-      "grad_norm": 491.9853820800781,
-      "learning_rate": 2.1127067481620013e-05,
-      "loss": 44.5126,
-      "step": 148640
-    },
-    {
-      "epoch": 0.6005648096898395,
-      "grad_norm": 737.5925903320312,
-      "learning_rate": 2.1123619002058155e-05,
-      "loss": 51.0688,
-      "step": 148650
-    },
-    {
-      "epoch": 0.6006052109552071,
-      "grad_norm": 823.9745483398438,
-      "learning_rate": 2.112017059806835e-05,
-      "loss": 35.5344,
-      "step": 148660
-    },
-    {
-      "epoch": 0.6006456122205748,
-      "grad_norm": 581.1257934570312,
-      "learning_rate": 2.111672226971781e-05,
-      "loss": 58.1983,
-      "step": 148670
-    },
-    {
-      "epoch": 0.6006860134859424,
-      "grad_norm": 1033.891845703125,
-      "learning_rate": 2.1113274017073774e-05,
-      "loss": 51.1567,
-      "step": 148680
-    },
-    {
-      "epoch": 0.6007264147513101,
-      "grad_norm": 786.7816772460938,
-      "learning_rate": 2.1109825840203464e-05,
-      "loss": 69.1057,
-      "step": 148690
-    },
-    {
-      "epoch": 0.6007668160166777,
-      "grad_norm": 376.516357421875,
-      "learning_rate": 2.1106377739174098e-05,
-      "loss": 79.144,
-      "step": 148700
-    },
-    {
-      "epoch": 0.6008072172820452,
-      "grad_norm": 815.7255859375,
-      "learning_rate": 2.1102929714052908e-05,
-      "loss": 45.9788,
-      "step": 148710
-    },
-    {
-      "epoch": 0.6008476185474129,
-      "grad_norm": 1409.89794921875,
-      "learning_rate": 2.1099481764907108e-05,
-      "loss": 83.6714,
-      "step": 148720
-    },
-    {
-      "epoch": 0.6008880198127805,
-      "grad_norm": 766.6243286132812,
-      "learning_rate": 2.109603389180391e-05,
-      "loss": 60.9154,
-      "step": 148730
-    },
-    {
-      "epoch": 0.6009284210781481,
-      "grad_norm": 440.70526123046875,
-      "learning_rate": 2.1092586094810553e-05,
-      "loss": 46.9509,
-      "step": 148740
-    },
-    {
-      "epoch": 0.6009688223435158,
-      "grad_norm": 544.2029418945312,
-      "learning_rate": 2.1089138373994223e-05,
-      "loss": 28.7837,
-      "step": 148750
-    },
-    {
-      "epoch": 0.6010092236088834,
-      "grad_norm": 557.417236328125,
-      "learning_rate": 2.108569072942217e-05,
-      "loss": 50.8612,
-      "step": 148760
-    },
-    {
-      "epoch": 0.6010496248742511,
-      "grad_norm": 1011.0460205078125,
-      "learning_rate": 2.1082243161161595e-05,
-      "loss": 71.0695,
-      "step": 148770
-    },
-    {
-      "epoch": 0.6010900261396187,
-      "grad_norm": 582.4788208007812,
-      "learning_rate": 2.10787956692797e-05,
-      "loss": 48.1233,
-      "step": 148780
-    },
-    {
-      "epoch": 0.6011304274049863,
-      "grad_norm": 3780.80419921875,
-      "learning_rate": 2.10753482538437e-05,
-      "loss": 88.0165,
-      "step": 148790
-    },
-    {
-      "epoch": 0.601170828670354,
-      "grad_norm": 610.1556396484375,
-      "learning_rate": 2.1071900914920816e-05,
-      "loss": 40.7018,
-      "step": 148800
-    },
-    {
-      "epoch": 0.6012112299357216,
-      "grad_norm": 762.4368896484375,
-      "learning_rate": 2.106845365257823e-05,
-      "loss": 54.8154,
-      "step": 148810
-    },
-    {
-      "epoch": 0.6012516312010893,
-      "grad_norm": 200.17686462402344,
-      "learning_rate": 2.1065006466883177e-05,
-      "loss": 43.5993,
-      "step": 148820
-    },
-    {
-      "epoch": 0.6012920324664568,
-      "grad_norm": 2111.885498046875,
-      "learning_rate": 2.1061559357902838e-05,
-      "loss": 42.0404,
-      "step": 148830
-    },
-    {
-      "epoch": 0.6013324337318244,
-      "grad_norm": 846.29833984375,
-      "learning_rate": 2.1058112325704436e-05,
-      "loss": 53.8223,
-      "step": 148840
-    },
-    {
-      "epoch": 0.6013728349971921,
-      "grad_norm": 426.0714416503906,
-      "learning_rate": 2.1054665370355166e-05,
-      "loss": 43.2251,
-      "step": 148850
-    },
-    {
-      "epoch": 0.6014132362625597,
-      "grad_norm": 958.9960327148438,
-      "learning_rate": 2.1051218491922216e-05,
-      "loss": 84.0954,
-      "step": 148860
-    },
-    {
-      "epoch": 0.6014536375279274,
-      "grad_norm": 308.5674743652344,
-      "learning_rate": 2.1047771690472804e-05,
-      "loss": 40.4277,
-      "step": 148870
-    },
-    {
-      "epoch": 0.601494038793295,
-      "grad_norm": 955.7950439453125,
-      "learning_rate": 2.1044324966074104e-05,
-      "loss": 66.7933,
-      "step": 148880
-    },
-    {
-      "epoch": 0.6015344400586626,
-      "grad_norm": 2592.542724609375,
-      "learning_rate": 2.104087831879334e-05,
-      "loss": 59.7257,
-      "step": 148890
-    },
-    {
-      "epoch": 0.6015748413240303,
-      "grad_norm": 468.540771484375,
-      "learning_rate": 2.1037431748697688e-05,
-      "loss": 57.7334,
-      "step": 148900
-    },
-    {
-      "epoch": 0.6016152425893979,
-      "grad_norm": 1156.9520263671875,
-      "learning_rate": 2.1033985255854336e-05,
-      "loss": 95.6471,
-      "step": 148910
-    },
-    {
-      "epoch": 0.6016556438547656,
-      "grad_norm": 992.9965209960938,
-      "learning_rate": 2.103053884033049e-05,
-      "loss": 54.1949,
-      "step": 148920
-    },
-    {
-      "epoch": 0.6016960451201332,
-      "grad_norm": 868.084228515625,
-      "learning_rate": 2.1027092502193334e-05,
-      "loss": 46.5351,
-      "step": 148930
-    },
-    {
-      "epoch": 0.6017364463855008,
-      "grad_norm": 575.07666015625,
-      "learning_rate": 2.1023646241510048e-05,
-      "loss": 39.6612,
-      "step": 148940
-    },
-    {
-      "epoch": 0.6017768476508685,
-      "grad_norm": 750.913330078125,
-      "learning_rate": 2.1020200058347833e-05,
-      "loss": 68.0529,
-      "step": 148950
-    },
-    {
-      "epoch": 0.601817248916236,
-      "grad_norm": 508.6002502441406,
-      "learning_rate": 2.1016753952773867e-05,
-      "loss": 33.4746,
-      "step": 148960
-    },
-    {
-      "epoch": 0.6018576501816036,
-      "grad_norm": 769.2796020507812,
-      "learning_rate": 2.1013307924855333e-05,
-      "loss": 50.7453,
-      "step": 148970
-    },
-    {
-      "epoch": 0.6018980514469713,
-      "grad_norm": 458.1070861816406,
-      "learning_rate": 2.1009861974659413e-05,
-      "loss": 51.415,
-      "step": 148980
-    },
-    {
-      "epoch": 0.6019384527123389,
-      "grad_norm": 310.19415283203125,
-      "learning_rate": 2.100641610225328e-05,
-      "loss": 74.4553,
-      "step": 148990
-    },
-    {
-      "epoch": 0.6019788539777066,
-      "grad_norm": 488.74908447265625,
-      "learning_rate": 2.1002970307704132e-05,
-      "loss": 50.668,
-      "step": 149000
-    },
-    {
-      "epoch": 0.6020192552430742,
-      "grad_norm": 1650.5919189453125,
-      "learning_rate": 2.0999524591079133e-05,
-      "loss": 64.8103,
-      "step": 149010
-    },
-    {
-      "epoch": 0.6020596565084418,
-      "grad_norm": 556.3172607421875,
-      "learning_rate": 2.0996078952445452e-05,
-      "loss": 75.6496,
-      "step": 149020
-    },
-    {
-      "epoch": 0.6021000577738095,
-      "grad_norm": 964.8404541015625,
-      "learning_rate": 2.0992633391870286e-05,
-      "loss": 53.5392,
-      "step": 149030
-    },
-    {
-      "epoch": 0.6021404590391771,
-      "grad_norm": 740.5578002929688,
-      "learning_rate": 2.0989187909420786e-05,
-      "loss": 55.9263,
-      "step": 149040
-    },
-    {
-      "epoch": 0.6021808603045448,
-      "grad_norm": 771.8296508789062,
-      "learning_rate": 2.0985742505164144e-05,
-      "loss": 65.7144,
-      "step": 149050
-    },
-    {
-      "epoch": 0.6022212615699124,
-      "grad_norm": 998.5512084960938,
-      "learning_rate": 2.0982297179167515e-05,
-      "loss": 44.6573,
-      "step": 149060
-    },
-    {
-      "epoch": 0.60226166283528,
-      "grad_norm": 330.9595947265625,
-      "learning_rate": 2.097885193149806e-05,
-      "loss": 37.0711,
-      "step": 149070
-    },
-    {
-      "epoch": 0.6023020641006477,
-      "grad_norm": 484.8772888183594,
-      "learning_rate": 2.0975406762222966e-05,
-      "loss": 63.3068,
-      "step": 149080
-    },
-    {
-      "epoch": 0.6023424653660152,
-      "grad_norm": 790.2227172851562,
-      "learning_rate": 2.0971961671409392e-05,
-      "loss": 34.2787,
-      "step": 149090
-    },
-    {
-      "epoch": 0.6023828666313829,
-      "grad_norm": 723.9091186523438,
-      "learning_rate": 2.0968516659124486e-05,
-      "loss": 26.5191,
-      "step": 149100
-    },
-    {
-      "epoch": 0.6024232678967505,
-      "grad_norm": 1749.529052734375,
-      "learning_rate": 2.0965071725435436e-05,
-      "loss": 41.8332,
-      "step": 149110
-    },
-    {
-      "epoch": 0.6024636691621181,
-      "grad_norm": 528.8098754882812,
-      "learning_rate": 2.0961626870409383e-05,
-      "loss": 39.1988,
-      "step": 149120
-    },
-    {
-      "epoch": 0.6025040704274858,
-      "grad_norm": 985.6099243164062,
-      "learning_rate": 2.0958182094113498e-05,
-      "loss": 51.4806,
-      "step": 149130
-    },
-    {
-      "epoch": 0.6025444716928534,
-      "grad_norm": 532.384033203125,
-      "learning_rate": 2.0954737396614937e-05,
-      "loss": 42.4193,
-      "step": 149140
-    },
-    {
-      "epoch": 0.602584872958221,
-      "grad_norm": 533.1931762695312,
-      "learning_rate": 2.095129277798084e-05,
-      "loss": 55.3392,
-      "step": 149150
-    },
-    {
-      "epoch": 0.6026252742235887,
-      "grad_norm": 636.3084106445312,
-      "learning_rate": 2.0947848238278385e-05,
-      "loss": 51.3045,
-      "step": 149160
-    },
-    {
-      "epoch": 0.6026656754889563,
-      "grad_norm": 726.8486938476562,
-      "learning_rate": 2.0944403777574718e-05,
-      "loss": 46.1654,
-      "step": 149170
-    },
-    {
-      "epoch": 0.602706076754324,
-      "grad_norm": 1090.1893310546875,
-      "learning_rate": 2.0940959395936975e-05,
-      "loss": 54.1789,
-      "step": 149180
-    },
-    {
-      "epoch": 0.6027464780196916,
-      "grad_norm": 727.7335205078125,
-      "learning_rate": 2.0937515093432327e-05,
-      "loss": 40.0967,
-      "step": 149190
-    },
-    {
-      "epoch": 0.6027868792850593,
-      "grad_norm": 450.4414367675781,
-      "learning_rate": 2.0934070870127912e-05,
-      "loss": 65.3961,
-      "step": 149200
-    },
-    {
-      "epoch": 0.6028272805504269,
-      "grad_norm": 1227.625732421875,
-      "learning_rate": 2.0930626726090883e-05,
-      "loss": 49.5206,
-      "step": 149210
-    },
-    {
-      "epoch": 0.6028676818157944,
-      "grad_norm": 586.02001953125,
-      "learning_rate": 2.0927182661388373e-05,
-      "loss": 51.2468,
-      "step": 149220
-    },
-    {
-      "epoch": 0.6029080830811621,
-      "grad_norm": 452.15679931640625,
-      "learning_rate": 2.0923738676087534e-05,
-      "loss": 60.5171,
-      "step": 149230
-    },
-    {
-      "epoch": 0.6029484843465297,
-      "grad_norm": 593.4280395507812,
-      "learning_rate": 2.0920294770255517e-05,
-      "loss": 54.6655,
-      "step": 149240
-    },
-    {
-      "epoch": 0.6029888856118973,
-      "grad_norm": 536.1670532226562,
-      "learning_rate": 2.0916850943959452e-05,
-      "loss": 54.9936,
-      "step": 149250
-    },
-    {
-      "epoch": 0.603029286877265,
-      "grad_norm": 498.478515625,
-      "learning_rate": 2.091340719726647e-05,
-      "loss": 44.6644,
-      "step": 149260
-    },
-    {
-      "epoch": 0.6030696881426326,
-      "grad_norm": 610.2139892578125,
-      "learning_rate": 2.0909963530243726e-05,
-      "loss": 47.0653,
-      "step": 149270
-    },
-    {
-      "epoch": 0.6031100894080003,
-      "grad_norm": 909.2549438476562,
-      "learning_rate": 2.0906519942958347e-05,
-      "loss": 37.5893,
-      "step": 149280
-    },
-    {
-      "epoch": 0.6031504906733679,
-      "grad_norm": 1333.1292724609375,
-      "learning_rate": 2.0903076435477467e-05,
-      "loss": 59.148,
-      "step": 149290
-    },
-    {
-      "epoch": 0.6031908919387355,
-      "grad_norm": 560.1817016601562,
-      "learning_rate": 2.0899633007868227e-05,
-      "loss": 72.2637,
-      "step": 149300
-    },
-    {
-      "epoch": 0.6032312932041032,
-      "grad_norm": 931.7703857421875,
-      "learning_rate": 2.089618966019775e-05,
-      "loss": 46.3886,
-      "step": 149310
-    },
-    {
-      "epoch": 0.6032716944694708,
-      "grad_norm": 628.0271606445312,
-      "learning_rate": 2.089274639253317e-05,
-      "loss": 56.8358,
-      "step": 149320
-    },
-    {
-      "epoch": 0.6033120957348385,
-      "grad_norm": 679.5213623046875,
-      "learning_rate": 2.0889303204941615e-05,
-      "loss": 50.709,
-      "step": 149330
-    },
-    {
-      "epoch": 0.6033524970002061,
-      "grad_norm": 473.5839538574219,
-      "learning_rate": 2.0885860097490202e-05,
-      "loss": 61.7202,
-      "step": 149340
-    },
-    {
-      "epoch": 0.6033928982655736,
-      "grad_norm": 786.1121215820312,
-      "learning_rate": 2.088241707024607e-05,
-      "loss": 36.438,
-      "step": 149350
-    },
-    {
-      "epoch": 0.6034332995309413,
-      "grad_norm": 739.2479858398438,
-      "learning_rate": 2.0878974123276328e-05,
-      "loss": 40.9667,
-      "step": 149360
-    },
-    {
-      "epoch": 0.6034737007963089,
-      "grad_norm": 310.1951599121094,
-      "learning_rate": 2.0875531256648122e-05,
-      "loss": 57.812,
-      "step": 149370
-    },
-    {
-      "epoch": 0.6035141020616765,
-      "grad_norm": 1145.41748046875,
-      "learning_rate": 2.0872088470428553e-05,
-      "loss": 59.6196,
-      "step": 149380
-    },
-    {
-      "epoch": 0.6035545033270442,
-      "grad_norm": 484.976806640625,
-      "learning_rate": 2.086864576468474e-05,
-      "loss": 55.1442,
-      "step": 149390
-    },
-    {
-      "epoch": 0.6035949045924118,
-      "grad_norm": 0.0,
-      "learning_rate": 2.0865203139483812e-05,
-      "loss": 58.5609,
-      "step": 149400
-    },
-    {
-      "epoch": 0.6036353058577795,
-      "grad_norm": 340.4298400878906,
-      "learning_rate": 2.0861760594892867e-05,
-      "loss": 71.7529,
-      "step": 149410
-    },
-    {
-      "epoch": 0.6036757071231471,
-      "grad_norm": 583.3339233398438,
-      "learning_rate": 2.0858318130979038e-05,
-      "loss": 46.8282,
-      "step": 149420
-    },
-    {
-      "epoch": 0.6037161083885147,
-      "grad_norm": 696.9962768554688,
-      "learning_rate": 2.0854875747809428e-05,
-      "loss": 36.0242,
-      "step": 149430
-    },
-    {
-      "epoch": 0.6037565096538824,
-      "grad_norm": 530.3546752929688,
-      "learning_rate": 2.085143344545114e-05,
-      "loss": 70.6566,
-      "step": 149440
-    },
-    {
-      "epoch": 0.60379691091925,
-      "grad_norm": 764.704345703125,
-      "learning_rate": 2.0847991223971306e-05,
-      "loss": 51.4275,
-      "step": 149450
-    },
-    {
-      "epoch": 0.6038373121846177,
-      "grad_norm": 489.4782409667969,
-      "learning_rate": 2.0844549083437022e-05,
-      "loss": 69.5268,
-      "step": 149460
-    },
-    {
-      "epoch": 0.6038777134499852,
-      "grad_norm": 779.9122924804688,
-      "learning_rate": 2.084110702391538e-05,
-      "loss": 46.3204,
-      "step": 149470
-    },
-    {
-      "epoch": 0.6039181147153528,
-      "grad_norm": 636.275146484375,
-      "learning_rate": 2.083766504547351e-05,
-      "loss": 46.5596,
-      "step": 149480
-    },
-    {
-      "epoch": 0.6039585159807205,
-      "grad_norm": 759.5249633789062,
-      "learning_rate": 2.0834223148178488e-05,
-      "loss": 46.3107,
-      "step": 149490
-    },
-    {
-      "epoch": 0.6039989172460881,
-      "grad_norm": 575.3982543945312,
-      "learning_rate": 2.0830781332097446e-05,
-      "loss": 81.1673,
-      "step": 149500
-    },
-    {
-      "epoch": 0.6040393185114558,
-      "grad_norm": 952.708740234375,
-      "learning_rate": 2.0827339597297466e-05,
-      "loss": 39.2819,
-      "step": 149510
-    },
-    {
-      "epoch": 0.6040797197768234,
-      "grad_norm": 628.7777099609375,
-      "learning_rate": 2.082389794384564e-05,
-      "loss": 51.4413,
-      "step": 149520
-    },
-    {
-      "epoch": 0.604120121042191,
-      "grad_norm": 675.4315795898438,
-      "learning_rate": 2.0820456371809078e-05,
-      "loss": 49.7467,
-      "step": 149530
-    },
-    {
-      "epoch": 0.6041605223075587,
-      "grad_norm": 915.1492919921875,
-      "learning_rate": 2.0817014881254876e-05,
-      "loss": 53.7017,
-      "step": 149540
-    },
-    {
-      "epoch": 0.6042009235729263,
-      "grad_norm": 1439.67431640625,
-      "learning_rate": 2.0813573472250114e-05,
-      "loss": 49.7932,
-      "step": 149550
-    },
-    {
-      "epoch": 0.604241324838294,
-      "grad_norm": 819.9117431640625,
-      "learning_rate": 2.08101321448619e-05,
-      "loss": 54.5777,
-      "step": 149560
-    },
-    {
-      "epoch": 0.6042817261036616,
-      "grad_norm": 747.6610107421875,
-      "learning_rate": 2.0806690899157306e-05,
-      "loss": 79.415,
-      "step": 149570
-    },
-    {
-      "epoch": 0.6043221273690292,
-      "grad_norm": 2668.103759765625,
-      "learning_rate": 2.080324973520344e-05,
-      "loss": 90.0351,
-      "step": 149580
-    },
-    {
-      "epoch": 0.6043625286343969,
-      "grad_norm": 1174.483154296875,
-      "learning_rate": 2.079980865306739e-05,
-      "loss": 69.4175,
-      "step": 149590
-    },
-    {
-      "epoch": 0.6044029298997644,
-      "grad_norm": 467.97698974609375,
-      "learning_rate": 2.0796367652816213e-05,
-      "loss": 58.8468,
-      "step": 149600
-    },
-    {
-      "epoch": 0.604443331165132,
-      "grad_norm": 751.82177734375,
-      "learning_rate": 2.0792926734517024e-05,
-      "loss": 38.2991,
-      "step": 149610
-    },
-    {
-      "epoch": 0.6044837324304997,
-      "grad_norm": 512.2536010742188,
-      "learning_rate": 2.0789485898236896e-05,
-      "loss": 43.8234,
-      "step": 149620
-    },
-    {
-      "epoch": 0.6045241336958673,
-      "grad_norm": 1900.0654296875,
-      "learning_rate": 2.07860451440429e-05,
-      "loss": 62.5554,
-      "step": 149630
-    },
-    {
-      "epoch": 0.604564534961235,
-      "grad_norm": 802.1771850585938,
-      "learning_rate": 2.0782604472002128e-05,
-      "loss": 68.0368,
-      "step": 149640
-    },
-    {
-      "epoch": 0.6046049362266026,
-      "grad_norm": 1752.5911865234375,
-      "learning_rate": 2.0779163882181655e-05,
-      "loss": 66.3808,
-      "step": 149650
-    },
-    {
-      "epoch": 0.6046453374919702,
-      "grad_norm": 680.7577514648438,
-      "learning_rate": 2.0775723374648552e-05,
-      "loss": 48.9644,
-      "step": 149660
-    },
-    {
-      "epoch": 0.6046857387573379,
-      "grad_norm": 433.6819152832031,
-      "learning_rate": 2.0772282949469905e-05,
-      "loss": 41.0124,
-      "step": 149670
-    },
-    {
-      "epoch": 0.6047261400227055,
-      "grad_norm": 327.3218078613281,
-      "learning_rate": 2.076884260671276e-05,
-      "loss": 36.1269,
-      "step": 149680
-    },
-    {
-      "epoch": 0.6047665412880732,
-      "grad_norm": 310.0195617675781,
-      "learning_rate": 2.0765402346444226e-05,
-      "loss": 53.7047,
-      "step": 149690
-    },
-    {
-      "epoch": 0.6048069425534408,
-      "grad_norm": 1084.9659423828125,
-      "learning_rate": 2.076196216873135e-05,
-      "loss": 68.674,
-      "step": 149700
-    },
-    {
-      "epoch": 0.6048473438188084,
-      "grad_norm": 1128.5260009765625,
-      "learning_rate": 2.075852207364119e-05,
-      "loss": 73.8726,
-      "step": 149710
-    },
-    {
-      "epoch": 0.6048877450841761,
-      "grad_norm": 1129.6939697265625,
-      "learning_rate": 2.075508206124084e-05,
-      "loss": 45.7701,
-      "step": 149720
-    },
-    {
-      "epoch": 0.6049281463495436,
-      "grad_norm": 755.3470458984375,
-      "learning_rate": 2.0751642131597344e-05,
-      "loss": 40.1757,
-      "step": 149730
-    },
-    {
-      "epoch": 0.6049685476149113,
-      "grad_norm": 577.21435546875,
-      "learning_rate": 2.0748202284777777e-05,
-      "loss": 79.7054,
-      "step": 149740
-    },
-    {
-      "epoch": 0.6050089488802789,
-      "grad_norm": 755.05126953125,
-      "learning_rate": 2.0744762520849193e-05,
-      "loss": 68.8899,
-      "step": 149750
-    },
-    {
-      "epoch": 0.6050493501456465,
-      "grad_norm": 641.5477905273438,
-      "learning_rate": 2.0741322839878647e-05,
-      "loss": 64.6525,
-      "step": 149760
-    },
-    {
-      "epoch": 0.6050897514110142,
-      "grad_norm": 826.8701782226562,
-      "learning_rate": 2.0737883241933213e-05,
-      "loss": 63.9232,
-      "step": 149770
-    },
-    {
-      "epoch": 0.6051301526763818,
-      "grad_norm": 84.71437072753906,
-      "learning_rate": 2.0734443727079943e-05,
-      "loss": 50.706,
-      "step": 149780
-    },
-    {
-      "epoch": 0.6051705539417495,
-      "grad_norm": 494.0994567871094,
-      "learning_rate": 2.0731004295385874e-05,
-      "loss": 55.4391,
-      "step": 149790
-    },
-    {
-      "epoch": 0.6052109552071171,
-      "grad_norm": 783.1380004882812,
-      "learning_rate": 2.0727564946918087e-05,
-      "loss": 36.2091,
-      "step": 149800
-    },
-    {
-      "epoch": 0.6052513564724847,
-      "grad_norm": 1930.76416015625,
-      "learning_rate": 2.0724125681743618e-05,
-      "loss": 49.5212,
-      "step": 149810
-    },
-    {
-      "epoch": 0.6052917577378524,
-      "grad_norm": 949.4205322265625,
-      "learning_rate": 2.072068649992952e-05,
-      "loss": 50.3558,
-      "step": 149820
-    },
-    {
-      "epoch": 0.60533215900322,
-      "grad_norm": 826.8944091796875,
-      "learning_rate": 2.0717247401542844e-05,
-      "loss": 53.6828,
-      "step": 149830
-    },
-    {
-      "epoch": 0.6053725602685877,
-      "grad_norm": 732.9508666992188,
-      "learning_rate": 2.0713808386650625e-05,
-      "loss": 45.9292,
-      "step": 149840
-    },
-    {
-      "epoch": 0.6054129615339553,
-      "grad_norm": 828.0859985351562,
-      "learning_rate": 2.0710369455319928e-05,
-      "loss": 56.6847,
-      "step": 149850
-    },
-    {
-      "epoch": 0.6054533627993228,
-      "grad_norm": 1403.733642578125,
-      "learning_rate": 2.070693060761779e-05,
-      "loss": 57.0237,
-      "step": 149860
-    },
-    {
-      "epoch": 0.6054937640646905,
-      "grad_norm": 722.1435546875,
-      "learning_rate": 2.0703491843611234e-05,
-      "loss": 62.3158,
-      "step": 149870
-    },
-    {
-      "epoch": 0.6055341653300581,
-      "grad_norm": 853.1331787109375,
-      "learning_rate": 2.070005316336733e-05,
-      "loss": 61.9291,
-      "step": 149880
-    },
-    {
-      "epoch": 0.6055745665954257,
-      "grad_norm": 985.730712890625,
-      "learning_rate": 2.06966145669531e-05,
-      "loss": 62.4463,
-      "step": 149890
-    },
-    {
-      "epoch": 0.6056149678607934,
-      "grad_norm": 261.92486572265625,
-      "learning_rate": 2.0693176054435587e-05,
-      "loss": 58.0517,
-      "step": 149900
-    },
-    {
-      "epoch": 0.605655369126161,
-      "grad_norm": 1041.611083984375,
-      "learning_rate": 2.068973762588182e-05,
-      "loss": 61.4181,
-      "step": 149910
-    },
-    {
-      "epoch": 0.6056957703915287,
-      "grad_norm": 319.9795227050781,
-      "learning_rate": 2.0686299281358835e-05,
-      "loss": 57.8208,
-      "step": 149920
-    },
-    {
-      "epoch": 0.6057361716568963,
-      "grad_norm": 975.4109497070312,
-      "learning_rate": 2.0682861020933675e-05,
-      "loss": 69.4663,
-      "step": 149930
-    },
-    {
-      "epoch": 0.605776572922264,
-      "grad_norm": 1784.4403076171875,
-      "learning_rate": 2.0679422844673348e-05,
-      "loss": 50.2619,
-      "step": 149940
-    },
-    {
-      "epoch": 0.6058169741876316,
-      "grad_norm": 855.6481323242188,
-      "learning_rate": 2.067598475264491e-05,
-      "loss": 33.051,
-      "step": 149950
-    },
-    {
-      "epoch": 0.6058573754529992,
-      "grad_norm": 934.8955078125,
-      "learning_rate": 2.067254674491538e-05,
-      "loss": 69.8998,
-      "step": 149960
-    },
-    {
-      "epoch": 0.6058977767183669,
-      "grad_norm": 446.6639709472656,
-      "learning_rate": 2.0669108821551757e-05,
-      "loss": 57.6521,
-      "step": 149970
-    },
-    {
-      "epoch": 0.6059381779837345,
-      "grad_norm": 1895.581298828125,
-      "learning_rate": 2.0665670982621105e-05,
-      "loss": 64.6645,
-      "step": 149980
-    },
-    {
-      "epoch": 0.605978579249102,
-      "grad_norm": 645.7675170898438,
-      "learning_rate": 2.066223322819043e-05,
-      "loss": 60.931,
-      "step": 149990
-    },
-    {
-      "epoch": 0.6060189805144697,
-      "grad_norm": 384.6867980957031,
-      "learning_rate": 2.0658795558326743e-05,
-      "loss": 61.6462,
-      "step": 150000
-    },
-    {
-      "epoch": 0.6060593817798373,
-      "grad_norm": 0.0,
-      "learning_rate": 2.065535797309708e-05,
-      "loss": 58.9445,
-      "step": 150010
-    },
-    {
-      "epoch": 0.606099783045205,
-      "grad_norm": 629.1178588867188,
-      "learning_rate": 2.0651920472568435e-05,
-      "loss": 32.2629,
-      "step": 150020
-    },
-    {
-      "epoch": 0.6061401843105726,
-      "grad_norm": 704.248291015625,
-      "learning_rate": 2.064848305680785e-05,
-      "loss": 60.3656,
-      "step": 150030
-    },
-    {
-      "epoch": 0.6061805855759402,
-      "grad_norm": 662.4960327148438,
-      "learning_rate": 2.0645045725882332e-05,
-      "loss": 74.1694,
-      "step": 150040
-    },
-    {
-      "epoch": 0.6062209868413079,
-      "grad_norm": 374.74383544921875,
-      "learning_rate": 2.0641608479858877e-05,
-      "loss": 42.1155,
-      "step": 150050
-    },
-    {
-      "epoch": 0.6062613881066755,
-      "grad_norm": 195.5003204345703,
-      "learning_rate": 2.0638171318804518e-05,
-      "loss": 60.8555,
-      "step": 150060
-    },
-    {
-      "epoch": 0.6063017893720432,
-      "grad_norm": 248.31491088867188,
-      "learning_rate": 2.0634734242786257e-05,
-      "loss": 46.3665,
-      "step": 150070
-    },
-    {
-      "epoch": 0.6063421906374108,
-      "grad_norm": 615.802734375,
-      "learning_rate": 2.0631297251871093e-05,
-      "loss": 42.9527,
-      "step": 150080
-    },
-    {
-      "epoch": 0.6063825919027784,
-      "grad_norm": 783.1875610351562,
-      "learning_rate": 2.0627860346126045e-05,
-      "loss": 42.422,
-      "step": 150090
-    },
-    {
-      "epoch": 0.6064229931681461,
-      "grad_norm": 728.5169067382812,
-      "learning_rate": 2.0624423525618098e-05,
-      "loss": 56.8203,
-      "step": 150100
-    },
-    {
-      "epoch": 0.6064633944335136,
-      "grad_norm": 745.2196655273438,
-      "learning_rate": 2.0620986790414276e-05,
-      "loss": 50.3501,
-      "step": 150110
-    },
-    {
-      "epoch": 0.6065037956988812,
-      "grad_norm": 662.663330078125,
-      "learning_rate": 2.0617550140581578e-05,
-      "loss": 54.6142,
-      "step": 150120
-    },
-    {
-      "epoch": 0.6065441969642489,
-      "grad_norm": 1055.7906494140625,
-      "learning_rate": 2.0614113576186978e-05,
-      "loss": 44.9255,
-      "step": 150130
-    },
-    {
-      "epoch": 0.6065845982296165,
-      "grad_norm": 2168.784912109375,
-      "learning_rate": 2.0610677097297505e-05,
-      "loss": 57.7951,
-      "step": 150140
-    },
-    {
-      "epoch": 0.6066249994949842,
-      "grad_norm": 1844.6190185546875,
-      "learning_rate": 2.0607240703980142e-05,
-      "loss": 57.9656,
-      "step": 150150
-    },
-    {
-      "epoch": 0.6066654007603518,
-      "grad_norm": 577.7366333007812,
-      "learning_rate": 2.0603804396301876e-05,
-      "loss": 68.4628,
-      "step": 150160
-    },
-    {
-      "epoch": 0.6067058020257194,
-      "grad_norm": 529.1781005859375,
-      "learning_rate": 2.0600368174329714e-05,
-      "loss": 44.7865,
-      "step": 150170
-    },
-    {
-      "epoch": 0.6067462032910871,
-      "grad_norm": 867.0560913085938,
-      "learning_rate": 2.0596932038130628e-05,
-      "loss": 70.4242,
-      "step": 150180
-    },
-    {
-      "epoch": 0.6067866045564547,
-      "grad_norm": 436.40948486328125,
-      "learning_rate": 2.0593495987771634e-05,
-      "loss": 41.8435,
-      "step": 150190
-    },
-    {
-      "epoch": 0.6068270058218224,
-      "grad_norm": 644.1611328125,
-      "learning_rate": 2.0590060023319696e-05,
-      "loss": 34.4137,
-      "step": 150200
-    },
-    {
-      "epoch": 0.60686740708719,
-      "grad_norm": 1544.953857421875,
-      "learning_rate": 2.0586624144841803e-05,
-      "loss": 57.2875,
-      "step": 150210
-    },
-    {
-      "epoch": 0.6069078083525576,
-      "grad_norm": 762.7637939453125,
-      "learning_rate": 2.0583188352404954e-05,
-      "loss": 77.1049,
-      "step": 150220
-    },
-    {
-      "epoch": 0.6069482096179253,
-      "grad_norm": 1265.564453125,
-      "learning_rate": 2.0579752646076123e-05,
-      "loss": 47.9961,
-      "step": 150230
-    },
-    {
-      "epoch": 0.6069886108832928,
-      "grad_norm": 1833.8243408203125,
-      "learning_rate": 2.0576317025922283e-05,
-      "loss": 65.6349,
-      "step": 150240
-    },
-    {
-      "epoch": 0.6070290121486605,
-      "grad_norm": 1576.8372802734375,
-      "learning_rate": 2.057288149201042e-05,
-      "loss": 49.3936,
-      "step": 150250
-    },
-    {
-      "epoch": 0.6070694134140281,
-      "grad_norm": 637.5480346679688,
-      "learning_rate": 2.056944604440751e-05,
-      "loss": 61.4114,
-      "step": 150260
-    },
-    {
-      "epoch": 0.6071098146793957,
-      "grad_norm": 895.0432739257812,
-      "learning_rate": 2.056601068318054e-05,
-      "loss": 39.7264,
-      "step": 150270
-    },
-    {
-      "epoch": 0.6071502159447634,
-      "grad_norm": 600.3475952148438,
-      "learning_rate": 2.056257540839647e-05,
-      "loss": 47.7371,
-      "step": 150280
-    },
-    {
-      "epoch": 0.607190617210131,
-      "grad_norm": 985.7539672851562,
-      "learning_rate": 2.055914022012227e-05,
-      "loss": 68.4636,
-      "step": 150290
-    },
-    {
-      "epoch": 0.6072310184754987,
-      "grad_norm": 838.5362548828125,
-      "learning_rate": 2.0555705118424927e-05,
-      "loss": 39.3113,
-      "step": 150300
-    },
-    {
-      "epoch": 0.6072714197408663,
-      "grad_norm": 802.4273071289062,
-      "learning_rate": 2.05522701033714e-05,
-      "loss": 54.2198,
-      "step": 150310
-    },
-    {
-      "epoch": 0.6073118210062339,
-      "grad_norm": 535.5287475585938,
-      "learning_rate": 2.0548835175028647e-05,
-      "loss": 43.866,
-      "step": 150320
-    },
-    {
-      "epoch": 0.6073522222716016,
-      "grad_norm": 556.2651977539062,
-      "learning_rate": 2.0545400333463656e-05,
-      "loss": 28.3161,
-      "step": 150330
-    },
-    {
-      "epoch": 0.6073926235369692,
-      "grad_norm": 589.1482543945312,
-      "learning_rate": 2.054196557874337e-05,
-      "loss": 63.3741,
-      "step": 150340
-    },
-    {
-      "epoch": 0.6074330248023369,
-      "grad_norm": 1289.090576171875,
-      "learning_rate": 2.0538530910934768e-05,
-      "loss": 53.0286,
-      "step": 150350
-    },
-    {
-      "epoch": 0.6074734260677045,
-      "grad_norm": 502.2218322753906,
-      "learning_rate": 2.0535096330104804e-05,
-      "loss": 45.3578,
-      "step": 150360
-    },
-    {
-      "epoch": 0.607513827333072,
-      "grad_norm": 598.0867309570312,
-      "learning_rate": 2.0531661836320422e-05,
-      "loss": 39.5945,
-      "step": 150370
-    },
-    {
-      "epoch": 0.6075542285984397,
-      "grad_norm": 724.6481323242188,
-      "learning_rate": 2.0528227429648604e-05,
-      "loss": 39.6854,
-      "step": 150380
-    },
-    {
-      "epoch": 0.6075946298638073,
-      "grad_norm": 962.4512939453125,
-      "learning_rate": 2.0524793110156297e-05,
-      "loss": 86.767,
-      "step": 150390
-    },
-    {
-      "epoch": 0.6076350311291749,
-      "grad_norm": 635.9224853515625,
-      "learning_rate": 2.0521358877910444e-05,
-      "loss": 44.4046,
-      "step": 150400
-    },
-    {
-      "epoch": 0.6076754323945426,
-      "grad_norm": 891.631103515625,
-      "learning_rate": 2.051792473297801e-05,
-      "loss": 49.1075,
-      "step": 150410
-    },
-    {
-      "epoch": 0.6077158336599102,
-      "grad_norm": 374.13427734375,
-      "learning_rate": 2.051449067542594e-05,
-      "loss": 31.0687,
-      "step": 150420
-    },
-    {
-      "epoch": 0.6077562349252779,
-      "grad_norm": 596.3746948242188,
-      "learning_rate": 2.0511056705321185e-05,
-      "loss": 47.9743,
-      "step": 150430
-    },
-    {
-      "epoch": 0.6077966361906455,
-      "grad_norm": 262.538818359375,
-      "learning_rate": 2.0507622822730695e-05,
-      "loss": 42.9091,
-      "step": 150440
-    },
-    {
-      "epoch": 0.6078370374560131,
-      "grad_norm": 1468.2989501953125,
-      "learning_rate": 2.0504189027721395e-05,
-      "loss": 57.1524,
-      "step": 150450
-    },
-    {
-      "epoch": 0.6078774387213808,
-      "grad_norm": 646.4707641601562,
-      "learning_rate": 2.050075532036026e-05,
-      "loss": 67.7372,
-      "step": 150460
-    },
-    {
-      "epoch": 0.6079178399867484,
-      "grad_norm": 5379.966796875,
-      "learning_rate": 2.0497321700714204e-05,
-      "loss": 67.6201,
-      "step": 150470
-    },
-    {
-      "epoch": 0.6079582412521161,
-      "grad_norm": 717.5620727539062,
-      "learning_rate": 2.0493888168850188e-05,
-      "loss": 44.0507,
-      "step": 150480
-    },
-    {
-      "epoch": 0.6079986425174837,
-      "grad_norm": 1433.787109375,
-      "learning_rate": 2.0490454724835147e-05,
-      "loss": 84.3167,
-      "step": 150490
-    },
-    {
-      "epoch": 0.6080390437828512,
-      "grad_norm": 862.51611328125,
-      "learning_rate": 2.0487021368736003e-05,
-      "loss": 47.1453,
-      "step": 150500
-    },
-    {
-      "epoch": 0.6080794450482189,
-      "grad_norm": 1338.2392578125,
-      "learning_rate": 2.0483588100619707e-05,
-      "loss": 59.3567,
-      "step": 150510
-    },
-    {
-      "epoch": 0.6081198463135865,
-      "grad_norm": 1019.3951416015625,
-      "learning_rate": 2.0480154920553186e-05,
-      "loss": 41.7931,
-      "step": 150520
-    },
-    {
-      "epoch": 0.6081602475789541,
-      "grad_norm": 778.1670532226562,
-      "learning_rate": 2.047672182860336e-05,
-      "loss": 56.9556,
-      "step": 150530
-    },
-    {
-      "epoch": 0.6082006488443218,
-      "grad_norm": 594.0452880859375,
-      "learning_rate": 2.0473288824837187e-05,
-      "loss": 45.3777,
-      "step": 150540
-    },
-    {
-      "epoch": 0.6082410501096894,
-      "grad_norm": 1647.530029296875,
-      "learning_rate": 2.0469855909321564e-05,
-      "loss": 58.5577,
-      "step": 150550
-    },
-    {
-      "epoch": 0.6082814513750571,
-      "grad_norm": 654.25390625,
-      "learning_rate": 2.0466423082123443e-05,
-      "loss": 39.5328,
-      "step": 150560
-    },
-    {
-      "epoch": 0.6083218526404247,
-      "grad_norm": 563.961181640625,
-      "learning_rate": 2.046299034330974e-05,
-      "loss": 56.6749,
-      "step": 150570
-    },
-    {
-      "epoch": 0.6083622539057923,
-      "grad_norm": 900.1375732421875,
-      "learning_rate": 2.0459557692947367e-05,
-      "loss": 40.388,
-      "step": 150580
-    },
-    {
-      "epoch": 0.60840265517116,
-      "grad_norm": 814.3180541992188,
-      "learning_rate": 2.0456125131103255e-05,
-      "loss": 46.5721,
-      "step": 150590
-    },
-    {
-      "epoch": 0.6084430564365276,
-      "grad_norm": 1955.17919921875,
-      "learning_rate": 2.0452692657844333e-05,
-      "loss": 53.1568,
-      "step": 150600
-    },
-    {
-      "epoch": 0.6084834577018953,
-      "grad_norm": 2265.74951171875,
-      "learning_rate": 2.04492602732375e-05,
-      "loss": 68.3158,
-      "step": 150610
-    },
-    {
-      "epoch": 0.6085238589672629,
-      "grad_norm": 405.62066650390625,
-      "learning_rate": 2.0445827977349685e-05,
-      "loss": 41.4512,
-      "step": 150620
-    },
-    {
-      "epoch": 0.6085642602326304,
-      "grad_norm": 762.5441284179688,
-      "learning_rate": 2.044239577024779e-05,
-      "loss": 44.7131,
-      "step": 150630
-    },
-    {
-      "epoch": 0.6086046614979981,
-      "grad_norm": 1103.0228271484375,
-      "learning_rate": 2.0438963651998747e-05,
-      "loss": 59.9257,
-      "step": 150640
-    },
-    {
-      "epoch": 0.6086450627633657,
-      "grad_norm": 773.6843872070312,
-      "learning_rate": 2.0435531622669453e-05,
-      "loss": 47.6074,
-      "step": 150650
-    },
-    {
-      "epoch": 0.6086854640287334,
-      "grad_norm": 500.89013671875,
-      "learning_rate": 2.0432099682326812e-05,
-      "loss": 32.2808,
-      "step": 150660
-    },
-    {
-      "epoch": 0.608725865294101,
-      "grad_norm": 907.202392578125,
-      "learning_rate": 2.042866783103775e-05,
-      "loss": 49.861,
-      "step": 150670
-    },
-    {
-      "epoch": 0.6087662665594686,
-      "grad_norm": 459.9265441894531,
-      "learning_rate": 2.042523606886916e-05,
-      "loss": 34.1603,
-      "step": 150680
-    },
-    {
-      "epoch": 0.6088066678248363,
-      "grad_norm": 915.3919067382812,
-      "learning_rate": 2.0421804395887942e-05,
-      "loss": 58.615,
-      "step": 150690
-    },
-    {
-      "epoch": 0.6088470690902039,
-      "grad_norm": 642.7056274414062,
-      "learning_rate": 2.0418372812161012e-05,
-      "loss": 65.4354,
-      "step": 150700
-    },
-    {
-      "epoch": 0.6088874703555716,
-      "grad_norm": 433.1236572265625,
-      "learning_rate": 2.041494131775525e-05,
-      "loss": 40.9949,
-      "step": 150710
-    },
-    {
-      "epoch": 0.6089278716209392,
-      "grad_norm": 1081.02880859375,
-      "learning_rate": 2.041150991273758e-05,
-      "loss": 43.7319,
-      "step": 150720
-    },
-    {
-      "epoch": 0.6089682728863068,
-      "grad_norm": 701.385498046875,
-      "learning_rate": 2.0408078597174886e-05,
-      "loss": 32.3251,
-      "step": 150730
-    },
-    {
-      "epoch": 0.6090086741516745,
-      "grad_norm": 731.4369506835938,
-      "learning_rate": 2.0404647371134055e-05,
-      "loss": 45.6466,
-      "step": 150740
-    },
-    {
-      "epoch": 0.609049075417042,
-      "grad_norm": 523.586181640625,
-      "learning_rate": 2.0401216234681995e-05,
-      "loss": 27.7691,
-      "step": 150750
-    },
-    {
-      "epoch": 0.6090894766824096,
-      "grad_norm": 551.3516845703125,
-      "learning_rate": 2.0397785187885598e-05,
-      "loss": 48.8377,
-      "step": 150760
-    },
-    {
-      "epoch": 0.6091298779477773,
-      "grad_norm": 513.434814453125,
-      "learning_rate": 2.039435423081174e-05,
-      "loss": 51.5516,
-      "step": 150770
-    },
-    {
-      "epoch": 0.6091702792131449,
-      "grad_norm": 1080.2301025390625,
-      "learning_rate": 2.039092336352732e-05,
-      "loss": 75.6299,
-      "step": 150780
-    },
-    {
-      "epoch": 0.6092106804785126,
-      "grad_norm": 568.66796875,
-      "learning_rate": 2.038749258609922e-05,
-      "loss": 44.2861,
-      "step": 150790
-    },
-    {
-      "epoch": 0.6092510817438802,
-      "grad_norm": 1709.7052001953125,
-      "learning_rate": 2.038406189859433e-05,
-      "loss": 62.1356,
-      "step": 150800
-    },
-    {
-      "epoch": 0.6092914830092478,
-      "grad_norm": 1051.6448974609375,
-      "learning_rate": 2.0380631301079532e-05,
-      "loss": 52.0556,
-      "step": 150810
-    },
-    {
-      "epoch": 0.6093318842746155,
-      "grad_norm": 643.7091674804688,
-      "learning_rate": 2.037720079362169e-05,
-      "loss": 34.8519,
-      "step": 150820
-    },
-    {
-      "epoch": 0.6093722855399831,
-      "grad_norm": 995.33642578125,
-      "learning_rate": 2.0373770376287715e-05,
-      "loss": 45.4539,
-      "step": 150830
-    },
-    {
-      "epoch": 0.6094126868053508,
-      "grad_norm": 871.22705078125,
-      "learning_rate": 2.037034004914447e-05,
-      "loss": 86.5264,
-      "step": 150840
-    },
-    {
-      "epoch": 0.6094530880707184,
-      "grad_norm": 291.0818786621094,
-      "learning_rate": 2.0366909812258817e-05,
-      "loss": 45.437,
-      "step": 150850
-    },
-    {
-      "epoch": 0.609493489336086,
-      "grad_norm": 370.9341125488281,
-      "learning_rate": 2.0363479665697652e-05,
-      "loss": 41.5995,
-      "step": 150860
-    },
-    {
-      "epoch": 0.6095338906014537,
-      "grad_norm": 993.7173461914062,
-      "learning_rate": 2.0360049609527825e-05,
-      "loss": 33.4351,
-      "step": 150870
-    },
-    {
-      "epoch": 0.6095742918668212,
-      "grad_norm": 269.01416015625,
-      "learning_rate": 2.0356619643816234e-05,
-      "loss": 35.0631,
-      "step": 150880
-    },
-    {
-      "epoch": 0.6096146931321889,
-      "grad_norm": 5912.23828125,
-      "learning_rate": 2.035318976862973e-05,
-      "loss": 54.3141,
-      "step": 150890
-    },
-    {
-      "epoch": 0.6096550943975565,
-      "grad_norm": 536.7326049804688,
-      "learning_rate": 2.034975998403517e-05,
-      "loss": 62.3261,
-      "step": 150900
-    },
-    {
-      "epoch": 0.6096954956629241,
-      "grad_norm": 735.4278564453125,
-      "learning_rate": 2.034633029009945e-05,
-      "loss": 50.3085,
-      "step": 150910
-    },
-    {
-      "epoch": 0.6097358969282918,
-      "grad_norm": 1243.7513427734375,
-      "learning_rate": 2.034290068688941e-05,
-      "loss": 58.6372,
-      "step": 150920
-    },
-    {
-      "epoch": 0.6097762981936594,
-      "grad_norm": 1334.8585205078125,
-      "learning_rate": 2.033947117447192e-05,
-      "loss": 61.1152,
-      "step": 150930
-    },
-    {
-      "epoch": 0.6098166994590271,
-      "grad_norm": 1525.87939453125,
-      "learning_rate": 2.0336041752913843e-05,
-      "loss": 51.0284,
-      "step": 150940
-    },
-    {
-      "epoch": 0.6098571007243947,
-      "grad_norm": 557.0948486328125,
-      "learning_rate": 2.0332612422282027e-05,
-      "loss": 64.3433,
-      "step": 150950
-    },
-    {
-      "epoch": 0.6098975019897623,
-      "grad_norm": 390.4165344238281,
-      "learning_rate": 2.032918318264334e-05,
-      "loss": 60.4239,
-      "step": 150960
-    },
-    {
-      "epoch": 0.60993790325513,
-      "grad_norm": 876.21044921875,
-      "learning_rate": 2.032575403406463e-05,
-      "loss": 41.9331,
-      "step": 150970
-    },
-    {
-      "epoch": 0.6099783045204976,
-      "grad_norm": 296.02685546875,
-      "learning_rate": 2.0322324976612745e-05,
-      "loss": 48.232,
-      "step": 150980
-    },
-    {
-      "epoch": 0.6100187057858653,
-      "grad_norm": 633.3362426757812,
-      "learning_rate": 2.0318896010354553e-05,
-      "loss": 58.8014,
-      "step": 150990
-    },
-    {
-      "epoch": 0.6100591070512329,
-      "grad_norm": 716.9625244140625,
-      "learning_rate": 2.031546713535688e-05,
-      "loss": 63.4572,
-      "step": 151000
-    },
-    {
-      "epoch": 0.6100995083166004,
-      "grad_norm": 639.0398559570312,
-      "learning_rate": 2.0312038351686598e-05,
-      "loss": 51.1365,
-      "step": 151010
-    },
-    {
-      "epoch": 0.6101399095819681,
-      "grad_norm": 1237.0989990234375,
-      "learning_rate": 2.0308609659410544e-05,
-      "loss": 52.9566,
-      "step": 151020
-    },
-    {
-      "epoch": 0.6101803108473357,
-      "grad_norm": 675.1341552734375,
-      "learning_rate": 2.0305181058595552e-05,
-      "loss": 57.4781,
-      "step": 151030
-    },
-    {
-      "epoch": 0.6102207121127033,
-      "grad_norm": 1143.0931396484375,
-      "learning_rate": 2.030175254930848e-05,
-      "loss": 49.0002,
-      "step": 151040
-    },
-    {
-      "epoch": 0.610261113378071,
-      "grad_norm": 784.3997802734375,
-      "learning_rate": 2.0298324131616158e-05,
-      "loss": 55.3003,
-      "step": 151050
-    },
-    {
-      "epoch": 0.6103015146434386,
-      "grad_norm": 360.4586181640625,
-      "learning_rate": 2.029489580558542e-05,
-      "loss": 37.6115,
-      "step": 151060
-    },
-    {
-      "epoch": 0.6103419159088063,
-      "grad_norm": 2819.0810546875,
-      "learning_rate": 2.029146757128312e-05,
-      "loss": 59.4453,
-      "step": 151070
-    },
-    {
-      "epoch": 0.6103823171741739,
-      "grad_norm": 975.2780151367188,
-      "learning_rate": 2.0288039428776073e-05,
-      "loss": 62.0133,
-      "step": 151080
-    },
-    {
-      "epoch": 0.6104227184395415,
-      "grad_norm": 1327.5264892578125,
-      "learning_rate": 2.0284611378131136e-05,
-      "loss": 67.0731,
-      "step": 151090
-    },
-    {
-      "epoch": 0.6104631197049092,
-      "grad_norm": 405.42041015625,
-      "learning_rate": 2.0281183419415125e-05,
-      "loss": 48.7321,
-      "step": 151100
-    },
-    {
-      "epoch": 0.6105035209702768,
-      "grad_norm": 507.772705078125,
-      "learning_rate": 2.0277755552694867e-05,
-      "loss": 53.5197,
-      "step": 151110
-    },
-    {
-      "epoch": 0.6105439222356445,
-      "grad_norm": 566.7341918945312,
-      "learning_rate": 2.02743277780372e-05,
-      "loss": 45.4196,
-      "step": 151120
-    },
-    {
-      "epoch": 0.6105843235010121,
-      "grad_norm": 907.8681030273438,
-      "learning_rate": 2.027090009550895e-05,
-      "loss": 43.463,
-      "step": 151130
-    },
-    {
-      "epoch": 0.6106247247663796,
-      "grad_norm": 753.9053955078125,
-      "learning_rate": 2.0267472505176927e-05,
-      "loss": 51.493,
-      "step": 151140
-    },
-    {
-      "epoch": 0.6106651260317473,
-      "grad_norm": 456.0301818847656,
-      "learning_rate": 2.0264045007107973e-05,
-      "loss": 58.2713,
-      "step": 151150
-    },
-    {
-      "epoch": 0.6107055272971149,
-      "grad_norm": 894.2009887695312,
-      "learning_rate": 2.0260617601368886e-05,
-      "loss": 64.6016,
-      "step": 151160
-    },
-    {
-      "epoch": 0.6107459285624826,
-      "grad_norm": 613.248046875,
-      "learning_rate": 2.0257190288026516e-05,
-      "loss": 46.2595,
-      "step": 151170
-    },
-    {
-      "epoch": 0.6107863298278502,
-      "grad_norm": 976.0096435546875,
-      "learning_rate": 2.0253763067147657e-05,
-      "loss": 60.338,
-      "step": 151180
-    },
-    {
-      "epoch": 0.6108267310932178,
-      "grad_norm": 844.7362060546875,
-      "learning_rate": 2.025033593879913e-05,
-      "loss": 41.8549,
-      "step": 151190
-    },
-    {
-      "epoch": 0.6108671323585855,
-      "grad_norm": 1008.4219970703125,
-      "learning_rate": 2.024690890304775e-05,
-      "loss": 67.3454,
-      "step": 151200
-    },
-    {
-      "epoch": 0.6109075336239531,
-      "grad_norm": 652.0379028320312,
-      "learning_rate": 2.0243481959960327e-05,
-      "loss": 60.3732,
-      "step": 151210
-    },
-    {
-      "epoch": 0.6109479348893208,
-      "grad_norm": 252.4768829345703,
-      "learning_rate": 2.0240055109603668e-05,
-      "loss": 43.6406,
-      "step": 151220
-    },
-    {
-      "epoch": 0.6109883361546884,
-      "grad_norm": 456.96221923828125,
-      "learning_rate": 2.023662835204459e-05,
-      "loss": 54.4126,
-      "step": 151230
-    },
-    {
-      "epoch": 0.611028737420056,
-      "grad_norm": 1217.955322265625,
-      "learning_rate": 2.0233201687349887e-05,
-      "loss": 74.467,
-      "step": 151240
-    },
-    {
-      "epoch": 0.6110691386854237,
-      "grad_norm": 1050.6669921875,
-      "learning_rate": 2.022977511558638e-05,
-      "loss": 53.1463,
-      "step": 151250
-    },
-    {
-      "epoch": 0.6111095399507913,
-      "grad_norm": 345.2624206542969,
-      "learning_rate": 2.0226348636820865e-05,
-      "loss": 43.0883,
-      "step": 151260
-    },
-    {
-      "epoch": 0.6111499412161588,
-      "grad_norm": 811.0917358398438,
-      "learning_rate": 2.022292225112013e-05,
-      "loss": 46.1009,
-      "step": 151270
-    },
-    {
-      "epoch": 0.6111903424815265,
-      "grad_norm": 419.13983154296875,
-      "learning_rate": 2.0219495958550992e-05,
-      "loss": 42.6573,
-      "step": 151280
-    },
-    {
-      "epoch": 0.6112307437468941,
-      "grad_norm": 572.9585571289062,
-      "learning_rate": 2.0216069759180248e-05,
-      "loss": 64.8317,
-      "step": 151290
-    },
-    {
-      "epoch": 0.6112711450122618,
-      "grad_norm": 1485.81640625,
-      "learning_rate": 2.021264365307468e-05,
-      "loss": 63.413,
-      "step": 151300
-    },
-    {
-      "epoch": 0.6113115462776294,
-      "grad_norm": 856.283447265625,
-      "learning_rate": 2.0209217640301088e-05,
-      "loss": 40.8531,
-      "step": 151310
-    },
-    {
-      "epoch": 0.611351947542997,
-      "grad_norm": 429.21636962890625,
-      "learning_rate": 2.020579172092626e-05,
-      "loss": 75.7096,
-      "step": 151320
-    },
-    {
-      "epoch": 0.6113923488083647,
-      "grad_norm": 770.0364990234375,
-      "learning_rate": 2.0202365895017e-05,
-      "loss": 90.2186,
-      "step": 151330
-    },
-    {
-      "epoch": 0.6114327500737323,
-      "grad_norm": 775.0713500976562,
-      "learning_rate": 2.0198940162640084e-05,
-      "loss": 55.38,
-      "step": 151340
-    },
-    {
-      "epoch": 0.6114731513391,
-      "grad_norm": 341.93707275390625,
-      "learning_rate": 2.0195514523862293e-05,
-      "loss": 43.4222,
-      "step": 151350
-    },
-    {
-      "epoch": 0.6115135526044676,
-      "grad_norm": 1337.871826171875,
-      "learning_rate": 2.0192088978750433e-05,
-      "loss": 59.2981,
-      "step": 151360
-    },
-    {
-      "epoch": 0.6115539538698352,
-      "grad_norm": 847.0081787109375,
-      "learning_rate": 2.0188663527371272e-05,
-      "loss": 53.3189,
-      "step": 151370
-    },
-    {
-      "epoch": 0.6115943551352029,
-      "grad_norm": 903.5023193359375,
-      "learning_rate": 2.0185238169791585e-05,
-      "loss": 46.3016,
-      "step": 151380
-    },
-    {
-      "epoch": 0.6116347564005704,
-      "grad_norm": 691.9161376953125,
-      "learning_rate": 2.0181812906078164e-05,
-      "loss": 55.7604,
-      "step": 151390
-    },
-    {
-      "epoch": 0.611675157665938,
-      "grad_norm": 793.9723510742188,
-      "learning_rate": 2.0178387736297773e-05,
-      "loss": 62.8434,
-      "step": 151400
-    },
-    {
-      "epoch": 0.6117155589313057,
-      "grad_norm": 764.1298217773438,
-      "learning_rate": 2.0174962660517206e-05,
-      "loss": 71.301,
-      "step": 151410
-    },
-    {
-      "epoch": 0.6117559601966733,
-      "grad_norm": 990.5913696289062,
-      "learning_rate": 2.0171537678803225e-05,
-      "loss": 42.5648,
-      "step": 151420
-    },
-    {
-      "epoch": 0.611796361462041,
-      "grad_norm": 670.5265502929688,
-      "learning_rate": 2.016811279122259e-05,
-      "loss": 48.3127,
-      "step": 151430
-    },
-    {
-      "epoch": 0.6118367627274086,
-      "grad_norm": 597.8585815429688,
-      "learning_rate": 2.0164687997842096e-05,
-      "loss": 43.7411,
-      "step": 151440
-    },
-    {
-      "epoch": 0.6118771639927763,
-      "grad_norm": 533.0531616210938,
-      "learning_rate": 2.0161263298728495e-05,
-      "loss": 50.8494,
-      "step": 151450
-    },
-    {
-      "epoch": 0.6119175652581439,
-      "grad_norm": 334.20184326171875,
-      "learning_rate": 2.015783869394856e-05,
-      "loss": 52.9698,
-      "step": 151460
-    },
-    {
-      "epoch": 0.6119579665235115,
-      "grad_norm": 557.2747802734375,
-      "learning_rate": 2.0154414183569055e-05,
-      "loss": 31.3677,
-      "step": 151470
-    },
-    {
-      "epoch": 0.6119983677888792,
-      "grad_norm": 513.8084106445312,
-      "learning_rate": 2.0150989767656728e-05,
-      "loss": 67.8341,
-      "step": 151480
-    },
-    {
-      "epoch": 0.6120387690542468,
-      "grad_norm": 775.8402099609375,
-      "learning_rate": 2.0147565446278364e-05,
-      "loss": 54.9718,
-      "step": 151490
-    },
-    {
-      "epoch": 0.6120791703196145,
-      "grad_norm": 209.1566925048828,
-      "learning_rate": 2.0144141219500705e-05,
-      "loss": 34.7493,
-      "step": 151500
-    },
-    {
-      "epoch": 0.6121195715849821,
-      "grad_norm": 761.5418090820312,
-      "learning_rate": 2.014071708739051e-05,
-      "loss": 52.2124,
-      "step": 151510
-    },
-    {
-      "epoch": 0.6121599728503496,
-      "grad_norm": 764.025634765625,
-      "learning_rate": 2.013729305001454e-05,
-      "loss": 57.659,
-      "step": 151520
-    },
-    {
-      "epoch": 0.6122003741157173,
-      "grad_norm": 160.14700317382812,
-      "learning_rate": 2.0133869107439545e-05,
-      "loss": 68.8839,
-      "step": 151530
-    },
-    {
-      "epoch": 0.6122407753810849,
-      "grad_norm": 546.741943359375,
-      "learning_rate": 2.0130445259732285e-05,
-      "loss": 48.702,
-      "step": 151540
-    },
-    {
-      "epoch": 0.6122811766464525,
-      "grad_norm": 556.5968627929688,
-      "learning_rate": 2.0127021506959488e-05,
-      "loss": 34.9057,
-      "step": 151550
-    },
-    {
-      "epoch": 0.6123215779118202,
-      "grad_norm": 1158.7283935546875,
-      "learning_rate": 2.012359784918792e-05,
-      "loss": 31.9852,
-      "step": 151560
-    },
-    {
-      "epoch": 0.6123619791771878,
-      "grad_norm": 207.57945251464844,
-      "learning_rate": 2.012017428648433e-05,
-      "loss": 44.4616,
-      "step": 151570
-    },
-    {
-      "epoch": 0.6124023804425555,
-      "grad_norm": 1311.223388671875,
-      "learning_rate": 2.011675081891545e-05,
-      "loss": 46.3915,
-      "step": 151580
-    },
-    {
-      "epoch": 0.6124427817079231,
-      "grad_norm": 539.1629028320312,
-      "learning_rate": 2.011332744654802e-05,
-      "loss": 61.3504,
-      "step": 151590
-    },
-    {
-      "epoch": 0.6124831829732907,
-      "grad_norm": 905.910400390625,
-      "learning_rate": 2.01099041694488e-05,
-      "loss": 50.0863,
-      "step": 151600
-    },
-    {
-      "epoch": 0.6125235842386584,
-      "grad_norm": 319.6258850097656,
-      "learning_rate": 2.01064809876845e-05,
-      "loss": 50.1784,
-      "step": 151610
-    },
-    {
-      "epoch": 0.612563985504026,
-      "grad_norm": 879.3612670898438,
-      "learning_rate": 2.0103057901321884e-05,
-      "loss": 63.5632,
-      "step": 151620
-    },
-    {
-      "epoch": 0.6126043867693937,
-      "grad_norm": 445.58953857421875,
-      "learning_rate": 2.0099634910427678e-05,
-      "loss": 31.5658,
-      "step": 151630
-    },
-    {
-      "epoch": 0.6126447880347613,
-      "grad_norm": 766.837890625,
-      "learning_rate": 2.0096212015068606e-05,
-      "loss": 48.1527,
-      "step": 151640
-    },
-    {
-      "epoch": 0.6126851893001288,
-      "grad_norm": 1137.3616943359375,
-      "learning_rate": 2.009278921531141e-05,
-      "loss": 78.9928,
-      "step": 151650
-    },
-    {
-      "epoch": 0.6127255905654965,
-      "grad_norm": 296.29620361328125,
-      "learning_rate": 2.0089366511222813e-05,
-      "loss": 56.172,
-      "step": 151660
-    },
-    {
-      "epoch": 0.6127659918308641,
-      "grad_norm": 926.5879516601562,
-      "learning_rate": 2.0085943902869537e-05,
-      "loss": 48.9301,
-      "step": 151670
-    },
-    {
-      "epoch": 0.6128063930962317,
-      "grad_norm": 633.228271484375,
-      "learning_rate": 2.0082521390318322e-05,
-      "loss": 41.1792,
-      "step": 151680
-    },
-    {
-      "epoch": 0.6128467943615994,
-      "grad_norm": 701.0936279296875,
-      "learning_rate": 2.0079098973635872e-05,
-      "loss": 33.9974,
-      "step": 151690
-    },
-    {
-      "epoch": 0.612887195626967,
-      "grad_norm": 275.7187805175781,
-      "learning_rate": 2.0075676652888936e-05,
-      "loss": 48.2116,
-      "step": 151700
-    },
-    {
-      "epoch": 0.6129275968923347,
-      "grad_norm": 1368.0904541015625,
-      "learning_rate": 2.0072254428144216e-05,
-      "loss": 64.7694,
-      "step": 151710
-    },
-    {
-      "epoch": 0.6129679981577023,
-      "grad_norm": 623.418701171875,
-      "learning_rate": 2.0068832299468428e-05,
-      "loss": 42.1006,
-      "step": 151720
-    },
-    {
-      "epoch": 0.61300839942307,
-      "grad_norm": 870.45556640625,
-      "learning_rate": 2.00654102669283e-05,
-      "loss": 47.7842,
-      "step": 151730
-    },
-    {
-      "epoch": 0.6130488006884376,
-      "grad_norm": 588.7037353515625,
-      "learning_rate": 2.0061988330590538e-05,
-      "loss": 83.3482,
-      "step": 151740
-    },
-    {
-      "epoch": 0.6130892019538052,
-      "grad_norm": 657.9315795898438,
-      "learning_rate": 2.0058566490521847e-05,
-      "loss": 43.4999,
-      "step": 151750
-    },
-    {
-      "epoch": 0.6131296032191729,
-      "grad_norm": 708.498779296875,
-      "learning_rate": 2.0055144746788957e-05,
-      "loss": 68.977,
-      "step": 151760
-    },
-    {
-      "epoch": 0.6131700044845405,
-      "grad_norm": 1133.77490234375,
-      "learning_rate": 2.0051723099458554e-05,
-      "loss": 48.2999,
-      "step": 151770
-    },
-    {
-      "epoch": 0.613210405749908,
-      "grad_norm": 302.3292236328125,
-      "learning_rate": 2.0048301548597363e-05,
-      "loss": 40.6427,
-      "step": 151780
-    },
-    {
-      "epoch": 0.6132508070152757,
-      "grad_norm": 404.0072937011719,
-      "learning_rate": 2.004488009427209e-05,
-      "loss": 32.1956,
-      "step": 151790
-    },
-    {
-      "epoch": 0.6132912082806433,
-      "grad_norm": 1267.0899658203125,
-      "learning_rate": 2.004145873654942e-05,
-      "loss": 67.8385,
-      "step": 151800
-    },
-    {
-      "epoch": 0.613331609546011,
-      "grad_norm": 446.0968322753906,
-      "learning_rate": 2.0038037475496075e-05,
-      "loss": 56.1292,
-      "step": 151810
-    },
-    {
-      "epoch": 0.6133720108113786,
-      "grad_norm": 1070.377197265625,
-      "learning_rate": 2.003461631117874e-05,
-      "loss": 60.586,
-      "step": 151820
-    },
-    {
-      "epoch": 0.6134124120767462,
-      "grad_norm": 430.245361328125,
-      "learning_rate": 2.003119524366411e-05,
-      "loss": 82.9556,
-      "step": 151830
-    },
-    {
-      "epoch": 0.6134528133421139,
-      "grad_norm": 566.5420532226562,
-      "learning_rate": 2.0027774273018892e-05,
-      "loss": 35.1446,
-      "step": 151840
-    },
-    {
-      "epoch": 0.6134932146074815,
-      "grad_norm": 570.1262817382812,
-      "learning_rate": 2.0024353399309765e-05,
-      "loss": 46.2369,
-      "step": 151850
-    },
-    {
-      "epoch": 0.6135336158728492,
-      "grad_norm": 409.9158935546875,
-      "learning_rate": 2.0020932622603444e-05,
-      "loss": 62.6959,
-      "step": 151860
-    },
-    {
-      "epoch": 0.6135740171382168,
-      "grad_norm": 869.2930908203125,
-      "learning_rate": 2.00175119429666e-05,
-      "loss": 42.3853,
-      "step": 151870
-    },
-    {
-      "epoch": 0.6136144184035844,
-      "grad_norm": 733.9028930664062,
-      "learning_rate": 2.0014091360465927e-05,
-      "loss": 90.8488,
-      "step": 151880
-    },
-    {
-      "epoch": 0.6136548196689521,
-      "grad_norm": 806.4086303710938,
-      "learning_rate": 2.00106708751681e-05,
-      "loss": 73.2679,
-      "step": 151890
-    },
-    {
-      "epoch": 0.6136952209343196,
-      "grad_norm": 1075.563232421875,
-      "learning_rate": 2.000725048713983e-05,
-      "loss": 41.3077,
-      "step": 151900
-    },
-    {
-      "epoch": 0.6137356221996872,
-      "grad_norm": 1005.3833618164062,
-      "learning_rate": 2.000383019644777e-05,
-      "loss": 66.0867,
-      "step": 151910
-    },
-    {
-      "epoch": 0.6137760234650549,
-      "grad_norm": 1066.146484375,
-      "learning_rate": 2.000041000315862e-05,
-      "loss": 43.0172,
-      "step": 151920
-    },
-    {
-      "epoch": 0.6138164247304225,
-      "grad_norm": 379.4106140136719,
-      "learning_rate": 1.999698990733904e-05,
-      "loss": 25.9617,
-      "step": 151930
-    },
-    {
-      "epoch": 0.6138568259957902,
-      "grad_norm": 494.0150146484375,
-      "learning_rate": 1.9993569909055725e-05,
-      "loss": 57.3421,
-      "step": 151940
-    },
-    {
-      "epoch": 0.6138972272611578,
-      "grad_norm": 467.149169921875,
-      "learning_rate": 1.9990150008375347e-05,
-      "loss": 41.3854,
-      "step": 151950
-    },
-    {
-      "epoch": 0.6139376285265254,
-      "grad_norm": 1409.2052001953125,
-      "learning_rate": 1.998673020536456e-05,
-      "loss": 58.1735,
-      "step": 151960
-    },
-    {
-      "epoch": 0.6139780297918931,
-      "grad_norm": 736.8629760742188,
-      "learning_rate": 1.998331050009006e-05,
-      "loss": 52.9153,
-      "step": 151970
-    },
-    {
-      "epoch": 0.6140184310572607,
-      "grad_norm": 791.1087036132812,
-      "learning_rate": 1.99798908926185e-05,
-      "loss": 64.018,
-      "step": 151980
-    },
-    {
-      "epoch": 0.6140588323226284,
-      "grad_norm": 990.827392578125,
-      "learning_rate": 1.9976471383016557e-05,
-      "loss": 56.9068,
-      "step": 151990
-    },
-    {
-      "epoch": 0.614099233587996,
-      "grad_norm": 720.25390625,
-      "learning_rate": 1.9973051971350888e-05,
-      "loss": 59.2837,
-      "step": 152000
-    },
-    {
-      "epoch": 0.6141396348533636,
-      "grad_norm": 877.4662475585938,
-      "learning_rate": 1.9969632657688155e-05,
-      "loss": 48.7033,
-      "step": 152010
-    },
-    {
-      "epoch": 0.6141800361187313,
-      "grad_norm": 656.0659790039062,
-      "learning_rate": 1.9966213442095028e-05,
-      "loss": 57.7444,
-      "step": 152020
-    },
-    {
-      "epoch": 0.6142204373840988,
-      "grad_norm": 1309.908447265625,
-      "learning_rate": 1.9962794324638162e-05,
-      "loss": 72.4926,
-      "step": 152030
-    },
-    {
-      "epoch": 0.6142608386494665,
-      "grad_norm": 421.1377868652344,
-      "learning_rate": 1.9959375305384203e-05,
-      "loss": 39.82,
-      "step": 152040
-    },
-    {
-      "epoch": 0.6143012399148341,
-      "grad_norm": 892.0406494140625,
-      "learning_rate": 1.9955956384399828e-05,
-      "loss": 65.1622,
-      "step": 152050
-    },
-    {
-      "epoch": 0.6143416411802017,
-      "grad_norm": 984.6168823242188,
-      "learning_rate": 1.995253756175168e-05,
-      "loss": 65.7813,
-      "step": 152060
-    },
-    {
-      "epoch": 0.6143820424455694,
-      "grad_norm": 821.1974487304688,
-      "learning_rate": 1.994911883750641e-05,
-      "loss": 57.6058,
-      "step": 152070
-    },
-    {
-      "epoch": 0.614422443710937,
-      "grad_norm": 419.92474365234375,
-      "learning_rate": 1.994570021173067e-05,
-      "loss": 69.8375,
-      "step": 152080
-    },
-    {
-      "epoch": 0.6144628449763047,
-      "grad_norm": 415.35400390625,
-      "learning_rate": 1.9942281684491098e-05,
-      "loss": 44.2016,
-      "step": 152090
-    },
-    {
-      "epoch": 0.6145032462416723,
-      "grad_norm": 475.3173522949219,
-      "learning_rate": 1.9938863255854357e-05,
-      "loss": 79.3213,
-      "step": 152100
-    },
-    {
-      "epoch": 0.6145436475070399,
-      "grad_norm": 1137.9266357421875,
-      "learning_rate": 1.9935444925887082e-05,
-      "loss": 56.144,
-      "step": 152110
-    },
-    {
-      "epoch": 0.6145840487724076,
-      "grad_norm": 545.95556640625,
-      "learning_rate": 1.9932026694655907e-05,
-      "loss": 65.3321,
-      "step": 152120
-    },
-    {
-      "epoch": 0.6146244500377752,
-      "grad_norm": 1686.6630859375,
-      "learning_rate": 1.992860856222749e-05,
-      "loss": 64.3524,
-      "step": 152130
-    },
-    {
-      "epoch": 0.6146648513031429,
-      "grad_norm": 951.417236328125,
-      "learning_rate": 1.992519052866845e-05,
-      "loss": 60.3568,
-      "step": 152140
-    },
-    {
-      "epoch": 0.6147052525685105,
-      "grad_norm": 941.8363647460938,
-      "learning_rate": 1.992177259404545e-05,
-      "loss": 62.5279,
-      "step": 152150
-    },
-    {
-      "epoch": 0.614745653833878,
-      "grad_norm": 1050.4307861328125,
-      "learning_rate": 1.99183547584251e-05,
-      "loss": 47.3056,
-      "step": 152160
-    },
-    {
-      "epoch": 0.6147860550992457,
-      "grad_norm": 1039.5369873046875,
-      "learning_rate": 1.9914937021874032e-05,
-      "loss": 48.0764,
-      "step": 152170
-    },
-    {
-      "epoch": 0.6148264563646133,
-      "grad_norm": 406.83685302734375,
-      "learning_rate": 1.9911519384458893e-05,
-      "loss": 52.5042,
-      "step": 152180
-    },
-    {
-      "epoch": 0.614866857629981,
-      "grad_norm": 560.0790405273438,
-      "learning_rate": 1.9908101846246304e-05,
-      "loss": 44.851,
-      "step": 152190
-    },
-    {
-      "epoch": 0.6149072588953486,
-      "grad_norm": 795.4417724609375,
-      "learning_rate": 1.9904684407302883e-05,
-      "loss": 86.3452,
-      "step": 152200
-    },
-    {
-      "epoch": 0.6149476601607162,
-      "grad_norm": 578.105712890625,
-      "learning_rate": 1.990126706769527e-05,
-      "loss": 30.9948,
-      "step": 152210
-    },
-    {
-      "epoch": 0.6149880614260839,
-      "grad_norm": 638.46044921875,
-      "learning_rate": 1.989784982749008e-05,
-      "loss": 39.7895,
-      "step": 152220
-    },
-    {
-      "epoch": 0.6150284626914515,
-      "grad_norm": 708.2172241210938,
-      "learning_rate": 1.989443268675393e-05,
-      "loss": 99.4368,
-      "step": 152230
-    },
-    {
-      "epoch": 0.6150688639568191,
-      "grad_norm": 1003.4779663085938,
-      "learning_rate": 1.989101564555345e-05,
-      "loss": 52.786,
-      "step": 152240
-    },
-    {
-      "epoch": 0.6151092652221868,
-      "grad_norm": 658.6339721679688,
-      "learning_rate": 1.9887598703955242e-05,
-      "loss": 53.8698,
-      "step": 152250
-    },
-    {
-      "epoch": 0.6151496664875544,
-      "grad_norm": 1459.9710693359375,
-      "learning_rate": 1.988418186202594e-05,
-      "loss": 56.2961,
-      "step": 152260
-    },
-    {
-      "epoch": 0.6151900677529221,
-      "grad_norm": 706.1653442382812,
-      "learning_rate": 1.988076511983214e-05,
-      "loss": 51.5391,
-      "step": 152270
-    },
-    {
-      "epoch": 0.6152304690182897,
-      "grad_norm": 863.3594360351562,
-      "learning_rate": 1.9877348477440456e-05,
-      "loss": 63.8783,
-      "step": 152280
-    },
-    {
-      "epoch": 0.6152708702836572,
-      "grad_norm": 1135.580322265625,
-      "learning_rate": 1.9873931934917506e-05,
-      "loss": 66.3226,
-      "step": 152290
-    },
-    {
-      "epoch": 0.6153112715490249,
-      "grad_norm": 656.6669311523438,
-      "learning_rate": 1.9870515492329884e-05,
-      "loss": 46.0662,
-      "step": 152300
-    },
-    {
-      "epoch": 0.6153516728143925,
-      "grad_norm": 1270.214599609375,
-      "learning_rate": 1.9867099149744213e-05,
-      "loss": 53.8983,
-      "step": 152310
-    },
-    {
-      "epoch": 0.6153920740797602,
-      "grad_norm": 1297.236328125,
-      "learning_rate": 1.9863682907227088e-05,
-      "loss": 53.0378,
-      "step": 152320
-    },
-    {
-      "epoch": 0.6154324753451278,
-      "grad_norm": 834.4827880859375,
-      "learning_rate": 1.98602667648451e-05,
-      "loss": 59.2501,
-      "step": 152330
-    },
-    {
-      "epoch": 0.6154728766104954,
-      "grad_norm": 380.9190368652344,
-      "learning_rate": 1.9856850722664864e-05,
-      "loss": 48.5186,
-      "step": 152340
-    },
-    {
-      "epoch": 0.6155132778758631,
-      "grad_norm": 661.6729125976562,
-      "learning_rate": 1.9853434780752973e-05,
-      "loss": 43.4546,
-      "step": 152350
-    },
-    {
-      "epoch": 0.6155536791412307,
-      "grad_norm": 0.0,
-      "learning_rate": 1.9850018939176014e-05,
-      "loss": 43.309,
-      "step": 152360
-    },
-    {
-      "epoch": 0.6155940804065984,
-      "grad_norm": 1031.6661376953125,
-      "learning_rate": 1.984660319800059e-05,
-      "loss": 58.6191,
-      "step": 152370
-    },
-    {
-      "epoch": 0.615634481671966,
-      "grad_norm": 544.71337890625,
-      "learning_rate": 1.9843187557293284e-05,
-      "loss": 59.7559,
-      "step": 152380
-    },
-    {
-      "epoch": 0.6156748829373336,
-      "grad_norm": 1013.8038330078125,
-      "learning_rate": 1.98397720171207e-05,
-      "loss": 56.4807,
-      "step": 152390
-    },
-    {
-      "epoch": 0.6157152842027013,
-      "grad_norm": 952.9639892578125,
-      "learning_rate": 1.983635657754942e-05,
-      "loss": 59.5986,
-      "step": 152400
-    },
-    {
-      "epoch": 0.6157556854680689,
-      "grad_norm": 650.10107421875,
-      "learning_rate": 1.983294123864602e-05,
-      "loss": 60.1285,
-      "step": 152410
-    },
-    {
-      "epoch": 0.6157960867334364,
-      "grad_norm": 1023.7354125976562,
-      "learning_rate": 1.9829526000477095e-05,
-      "loss": 63.0215,
-      "step": 152420
-    },
-    {
-      "epoch": 0.6158364879988041,
-      "grad_norm": 1077.8424072265625,
-      "learning_rate": 1.982611086310922e-05,
-      "loss": 50.8245,
-      "step": 152430
-    },
-    {
-      "epoch": 0.6158768892641717,
-      "grad_norm": 702.2991333007812,
-      "learning_rate": 1.9822695826608972e-05,
-      "loss": 102.1233,
-      "step": 152440
-    },
-    {
-      "epoch": 0.6159172905295394,
-      "grad_norm": 432.0490417480469,
-      "learning_rate": 1.981928089104294e-05,
-      "loss": 31.7835,
-      "step": 152450
-    },
-    {
-      "epoch": 0.615957691794907,
-      "grad_norm": 942.7676391601562,
-      "learning_rate": 1.981586605647769e-05,
-      "loss": 51.312,
-      "step": 152460
-    },
-    {
-      "epoch": 0.6159980930602746,
-      "grad_norm": 485.92242431640625,
-      "learning_rate": 1.9812451322979805e-05,
-      "loss": 54.0049,
-      "step": 152470
-    },
-    {
-      "epoch": 0.6160384943256423,
-      "grad_norm": 2057.892578125,
-      "learning_rate": 1.9809036690615853e-05,
-      "loss": 57.5523,
-      "step": 152480
-    },
-    {
-      "epoch": 0.6160788955910099,
-      "grad_norm": 1027.524658203125,
-      "learning_rate": 1.98056221594524e-05,
-      "loss": 64.4764,
-      "step": 152490
-    },
-    {
-      "epoch": 0.6161192968563776,
-      "grad_norm": 434.5711975097656,
-      "learning_rate": 1.980220772955602e-05,
-      "loss": 43.7486,
-      "step": 152500
-    },
-    {
-      "epoch": 0.6161596981217452,
-      "grad_norm": 1201.443603515625,
-      "learning_rate": 1.979879340099327e-05,
-      "loss": 40.2573,
-      "step": 152510
-    },
-    {
-      "epoch": 0.6162000993871128,
-      "grad_norm": 1387.3876953125,
-      "learning_rate": 1.979537917383073e-05,
-      "loss": 44.6969,
-      "step": 152520
-    },
-    {
-      "epoch": 0.6162405006524805,
-      "grad_norm": 564.0380249023438,
-      "learning_rate": 1.979196504813495e-05,
-      "loss": 36.653,
-      "step": 152530
-    },
-    {
-      "epoch": 0.616280901917848,
-      "grad_norm": 824.3677978515625,
-      "learning_rate": 1.9788551023972484e-05,
-      "loss": 43.8623,
-      "step": 152540
-    },
-    {
-      "epoch": 0.6163213031832157,
-      "grad_norm": 531.8292236328125,
-      "learning_rate": 1.9785137101409908e-05,
-      "loss": 57.3038,
-      "step": 152550
-    },
-    {
-      "epoch": 0.6163617044485833,
-      "grad_norm": 276.3924865722656,
-      "learning_rate": 1.9781723280513768e-05,
-      "loss": 42.5384,
-      "step": 152560
-    },
-    {
-      "epoch": 0.6164021057139509,
-      "grad_norm": 521.0140380859375,
-      "learning_rate": 1.9778309561350616e-05,
-      "loss": 54.9165,
-      "step": 152570
-    },
-    {
-      "epoch": 0.6164425069793186,
-      "grad_norm": 4706.6787109375,
-      "learning_rate": 1.9774895943987007e-05,
-      "loss": 70.1737,
-      "step": 152580
-    },
-    {
-      "epoch": 0.6164829082446862,
-      "grad_norm": 1025.7991943359375,
-      "learning_rate": 1.977148242848949e-05,
-      "loss": 51.8512,
-      "step": 152590
-    },
-    {
-      "epoch": 0.6165233095100539,
-      "grad_norm": 1082.083984375,
-      "learning_rate": 1.9768069014924622e-05,
-      "loss": 45.9668,
-      "step": 152600
-    },
-    {
-      "epoch": 0.6165637107754215,
-      "grad_norm": 820.9380493164062,
-      "learning_rate": 1.9764655703358945e-05,
-      "loss": 57.365,
-      "step": 152610
-    },
-    {
-      "epoch": 0.6166041120407891,
-      "grad_norm": 500.4762878417969,
-      "learning_rate": 1.9761242493858987e-05,
-      "loss": 60.4964,
-      "step": 152620
-    },
-    {
-      "epoch": 0.6166445133061568,
-      "grad_norm": 767.7699584960938,
-      "learning_rate": 1.975782938649131e-05,
-      "loss": 44.4775,
-      "step": 152630
-    },
-    {
-      "epoch": 0.6166849145715244,
-      "grad_norm": 730.7559204101562,
-      "learning_rate": 1.9754416381322455e-05,
-      "loss": 61.5032,
-      "step": 152640
-    },
-    {
-      "epoch": 0.616725315836892,
-      "grad_norm": 621.5751342773438,
-      "learning_rate": 1.975100347841894e-05,
-      "loss": 45.6827,
-      "step": 152650
-    },
-    {
-      "epoch": 0.6167657171022597,
-      "grad_norm": 1287.0096435546875,
-      "learning_rate": 1.974759067784732e-05,
-      "loss": 53.478,
-      "step": 152660
-    },
-    {
-      "epoch": 0.6168061183676272,
-      "grad_norm": 415.4209899902344,
-      "learning_rate": 1.974417797967413e-05,
-      "loss": 28.9644,
-      "step": 152670
-    },
-    {
-      "epoch": 0.6168465196329949,
-      "grad_norm": 277.8809509277344,
-      "learning_rate": 1.9740765383965893e-05,
-      "loss": 55.9292,
-      "step": 152680
-    },
-    {
-      "epoch": 0.6168869208983625,
-      "grad_norm": 597.7545776367188,
-      "learning_rate": 1.9737352890789142e-05,
-      "loss": 35.1805,
-      "step": 152690
-    },
-    {
-      "epoch": 0.6169273221637301,
-      "grad_norm": 395.4619140625,
-      "learning_rate": 1.9733940500210398e-05,
-      "loss": 64.5738,
-      "step": 152700
-    },
-    {
-      "epoch": 0.6169677234290978,
-      "grad_norm": 606.8431396484375,
-      "learning_rate": 1.9730528212296208e-05,
-      "loss": 47.9119,
-      "step": 152710
-    },
-    {
-      "epoch": 0.6170081246944654,
-      "grad_norm": 919.1580810546875,
-      "learning_rate": 1.9727116027113077e-05,
-      "loss": 42.3338,
-      "step": 152720
-    },
-    {
-      "epoch": 0.6170485259598331,
-      "grad_norm": 1222.4852294921875,
-      "learning_rate": 1.972370394472753e-05,
-      "loss": 89.0341,
-      "step": 152730
-    },
-    {
-      "epoch": 0.6170889272252007,
-      "grad_norm": 975.9841918945312,
-      "learning_rate": 1.9720291965206095e-05,
-      "loss": 40.6854,
-      "step": 152740
-    },
-    {
-      "epoch": 0.6171293284905683,
-      "grad_norm": 815.9156494140625,
-      "learning_rate": 1.9716880088615285e-05,
-      "loss": 67.8414,
-      "step": 152750
-    },
-    {
-      "epoch": 0.617169729755936,
-      "grad_norm": 587.384033203125,
-      "learning_rate": 1.9713468315021622e-05,
-      "loss": 60.1556,
-      "step": 152760
-    },
-    {
-      "epoch": 0.6172101310213036,
-      "grad_norm": 369.40081787109375,
-      "learning_rate": 1.9710056644491614e-05,
-      "loss": 46.0275,
-      "step": 152770
-    },
-    {
-      "epoch": 0.6172505322866713,
-      "grad_norm": 611.0215454101562,
-      "learning_rate": 1.9706645077091767e-05,
-      "loss": 50.6546,
-      "step": 152780
-    },
-    {
-      "epoch": 0.6172909335520389,
-      "grad_norm": 170.70001220703125,
-      "learning_rate": 1.970323361288861e-05,
-      "loss": 23.4807,
-      "step": 152790
-    },
-    {
-      "epoch": 0.6173313348174064,
-      "grad_norm": 473.4237976074219,
-      "learning_rate": 1.969982225194864e-05,
-      "loss": 63.5025,
-      "step": 152800
-    },
-    {
-      "epoch": 0.6173717360827741,
-      "grad_norm": 773.618408203125,
-      "learning_rate": 1.9696410994338354e-05,
-      "loss": 52.0266,
-      "step": 152810
-    },
-    {
-      "epoch": 0.6174121373481417,
-      "grad_norm": 649.8694458007812,
-      "learning_rate": 1.9692999840124275e-05,
-      "loss": 54.1027,
-      "step": 152820
-    },
-    {
-      "epoch": 0.6174525386135093,
-      "grad_norm": 1274.1107177734375,
-      "learning_rate": 1.9689588789372896e-05,
-      "loss": 67.0911,
-      "step": 152830
-    },
-    {
-      "epoch": 0.617492939878877,
-      "grad_norm": 541.9605102539062,
-      "learning_rate": 1.9686177842150715e-05,
-      "loss": 68.1363,
-      "step": 152840
-    },
-    {
-      "epoch": 0.6175333411442446,
-      "grad_norm": 1026.8818359375,
-      "learning_rate": 1.968276699852424e-05,
-      "loss": 53.6045,
-      "step": 152850
-    },
-    {
-      "epoch": 0.6175737424096123,
-      "grad_norm": 864.9629516601562,
-      "learning_rate": 1.9679356258559944e-05,
-      "loss": 79.2347,
-      "step": 152860
-    },
-    {
-      "epoch": 0.6176141436749799,
-      "grad_norm": 593.311767578125,
-      "learning_rate": 1.967594562232435e-05,
-      "loss": 42.9782,
-      "step": 152870
-    },
-    {
-      "epoch": 0.6176545449403475,
-      "grad_norm": 664.3198852539062,
-      "learning_rate": 1.967253508988394e-05,
-      "loss": 52.8853,
-      "step": 152880
-    },
-    {
-      "epoch": 0.6176949462057152,
-      "grad_norm": 1579.415283203125,
-      "learning_rate": 1.9669124661305185e-05,
-      "loss": 75.1785,
-      "step": 152890
-    },
-    {
-      "epoch": 0.6177353474710828,
-      "grad_norm": 1345.0396728515625,
-      "learning_rate": 1.9665714336654602e-05,
-      "loss": 46.0577,
-      "step": 152900
-    },
-    {
-      "epoch": 0.6177757487364505,
-      "grad_norm": 811.4996948242188,
-      "learning_rate": 1.966230411599866e-05,
-      "loss": 55.0498,
-      "step": 152910
-    },
-    {
-      "epoch": 0.6178161500018181,
-      "grad_norm": 1309.7557373046875,
-      "learning_rate": 1.9658893999403847e-05,
-      "loss": 51.9295,
-      "step": 152920
-    },
-    {
-      "epoch": 0.6178565512671856,
-      "grad_norm": 936.4150390625,
-      "learning_rate": 1.9655483986936653e-05,
-      "loss": 45.001,
-      "step": 152930
-    },
-    {
-      "epoch": 0.6178969525325533,
-      "grad_norm": 293.4042663574219,
-      "learning_rate": 1.965207407866354e-05,
-      "loss": 54.548,
-      "step": 152940
-    },
-    {
-      "epoch": 0.6179373537979209,
-      "grad_norm": 679.14306640625,
-      "learning_rate": 1.9648664274651e-05,
-      "loss": 48.1192,
-      "step": 152950
-    },
-    {
-      "epoch": 0.6179777550632886,
-      "grad_norm": 677.8344116210938,
-      "learning_rate": 1.964525457496551e-05,
-      "loss": 55.2474,
-      "step": 152960
-    },
-    {
-      "epoch": 0.6180181563286562,
-      "grad_norm": 584.7177734375,
-      "learning_rate": 1.964184497967353e-05,
-      "loss": 52.2991,
-      "step": 152970
-    },
-    {
-      "epoch": 0.6180585575940238,
-      "grad_norm": 619.6227416992188,
-      "learning_rate": 1.9638435488841546e-05,
-      "loss": 38.7748,
-      "step": 152980
-    },
-    {
-      "epoch": 0.6180989588593915,
-      "grad_norm": 824.2867431640625,
-      "learning_rate": 1.9635026102536014e-05,
-      "loss": 41.0035,
-      "step": 152990
-    },
-    {
-      "epoch": 0.6181393601247591,
-      "grad_norm": 825.877685546875,
-      "learning_rate": 1.963161682082342e-05,
-      "loss": 77.0988,
-      "step": 153000
-    },
-    {
-      "epoch": 0.6181797613901268,
-      "grad_norm": 733.9141235351562,
-      "learning_rate": 1.9628207643770223e-05,
-      "loss": 42.6838,
-      "step": 153010
-    },
-    {
-      "epoch": 0.6182201626554944,
-      "grad_norm": 1160.487060546875,
-      "learning_rate": 1.9624798571442873e-05,
-      "loss": 46.4142,
-      "step": 153020
-    },
-    {
-      "epoch": 0.618260563920862,
-      "grad_norm": 331.2978210449219,
-      "learning_rate": 1.9621389603907852e-05,
-      "loss": 29.5836,
-      "step": 153030
-    },
-    {
-      "epoch": 0.6183009651862297,
-      "grad_norm": 704.036376953125,
-      "learning_rate": 1.96179807412316e-05,
-      "loss": 51.0001,
-      "step": 153040
-    },
-    {
-      "epoch": 0.6183413664515973,
-      "grad_norm": 751.7705078125,
-      "learning_rate": 1.961457198348059e-05,
-      "loss": 53.3635,
-      "step": 153050
-    },
-    {
-      "epoch": 0.6183817677169648,
-      "grad_norm": 2129.126708984375,
-      "learning_rate": 1.9611163330721275e-05,
-      "loss": 66.4268,
-      "step": 153060
-    },
-    {
-      "epoch": 0.6184221689823325,
-      "grad_norm": 1071.379150390625,
-      "learning_rate": 1.9607754783020092e-05,
-      "loss": 75.9926,
-      "step": 153070
-    },
-    {
-      "epoch": 0.6184625702477001,
-      "grad_norm": 600.6515502929688,
-      "learning_rate": 1.9604346340443518e-05,
-      "loss": 49.0746,
-      "step": 153080
-    },
-    {
-      "epoch": 0.6185029715130678,
-      "grad_norm": 635.5545043945312,
-      "learning_rate": 1.9600938003057994e-05,
-      "loss": 50.7416,
-      "step": 153090
-    },
-    {
-      "epoch": 0.6185433727784354,
-      "grad_norm": 392.02178955078125,
-      "learning_rate": 1.959752977092995e-05,
-      "loss": 98.5631,
-      "step": 153100
-    },
-    {
-      "epoch": 0.618583774043803,
-      "grad_norm": 811.9187622070312,
-      "learning_rate": 1.9594121644125852e-05,
-      "loss": 55.1491,
-      "step": 153110
-    },
-    {
-      "epoch": 0.6186241753091707,
-      "grad_norm": 912.7703247070312,
-      "learning_rate": 1.9590713622712132e-05,
-      "loss": 69.0469,
-      "step": 153120
-    },
-    {
-      "epoch": 0.6186645765745383,
-      "grad_norm": 588.3534545898438,
-      "learning_rate": 1.9587305706755236e-05,
-      "loss": 54.1767,
-      "step": 153130
-    },
-    {
-      "epoch": 0.618704977839906,
-      "grad_norm": 878.7616577148438,
-      "learning_rate": 1.9583897896321607e-05,
-      "loss": 68.2562,
-      "step": 153140
-    },
-    {
-      "epoch": 0.6187453791052736,
-      "grad_norm": 334.1783142089844,
-      "learning_rate": 1.958049019147767e-05,
-      "loss": 40.0663,
-      "step": 153150
-    },
-    {
-      "epoch": 0.6187857803706412,
-      "grad_norm": 619.4424438476562,
-      "learning_rate": 1.957708259228987e-05,
-      "loss": 35.9437,
-      "step": 153160
-    },
-    {
-      "epoch": 0.6188261816360089,
-      "grad_norm": 1056.420166015625,
-      "learning_rate": 1.957367509882464e-05,
-      "loss": 54.9157,
-      "step": 153170
-    },
-    {
-      "epoch": 0.6188665829013764,
-      "grad_norm": 1181.443115234375,
-      "learning_rate": 1.9570267711148403e-05,
-      "loss": 50.0268,
-      "step": 153180
-    },
-    {
-      "epoch": 0.6189069841667441,
-      "grad_norm": 431.3962097167969,
-      "learning_rate": 1.9566860429327595e-05,
-      "loss": 47.6178,
-      "step": 153190
-    },
-    {
-      "epoch": 0.6189473854321117,
-      "grad_norm": 490.9452819824219,
-      "learning_rate": 1.956345325342863e-05,
-      "loss": 48.6404,
-      "step": 153200
-    },
-    {
-      "epoch": 0.6189877866974793,
-      "grad_norm": 572.4602661132812,
-      "learning_rate": 1.9560046183517953e-05,
-      "loss": 50.4382,
-      "step": 153210
-    },
-    {
-      "epoch": 0.619028187962847,
-      "grad_norm": 801.1991577148438,
-      "learning_rate": 1.955663921966198e-05,
-      "loss": 41.4329,
-      "step": 153220
-    },
-    {
-      "epoch": 0.6190685892282146,
-      "grad_norm": 290.9499206542969,
-      "learning_rate": 1.9553232361927114e-05,
-      "loss": 56.9374,
-      "step": 153230
-    },
-    {
-      "epoch": 0.6191089904935823,
-      "grad_norm": 657.2138671875,
-      "learning_rate": 1.95498256103798e-05,
-      "loss": 58.0168,
-      "step": 153240
-    },
-    {
-      "epoch": 0.6191493917589499,
-      "grad_norm": 1027.6114501953125,
-      "learning_rate": 1.9546418965086442e-05,
-      "loss": 54.518,
-      "step": 153250
-    },
-    {
-      "epoch": 0.6191897930243175,
-      "grad_norm": 748.5098266601562,
-      "learning_rate": 1.954301242611344e-05,
-      "loss": 43.2595,
-      "step": 153260
-    },
-    {
-      "epoch": 0.6192301942896852,
-      "grad_norm": 648.7311401367188,
-      "learning_rate": 1.9539605993527237e-05,
-      "loss": 59.3503,
-      "step": 153270
-    },
-    {
-      "epoch": 0.6192705955550528,
-      "grad_norm": 344.7987365722656,
-      "learning_rate": 1.9536199667394215e-05,
-      "loss": 61.7763,
-      "step": 153280
-    },
-    {
-      "epoch": 0.6193109968204205,
-      "grad_norm": 451.1872863769531,
-      "learning_rate": 1.95327934477808e-05,
-      "loss": 84.8409,
-      "step": 153290
-    },
-    {
-      "epoch": 0.6193513980857881,
-      "grad_norm": 830.3189697265625,
-      "learning_rate": 1.9529387334753395e-05,
-      "loss": 43.1897,
-      "step": 153300
-    },
-    {
-      "epoch": 0.6193917993511556,
-      "grad_norm": 514.0184326171875,
-      "learning_rate": 1.9525981328378384e-05,
-      "loss": 53.5215,
-      "step": 153310
-    },
-    {
-      "epoch": 0.6194322006165233,
-      "grad_norm": 357.33038330078125,
-      "learning_rate": 1.95225754287222e-05,
-      "loss": 53.0679,
-      "step": 153320
-    },
-    {
-      "epoch": 0.6194726018818909,
-      "grad_norm": 453.4263610839844,
-      "learning_rate": 1.9519169635851224e-05,
-      "loss": 52.7831,
-      "step": 153330
-    },
-    {
-      "epoch": 0.6195130031472585,
-      "grad_norm": 1120.7613525390625,
-      "learning_rate": 1.951576394983185e-05,
-      "loss": 87.4061,
-      "step": 153340
-    },
-    {
-      "epoch": 0.6195534044126262,
-      "grad_norm": 1069.5909423828125,
-      "learning_rate": 1.9512358370730493e-05,
-      "loss": 63.5944,
-      "step": 153350
-    },
-    {
-      "epoch": 0.6195938056779938,
-      "grad_norm": 665.5206909179688,
-      "learning_rate": 1.9508952898613528e-05,
-      "loss": 50.0496,
-      "step": 153360
-    },
-    {
-      "epoch": 0.6196342069433615,
-      "grad_norm": 509.3944091796875,
-      "learning_rate": 1.9505547533547358e-05,
-      "loss": 50.3182,
-      "step": 153370
-    },
-    {
-      "epoch": 0.6196746082087291,
-      "grad_norm": 457.98583984375,
-      "learning_rate": 1.950214227559837e-05,
-      "loss": 39.0197,
-      "step": 153380
-    },
-    {
-      "epoch": 0.6197150094740967,
-      "grad_norm": 1662.1475830078125,
-      "learning_rate": 1.9498737124832936e-05,
-      "loss": 41.9308,
-      "step": 153390
-    },
-    {
-      "epoch": 0.6197554107394644,
-      "grad_norm": 1056.866455078125,
-      "learning_rate": 1.9495332081317464e-05,
-      "loss": 43.6751,
-      "step": 153400
-    },
-    {
-      "epoch": 0.619795812004832,
-      "grad_norm": 1036.8895263671875,
-      "learning_rate": 1.949192714511833e-05,
-      "loss": 56.6728,
-      "step": 153410
-    },
-    {
-      "epoch": 0.6198362132701997,
-      "grad_norm": 950.1953735351562,
-      "learning_rate": 1.9488522316301898e-05,
-      "loss": 84.3255,
-      "step": 153420
-    },
-    {
-      "epoch": 0.6198766145355673,
-      "grad_norm": 740.6848754882812,
-      "learning_rate": 1.9485117594934574e-05,
-      "loss": 51.3797,
-      "step": 153430
-    },
-    {
-      "epoch": 0.6199170158009348,
-      "grad_norm": 798.8438720703125,
-      "learning_rate": 1.9481712981082714e-05,
-      "loss": 72.0278,
-      "step": 153440
-    },
-    {
-      "epoch": 0.6199574170663025,
-      "grad_norm": 735.481689453125,
-      "learning_rate": 1.947830847481271e-05,
-      "loss": 39.8441,
-      "step": 153450
-    },
-    {
-      "epoch": 0.6199978183316701,
-      "grad_norm": 574.896240234375,
-      "learning_rate": 1.947490407619092e-05,
-      "loss": 59.1198,
-      "step": 153460
-    },
-    {
-      "epoch": 0.6200382195970378,
-      "grad_norm": 802.9642944335938,
-      "learning_rate": 1.9471499785283712e-05,
-      "loss": 58.8451,
-      "step": 153470
-    },
-    {
-      "epoch": 0.6200786208624054,
-      "grad_norm": 601.0955810546875,
-      "learning_rate": 1.946809560215747e-05,
-      "loss": 53.9513,
-      "step": 153480
-    },
-    {
-      "epoch": 0.620119022127773,
-      "grad_norm": 844.2098999023438,
-      "learning_rate": 1.9464691526878555e-05,
-      "loss": 56.864,
-      "step": 153490
-    },
-    {
-      "epoch": 0.6201594233931407,
-      "grad_norm": 836.8554077148438,
-      "learning_rate": 1.946128755951332e-05,
-      "loss": 48.7098,
-      "step": 153500
-    },
-    {
-      "epoch": 0.6201998246585083,
-      "grad_norm": 1133.341552734375,
-      "learning_rate": 1.945788370012814e-05,
-      "loss": 60.1469,
-      "step": 153510
-    },
-    {
-      "epoch": 0.620240225923876,
-      "grad_norm": 597.8056640625,
-      "learning_rate": 1.945447994878937e-05,
-      "loss": 52.2197,
-      "step": 153520
-    },
-    {
-      "epoch": 0.6202806271892436,
-      "grad_norm": 370.60791015625,
-      "learning_rate": 1.945107630556337e-05,
-      "loss": 30.6839,
-      "step": 153530
-    },
-    {
-      "epoch": 0.6203210284546112,
-      "grad_norm": 2185.733642578125,
-      "learning_rate": 1.9447672770516494e-05,
-      "loss": 55.9894,
-      "step": 153540
-    },
-    {
-      "epoch": 0.6203614297199789,
-      "grad_norm": 950.8214721679688,
-      "learning_rate": 1.9444269343715092e-05,
-      "loss": 50.5676,
-      "step": 153550
-    },
-    {
-      "epoch": 0.6204018309853465,
-      "grad_norm": 982.4635009765625,
-      "learning_rate": 1.9440866025225525e-05,
-      "loss": 50.9509,
-      "step": 153560
-    },
-    {
-      "epoch": 0.620442232250714,
-      "grad_norm": 1476.201171875,
-      "learning_rate": 1.9437462815114128e-05,
-      "loss": 50.0211,
-      "step": 153570
-    },
-    {
-      "epoch": 0.6204826335160817,
-      "grad_norm": 811.0206298828125,
-      "learning_rate": 1.9434059713447265e-05,
-      "loss": 67.0507,
-      "step": 153580
-    },
-    {
-      "epoch": 0.6205230347814493,
-      "grad_norm": 753.006591796875,
-      "learning_rate": 1.9430656720291275e-05,
-      "loss": 53.868,
-      "step": 153590
-    },
-    {
-      "epoch": 0.620563436046817,
-      "grad_norm": 937.0206909179688,
-      "learning_rate": 1.942725383571249e-05,
-      "loss": 55.5334,
-      "step": 153600
-    },
-    {
-      "epoch": 0.6206038373121846,
-      "grad_norm": 889.5717163085938,
-      "learning_rate": 1.942385105977727e-05,
-      "loss": 70.7559,
-      "step": 153610
-    },
-    {
-      "epoch": 0.6206442385775522,
-      "grad_norm": 412.93267822265625,
-      "learning_rate": 1.9420448392551943e-05,
-      "loss": 36.8532,
-      "step": 153620
-    },
-    {
-      "epoch": 0.6206846398429199,
-      "grad_norm": 654.7472534179688,
-      "learning_rate": 1.9417045834102844e-05,
-      "loss": 48.5801,
-      "step": 153630
-    },
-    {
-      "epoch": 0.6207250411082875,
-      "grad_norm": 586.552490234375,
-      "learning_rate": 1.9413643384496316e-05,
-      "loss": 54.077,
-      "step": 153640
-    },
-    {
-      "epoch": 0.6207654423736552,
-      "grad_norm": 811.3733520507812,
-      "learning_rate": 1.9410241043798676e-05,
-      "loss": 63.7313,
-      "step": 153650
-    },
-    {
-      "epoch": 0.6208058436390228,
-      "grad_norm": 444.1869812011719,
-      "learning_rate": 1.9406838812076273e-05,
-      "loss": 46.0719,
-      "step": 153660
-    },
-    {
-      "epoch": 0.6208462449043904,
-      "grad_norm": 811.6187744140625,
-      "learning_rate": 1.9403436689395426e-05,
-      "loss": 62.3275,
-      "step": 153670
-    },
-    {
-      "epoch": 0.6208866461697581,
-      "grad_norm": 504.3353271484375,
-      "learning_rate": 1.9400034675822452e-05,
-      "loss": 42.1343,
-      "step": 153680
-    },
-    {
-      "epoch": 0.6209270474351257,
-      "grad_norm": 799.2766723632812,
-      "learning_rate": 1.93966327714237e-05,
-      "loss": 52.7461,
-      "step": 153690
-    },
-    {
-      "epoch": 0.6209674487004933,
-      "grad_norm": 461.0577392578125,
-      "learning_rate": 1.9393230976265473e-05,
-      "loss": 59.0141,
-      "step": 153700
-    },
-    {
-      "epoch": 0.6210078499658609,
-      "grad_norm": 828.9132690429688,
-      "learning_rate": 1.938982929041409e-05,
-      "loss": 46.2773,
-      "step": 153710
-    },
-    {
-      "epoch": 0.6210482512312285,
-      "grad_norm": 354.5053405761719,
-      "learning_rate": 1.938642771393588e-05,
-      "loss": 53.5887,
-      "step": 153720
-    },
-    {
-      "epoch": 0.6210886524965962,
-      "grad_norm": 144.50497436523438,
-      "learning_rate": 1.9383026246897143e-05,
-      "loss": 37.4983,
-      "step": 153730
-    },
-    {
-      "epoch": 0.6211290537619638,
-      "grad_norm": 783.1920776367188,
-      "learning_rate": 1.937962488936421e-05,
-      "loss": 38.912,
-      "step": 153740
-    },
-    {
-      "epoch": 0.6211694550273315,
-      "grad_norm": 666.5381469726562,
-      "learning_rate": 1.937622364140338e-05,
-      "loss": 69.1318,
-      "step": 153750
-    },
-    {
-      "epoch": 0.6212098562926991,
-      "grad_norm": 760.7601318359375,
-      "learning_rate": 1.9372822503080957e-05,
-      "loss": 40.2389,
-      "step": 153760
-    },
-    {
-      "epoch": 0.6212502575580667,
-      "grad_norm": 508.2515563964844,
-      "learning_rate": 1.9369421474463268e-05,
-      "loss": 42.0188,
-      "step": 153770
-    },
-    {
-      "epoch": 0.6212906588234344,
-      "grad_norm": 610.3115844726562,
-      "learning_rate": 1.9366020555616603e-05,
-      "loss": 43.7602,
-      "step": 153780
-    },
-    {
-      "epoch": 0.621331060088802,
-      "grad_norm": 336.53662109375,
-      "learning_rate": 1.936261974660727e-05,
-      "loss": 44.282,
-      "step": 153790
-    },
-    {
-      "epoch": 0.6213714613541697,
-      "grad_norm": 952.2464599609375,
-      "learning_rate": 1.9359219047501565e-05,
-      "loss": 81.2008,
-      "step": 153800
-    },
-    {
-      "epoch": 0.6214118626195373,
-      "grad_norm": 365.46612548828125,
-      "learning_rate": 1.9355818458365782e-05,
-      "loss": 37.9822,
-      "step": 153810
-    },
-    {
-      "epoch": 0.6214522638849048,
-      "grad_norm": 697.7381591796875,
-      "learning_rate": 1.9352417979266233e-05,
-      "loss": 66.52,
-      "step": 153820
-    },
-    {
-      "epoch": 0.6214926651502725,
-      "grad_norm": 800.6105346679688,
-      "learning_rate": 1.93490176102692e-05,
-      "loss": 62.5696,
-      "step": 153830
-    },
-    {
-      "epoch": 0.6215330664156401,
-      "grad_norm": 527.7947387695312,
-      "learning_rate": 1.9345617351440973e-05,
-      "loss": 79.746,
-      "step": 153840
-    },
-    {
-      "epoch": 0.6215734676810077,
-      "grad_norm": 1173.56689453125,
-      "learning_rate": 1.9342217202847856e-05,
-      "loss": 59.6406,
-      "step": 153850
-    },
-    {
-      "epoch": 0.6216138689463754,
-      "grad_norm": 1889.8114013671875,
-      "learning_rate": 1.9338817164556128e-05,
-      "loss": 65.9831,
-      "step": 153860
-    },
-    {
-      "epoch": 0.621654270211743,
-      "grad_norm": 1583.56591796875,
-      "learning_rate": 1.9335417236632065e-05,
-      "loss": 48.6368,
-      "step": 153870
-    },
-    {
-      "epoch": 0.6216946714771107,
-      "grad_norm": 379.38201904296875,
-      "learning_rate": 1.9332017419141962e-05,
-      "loss": 33.5321,
-      "step": 153880
-    },
-    {
-      "epoch": 0.6217350727424783,
-      "grad_norm": 617.2946166992188,
-      "learning_rate": 1.9328617712152098e-05,
-      "loss": 60.8888,
-      "step": 153890
-    },
-    {
-      "epoch": 0.6217754740078459,
-      "grad_norm": 685.6942138671875,
-      "learning_rate": 1.9325218115728755e-05,
-      "loss": 48.9418,
-      "step": 153900
-    },
-    {
-      "epoch": 0.6218158752732136,
-      "grad_norm": 514.7144165039062,
-      "learning_rate": 1.9321818629938208e-05,
-      "loss": 67.4029,
-      "step": 153910
-    },
-    {
-      "epoch": 0.6218562765385812,
-      "grad_norm": 1621.0958251953125,
-      "learning_rate": 1.9318419254846718e-05,
-      "loss": 42.5269,
-      "step": 153920
-    },
-    {
-      "epoch": 0.6218966778039489,
-      "grad_norm": 954.4661254882812,
-      "learning_rate": 1.9315019990520582e-05,
-      "loss": 60.1667,
-      "step": 153930
-    },
-    {
-      "epoch": 0.6219370790693165,
-      "grad_norm": 1003.4674682617188,
-      "learning_rate": 1.9311620837026057e-05,
-      "loss": 68.499,
-      "step": 153940
-    },
-    {
-      "epoch": 0.621977480334684,
-      "grad_norm": 713.1990966796875,
-      "learning_rate": 1.9308221794429403e-05,
-      "loss": 50.8365,
-      "step": 153950
-    },
-    {
-      "epoch": 0.6220178816000517,
-      "grad_norm": 381.65618896484375,
-      "learning_rate": 1.9304822862796903e-05,
-      "loss": 34.9108,
-      "step": 153960
-    },
-    {
-      "epoch": 0.6220582828654193,
-      "grad_norm": 1819.873046875,
-      "learning_rate": 1.930142404219481e-05,
-      "loss": 46.8878,
-      "step": 153970
-    },
-    {
-      "epoch": 0.622098684130787,
-      "grad_norm": 444.01611328125,
-      "learning_rate": 1.9298025332689397e-05,
-      "loss": 55.4984,
-      "step": 153980
-    },
-    {
-      "epoch": 0.6221390853961546,
-      "grad_norm": 379.1543273925781,
-      "learning_rate": 1.9294626734346914e-05,
-      "loss": 73.1394,
-      "step": 153990
-    },
-    {
-      "epoch": 0.6221794866615222,
-      "grad_norm": 1029.5592041015625,
-      "learning_rate": 1.9291228247233605e-05,
-      "loss": 53.775,
-      "step": 154000
-    },
-    {
-      "epoch": 0.6222198879268899,
-      "grad_norm": 991.5259399414062,
-      "learning_rate": 1.9287829871415757e-05,
-      "loss": 55.4859,
-      "step": 154010
-    },
-    {
-      "epoch": 0.6222602891922575,
-      "grad_norm": 375.24810791015625,
-      "learning_rate": 1.9284431606959594e-05,
-      "loss": 47.712,
-      "step": 154020
-    },
-    {
-      "epoch": 0.6223006904576251,
-      "grad_norm": 706.0023193359375,
-      "learning_rate": 1.9281033453931388e-05,
-      "loss": 42.1541,
-      "step": 154030
-    },
-    {
-      "epoch": 0.6223410917229928,
-      "grad_norm": 842.9607543945312,
-      "learning_rate": 1.9277635412397383e-05,
-      "loss": 51.2036,
-      "step": 154040
-    },
-    {
-      "epoch": 0.6223814929883604,
-      "grad_norm": 0.0,
-      "learning_rate": 1.9274237482423814e-05,
-      "loss": 51.3722,
-      "step": 154050
-    },
-    {
-      "epoch": 0.6224218942537281,
-      "grad_norm": 1058.037841796875,
-      "learning_rate": 1.9270839664076936e-05,
-      "loss": 50.9369,
-      "step": 154060
-    },
-    {
-      "epoch": 0.6224622955190957,
-      "grad_norm": 908.2258911132812,
-      "learning_rate": 1.9267441957422994e-05,
-      "loss": 48.8932,
-      "step": 154070
-    },
-    {
-      "epoch": 0.6225026967844632,
-      "grad_norm": 783.5228271484375,
-      "learning_rate": 1.926404436252821e-05,
-      "loss": 48.0859,
-      "step": 154080
-    },
-    {
-      "epoch": 0.6225430980498309,
-      "grad_norm": 719.8128662109375,
-      "learning_rate": 1.926064687945884e-05,
-      "loss": 58.9963,
-      "step": 154090
-    },
-    {
-      "epoch": 0.6225834993151985,
-      "grad_norm": 573.966552734375,
-      "learning_rate": 1.9257249508281107e-05,
-      "loss": 50.9664,
-      "step": 154100
-    },
-    {
-      "epoch": 0.6226239005805662,
-      "grad_norm": 477.0748291015625,
-      "learning_rate": 1.925385224906126e-05,
-      "loss": 38.2241,
-      "step": 154110
-    },
-    {
-      "epoch": 0.6226643018459338,
-      "grad_norm": 405.0860290527344,
-      "learning_rate": 1.9250455101865526e-05,
-      "loss": 56.007,
-      "step": 154120
-    },
-    {
-      "epoch": 0.6227047031113014,
-      "grad_norm": 1051.1751708984375,
-      "learning_rate": 1.924705806676012e-05,
-      "loss": 67.0096,
-      "step": 154130
-    },
-    {
-      "epoch": 0.6227451043766691,
-      "grad_norm": 960.0936889648438,
-      "learning_rate": 1.9243661143811287e-05,
-      "loss": 39.5979,
-      "step": 154140
-    },
-    {
-      "epoch": 0.6227855056420367,
-      "grad_norm": 476.98992919921875,
-      "learning_rate": 1.9240264333085245e-05,
-      "loss": 69.4723,
-      "step": 154150
-    },
-    {
-      "epoch": 0.6228259069074044,
-      "grad_norm": 675.7411499023438,
-      "learning_rate": 1.92368676346482e-05,
-      "loss": 46.5527,
-      "step": 154160
-    },
-    {
-      "epoch": 0.622866308172772,
-      "grad_norm": 305.06793212890625,
-      "learning_rate": 1.92334710485664e-05,
-      "loss": 42.114,
-      "step": 154170
-    },
-    {
-      "epoch": 0.6229067094381396,
-      "grad_norm": 505.45458984375,
-      "learning_rate": 1.9230074574906042e-05,
-      "loss": 66.8375,
-      "step": 154180
-    },
-    {
-      "epoch": 0.6229471107035073,
-      "grad_norm": 799.241455078125,
-      "learning_rate": 1.9226678213733358e-05,
-      "loss": 59.1406,
-      "step": 154190
-    },
-    {
-      "epoch": 0.6229875119688749,
-      "grad_norm": 676.989013671875,
-      "learning_rate": 1.922328196511456e-05,
-      "loss": 33.6825,
-      "step": 154200
-    },
-    {
-      "epoch": 0.6230279132342424,
-      "grad_norm": 1138.4705810546875,
-      "learning_rate": 1.9219885829115843e-05,
-      "loss": 48.9602,
-      "step": 154210
-    },
-    {
-      "epoch": 0.6230683144996101,
-      "grad_norm": 401.67071533203125,
-      "learning_rate": 1.921648980580343e-05,
-      "loss": 54.2608,
-      "step": 154220
-    },
-    {
-      "epoch": 0.6231087157649777,
-      "grad_norm": 2265.824951171875,
-      "learning_rate": 1.921309389524353e-05,
-      "loss": 78.4544,
-      "step": 154230
-    },
-    {
-      "epoch": 0.6231491170303454,
-      "grad_norm": 764.9121704101562,
-      "learning_rate": 1.920969809750234e-05,
-      "loss": 58.4583,
-      "step": 154240
-    },
-    {
-      "epoch": 0.623189518295713,
-      "grad_norm": 914.537109375,
-      "learning_rate": 1.920630241264607e-05,
-      "loss": 41.7083,
-      "step": 154250
-    },
-    {
-      "epoch": 0.6232299195610806,
-      "grad_norm": 293.2212219238281,
-      "learning_rate": 1.9202906840740907e-05,
-      "loss": 43.5702,
-      "step": 154260
-    },
-    {
-      "epoch": 0.6232703208264483,
-      "grad_norm": 665.3999633789062,
-      "learning_rate": 1.9199511381853076e-05,
-      "loss": 47.1354,
-      "step": 154270
-    },
-    {
-      "epoch": 0.6233107220918159,
-      "grad_norm": 415.6860656738281,
-      "learning_rate": 1.919611603604875e-05,
-      "loss": 40.6379,
-      "step": 154280
-    },
-    {
-      "epoch": 0.6233511233571836,
-      "grad_norm": 528.6532592773438,
-      "learning_rate": 1.919272080339412e-05,
-      "loss": 42.3219,
-      "step": 154290
-    },
-    {
-      "epoch": 0.6233915246225512,
-      "grad_norm": 700.0554809570312,
-      "learning_rate": 1.91893256839554e-05,
-      "loss": 58.3909,
-      "step": 154300
-    },
-    {
-      "epoch": 0.6234319258879188,
-      "grad_norm": 941.929931640625,
-      "learning_rate": 1.918593067779877e-05,
-      "loss": 63.8411,
-      "step": 154310
-    },
-    {
-      "epoch": 0.6234723271532865,
-      "grad_norm": 319.9075622558594,
-      "learning_rate": 1.9182535784990403e-05,
-      "loss": 66.3877,
-      "step": 154320
-    },
-    {
-      "epoch": 0.6235127284186541,
-      "grad_norm": 357.0832824707031,
-      "learning_rate": 1.9179141005596505e-05,
-      "loss": 60.1818,
-      "step": 154330
-    },
-    {
-      "epoch": 0.6235531296840217,
-      "grad_norm": 571.7152709960938,
-      "learning_rate": 1.9175746339683244e-05,
-      "loss": 67.6065,
-      "step": 154340
-    },
-    {
-      "epoch": 0.6235935309493893,
-      "grad_norm": 501.433349609375,
-      "learning_rate": 1.917235178731681e-05,
-      "loss": 48.194,
-      "step": 154350
-    },
-    {
-      "epoch": 0.6236339322147569,
-      "grad_norm": 1153.9691162109375,
-      "learning_rate": 1.916895734856338e-05,
-      "loss": 49.62,
-      "step": 154360
-    },
-    {
-      "epoch": 0.6236743334801246,
-      "grad_norm": 777.2158813476562,
-      "learning_rate": 1.916556302348912e-05,
-      "loss": 62.7281,
-      "step": 154370
-    },
-    {
-      "epoch": 0.6237147347454922,
-      "grad_norm": 1199.586181640625,
-      "learning_rate": 1.9162168812160218e-05,
-      "loss": 50.8841,
-      "step": 154380
-    },
-    {
-      "epoch": 0.6237551360108599,
-      "grad_norm": 1093.0955810546875,
-      "learning_rate": 1.9158774714642845e-05,
-      "loss": 50.4877,
-      "step": 154390
-    },
-    {
-      "epoch": 0.6237955372762275,
-      "grad_norm": 514.9522705078125,
-      "learning_rate": 1.915538073100316e-05,
-      "loss": 35.3574,
-      "step": 154400
-    },
-    {
-      "epoch": 0.6238359385415951,
-      "grad_norm": 630.1919555664062,
-      "learning_rate": 1.9151986861307344e-05,
-      "loss": 46.5631,
-      "step": 154410
-    },
-    {
-      "epoch": 0.6238763398069628,
-      "grad_norm": 663.0952758789062,
-      "learning_rate": 1.914859310562154e-05,
-      "loss": 48.6632,
-      "step": 154420
-    },
-    {
-      "epoch": 0.6239167410723304,
-      "grad_norm": 3295.518798828125,
-      "learning_rate": 1.914519946401194e-05,
-      "loss": 92.438,
-      "step": 154430
-    },
-    {
-      "epoch": 0.6239571423376981,
-      "grad_norm": 3211.037353515625,
-      "learning_rate": 1.914180593654469e-05,
-      "loss": 51.1125,
-      "step": 154440
-    },
-    {
-      "epoch": 0.6239975436030657,
-      "grad_norm": 391.7280578613281,
-      "learning_rate": 1.9138412523285936e-05,
-      "loss": 52.6217,
-      "step": 154450
-    },
-    {
-      "epoch": 0.6240379448684332,
-      "grad_norm": 1007.160400390625,
-      "learning_rate": 1.9135019224301864e-05,
-      "loss": 62.9541,
-      "step": 154460
-    },
-    {
-      "epoch": 0.6240783461338009,
-      "grad_norm": 1041.9798583984375,
-      "learning_rate": 1.9131626039658607e-05,
-      "loss": 51.7518,
-      "step": 154470
-    },
-    {
-      "epoch": 0.6241187473991685,
-      "grad_norm": 690.2793579101562,
-      "learning_rate": 1.9128232969422315e-05,
-      "loss": 60.4521,
-      "step": 154480
-    },
-    {
-      "epoch": 0.6241591486645361,
-      "grad_norm": 874.3128662109375,
-      "learning_rate": 1.9124840013659155e-05,
-      "loss": 50.6818,
-      "step": 154490
-    },
-    {
-      "epoch": 0.6241995499299038,
-      "grad_norm": 504.86700439453125,
-      "learning_rate": 1.912144717243525e-05,
-      "loss": 39.3337,
-      "step": 154500
-    },
-    {
-      "epoch": 0.6242399511952714,
-      "grad_norm": 1537.7491455078125,
-      "learning_rate": 1.9118054445816767e-05,
-      "loss": 56.018,
-      "step": 154510
-    },
-    {
-      "epoch": 0.6242803524606391,
-      "grad_norm": 689.298095703125,
-      "learning_rate": 1.9114661833869847e-05,
-      "loss": 41.9827,
-      "step": 154520
-    },
-    {
-      "epoch": 0.6243207537260067,
-      "grad_norm": 1061.407470703125,
-      "learning_rate": 1.911126933666061e-05,
-      "loss": 50.2677,
-      "step": 154530
-    },
-    {
-      "epoch": 0.6243611549913743,
-      "grad_norm": 347.8809509277344,
-      "learning_rate": 1.9107876954255217e-05,
-      "loss": 44.7597,
-      "step": 154540
-    },
-    {
-      "epoch": 0.624401556256742,
-      "grad_norm": 1970.9144287109375,
-      "learning_rate": 1.9104484686719795e-05,
-      "loss": 58.5685,
-      "step": 154550
-    },
-    {
-      "epoch": 0.6244419575221096,
-      "grad_norm": 253.61517333984375,
-      "learning_rate": 1.9101092534120478e-05,
-      "loss": 29.4708,
-      "step": 154560
-    },
-    {
-      "epoch": 0.6244823587874773,
-      "grad_norm": 402.3465270996094,
-      "learning_rate": 1.9097700496523404e-05,
-      "loss": 50.5293,
-      "step": 154570
-    },
-    {
-      "epoch": 0.6245227600528449,
-      "grad_norm": 1666.6761474609375,
-      "learning_rate": 1.9094308573994692e-05,
-      "loss": 66.7547,
-      "step": 154580
-    },
-    {
-      "epoch": 0.6245631613182124,
-      "grad_norm": 272.4801940917969,
-      "learning_rate": 1.909091676660048e-05,
-      "loss": 28.7191,
-      "step": 154590
-    },
-    {
-      "epoch": 0.6246035625835801,
-      "grad_norm": 1157.2830810546875,
-      "learning_rate": 1.908752507440689e-05,
-      "loss": 75.1753,
-      "step": 154600
-    },
-    {
-      "epoch": 0.6246439638489477,
-      "grad_norm": 1568.1524658203125,
-      "learning_rate": 1.908413349748003e-05,
-      "loss": 51.9157,
-      "step": 154610
-    },
-    {
-      "epoch": 0.6246843651143154,
-      "grad_norm": 797.2478637695312,
-      "learning_rate": 1.9080742035886045e-05,
-      "loss": 46.6599,
-      "step": 154620
-    },
-    {
-      "epoch": 0.624724766379683,
-      "grad_norm": 253.48960876464844,
-      "learning_rate": 1.9077350689691032e-05,
-      "loss": 49.9191,
-      "step": 154630
-    },
-    {
-      "epoch": 0.6247651676450506,
-      "grad_norm": 780.4749755859375,
-      "learning_rate": 1.9073959458961125e-05,
-      "loss": 66.0298,
-      "step": 154640
-    },
-    {
-      "epoch": 0.6248055689104183,
-      "grad_norm": 413.2430419921875,
-      "learning_rate": 1.907056834376243e-05,
-      "loss": 50.5995,
-      "step": 154650
-    },
-    {
-      "epoch": 0.6248459701757859,
-      "grad_norm": 1501.6806640625,
-      "learning_rate": 1.906717734416105e-05,
-      "loss": 50.1577,
-      "step": 154660
-    },
-    {
-      "epoch": 0.6248863714411536,
-      "grad_norm": 499.1429443359375,
-      "learning_rate": 1.906378646022311e-05,
-      "loss": 52.356,
-      "step": 154670
-    },
-    {
-      "epoch": 0.6249267727065212,
-      "grad_norm": 959.724609375,
-      "learning_rate": 1.9060395692014708e-05,
-      "loss": 59.3793,
-      "step": 154680
-    },
-    {
-      "epoch": 0.6249671739718888,
-      "grad_norm": 436.2023620605469,
-      "learning_rate": 1.905700503960194e-05,
-      "loss": 51.1567,
-      "step": 154690
-    },
-    {
-      "epoch": 0.6250075752372565,
-      "grad_norm": 669.9522094726562,
-      "learning_rate": 1.9053614503050928e-05,
-      "loss": 55.7573,
-      "step": 154700
-    },
-    {
-      "epoch": 0.6250479765026241,
-      "grad_norm": 1216.4349365234375,
-      "learning_rate": 1.9050224082427753e-05,
-      "loss": 62.0031,
-      "step": 154710
-    },
-    {
-      "epoch": 0.6250883777679916,
-      "grad_norm": 698.4359741210938,
-      "learning_rate": 1.9046833777798533e-05,
-      "loss": 63.305,
-      "step": 154720
-    },
-    {
-      "epoch": 0.6251287790333593,
-      "grad_norm": 750.6848754882812,
-      "learning_rate": 1.9043443589229355e-05,
-      "loss": 44.9028,
-      "step": 154730
-    },
-    {
-      "epoch": 0.6251691802987269,
-      "grad_norm": 688.37109375,
-      "learning_rate": 1.9040053516786306e-05,
-      "loss": 57.3157,
-      "step": 154740
-    },
-    {
-      "epoch": 0.6252095815640946,
-      "grad_norm": 384.5275573730469,
-      "learning_rate": 1.9036663560535483e-05,
-      "loss": 63.4863,
-      "step": 154750
-    },
-    {
-      "epoch": 0.6252499828294622,
-      "grad_norm": 1952.0465087890625,
-      "learning_rate": 1.9033273720542975e-05,
-      "loss": 69.0948,
-      "step": 154760
-    },
-    {
-      "epoch": 0.6252903840948298,
-      "grad_norm": 1709.10888671875,
-      "learning_rate": 1.902988399687486e-05,
-      "loss": 119.0979,
-      "step": 154770
-    },
-    {
-      "epoch": 0.6253307853601975,
-      "grad_norm": 499.8750915527344,
-      "learning_rate": 1.9026494389597238e-05,
-      "loss": 36.8402,
-      "step": 154780
-    },
-    {
-      "epoch": 0.6253711866255651,
-      "grad_norm": 612.3027954101562,
-      "learning_rate": 1.9023104898776176e-05,
-      "loss": 34.4344,
-      "step": 154790
-    },
-    {
-      "epoch": 0.6254115878909328,
-      "grad_norm": 390.22845458984375,
-      "learning_rate": 1.9019715524477767e-05,
-      "loss": 71.2229,
-      "step": 154800
-    },
-    {
-      "epoch": 0.6254519891563004,
-      "grad_norm": 1915.9998779296875,
-      "learning_rate": 1.9016326266768088e-05,
-      "loss": 57.8614,
-      "step": 154810
-    },
-    {
-      "epoch": 0.625492390421668,
-      "grad_norm": 317.40777587890625,
-      "learning_rate": 1.90129371257132e-05,
-      "loss": 52.2577,
-      "step": 154820
-    },
-    {
-      "epoch": 0.6255327916870357,
-      "grad_norm": 685.293212890625,
-      "learning_rate": 1.9009548101379194e-05,
-      "loss": 56.8706,
-      "step": 154830
-    },
-    {
-      "epoch": 0.6255731929524033,
-      "grad_norm": 803.1461791992188,
-      "learning_rate": 1.9006159193832125e-05,
-      "loss": 67.1199,
-      "step": 154840
-    },
-    {
-      "epoch": 0.6256135942177709,
-      "grad_norm": 978.5604248046875,
-      "learning_rate": 1.9002770403138065e-05,
-      "loss": 55.7089,
-      "step": 154850
-    },
-    {
-      "epoch": 0.6256539954831385,
-      "grad_norm": 819.3570556640625,
-      "learning_rate": 1.899938172936309e-05,
-      "loss": 69.4563,
-      "step": 154860
-    },
-    {
-      "epoch": 0.6256943967485061,
-      "grad_norm": 326.7989196777344,
-      "learning_rate": 1.8995993172573253e-05,
-      "loss": 62.7371,
-      "step": 154870
-    },
-    {
-      "epoch": 0.6257347980138738,
-      "grad_norm": 1138.951416015625,
-      "learning_rate": 1.8992604732834623e-05,
-      "loss": 71.6288,
-      "step": 154880
-    },
-    {
-      "epoch": 0.6257751992792414,
-      "grad_norm": 1044.2083740234375,
-      "learning_rate": 1.898921641021326e-05,
-      "loss": 30.9312,
-      "step": 154890
-    },
-    {
-      "epoch": 0.625815600544609,
-      "grad_norm": 663.309326171875,
-      "learning_rate": 1.8985828204775206e-05,
-      "loss": 43.888,
-      "step": 154900
-    },
-    {
-      "epoch": 0.6258560018099767,
-      "grad_norm": 392.0646057128906,
-      "learning_rate": 1.898244011658654e-05,
-      "loss": 57.6133,
-      "step": 154910
-    },
-    {
-      "epoch": 0.6258964030753443,
-      "grad_norm": 2559.40771484375,
-      "learning_rate": 1.89790521457133e-05,
-      "loss": 56.647,
-      "step": 154920
-    },
-    {
-      "epoch": 0.625936804340712,
-      "grad_norm": 602.9083862304688,
-      "learning_rate": 1.8975664292221532e-05,
-      "loss": 49.7372,
-      "step": 154930
-    },
-    {
-      "epoch": 0.6259772056060796,
-      "grad_norm": 1236.2510986328125,
-      "learning_rate": 1.89722765561773e-05,
-      "loss": 50.0771,
-      "step": 154940
-    },
-    {
-      "epoch": 0.6260176068714473,
-      "grad_norm": 626.7944946289062,
-      "learning_rate": 1.8968888937646622e-05,
-      "loss": 47.5724,
-      "step": 154950
-    },
-    {
-      "epoch": 0.6260580081368149,
-      "grad_norm": 599.9056396484375,
-      "learning_rate": 1.8965501436695577e-05,
-      "loss": 62.9184,
-      "step": 154960
-    },
-    {
-      "epoch": 0.6260984094021824,
-      "grad_norm": 548.3302001953125,
-      "learning_rate": 1.8962114053390185e-05,
-      "loss": 26.0454,
-      "step": 154970
-    },
-    {
-      "epoch": 0.6261388106675501,
-      "grad_norm": 329.4832458496094,
-      "learning_rate": 1.8958726787796477e-05,
-      "loss": 29.8244,
-      "step": 154980
-    },
-    {
-      "epoch": 0.6261792119329177,
-      "grad_norm": 371.7790832519531,
-      "learning_rate": 1.8955339639980512e-05,
-      "loss": 46.4422,
-      "step": 154990
-    },
-    {
-      "epoch": 0.6262196131982853,
-      "grad_norm": 636.3258056640625,
-      "learning_rate": 1.895195261000831e-05,
-      "loss": 69.783,
-      "step": 155000
-    },
-    {
-      "epoch": 0.626260014463653,
-      "grad_norm": 468.350341796875,
-      "learning_rate": 1.8948565697945907e-05,
-      "loss": 76.4184,
-      "step": 155010
-    },
-    {
-      "epoch": 0.6263004157290206,
-      "grad_norm": 1748.3941650390625,
-      "learning_rate": 1.894517890385933e-05,
-      "loss": 72.7289,
-      "step": 155020
-    },
-    {
-      "epoch": 0.6263408169943883,
-      "grad_norm": 265.7440490722656,
-      "learning_rate": 1.8941792227814597e-05,
-      "loss": 94.1757,
-      "step": 155030
-    },
-    {
-      "epoch": 0.6263812182597559,
-      "grad_norm": 1297.6448974609375,
-      "learning_rate": 1.893840566987776e-05,
-      "loss": 45.0199,
-      "step": 155040
-    },
-    {
-      "epoch": 0.6264216195251235,
-      "grad_norm": 1067.8428955078125,
-      "learning_rate": 1.893501923011482e-05,
-      "loss": 73.8872,
-      "step": 155050
-    },
-    {
-      "epoch": 0.6264620207904912,
-      "grad_norm": 720.9359741210938,
-      "learning_rate": 1.8931632908591796e-05,
-      "loss": 40.9904,
-      "step": 155060
-    },
-    {
-      "epoch": 0.6265024220558588,
-      "grad_norm": 430.0225830078125,
-      "learning_rate": 1.892824670537472e-05,
-      "loss": 51.2785,
-      "step": 155070
-    },
-    {
-      "epoch": 0.6265428233212265,
-      "grad_norm": 952.5421752929688,
-      "learning_rate": 1.8924860620529594e-05,
-      "loss": 57.799,
-      "step": 155080
-    },
-    {
-      "epoch": 0.6265832245865941,
-      "grad_norm": 811.53857421875,
-      "learning_rate": 1.8921474654122444e-05,
-      "loss": 56.84,
-      "step": 155090
-    },
-    {
-      "epoch": 0.6266236258519616,
-      "grad_norm": 1707.3125,
-      "learning_rate": 1.891808880621928e-05,
-      "loss": 52.2367,
-      "step": 155100
-    },
-    {
-      "epoch": 0.6266640271173293,
-      "grad_norm": 683.6460571289062,
-      "learning_rate": 1.891470307688609e-05,
-      "loss": 75.291,
-      "step": 155110
-    },
-    {
-      "epoch": 0.6267044283826969,
-      "grad_norm": 809.841796875,
-      "learning_rate": 1.891131746618891e-05,
-      "loss": 71.0322,
-      "step": 155120
-    },
-    {
-      "epoch": 0.6267448296480645,
-      "grad_norm": 601.0204467773438,
-      "learning_rate": 1.8907931974193728e-05,
-      "loss": 55.0767,
-      "step": 155130
-    },
-    {
-      "epoch": 0.6267852309134322,
-      "grad_norm": 1440.623779296875,
-      "learning_rate": 1.890454660096654e-05,
-      "loss": 75.6987,
-      "step": 155140
-    },
-    {
-      "epoch": 0.6268256321787998,
-      "grad_norm": 258.63116455078125,
-      "learning_rate": 1.890116134657336e-05,
-      "loss": 61.5404,
-      "step": 155150
-    },
-    {
-      "epoch": 0.6268660334441675,
-      "grad_norm": 507.3258056640625,
-      "learning_rate": 1.8897776211080182e-05,
-      "loss": 55.1314,
-      "step": 155160
-    },
-    {
-      "epoch": 0.6269064347095351,
-      "grad_norm": 755.1796264648438,
-      "learning_rate": 1.8894391194552997e-05,
-      "loss": 53.2773,
-      "step": 155170
-    },
-    {
-      "epoch": 0.6269468359749027,
-      "grad_norm": 618.8252563476562,
-      "learning_rate": 1.8891006297057798e-05,
-      "loss": 107.6506,
-      "step": 155180
-    },
-    {
-      "epoch": 0.6269872372402704,
-      "grad_norm": 376.06964111328125,
-      "learning_rate": 1.8887621518660577e-05,
-      "loss": 51.3056,
-      "step": 155190
-    },
-    {
-      "epoch": 0.627027638505638,
-      "grad_norm": 472.5889892578125,
-      "learning_rate": 1.888423685942732e-05,
-      "loss": 56.5211,
-      "step": 155200
-    },
-    {
-      "epoch": 0.6270680397710057,
-      "grad_norm": 652.8603515625,
-      "learning_rate": 1.8880852319424018e-05,
-      "loss": 37.9688,
-      "step": 155210
-    },
-    {
-      "epoch": 0.6271084410363733,
-      "grad_norm": 3448.654052734375,
-      "learning_rate": 1.887746789871664e-05,
-      "loss": 59.2658,
-      "step": 155220
-    },
-    {
-      "epoch": 0.6271488423017408,
-      "grad_norm": 392.4132995605469,
-      "learning_rate": 1.887408359737119e-05,
-      "loss": 35.5017,
-      "step": 155230
-    },
-    {
-      "epoch": 0.6271892435671085,
-      "grad_norm": 1030.261474609375,
-      "learning_rate": 1.8870699415453627e-05,
-      "loss": 54.3306,
-      "step": 155240
-    },
-    {
-      "epoch": 0.6272296448324761,
-      "grad_norm": 599.7904663085938,
-      "learning_rate": 1.8867315353029935e-05,
-      "loss": 58.7418,
-      "step": 155250
-    },
-    {
-      "epoch": 0.6272700460978438,
-      "grad_norm": 595.4382934570312,
-      "learning_rate": 1.886393141016609e-05,
-      "loss": 64.527,
-      "step": 155260
-    },
-    {
-      "epoch": 0.6273104473632114,
-      "grad_norm": 762.1326293945312,
-      "learning_rate": 1.886054758692806e-05,
-      "loss": 62.5916,
-      "step": 155270
-    },
-    {
-      "epoch": 0.627350848628579,
-      "grad_norm": 1090.7520751953125,
-      "learning_rate": 1.885716388338182e-05,
-      "loss": 55.2791,
-      "step": 155280
-    },
-    {
-      "epoch": 0.6273912498939467,
-      "grad_norm": 1541.6251220703125,
-      "learning_rate": 1.8853780299593332e-05,
-      "loss": 64.5295,
-      "step": 155290
-    },
-    {
-      "epoch": 0.6274316511593143,
-      "grad_norm": 1311.8800048828125,
-      "learning_rate": 1.885039683562855e-05,
-      "loss": 78.9983,
-      "step": 155300
-    },
-    {
-      "epoch": 0.627472052424682,
-      "grad_norm": 399.0144958496094,
-      "learning_rate": 1.884701349155346e-05,
-      "loss": 46.4417,
-      "step": 155310
-    },
-    {
-      "epoch": 0.6275124536900496,
-      "grad_norm": 1750.837890625,
-      "learning_rate": 1.8843630267434e-05,
-      "loss": 63.7231,
-      "step": 155320
-    },
-    {
-      "epoch": 0.6275528549554172,
-      "grad_norm": 1599.4073486328125,
-      "learning_rate": 1.8840247163336143e-05,
-      "loss": 41.7063,
-      "step": 155330
-    },
-    {
-      "epoch": 0.6275932562207849,
-      "grad_norm": 479.2684326171875,
-      "learning_rate": 1.883686417932584e-05,
-      "loss": 49.5772,
-      "step": 155340
-    },
-    {
-      "epoch": 0.6276336574861525,
-      "grad_norm": 518.2067260742188,
-      "learning_rate": 1.8833481315469042e-05,
-      "loss": 41.1463,
-      "step": 155350
-    },
-    {
-      "epoch": 0.62767405875152,
-      "grad_norm": 1002.8010864257812,
-      "learning_rate": 1.8830098571831705e-05,
-      "loss": 45.5635,
-      "step": 155360
-    },
-    {
-      "epoch": 0.6277144600168877,
-      "grad_norm": 305.9928894042969,
-      "learning_rate": 1.882671594847977e-05,
-      "loss": 80.5473,
-      "step": 155370
-    },
-    {
-      "epoch": 0.6277548612822553,
-      "grad_norm": 324.7518615722656,
-      "learning_rate": 1.8823333445479174e-05,
-      "loss": 44.882,
-      "step": 155380
-    },
-    {
-      "epoch": 0.627795262547623,
-      "grad_norm": 1423.6795654296875,
-      "learning_rate": 1.8819951062895885e-05,
-      "loss": 62.9079,
-      "step": 155390
-    },
-    {
-      "epoch": 0.6278356638129906,
-      "grad_norm": 1083.814453125,
-      "learning_rate": 1.8816568800795822e-05,
-      "loss": 74.6381,
-      "step": 155400
-    },
-    {
-      "epoch": 0.6278760650783582,
-      "grad_norm": 621.3231811523438,
-      "learning_rate": 1.8813186659244943e-05,
-      "loss": 70.4143,
-      "step": 155410
-    },
-    {
-      "epoch": 0.6279164663437259,
-      "grad_norm": 490.48724365234375,
-      "learning_rate": 1.8809804638309177e-05,
-      "loss": 52.3341,
-      "step": 155420
-    },
-    {
-      "epoch": 0.6279568676090935,
-      "grad_norm": 816.6148681640625,
-      "learning_rate": 1.880642273805445e-05,
-      "loss": 36.432,
-      "step": 155430
-    },
-    {
-      "epoch": 0.6279972688744612,
-      "grad_norm": 910.425048828125,
-      "learning_rate": 1.8803040958546707e-05,
-      "loss": 46.4939,
-      "step": 155440
-    },
-    {
-      "epoch": 0.6280376701398288,
-      "grad_norm": 787.06640625,
-      "learning_rate": 1.879965929985187e-05,
-      "loss": 48.2395,
-      "step": 155450
-    },
-    {
-      "epoch": 0.6280780714051964,
-      "grad_norm": 634.6830444335938,
-      "learning_rate": 1.8796277762035856e-05,
-      "loss": 28.2061,
-      "step": 155460
-    },
-    {
-      "epoch": 0.6281184726705641,
-      "grad_norm": 300.8420715332031,
-      "learning_rate": 1.879289634516461e-05,
-      "loss": 57.9607,
-      "step": 155470
-    },
-    {
-      "epoch": 0.6281588739359317,
-      "grad_norm": 638.6072387695312,
-      "learning_rate": 1.8789515049304038e-05,
-      "loss": 48.7444,
-      "step": 155480
-    },
-    {
-      "epoch": 0.6281992752012993,
-      "grad_norm": 429.5278625488281,
-      "learning_rate": 1.8786133874520078e-05,
-      "loss": 46.0567,
-      "step": 155490
-    },
-    {
-      "epoch": 0.6282396764666669,
-      "grad_norm": 405.1448059082031,
-      "learning_rate": 1.8782752820878634e-05,
-      "loss": 48.6294,
-      "step": 155500
-    },
-    {
-      "epoch": 0.6282800777320345,
-      "grad_norm": 22.98813247680664,
-      "learning_rate": 1.8779371888445624e-05,
-      "loss": 41.636,
-      "step": 155510
-    },
-    {
-      "epoch": 0.6283204789974022,
-      "grad_norm": 445.4391174316406,
-      "learning_rate": 1.8775991077286965e-05,
-      "loss": 49.7633,
-      "step": 155520
-    },
-    {
-      "epoch": 0.6283608802627698,
-      "grad_norm": 905.4905395507812,
-      "learning_rate": 1.8772610387468555e-05,
-      "loss": 84.4601,
-      "step": 155530
-    },
-    {
-      "epoch": 0.6284012815281375,
-      "grad_norm": 1306.422119140625,
-      "learning_rate": 1.8769229819056315e-05,
-      "loss": 66.4575,
-      "step": 155540
-    },
-    {
-      "epoch": 0.6284416827935051,
-      "grad_norm": 857.094970703125,
-      "learning_rate": 1.8765849372116153e-05,
-      "loss": 54.3967,
-      "step": 155550
-    },
-    {
-      "epoch": 0.6284820840588727,
-      "grad_norm": 795.1137084960938,
-      "learning_rate": 1.8762469046713956e-05,
-      "loss": 75.3259,
-      "step": 155560
-    },
-    {
-      "epoch": 0.6285224853242404,
-      "grad_norm": 763.9700317382812,
-      "learning_rate": 1.8759088842915644e-05,
-      "loss": 48.5607,
-      "step": 155570
-    },
-    {
-      "epoch": 0.628562886589608,
-      "grad_norm": 655.9850463867188,
-      "learning_rate": 1.8755708760787113e-05,
-      "loss": 52.3753,
-      "step": 155580
-    },
-    {
-      "epoch": 0.6286032878549757,
-      "grad_norm": 505.8095703125,
-      "learning_rate": 1.8752328800394242e-05,
-      "loss": 63.009,
-      "step": 155590
-    },
-    {
-      "epoch": 0.6286436891203433,
-      "grad_norm": 500.4537353515625,
-      "learning_rate": 1.8748948961802948e-05,
-      "loss": 42.3465,
-      "step": 155600
-    },
-    {
-      "epoch": 0.6286840903857108,
-      "grad_norm": 431.3822021484375,
-      "learning_rate": 1.8745569245079104e-05,
-      "loss": 49.0103,
-      "step": 155610
-    },
-    {
-      "epoch": 0.6287244916510785,
-      "grad_norm": 1207.086669921875,
-      "learning_rate": 1.8742189650288615e-05,
-      "loss": 47.9568,
-      "step": 155620
-    },
-    {
-      "epoch": 0.6287648929164461,
-      "grad_norm": 440.156982421875,
-      "learning_rate": 1.8738810177497365e-05,
-      "loss": 38.8417,
-      "step": 155630
-    },
-    {
-      "epoch": 0.6288052941818137,
-      "grad_norm": 274.6180725097656,
-      "learning_rate": 1.873543082677122e-05,
-      "loss": 31.751,
-      "step": 155640
-    },
-    {
-      "epoch": 0.6288456954471814,
-      "grad_norm": 376.49639892578125,
-      "learning_rate": 1.8732051598176086e-05,
-      "loss": 63.3421,
-      "step": 155650
-    },
-    {
-      "epoch": 0.628886096712549,
-      "grad_norm": 2292.678955078125,
-      "learning_rate": 1.872867249177783e-05,
-      "loss": 93.0058,
-      "step": 155660
-    },
-    {
-      "epoch": 0.6289264979779167,
-      "grad_norm": 803.1299438476562,
-      "learning_rate": 1.872529350764233e-05,
-      "loss": 37.7863,
-      "step": 155670
-    },
-    {
-      "epoch": 0.6289668992432843,
-      "grad_norm": 1643.458984375,
-      "learning_rate": 1.872191464583547e-05,
-      "loss": 78.7152,
-      "step": 155680
-    },
-    {
-      "epoch": 0.629007300508652,
-      "grad_norm": 673.4326171875,
-      "learning_rate": 1.8718535906423106e-05,
-      "loss": 43.4824,
-      "step": 155690
-    },
-    {
-      "epoch": 0.6290477017740196,
-      "grad_norm": 939.1830444335938,
-      "learning_rate": 1.871515728947113e-05,
-      "loss": 47.4452,
-      "step": 155700
-    },
-    {
-      "epoch": 0.6290881030393872,
-      "grad_norm": 1257.9764404296875,
-      "learning_rate": 1.8711778795045398e-05,
-      "loss": 52.2282,
-      "step": 155710
-    },
-    {
-      "epoch": 0.6291285043047549,
-      "grad_norm": 939.0426025390625,
-      "learning_rate": 1.8708400423211764e-05,
-      "loss": 37.8956,
-      "step": 155720
-    },
-    {
-      "epoch": 0.6291689055701225,
-      "grad_norm": 613.3606567382812,
-      "learning_rate": 1.8705022174036114e-05,
-      "loss": 40.0705,
-      "step": 155730
-    },
-    {
-      "epoch": 0.62920930683549,
-      "grad_norm": 487.5195007324219,
-      "learning_rate": 1.8701644047584293e-05,
-      "loss": 30.0292,
-      "step": 155740
-    },
-    {
-      "epoch": 0.6292497081008577,
-      "grad_norm": 642.8336181640625,
-      "learning_rate": 1.869826604392216e-05,
-      "loss": 49.1987,
-      "step": 155750
-    },
-    {
-      "epoch": 0.6292901093662253,
-      "grad_norm": 501.4791564941406,
-      "learning_rate": 1.869488816311558e-05,
-      "loss": 42.2183,
-      "step": 155760
-    },
-    {
-      "epoch": 0.629330510631593,
-      "grad_norm": 763.4742431640625,
-      "learning_rate": 1.86915104052304e-05,
-      "loss": 67.459,
-      "step": 155770
-    },
-    {
-      "epoch": 0.6293709118969606,
-      "grad_norm": 464.489013671875,
-      "learning_rate": 1.8688132770332476e-05,
-      "loss": 60.1081,
-      "step": 155780
-    },
-    {
-      "epoch": 0.6294113131623282,
-      "grad_norm": 241.61941528320312,
-      "learning_rate": 1.8684755258487653e-05,
-      "loss": 76.872,
-      "step": 155790
-    },
-    {
-      "epoch": 0.6294517144276959,
-      "grad_norm": 514.704345703125,
-      "learning_rate": 1.868137786976177e-05,
-      "loss": 67.6247,
-      "step": 155800
-    },
-    {
-      "epoch": 0.6294921156930635,
-      "grad_norm": 685.1314086914062,
-      "learning_rate": 1.8678000604220683e-05,
-      "loss": 48.1277,
-      "step": 155810
-    },
-    {
-      "epoch": 0.6295325169584312,
-      "grad_norm": 363.3772277832031,
-      "learning_rate": 1.8674623461930233e-05,
-      "loss": 64.7407,
-      "step": 155820
-    },
-    {
-      "epoch": 0.6295729182237988,
-      "grad_norm": 1206.5020751953125,
-      "learning_rate": 1.8671246442956243e-05,
-      "loss": 47.2309,
-      "step": 155830
-    },
-    {
-      "epoch": 0.6296133194891664,
-      "grad_norm": 1087.311279296875,
-      "learning_rate": 1.8667869547364576e-05,
-      "loss": 55.6458,
-      "step": 155840
-    },
-    {
-      "epoch": 0.6296537207545341,
-      "grad_norm": 1133.510498046875,
-      "learning_rate": 1.8664492775221042e-05,
-      "loss": 56.3427,
-      "step": 155850
-    },
-    {
-      "epoch": 0.6296941220199017,
-      "grad_norm": 513.0890502929688,
-      "learning_rate": 1.866111612659149e-05,
-      "loss": 52.5945,
-      "step": 155860
-    },
-    {
-      "epoch": 0.6297345232852692,
-      "grad_norm": 1111.2569580078125,
-      "learning_rate": 1.865773960154174e-05,
-      "loss": 44.945,
-      "step": 155870
-    },
-    {
-      "epoch": 0.6297749245506369,
-      "grad_norm": 566.517333984375,
-      "learning_rate": 1.865436320013762e-05,
-      "loss": 41.3349,
-      "step": 155880
-    },
-    {
-      "epoch": 0.6298153258160045,
-      "grad_norm": 1226.1885986328125,
-      "learning_rate": 1.865098692244496e-05,
-      "loss": 61.058,
-      "step": 155890
-    },
-    {
-      "epoch": 0.6298557270813722,
-      "grad_norm": 642.2447509765625,
-      "learning_rate": 1.864761076852958e-05,
-      "loss": 55.2996,
-      "step": 155900
-    },
-    {
-      "epoch": 0.6298961283467398,
-      "grad_norm": 963.623291015625,
-      "learning_rate": 1.864423473845729e-05,
-      "loss": 75.762,
-      "step": 155910
-    },
-    {
-      "epoch": 0.6299365296121074,
-      "grad_norm": 656.4022216796875,
-      "learning_rate": 1.864085883229392e-05,
-      "loss": 30.6479,
-      "step": 155920
-    },
-    {
-      "epoch": 0.6299769308774751,
-      "grad_norm": 972.303466796875,
-      "learning_rate": 1.8637483050105274e-05,
-      "loss": 55.5703,
-      "step": 155930
-    },
-    {
-      "epoch": 0.6300173321428427,
-      "grad_norm": 0.0,
-      "learning_rate": 1.8634107391957186e-05,
-      "loss": 34.8945,
-      "step": 155940
-    },
-    {
-      "epoch": 0.6300577334082104,
-      "grad_norm": 655.4844360351562,
-      "learning_rate": 1.863073185791545e-05,
-      "loss": 32.5324,
-      "step": 155950
-    },
-    {
-      "epoch": 0.630098134673578,
-      "grad_norm": 3676.433349609375,
-      "learning_rate": 1.8627356448045867e-05,
-      "loss": 62.8763,
-      "step": 155960
-    },
-    {
-      "epoch": 0.6301385359389456,
-      "grad_norm": 1670.3465576171875,
-      "learning_rate": 1.8623981162414263e-05,
-      "loss": 54.1714,
-      "step": 155970
-    },
-    {
-      "epoch": 0.6301789372043133,
-      "grad_norm": 976.9774169921875,
-      "learning_rate": 1.862060600108642e-05,
-      "loss": 49.8334,
-      "step": 155980
-    },
-    {
-      "epoch": 0.6302193384696809,
-      "grad_norm": 180.9348602294922,
-      "learning_rate": 1.861723096412814e-05,
-      "loss": 39.5339,
-      "step": 155990
-    },
-    {
-      "epoch": 0.6302597397350485,
-      "grad_norm": 754.650634765625,
-      "learning_rate": 1.8613856051605243e-05,
-      "loss": 24.7535,
-      "step": 156000
-    },
-    {
-      "epoch": 0.6303001410004161,
-      "grad_norm": 121.3849105834961,
-      "learning_rate": 1.8610481263583496e-05,
-      "loss": 27.1816,
-      "step": 156010
-    },
-    {
-      "epoch": 0.6303405422657837,
-      "grad_norm": 372.6917724609375,
-      "learning_rate": 1.8607106600128715e-05,
-      "loss": 54.0215,
-      "step": 156020
-    },
-    {
-      "epoch": 0.6303809435311514,
-      "grad_norm": 767.4069213867188,
-      "learning_rate": 1.8603732061306683e-05,
-      "loss": 37.4301,
-      "step": 156030
-    },
-    {
-      "epoch": 0.630421344796519,
-      "grad_norm": 181.80638122558594,
-      "learning_rate": 1.8600357647183185e-05,
-      "loss": 99.4061,
-      "step": 156040
-    },
-    {
-      "epoch": 0.6304617460618867,
-      "grad_norm": 380.4937438964844,
-      "learning_rate": 1.8596983357824012e-05,
-      "loss": 39.366,
-      "step": 156050
-    },
-    {
-      "epoch": 0.6305021473272543,
-      "grad_norm": 316.6199951171875,
-      "learning_rate": 1.8593609193294947e-05,
-      "loss": 47.472,
-      "step": 156060
-    },
-    {
-      "epoch": 0.6305425485926219,
-      "grad_norm": 0.0,
-      "learning_rate": 1.8590235153661757e-05,
-      "loss": 47.0838,
-      "step": 156070
-    },
-    {
-      "epoch": 0.6305829498579896,
-      "grad_norm": 1608.3348388671875,
-      "learning_rate": 1.8586861238990244e-05,
-      "loss": 49.0129,
-      "step": 156080
-    },
-    {
-      "epoch": 0.6306233511233572,
-      "grad_norm": 1044.3524169921875,
-      "learning_rate": 1.858348744934616e-05,
-      "loss": 52.0046,
-      "step": 156090
-    },
-    {
-      "epoch": 0.6306637523887249,
-      "grad_norm": 1383.7996826171875,
-      "learning_rate": 1.8580113784795305e-05,
-      "loss": 70.3039,
-      "step": 156100
-    },
-    {
-      "epoch": 0.6307041536540925,
-      "grad_norm": 833.0409545898438,
-      "learning_rate": 1.857674024540344e-05,
-      "loss": 50.6667,
-      "step": 156110
-    },
-    {
-      "epoch": 0.6307445549194601,
-      "grad_norm": 1461.901611328125,
-      "learning_rate": 1.8573366831236323e-05,
-      "loss": 54.4854,
-      "step": 156120
-    },
-    {
-      "epoch": 0.6307849561848277,
-      "grad_norm": 676.2404174804688,
-      "learning_rate": 1.856999354235973e-05,
-      "loss": 72.485,
-      "step": 156130
-    },
-    {
-      "epoch": 0.6308253574501953,
-      "grad_norm": 532.0584716796875,
-      "learning_rate": 1.8566620378839417e-05,
-      "loss": 42.6068,
-      "step": 156140
-    },
-    {
-      "epoch": 0.6308657587155629,
-      "grad_norm": 513.3949584960938,
-      "learning_rate": 1.856324734074116e-05,
-      "loss": 93.1572,
-      "step": 156150
-    },
-    {
-      "epoch": 0.6309061599809306,
-      "grad_norm": 1914.272216796875,
-      "learning_rate": 1.8559874428130706e-05,
-      "loss": 65.1624,
-      "step": 156160
-    },
-    {
-      "epoch": 0.6309465612462982,
-      "grad_norm": 542.9099731445312,
-      "learning_rate": 1.8556501641073813e-05,
-      "loss": 49.7561,
-      "step": 156170
-    },
-    {
-      "epoch": 0.6309869625116659,
-      "grad_norm": 781.0393676757812,
-      "learning_rate": 1.8553128979636243e-05,
-      "loss": 47.1173,
-      "step": 156180
-    },
-    {
-      "epoch": 0.6310273637770335,
-      "grad_norm": 813.7147827148438,
-      "learning_rate": 1.8549756443883746e-05,
-      "loss": 39.8702,
-      "step": 156190
-    },
-    {
-      "epoch": 0.6310677650424011,
-      "grad_norm": 1207.70849609375,
-      "learning_rate": 1.8546384033882062e-05,
-      "loss": 84.3893,
-      "step": 156200
-    },
-    {
-      "epoch": 0.6311081663077688,
-      "grad_norm": 583.6326904296875,
-      "learning_rate": 1.8543011749696944e-05,
-      "loss": 72.3307,
-      "step": 156210
-    },
-    {
-      "epoch": 0.6311485675731364,
-      "grad_norm": 815.402587890625,
-      "learning_rate": 1.8539639591394133e-05,
-      "loss": 54.2483,
-      "step": 156220
-    },
-    {
-      "epoch": 0.6311889688385041,
-      "grad_norm": 412.4958190917969,
-      "learning_rate": 1.8536267559039384e-05,
-      "loss": 55.6868,
-      "step": 156230
-    },
-    {
-      "epoch": 0.6312293701038717,
-      "grad_norm": 780.7228393554688,
-      "learning_rate": 1.8532895652698422e-05,
-      "loss": 52.2046,
-      "step": 156240
-    },
-    {
-      "epoch": 0.6312697713692392,
-      "grad_norm": 401.5675048828125,
-      "learning_rate": 1.852952387243698e-05,
-      "loss": 46.8964,
-      "step": 156250
-    },
-    {
-      "epoch": 0.6313101726346069,
-      "grad_norm": 464.23828125,
-      "learning_rate": 1.8526152218320815e-05,
-      "loss": 44.0663,
-      "step": 156260
-    },
-    {
-      "epoch": 0.6313505738999745,
-      "grad_norm": 363.0769958496094,
-      "learning_rate": 1.852278069041564e-05,
-      "loss": 69.285,
-      "step": 156270
-    },
-    {
-      "epoch": 0.6313909751653421,
-      "grad_norm": 281.2511901855469,
-      "learning_rate": 1.851940928878718e-05,
-      "loss": 51.6398,
-      "step": 156280
-    },
-    {
-      "epoch": 0.6314313764307098,
-      "grad_norm": 965.007080078125,
-      "learning_rate": 1.8516038013501188e-05,
-      "loss": 43.5258,
-      "step": 156290
-    },
-    {
-      "epoch": 0.6314717776960774,
-      "grad_norm": 1698.4444580078125,
-      "learning_rate": 1.8512666864623365e-05,
-      "loss": 76.7266,
-      "step": 156300
-    },
-    {
-      "epoch": 0.6315121789614451,
-      "grad_norm": 494.7231140136719,
-      "learning_rate": 1.8509295842219448e-05,
-      "loss": 43.2274,
-      "step": 156310
-    },
-    {
-      "epoch": 0.6315525802268127,
-      "grad_norm": 723.9692993164062,
-      "learning_rate": 1.8505924946355147e-05,
-      "loss": 49.6911,
-      "step": 156320
-    },
-    {
-      "epoch": 0.6315929814921803,
-      "grad_norm": 632.7535400390625,
-      "learning_rate": 1.8502554177096177e-05,
-      "loss": 104.3508,
-      "step": 156330
-    },
-    {
-      "epoch": 0.631633382757548,
-      "grad_norm": 908.9913940429688,
-      "learning_rate": 1.8499183534508263e-05,
-      "loss": 58.2353,
-      "step": 156340
-    },
-    {
-      "epoch": 0.6316737840229156,
-      "grad_norm": 802.5706787109375,
-      "learning_rate": 1.8495813018657116e-05,
-      "loss": 40.5866,
-      "step": 156350
-    },
-    {
-      "epoch": 0.6317141852882833,
-      "grad_norm": 604.8616333007812,
-      "learning_rate": 1.8492442629608434e-05,
-      "loss": 33.7931,
-      "step": 156360
-    },
-    {
-      "epoch": 0.6317545865536509,
-      "grad_norm": 625.544189453125,
-      "learning_rate": 1.848907236742794e-05,
-      "loss": 43.9119,
-      "step": 156370
-    },
-    {
-      "epoch": 0.6317949878190184,
-      "grad_norm": 3475.124755859375,
-      "learning_rate": 1.848570223218133e-05,
-      "loss": 60.458,
-      "step": 156380
-    },
-    {
-      "epoch": 0.6318353890843861,
-      "grad_norm": 398.1371765136719,
-      "learning_rate": 1.8482332223934314e-05,
-      "loss": 62.6251,
-      "step": 156390
-    },
-    {
-      "epoch": 0.6318757903497537,
-      "grad_norm": 893.6590576171875,
-      "learning_rate": 1.8478962342752583e-05,
-      "loss": 50.1743,
-      "step": 156400
-    },
-    {
-      "epoch": 0.6319161916151214,
-      "grad_norm": 1160.157470703125,
-      "learning_rate": 1.847559258870183e-05,
-      "loss": 41.7392,
-      "step": 156410
-    },
-    {
-      "epoch": 0.631956592880489,
-      "grad_norm": 1044.7193603515625,
-      "learning_rate": 1.847222296184777e-05,
-      "loss": 69.7084,
-      "step": 156420
-    },
-    {
-      "epoch": 0.6319969941458566,
-      "grad_norm": 606.6983032226562,
-      "learning_rate": 1.8468853462256085e-05,
-      "loss": 34.135,
-      "step": 156430
-    },
-    {
-      "epoch": 0.6320373954112243,
-      "grad_norm": 560.6671752929688,
-      "learning_rate": 1.846548408999245e-05,
-      "loss": 48.3755,
-      "step": 156440
-    },
-    {
-      "epoch": 0.6320777966765919,
-      "grad_norm": 376.26239013671875,
-      "learning_rate": 1.846211484512258e-05,
-      "loss": 38.9917,
-      "step": 156450
-    },
-    {
-      "epoch": 0.6321181979419596,
-      "grad_norm": 395.11138916015625,
-      "learning_rate": 1.8458745727712144e-05,
-      "loss": 41.6774,
-      "step": 156460
-    },
-    {
-      "epoch": 0.6321585992073272,
-      "grad_norm": 710.415283203125,
-      "learning_rate": 1.845537673782683e-05,
-      "loss": 53.0831,
-      "step": 156470
-    },
-    {
-      "epoch": 0.6321990004726948,
-      "grad_norm": 686.17919921875,
-      "learning_rate": 1.8452007875532317e-05,
-      "loss": 99.8278,
-      "step": 156480
-    },
-    {
-      "epoch": 0.6322394017380625,
-      "grad_norm": 486.0894775390625,
-      "learning_rate": 1.844863914089427e-05,
-      "loss": 42.6447,
-      "step": 156490
-    },
-    {
-      "epoch": 0.6322798030034301,
-      "grad_norm": 505.011962890625,
-      "learning_rate": 1.8445270533978388e-05,
-      "loss": 28.8061,
-      "step": 156500
-    },
-    {
-      "epoch": 0.6323202042687976,
-      "grad_norm": 680.4307861328125,
-      "learning_rate": 1.844190205485033e-05,
-      "loss": 57.4097,
-      "step": 156510
-    },
-    {
-      "epoch": 0.6323606055341653,
-      "grad_norm": 515.7891845703125,
-      "learning_rate": 1.8438533703575754e-05,
-      "loss": 55.8497,
-      "step": 156520
-    },
-    {
-      "epoch": 0.6324010067995329,
-      "grad_norm": 468.45953369140625,
-      "learning_rate": 1.8435165480220356e-05,
-      "loss": 63.9403,
-      "step": 156530
-    },
-    {
-      "epoch": 0.6324414080649006,
-      "grad_norm": 472.6759948730469,
-      "learning_rate": 1.8431797384849783e-05,
-      "loss": 24.1151,
-      "step": 156540
-    },
-    {
-      "epoch": 0.6324818093302682,
-      "grad_norm": 1147.822509765625,
-      "learning_rate": 1.84284294175297e-05,
-      "loss": 57.033,
-      "step": 156550
-    },
-    {
-      "epoch": 0.6325222105956358,
-      "grad_norm": 2694.32666015625,
-      "learning_rate": 1.8425061578325772e-05,
-      "loss": 46.6325,
-      "step": 156560
-    },
-    {
-      "epoch": 0.6325626118610035,
-      "grad_norm": 987.6583251953125,
-      "learning_rate": 1.8421693867303653e-05,
-      "loss": 47.813,
-      "step": 156570
-    },
-    {
-      "epoch": 0.6326030131263711,
-      "grad_norm": 856.338134765625,
-      "learning_rate": 1.8418326284528996e-05,
-      "loss": 48.7433,
-      "step": 156580
-    },
-    {
-      "epoch": 0.6326434143917388,
-      "grad_norm": 519.7534790039062,
-      "learning_rate": 1.8414958830067464e-05,
-      "loss": 42.5235,
-      "step": 156590
-    },
-    {
-      "epoch": 0.6326838156571064,
-      "grad_norm": 1199.5218505859375,
-      "learning_rate": 1.841159150398469e-05,
-      "loss": 70.6071,
-      "step": 156600
-    },
-    {
-      "epoch": 0.632724216922474,
-      "grad_norm": 636.864990234375,
-      "learning_rate": 1.8408224306346335e-05,
-      "loss": 42.0584,
-      "step": 156610
-    },
-    {
-      "epoch": 0.6327646181878417,
-      "grad_norm": 765.359130859375,
-      "learning_rate": 1.8404857237218038e-05,
-      "loss": 65.0647,
-      "step": 156620
-    },
-    {
-      "epoch": 0.6328050194532093,
-      "grad_norm": 180.48377990722656,
-      "learning_rate": 1.8401490296665445e-05,
-      "loss": 50.7906,
-      "step": 156630
-    },
-    {
-      "epoch": 0.6328454207185769,
-      "grad_norm": 529.7195434570312,
-      "learning_rate": 1.8398123484754203e-05,
-      "loss": 43.8623,
-      "step": 156640
-    },
-    {
-      "epoch": 0.6328858219839445,
-      "grad_norm": 915.6985473632812,
-      "learning_rate": 1.839475680154994e-05,
-      "loss": 71.13,
-      "step": 156650
-    },
-    {
-      "epoch": 0.6329262232493121,
-      "grad_norm": 469.94805908203125,
-      "learning_rate": 1.8391390247118295e-05,
-      "loss": 45.6889,
-      "step": 156660
-    },
-    {
-      "epoch": 0.6329666245146798,
-      "grad_norm": 915.044189453125,
-      "learning_rate": 1.838802382152489e-05,
-      "loss": 50.4943,
-      "step": 156670
-    },
-    {
-      "epoch": 0.6330070257800474,
-      "grad_norm": 647.909423828125,
-      "learning_rate": 1.8384657524835376e-05,
-      "loss": 51.613,
-      "step": 156680
-    },
-    {
-      "epoch": 0.6330474270454151,
-      "grad_norm": 954.224365234375,
-      "learning_rate": 1.8381291357115367e-05,
-      "loss": 69.1072,
-      "step": 156690
-    },
-    {
-      "epoch": 0.6330878283107827,
-      "grad_norm": 1955.664306640625,
-      "learning_rate": 1.8377925318430477e-05,
-      "loss": 89.2116,
-      "step": 156700
-    },
-    {
-      "epoch": 0.6331282295761503,
-      "grad_norm": 177.62167358398438,
-      "learning_rate": 1.8374559408846357e-05,
-      "loss": 65.1419,
-      "step": 156710
-    },
-    {
-      "epoch": 0.633168630841518,
-      "grad_norm": 12293.146484375,
-      "learning_rate": 1.8371193628428613e-05,
-      "loss": 78.1634,
-      "step": 156720
-    },
-    {
-      "epoch": 0.6332090321068856,
-      "grad_norm": 689.4757690429688,
-      "learning_rate": 1.8367827977242858e-05,
-      "loss": 52.0371,
-      "step": 156730
-    },
-    {
-      "epoch": 0.6332494333722533,
-      "grad_norm": 527.7891845703125,
-      "learning_rate": 1.8364462455354714e-05,
-      "loss": 65.2768,
-      "step": 156740
-    },
-    {
-      "epoch": 0.6332898346376209,
-      "grad_norm": 976.8132934570312,
-      "learning_rate": 1.8361097062829778e-05,
-      "loss": 66.5185,
-      "step": 156750
-    },
-    {
-      "epoch": 0.6333302359029885,
-      "grad_norm": 612.609130859375,
-      "learning_rate": 1.8357731799733686e-05,
-      "loss": 37.7978,
-      "step": 156760
-    },
-    {
-      "epoch": 0.6333706371683561,
-      "grad_norm": 514.4799194335938,
-      "learning_rate": 1.8354366666132035e-05,
-      "loss": 45.4972,
-      "step": 156770
-    },
-    {
-      "epoch": 0.6334110384337237,
-      "grad_norm": 1488.8907470703125,
-      "learning_rate": 1.8351001662090412e-05,
-      "loss": 64.7894,
-      "step": 156780
-    },
-    {
-      "epoch": 0.6334514396990913,
-      "grad_norm": 492.0083923339844,
-      "learning_rate": 1.8347636787674442e-05,
-      "loss": 48.8563,
-      "step": 156790
-    },
-    {
-      "epoch": 0.633491840964459,
-      "grad_norm": 897.5473022460938,
-      "learning_rate": 1.8344272042949724e-05,
-      "loss": 43.1116,
-      "step": 156800
-    },
-    {
-      "epoch": 0.6335322422298266,
-      "grad_norm": 419.75213623046875,
-      "learning_rate": 1.8340907427981843e-05,
-      "loss": 54.2634,
-      "step": 156810
-    },
-    {
-      "epoch": 0.6335726434951943,
-      "grad_norm": 604.8643188476562,
-      "learning_rate": 1.8337542942836406e-05,
-      "loss": 53.386,
-      "step": 156820
-    },
-    {
-      "epoch": 0.6336130447605619,
-      "grad_norm": 620.7161254882812,
-      "learning_rate": 1.8334178587578988e-05,
-      "loss": 79.5961,
-      "step": 156830
-    },
-    {
-      "epoch": 0.6336534460259295,
-      "grad_norm": 888.98388671875,
-      "learning_rate": 1.8330814362275198e-05,
-      "loss": 29.9113,
-      "step": 156840
-    },
-    {
-      "epoch": 0.6336938472912972,
-      "grad_norm": 833.1226196289062,
-      "learning_rate": 1.8327450266990616e-05,
-      "loss": 72.1005,
-      "step": 156850
-    },
-    {
-      "epoch": 0.6337342485566648,
-      "grad_norm": 370.32244873046875,
-      "learning_rate": 1.832408630179082e-05,
-      "loss": 36.1181,
-      "step": 156860
-    },
-    {
-      "epoch": 0.6337746498220325,
-      "grad_norm": 430.3398132324219,
-      "learning_rate": 1.8320722466741404e-05,
-      "loss": 38.9571,
-      "step": 156870
-    },
-    {
-      "epoch": 0.6338150510874001,
-      "grad_norm": 531.71435546875,
-      "learning_rate": 1.8317358761907942e-05,
-      "loss": 30.0332,
-      "step": 156880
-    },
-    {
-      "epoch": 0.6338554523527676,
-      "grad_norm": 398.49224853515625,
-      "learning_rate": 1.8313995187356004e-05,
-      "loss": 49.0851,
-      "step": 156890
-    },
-    {
-      "epoch": 0.6338958536181353,
-      "grad_norm": 516.3076782226562,
-      "learning_rate": 1.8310631743151185e-05,
-      "loss": 41.1616,
-      "step": 156900
-    },
-    {
-      "epoch": 0.6339362548835029,
-      "grad_norm": 267.231201171875,
-      "learning_rate": 1.830726842935904e-05,
-      "loss": 62.2498,
-      "step": 156910
-    },
-    {
-      "epoch": 0.6339766561488706,
-      "grad_norm": 692.8582763671875,
-      "learning_rate": 1.8303905246045138e-05,
-      "loss": 38.564,
-      "step": 156920
-    },
-    {
-      "epoch": 0.6340170574142382,
-      "grad_norm": 1075.482666015625,
-      "learning_rate": 1.8300542193275057e-05,
-      "loss": 41.1674,
-      "step": 156930
-    },
-    {
-      "epoch": 0.6340574586796058,
-      "grad_norm": 692.3712768554688,
-      "learning_rate": 1.8297179271114346e-05,
-      "loss": 73.7916,
-      "step": 156940
-    },
-    {
-      "epoch": 0.6340978599449735,
-      "grad_norm": 613.9649047851562,
-      "learning_rate": 1.8293816479628583e-05,
-      "loss": 47.4327,
-      "step": 156950
-    },
-    {
-      "epoch": 0.6341382612103411,
-      "grad_norm": 472.95611572265625,
-      "learning_rate": 1.829045381888332e-05,
-      "loss": 51.9652,
-      "step": 156960
-    },
-    {
-      "epoch": 0.6341786624757088,
-      "grad_norm": 743.0994262695312,
-      "learning_rate": 1.828709128894411e-05,
-      "loss": 49.6867,
-      "step": 156970
-    },
-    {
-      "epoch": 0.6342190637410764,
-      "grad_norm": 777.2012939453125,
-      "learning_rate": 1.8283728889876513e-05,
-      "loss": 43.0576,
-      "step": 156980
-    },
-    {
-      "epoch": 0.634259465006444,
-      "grad_norm": 518.1323852539062,
-      "learning_rate": 1.828036662174608e-05,
-      "loss": 38.773,
-      "step": 156990
-    },
-    {
-      "epoch": 0.6342998662718117,
-      "grad_norm": 864.1792602539062,
-      "learning_rate": 1.827700448461836e-05,
-      "loss": 58.8282,
-      "step": 157000
-    },
-    {
-      "epoch": 0.6343402675371793,
-      "grad_norm": 938.5631713867188,
-      "learning_rate": 1.8273642478558903e-05,
-      "loss": 84.6849,
-      "step": 157010
-    },
-    {
-      "epoch": 0.6343806688025468,
-      "grad_norm": 0.0,
-      "learning_rate": 1.8270280603633234e-05,
-      "loss": 38.4407,
-      "step": 157020
-    },
-    {
-      "epoch": 0.6344210700679145,
-      "grad_norm": 757.0884399414062,
-      "learning_rate": 1.8266918859906922e-05,
-      "loss": 48.6319,
-      "step": 157030
-    },
-    {
-      "epoch": 0.6344614713332821,
-      "grad_norm": 1165.9722900390625,
-      "learning_rate": 1.8263557247445495e-05,
-      "loss": 76.0864,
-      "step": 157040
-    },
-    {
-      "epoch": 0.6345018725986498,
-      "grad_norm": 454.1170959472656,
-      "learning_rate": 1.8260195766314476e-05,
-      "loss": 44.9781,
-      "step": 157050
-    },
-    {
-      "epoch": 0.6345422738640174,
-      "grad_norm": 1561.450927734375,
-      "learning_rate": 1.825683441657942e-05,
-      "loss": 50.8106,
-      "step": 157060
-    },
-    {
-      "epoch": 0.634582675129385,
-      "grad_norm": 898.5762939453125,
-      "learning_rate": 1.8253473198305846e-05,
-      "loss": 27.4869,
-      "step": 157070
-    },
-    {
-      "epoch": 0.6346230763947527,
-      "grad_norm": 562.4152221679688,
-      "learning_rate": 1.8250112111559287e-05,
-      "loss": 62.3948,
-      "step": 157080
-    },
-    {
-      "epoch": 0.6346634776601203,
-      "grad_norm": 672.2083740234375,
-      "learning_rate": 1.824675115640527e-05,
-      "loss": 61.7726,
-      "step": 157090
-    },
-    {
-      "epoch": 0.634703878925488,
-      "grad_norm": 926.6039428710938,
-      "learning_rate": 1.8243390332909303e-05,
-      "loss": 79.0976,
-      "step": 157100
-    },
-    {
-      "epoch": 0.6347442801908556,
-      "grad_norm": 400.0118713378906,
-      "learning_rate": 1.8240029641136935e-05,
-      "loss": 34.7595,
-      "step": 157110
-    },
-    {
-      "epoch": 0.6347846814562232,
-      "grad_norm": 566.873291015625,
-      "learning_rate": 1.8236669081153657e-05,
-      "loss": 49.1967,
-      "step": 157120
-    },
-    {
-      "epoch": 0.6348250827215909,
-      "grad_norm": 396.2868347167969,
-      "learning_rate": 1.8233308653025006e-05,
-      "loss": 68.7327,
-      "step": 157130
-    },
-    {
-      "epoch": 0.6348654839869585,
-      "grad_norm": 463.4564514160156,
-      "learning_rate": 1.8229948356816486e-05,
-      "loss": 48.3621,
-      "step": 157140
-    },
-    {
-      "epoch": 0.634905885252326,
-      "grad_norm": 510.2298278808594,
-      "learning_rate": 1.8226588192593605e-05,
-      "loss": 41.7354,
-      "step": 157150
-    },
-    {
-      "epoch": 0.6349462865176937,
-      "grad_norm": 630.56689453125,
-      "learning_rate": 1.822322816042188e-05,
-      "loss": 51.7539,
-      "step": 157160
-    },
-    {
-      "epoch": 0.6349866877830613,
-      "grad_norm": 533.6231689453125,
-      "learning_rate": 1.8219868260366802e-05,
-      "loss": 35.9532,
-      "step": 157170
-    },
-    {
-      "epoch": 0.635027089048429,
-      "grad_norm": 819.6351318359375,
-      "learning_rate": 1.8216508492493886e-05,
-      "loss": 70.248,
-      "step": 157180
-    },
-    {
-      "epoch": 0.6350674903137966,
-      "grad_norm": 964.2184448242188,
-      "learning_rate": 1.8213148856868634e-05,
-      "loss": 55.642,
-      "step": 157190
-    },
-    {
-      "epoch": 0.6351078915791643,
-      "grad_norm": 669.3191528320312,
-      "learning_rate": 1.820978935355653e-05,
-      "loss": 46.5878,
-      "step": 157200
-    },
-    {
-      "epoch": 0.6351482928445319,
-      "grad_norm": 1667.268310546875,
-      "learning_rate": 1.8206429982623086e-05,
-      "loss": 59.5129,
-      "step": 157210
-    },
-    {
-      "epoch": 0.6351886941098995,
-      "grad_norm": 640.9869995117188,
-      "learning_rate": 1.820307074413379e-05,
-      "loss": 47.4462,
-      "step": 157220
-    },
-    {
-      "epoch": 0.6352290953752672,
-      "grad_norm": 817.10986328125,
-      "learning_rate": 1.8199711638154116e-05,
-      "loss": 74.9785,
-      "step": 157230
-    },
-    {
-      "epoch": 0.6352694966406348,
-      "grad_norm": 666.6509399414062,
-      "learning_rate": 1.8196352664749576e-05,
-      "loss": 54.3086,
-      "step": 157240
-    },
-    {
-      "epoch": 0.6353098979060025,
-      "grad_norm": 1946.0594482421875,
-      "learning_rate": 1.8192993823985643e-05,
-      "loss": 49.1748,
-      "step": 157250
-    },
-    {
-      "epoch": 0.6353502991713701,
-      "grad_norm": 852.0538330078125,
-      "learning_rate": 1.81896351159278e-05,
-      "loss": 61.9614,
-      "step": 157260
-    },
-    {
-      "epoch": 0.6353907004367377,
-      "grad_norm": 352.1046142578125,
-      "learning_rate": 1.8186276540641527e-05,
-      "loss": 31.2032,
-      "step": 157270
-    },
-    {
-      "epoch": 0.6354311017021053,
-      "grad_norm": 2632.97216796875,
-      "learning_rate": 1.818291809819229e-05,
-      "loss": 62.7898,
-      "step": 157280
-    },
-    {
-      "epoch": 0.6354715029674729,
-      "grad_norm": 795.597412109375,
-      "learning_rate": 1.8179559788645583e-05,
-      "loss": 55.3189,
-      "step": 157290
-    },
-    {
-      "epoch": 0.6355119042328405,
-      "grad_norm": 661.72314453125,
-      "learning_rate": 1.817620161206687e-05,
-      "loss": 72.9509,
-      "step": 157300
-    },
-    {
-      "epoch": 0.6355523054982082,
-      "grad_norm": 1691.134521484375,
-      "learning_rate": 1.8172843568521613e-05,
-      "loss": 67.871,
-      "step": 157310
-    },
-    {
-      "epoch": 0.6355927067635758,
-      "grad_norm": 601.0391235351562,
-      "learning_rate": 1.8169485658075298e-05,
-      "loss": 58.32,
-      "step": 157320
-    },
-    {
-      "epoch": 0.6356331080289435,
-      "grad_norm": 564.531005859375,
-      "learning_rate": 1.8166127880793372e-05,
-      "loss": 69.3806,
-      "step": 157330
-    },
-    {
-      "epoch": 0.6356735092943111,
-      "grad_norm": 731.717529296875,
-      "learning_rate": 1.8162770236741294e-05,
-      "loss": 51.7775,
-      "step": 157340
-    },
-    {
-      "epoch": 0.6357139105596787,
-      "grad_norm": 379.14654541015625,
-      "learning_rate": 1.8159412725984543e-05,
-      "loss": 36.2211,
-      "step": 157350
-    },
-    {
-      "epoch": 0.6357543118250464,
-      "grad_norm": 1579.4599609375,
-      "learning_rate": 1.8156055348588546e-05,
-      "loss": 76.9737,
-      "step": 157360
-    },
-    {
-      "epoch": 0.635794713090414,
-      "grad_norm": 266.6926574707031,
-      "learning_rate": 1.8152698104618785e-05,
-      "loss": 53.4006,
-      "step": 157370
-    },
-    {
-      "epoch": 0.6358351143557817,
-      "grad_norm": 561.2648315429688,
-      "learning_rate": 1.8149340994140702e-05,
-      "loss": 62.8926,
-      "step": 157380
-    },
-    {
-      "epoch": 0.6358755156211493,
-      "grad_norm": 284.5917053222656,
-      "learning_rate": 1.814598401721973e-05,
-      "loss": 51.397,
-      "step": 157390
-    },
-    {
-      "epoch": 0.6359159168865169,
-      "grad_norm": 1907.217529296875,
-      "learning_rate": 1.8142627173921338e-05,
-      "loss": 55.0741,
-      "step": 157400
-    },
-    {
-      "epoch": 0.6359563181518845,
-      "grad_norm": 357.5486145019531,
-      "learning_rate": 1.813927046431096e-05,
-      "loss": 71.2046,
-      "step": 157410
-    },
-    {
-      "epoch": 0.6359967194172521,
-      "grad_norm": 526.5702514648438,
-      "learning_rate": 1.8135913888454033e-05,
-      "loss": 24.1797,
-      "step": 157420
-    },
-    {
-      "epoch": 0.6360371206826197,
-      "grad_norm": 672.9292602539062,
-      "learning_rate": 1.8132557446416005e-05,
-      "loss": 44.7071,
-      "step": 157430
-    },
-    {
-      "epoch": 0.6360775219479874,
-      "grad_norm": 344.2654113769531,
-      "learning_rate": 1.812920113826229e-05,
-      "loss": 56.6532,
-      "step": 157440
-    },
-    {
-      "epoch": 0.636117923213355,
-      "grad_norm": 781.5308227539062,
-      "learning_rate": 1.8125844964058354e-05,
-      "loss": 50.7511,
-      "step": 157450
-    },
-    {
-      "epoch": 0.6361583244787227,
-      "grad_norm": 2054.18408203125,
-      "learning_rate": 1.8122488923869605e-05,
-      "loss": 55.7488,
-      "step": 157460
-    },
-    {
-      "epoch": 0.6361987257440903,
-      "grad_norm": 585.6814575195312,
-      "learning_rate": 1.8119133017761465e-05,
-      "loss": 48.2461,
-      "step": 157470
-    },
-    {
-      "epoch": 0.636239127009458,
-      "grad_norm": 720.3282470703125,
-      "learning_rate": 1.811577724579938e-05,
-      "loss": 42.073,
-      "step": 157480
-    },
-    {
-      "epoch": 0.6362795282748256,
-      "grad_norm": 479.0511169433594,
-      "learning_rate": 1.8112421608048765e-05,
-      "loss": 53.2484,
-      "step": 157490
-    },
-    {
-      "epoch": 0.6363199295401932,
-      "grad_norm": 179.38345336914062,
-      "learning_rate": 1.8109066104575023e-05,
-      "loss": 70.5386,
-      "step": 157500
-    },
-    {
-      "epoch": 0.6363603308055609,
-      "grad_norm": 1411.388916015625,
-      "learning_rate": 1.8105710735443593e-05,
-      "loss": 67.7427,
-      "step": 157510
-    },
-    {
-      "epoch": 0.6364007320709285,
-      "grad_norm": 452.4316101074219,
-      "learning_rate": 1.8102355500719876e-05,
-      "loss": 43.8855,
-      "step": 157520
-    },
-    {
-      "epoch": 0.636441133336296,
-      "grad_norm": 1131.415283203125,
-      "learning_rate": 1.8099000400469297e-05,
-      "loss": 49.2478,
-      "step": 157530
-    },
-    {
-      "epoch": 0.6364815346016637,
-      "grad_norm": 1270.067138671875,
-      "learning_rate": 1.809564543475726e-05,
-      "loss": 48.759,
-      "step": 157540
-    },
-    {
-      "epoch": 0.6365219358670313,
-      "grad_norm": 512.9644775390625,
-      "learning_rate": 1.809229060364916e-05,
-      "loss": 33.8024,
-      "step": 157550
-    },
-    {
-      "epoch": 0.636562337132399,
-      "grad_norm": 1274.902587890625,
-      "learning_rate": 1.8088935907210418e-05,
-      "loss": 51.3391,
-      "step": 157560
-    },
-    {
-      "epoch": 0.6366027383977666,
-      "grad_norm": 881.6553955078125,
-      "learning_rate": 1.808558134550643e-05,
-      "loss": 64.7102,
-      "step": 157570
-    },
-    {
-      "epoch": 0.6366431396631342,
-      "grad_norm": 1123.6348876953125,
-      "learning_rate": 1.8082226918602585e-05,
-      "loss": 40.6092,
-      "step": 157580
-    },
-    {
-      "epoch": 0.6366835409285019,
-      "grad_norm": 441.2737731933594,
-      "learning_rate": 1.8078872626564295e-05,
-      "loss": 46.9466,
-      "step": 157590
-    },
-    {
-      "epoch": 0.6367239421938695,
-      "grad_norm": 999.4472045898438,
-      "learning_rate": 1.807551846945694e-05,
-      "loss": 77.4907,
-      "step": 157600
-    },
-    {
-      "epoch": 0.6367643434592372,
-      "grad_norm": 464.4021301269531,
-      "learning_rate": 1.8072164447345927e-05,
-      "loss": 60.3902,
-      "step": 157610
-    },
-    {
-      "epoch": 0.6368047447246048,
-      "grad_norm": 580.01953125,
-      "learning_rate": 1.8068810560296634e-05,
-      "loss": 66.6603,
-      "step": 157620
-    },
-    {
-      "epoch": 0.6368451459899724,
-      "grad_norm": 305.635498046875,
-      "learning_rate": 1.8065456808374433e-05,
-      "loss": 48.442,
-      "step": 157630
-    },
-    {
-      "epoch": 0.6368855472553401,
-      "grad_norm": 829.7825927734375,
-      "learning_rate": 1.8062103191644737e-05,
-      "loss": 54.7456,
-      "step": 157640
-    },
-    {
-      "epoch": 0.6369259485207077,
-      "grad_norm": 1587.976806640625,
-      "learning_rate": 1.80587497101729e-05,
-      "loss": 59.605,
-      "step": 157650
-    },
-    {
-      "epoch": 0.6369663497860752,
-      "grad_norm": 162.47715759277344,
-      "learning_rate": 1.8055396364024317e-05,
-      "loss": 39.8688,
-      "step": 157660
-    },
-    {
-      "epoch": 0.6370067510514429,
-      "grad_norm": 501.1140441894531,
-      "learning_rate": 1.8052043153264364e-05,
-      "loss": 39.7927,
-      "step": 157670
-    },
-    {
-      "epoch": 0.6370471523168105,
-      "grad_norm": 505.4931945800781,
-      "learning_rate": 1.8048690077958397e-05,
-      "loss": 42.7583,
-      "step": 157680
-    },
-    {
-      "epoch": 0.6370875535821782,
-      "grad_norm": 776.6007690429688,
-      "learning_rate": 1.80453371381718e-05,
-      "loss": 45.7655,
-      "step": 157690
-    },
-    {
-      "epoch": 0.6371279548475458,
-      "grad_norm": 1815.0123291015625,
-      "learning_rate": 1.804198433396994e-05,
-      "loss": 39.881,
-      "step": 157700
-    },
-    {
-      "epoch": 0.6371683561129134,
-      "grad_norm": 483.6055908203125,
-      "learning_rate": 1.8038631665418165e-05,
-      "loss": 58.2693,
-      "step": 157710
-    },
-    {
-      "epoch": 0.6372087573782811,
-      "grad_norm": 923.490234375,
-      "learning_rate": 1.803527913258186e-05,
-      "loss": 72.5108,
-      "step": 157720
-    },
-    {
-      "epoch": 0.6372491586436487,
-      "grad_norm": 487.7062072753906,
-      "learning_rate": 1.8031926735526363e-05,
-      "loss": 81.6203,
-      "step": 157730
-    },
-    {
-      "epoch": 0.6372895599090164,
-      "grad_norm": 775.66845703125,
-      "learning_rate": 1.8028574474317052e-05,
-      "loss": 41.2521,
-      "step": 157740
-    },
-    {
-      "epoch": 0.637329961174384,
-      "grad_norm": 435.292236328125,
-      "learning_rate": 1.802522234901927e-05,
-      "loss": 46.5016,
-      "step": 157750
-    },
-    {
-      "epoch": 0.6373703624397516,
-      "grad_norm": 422.0650634765625,
-      "learning_rate": 1.8021870359698368e-05,
-      "loss": 60.5132,
-      "step": 157760
-    },
-    {
-      "epoch": 0.6374107637051193,
-      "grad_norm": 808.7423095703125,
-      "learning_rate": 1.8018518506419698e-05,
-      "loss": 55.5142,
-      "step": 157770
-    },
-    {
-      "epoch": 0.6374511649704869,
-      "grad_norm": 885.0738525390625,
-      "learning_rate": 1.8015166789248604e-05,
-      "loss": 50.6987,
-      "step": 157780
-    },
-    {
-      "epoch": 0.6374915662358545,
-      "grad_norm": 515.9467163085938,
-      "learning_rate": 1.801181520825042e-05,
-      "loss": 43.7261,
-      "step": 157790
-    },
-    {
-      "epoch": 0.6375319675012221,
-      "grad_norm": 521.8585815429688,
-      "learning_rate": 1.800846376349051e-05,
-      "loss": 49.786,
-      "step": 157800
-    },
-    {
-      "epoch": 0.6375723687665897,
-      "grad_norm": 511.0011291503906,
-      "learning_rate": 1.800511245503418e-05,
-      "loss": 40.6436,
-      "step": 157810
-    },
-    {
-      "epoch": 0.6376127700319574,
-      "grad_norm": 320.0465087890625,
-      "learning_rate": 1.80017612829468e-05,
-      "loss": 50.3526,
-      "step": 157820
-    },
-    {
-      "epoch": 0.637653171297325,
-      "grad_norm": 638.8887939453125,
-      "learning_rate": 1.7998410247293686e-05,
-      "loss": 40.4759,
-      "step": 157830
-    },
-    {
-      "epoch": 0.6376935725626927,
-      "grad_norm": 1001.0759887695312,
-      "learning_rate": 1.7995059348140165e-05,
-      "loss": 52.3484,
-      "step": 157840
-    },
-    {
-      "epoch": 0.6377339738280603,
-      "grad_norm": 513.6463623046875,
-      "learning_rate": 1.7991708585551563e-05,
-      "loss": 42.4503,
-      "step": 157850
-    },
-    {
-      "epoch": 0.6377743750934279,
-      "grad_norm": 489.6709899902344,
-      "learning_rate": 1.7988357959593222e-05,
-      "loss": 34.6381,
-      "step": 157860
-    },
-    {
-      "epoch": 0.6378147763587956,
-      "grad_norm": 661.804443359375,
-      "learning_rate": 1.7985007470330444e-05,
-      "loss": 79.6513,
-      "step": 157870
-    },
-    {
-      "epoch": 0.6378551776241632,
-      "grad_norm": 427.7678527832031,
-      "learning_rate": 1.798165711782856e-05,
-      "loss": 56.3009,
-      "step": 157880
-    },
-    {
-      "epoch": 0.6378955788895309,
-      "grad_norm": 858.1063232421875,
-      "learning_rate": 1.7978306902152878e-05,
-      "loss": 39.8254,
-      "step": 157890
-    },
-    {
-      "epoch": 0.6379359801548985,
-      "grad_norm": 666.5377197265625,
-      "learning_rate": 1.7974956823368727e-05,
-      "loss": 49.063,
-      "step": 157900
-    },
-    {
-      "epoch": 0.6379763814202661,
-      "grad_norm": 1005.5430297851562,
-      "learning_rate": 1.7971606881541407e-05,
-      "loss": 61.1597,
-      "step": 157910
-    },
-    {
-      "epoch": 0.6380167826856337,
-      "grad_norm": 711.9946899414062,
-      "learning_rate": 1.796825707673622e-05,
-      "loss": 66.6129,
-      "step": 157920
-    },
-    {
-      "epoch": 0.6380571839510013,
-      "grad_norm": 674.28759765625,
-      "learning_rate": 1.7964907409018496e-05,
-      "loss": 48.3361,
-      "step": 157930
-    },
-    {
-      "epoch": 0.638097585216369,
-      "grad_norm": 895.3465576171875,
-      "learning_rate": 1.7961557878453516e-05,
-      "loss": 53.9553,
-      "step": 157940
-    },
-    {
-      "epoch": 0.6381379864817366,
-      "grad_norm": 206.62802124023438,
-      "learning_rate": 1.7958208485106586e-05,
-      "loss": 66.486,
-      "step": 157950
-    },
-    {
-      "epoch": 0.6381783877471042,
-      "grad_norm": 1281.931640625,
-      "learning_rate": 1.7954859229043016e-05,
-      "loss": 51.8759,
-      "step": 157960
-    },
-    {
-      "epoch": 0.6382187890124719,
-      "grad_norm": 420.1695861816406,
-      "learning_rate": 1.7951510110328077e-05,
-      "loss": 31.3493,
-      "step": 157970
-    },
-    {
-      "epoch": 0.6382591902778395,
-      "grad_norm": 1604.4742431640625,
-      "learning_rate": 1.7948161129027094e-05,
-      "loss": 79.5504,
-      "step": 157980
-    },
-    {
-      "epoch": 0.6382995915432071,
-      "grad_norm": 798.0696411132812,
-      "learning_rate": 1.7944812285205335e-05,
-      "loss": 45.0031,
-      "step": 157990
-    },
-    {
-      "epoch": 0.6383399928085748,
-      "grad_norm": 812.3040771484375,
-      "learning_rate": 1.7941463578928086e-05,
-      "loss": 47.2854,
-      "step": 158000
-    },
-    {
-      "epoch": 0.6383803940739424,
-      "grad_norm": 858.7913208007812,
-      "learning_rate": 1.7938115010260647e-05,
-      "loss": 60.8041,
-      "step": 158010
-    },
-    {
-      "epoch": 0.6384207953393101,
-      "grad_norm": 363.1507263183594,
-      "learning_rate": 1.7934766579268292e-05,
-      "loss": 67.6859,
-      "step": 158020
-    },
-    {
-      "epoch": 0.6384611966046777,
-      "grad_norm": 1177.5140380859375,
-      "learning_rate": 1.7931418286016292e-05,
-      "loss": 71.1444,
-      "step": 158030
-    },
-    {
-      "epoch": 0.6385015978700452,
-      "grad_norm": 682.765869140625,
-      "learning_rate": 1.7928070130569942e-05,
-      "loss": 41.8368,
-      "step": 158040
-    },
-    {
-      "epoch": 0.6385419991354129,
-      "grad_norm": 1204.63427734375,
-      "learning_rate": 1.7924722112994495e-05,
-      "loss": 44.7631,
-      "step": 158050
-    },
-    {
-      "epoch": 0.6385824004007805,
-      "grad_norm": 0.0,
-      "learning_rate": 1.792137423335524e-05,
-      "loss": 50.4615,
-      "step": 158060
-    },
-    {
-      "epoch": 0.6386228016661482,
-      "grad_norm": 1152.09521484375,
-      "learning_rate": 1.7918026491717437e-05,
-      "loss": 72.6507,
-      "step": 158070
-    },
-    {
-      "epoch": 0.6386632029315158,
-      "grad_norm": 551.220458984375,
-      "learning_rate": 1.7914678888146347e-05,
-      "loss": 49.2288,
-      "step": 158080
-    },
-    {
-      "epoch": 0.6387036041968834,
-      "grad_norm": 1303.2149658203125,
-      "learning_rate": 1.791133142270725e-05,
-      "loss": 45.762,
-      "step": 158090
-    },
-    {
-      "epoch": 0.6387440054622511,
-      "grad_norm": 623.5195922851562,
-      "learning_rate": 1.7907984095465395e-05,
-      "loss": 53.7673,
-      "step": 158100
-    },
-    {
-      "epoch": 0.6387844067276187,
-      "grad_norm": 799.129150390625,
-      "learning_rate": 1.7904636906486037e-05,
-      "loss": 38.7584,
-      "step": 158110
-    },
-    {
-      "epoch": 0.6388248079929864,
-      "grad_norm": 536.8320922851562,
-      "learning_rate": 1.790128985583444e-05,
-      "loss": 63.6297,
-      "step": 158120
-    },
-    {
-      "epoch": 0.638865209258354,
-      "grad_norm": 885.3190307617188,
-      "learning_rate": 1.7897942943575843e-05,
-      "loss": 65.513,
-      "step": 158130
-    },
-    {
-      "epoch": 0.6389056105237216,
-      "grad_norm": 729.3671875,
-      "learning_rate": 1.7894596169775512e-05,
-      "loss": 41.7076,
-      "step": 158140
-    },
-    {
-      "epoch": 0.6389460117890893,
-      "grad_norm": 540.5570678710938,
-      "learning_rate": 1.789124953449869e-05,
-      "loss": 50.7414,
-      "step": 158150
-    },
-    {
-      "epoch": 0.6389864130544569,
-      "grad_norm": 569.721435546875,
-      "learning_rate": 1.788790303781061e-05,
-      "loss": 78.4995,
-      "step": 158160
-    },
-    {
-      "epoch": 0.6390268143198244,
-      "grad_norm": 1166.6480712890625,
-      "learning_rate": 1.7884556679776525e-05,
-      "loss": 59.4203,
-      "step": 158170
-    },
-    {
-      "epoch": 0.6390672155851921,
-      "grad_norm": 601.839599609375,
-      "learning_rate": 1.788121046046167e-05,
-      "loss": 48.1814,
-      "step": 158180
-    },
-    {
-      "epoch": 0.6391076168505597,
-      "grad_norm": 716.4414672851562,
-      "learning_rate": 1.7877864379931287e-05,
-      "loss": 67.2857,
-      "step": 158190
-    },
-    {
-      "epoch": 0.6391480181159274,
-      "grad_norm": 395.68963623046875,
-      "learning_rate": 1.7874518438250597e-05,
-      "loss": 46.7157,
-      "step": 158200
-    },
-    {
-      "epoch": 0.639188419381295,
-      "grad_norm": 1754.9854736328125,
-      "learning_rate": 1.7871172635484842e-05,
-      "loss": 84.133,
-      "step": 158210
-    },
-    {
-      "epoch": 0.6392288206466626,
-      "grad_norm": 439.41949462890625,
-      "learning_rate": 1.7867826971699253e-05,
-      "loss": 43.6031,
-      "step": 158220
-    },
-    {
-      "epoch": 0.6392692219120303,
-      "grad_norm": 280.5320739746094,
-      "learning_rate": 1.7864481446959045e-05,
-      "loss": 60.2605,
-      "step": 158230
-    },
-    {
-      "epoch": 0.6393096231773979,
-      "grad_norm": 386.1781005859375,
-      "learning_rate": 1.7861136061329437e-05,
-      "loss": 32.9864,
-      "step": 158240
-    },
-    {
-      "epoch": 0.6393500244427656,
-      "grad_norm": 915.8598022460938,
-      "learning_rate": 1.7857790814875663e-05,
-      "loss": 38.008,
-      "step": 158250
-    },
-    {
-      "epoch": 0.6393904257081332,
-      "grad_norm": 678.6073608398438,
-      "learning_rate": 1.7854445707662928e-05,
-      "loss": 43.3288,
-      "step": 158260
-    },
-    {
-      "epoch": 0.6394308269735008,
-      "grad_norm": 853.765380859375,
-      "learning_rate": 1.785110073975646e-05,
-      "loss": 41.3135,
-      "step": 158270
-    },
-    {
-      "epoch": 0.6394712282388685,
-      "grad_norm": 773.3751220703125,
-      "learning_rate": 1.7847755911221466e-05,
-      "loss": 41.4973,
-      "step": 158280
-    },
-    {
-      "epoch": 0.6395116295042361,
-      "grad_norm": 665.0879516601562,
-      "learning_rate": 1.7844411222123147e-05,
-      "loss": 45.4086,
-      "step": 158290
-    },
-    {
-      "epoch": 0.6395520307696037,
-      "grad_norm": 616.668701171875,
-      "learning_rate": 1.7841066672526723e-05,
-      "loss": 29.242,
-      "step": 158300
-    },
-    {
-      "epoch": 0.6395924320349713,
-      "grad_norm": 2878.92724609375,
-      "learning_rate": 1.7837722262497385e-05,
-      "loss": 74.1233,
-      "step": 158310
-    },
-    {
-      "epoch": 0.6396328333003389,
-      "grad_norm": 495.0466613769531,
-      "learning_rate": 1.7834377992100333e-05,
-      "loss": 61.9444,
-      "step": 158320
-    },
-    {
-      "epoch": 0.6396732345657066,
-      "grad_norm": 598.6763305664062,
-      "learning_rate": 1.7831033861400777e-05,
-      "loss": 62.46,
-      "step": 158330
-    },
-    {
-      "epoch": 0.6397136358310742,
-      "grad_norm": 579.919921875,
-      "learning_rate": 1.78276898704639e-05,
-      "loss": 54.3735,
-      "step": 158340
-    },
-    {
-      "epoch": 0.6397540370964419,
-      "grad_norm": 641.3147583007812,
-      "learning_rate": 1.782434601935491e-05,
-      "loss": 62.4215,
-      "step": 158350
-    },
-    {
-      "epoch": 0.6397944383618095,
-      "grad_norm": 511.64569091796875,
-      "learning_rate": 1.782100230813899e-05,
-      "loss": 36.6608,
-      "step": 158360
-    },
-    {
-      "epoch": 0.6398348396271771,
-      "grad_norm": 1257.087158203125,
-      "learning_rate": 1.7817658736881322e-05,
-      "loss": 54.9247,
-      "step": 158370
-    },
-    {
-      "epoch": 0.6398752408925448,
-      "grad_norm": 735.0720825195312,
-      "learning_rate": 1.7814315305647093e-05,
-      "loss": 38.0215,
-      "step": 158380
-    },
-    {
-      "epoch": 0.6399156421579124,
-      "grad_norm": 1210.0875244140625,
-      "learning_rate": 1.781097201450149e-05,
-      "loss": 50.9138,
-      "step": 158390
-    },
-    {
-      "epoch": 0.63995604342328,
-      "grad_norm": 765.9678955078125,
-      "learning_rate": 1.7807628863509685e-05,
-      "loss": 38.4157,
-      "step": 158400
-    },
-    {
-      "epoch": 0.6399964446886477,
-      "grad_norm": 719.0951538085938,
-      "learning_rate": 1.7804285852736864e-05,
-      "loss": 73.3833,
-      "step": 158410
-    },
-    {
-      "epoch": 0.6400368459540153,
-      "grad_norm": 925.130615234375,
-      "learning_rate": 1.780094298224818e-05,
-      "loss": 60.2505,
-      "step": 158420
-    },
-    {
-      "epoch": 0.6400772472193829,
-      "grad_norm": 665.5335693359375,
-      "learning_rate": 1.7797600252108834e-05,
-      "loss": 45.8719,
-      "step": 158430
-    },
-    {
-      "epoch": 0.6401176484847505,
-      "grad_norm": 772.93603515625,
-      "learning_rate": 1.779425766238398e-05,
-      "loss": 70.8536,
-      "step": 158440
-    },
-    {
-      "epoch": 0.6401580497501181,
-      "grad_norm": 519.6685180664062,
-      "learning_rate": 1.7790915213138776e-05,
-      "loss": 72.4601,
-      "step": 158450
-    },
-    {
-      "epoch": 0.6401984510154858,
-      "grad_norm": 516.3569946289062,
-      "learning_rate": 1.7787572904438392e-05,
-      "loss": 38.114,
-      "step": 158460
-    },
-    {
-      "epoch": 0.6402388522808534,
-      "grad_norm": 633.9620971679688,
-      "learning_rate": 1.778423073634799e-05,
-      "loss": 72.208,
-      "step": 158470
-    },
-    {
-      "epoch": 0.6402792535462211,
-      "grad_norm": 568.24609375,
-      "learning_rate": 1.7780888708932718e-05,
-      "loss": 53.3362,
-      "step": 158480
-    },
-    {
-      "epoch": 0.6403196548115887,
-      "grad_norm": 2042.0194091796875,
-      "learning_rate": 1.7777546822257748e-05,
-      "loss": 59.6458,
-      "step": 158490
-    },
-    {
-      "epoch": 0.6403600560769563,
-      "grad_norm": 682.5473022460938,
-      "learning_rate": 1.7774205076388206e-05,
-      "loss": 31.9435,
-      "step": 158500
-    },
-    {
-      "epoch": 0.640400457342324,
-      "grad_norm": 708.43310546875,
-      "learning_rate": 1.777086347138927e-05,
-      "loss": 42.089,
-      "step": 158510
-    },
-    {
-      "epoch": 0.6404408586076916,
-      "grad_norm": 1302.8370361328125,
-      "learning_rate": 1.7767522007326072e-05,
-      "loss": 71.7163,
-      "step": 158520
-    },
-    {
-      "epoch": 0.6404812598730593,
-      "grad_norm": 904.3488159179688,
-      "learning_rate": 1.7764180684263752e-05,
-      "loss": 72.1357,
-      "step": 158530
-    },
-    {
-      "epoch": 0.6405216611384269,
-      "grad_norm": 854.4265747070312,
-      "learning_rate": 1.7760839502267452e-05,
-      "loss": 62.3567,
-      "step": 158540
-    },
-    {
-      "epoch": 0.6405620624037945,
-      "grad_norm": 639.02490234375,
-      "learning_rate": 1.7757498461402318e-05,
-      "loss": 49.8052,
-      "step": 158550
-    },
-    {
-      "epoch": 0.6406024636691621,
-      "grad_norm": 585.0482177734375,
-      "learning_rate": 1.7754157561733476e-05,
-      "loss": 42.8153,
-      "step": 158560
-    },
-    {
-      "epoch": 0.6406428649345297,
-      "grad_norm": 810.7124633789062,
-      "learning_rate": 1.775081680332607e-05,
-      "loss": 28.1556,
-      "step": 158570
-    },
-    {
-      "epoch": 0.6406832661998973,
-      "grad_norm": 1076.5986328125,
-      "learning_rate": 1.774747618624521e-05,
-      "loss": 77.0871,
-      "step": 158580
-    },
-    {
-      "epoch": 0.640723667465265,
-      "grad_norm": 524.3070068359375,
-      "learning_rate": 1.7744135710556045e-05,
-      "loss": 39.2759,
-      "step": 158590
-    },
-    {
-      "epoch": 0.6407640687306326,
-      "grad_norm": 260.4174499511719,
-      "learning_rate": 1.7740795376323692e-05,
-      "loss": 43.6856,
-      "step": 158600
-    },
-    {
-      "epoch": 0.6408044699960003,
-      "grad_norm": 441.21075439453125,
-      "learning_rate": 1.7737455183613258e-05,
-      "loss": 40.3064,
-      "step": 158610
-    },
-    {
-      "epoch": 0.6408448712613679,
-      "grad_norm": 1924.186767578125,
-      "learning_rate": 1.7734115132489886e-05,
-      "loss": 51.7431,
-      "step": 158620
-    },
-    {
-      "epoch": 0.6408852725267355,
-      "grad_norm": 352.7893371582031,
-      "learning_rate": 1.773077522301868e-05,
-      "loss": 55.666,
-      "step": 158630
-    },
-    {
-      "epoch": 0.6409256737921032,
-      "grad_norm": 381.9687805175781,
-      "learning_rate": 1.7727435455264747e-05,
-      "loss": 53.1297,
-      "step": 158640
-    },
-    {
-      "epoch": 0.6409660750574708,
-      "grad_norm": 1260.5701904296875,
-      "learning_rate": 1.772409582929321e-05,
-      "loss": 59.0081,
-      "step": 158650
-    },
-    {
-      "epoch": 0.6410064763228385,
-      "grad_norm": 442.5443420410156,
-      "learning_rate": 1.772075634516916e-05,
-      "loss": 28.4182,
-      "step": 158660
-    },
-    {
-      "epoch": 0.6410468775882061,
-      "grad_norm": 776.69873046875,
-      "learning_rate": 1.7717417002957725e-05,
-      "loss": 93.5575,
-      "step": 158670
-    },
-    {
-      "epoch": 0.6410872788535736,
-      "grad_norm": 983.2738037109375,
-      "learning_rate": 1.7714077802723994e-05,
-      "loss": 59.6079,
-      "step": 158680
-    },
-    {
-      "epoch": 0.6411276801189413,
-      "grad_norm": 338.20379638671875,
-      "learning_rate": 1.7710738744533057e-05,
-      "loss": 55.5007,
-      "step": 158690
-    },
-    {
-      "epoch": 0.6411680813843089,
-      "grad_norm": 872.261962890625,
-      "learning_rate": 1.7707399828450027e-05,
-      "loss": 57.6118,
-      "step": 158700
-    },
-    {
-      "epoch": 0.6412084826496766,
-      "grad_norm": 441.12628173828125,
-      "learning_rate": 1.7704061054539993e-05,
-      "loss": 40.5436,
-      "step": 158710
-    },
-    {
-      "epoch": 0.6412488839150442,
-      "grad_norm": 655.5393676757812,
-      "learning_rate": 1.7700722422868048e-05,
-      "loss": 57.556,
-      "step": 158720
-    },
-    {
-      "epoch": 0.6412892851804118,
-      "grad_norm": 864.171875,
-      "learning_rate": 1.7697383933499274e-05,
-      "loss": 35.3705,
-      "step": 158730
-    },
-    {
-      "epoch": 0.6413296864457795,
-      "grad_norm": 502.4867248535156,
-      "learning_rate": 1.7694045586498752e-05,
-      "loss": 38.337,
-      "step": 158740
-    },
-    {
-      "epoch": 0.6413700877111471,
-      "grad_norm": 961.3887329101562,
-      "learning_rate": 1.7690707381931583e-05,
-      "loss": 61.8364,
-      "step": 158750
-    },
-    {
-      "epoch": 0.6414104889765148,
-      "grad_norm": 2974.65087890625,
-      "learning_rate": 1.768736931986284e-05,
-      "loss": 53.8332,
-      "step": 158760
-    },
-    {
-      "epoch": 0.6414508902418824,
-      "grad_norm": 1488.987060546875,
-      "learning_rate": 1.768403140035758e-05,
-      "loss": 49.7502,
-      "step": 158770
-    },
-    {
-      "epoch": 0.64149129150725,
-      "grad_norm": 713.788818359375,
-      "learning_rate": 1.768069362348091e-05,
-      "loss": 48.6836,
-      "step": 158780
-    },
-    {
-      "epoch": 0.6415316927726177,
-      "grad_norm": 403.7445373535156,
-      "learning_rate": 1.767735598929788e-05,
-      "loss": 70.3594,
-      "step": 158790
-    },
-    {
-      "epoch": 0.6415720940379853,
-      "grad_norm": 0.0,
-      "learning_rate": 1.767401849787357e-05,
-      "loss": 34.1432,
-      "step": 158800
-    },
-    {
-      "epoch": 0.6416124953033528,
-      "grad_norm": 4756.49951171875,
-      "learning_rate": 1.7670681149273038e-05,
-      "loss": 64.6012,
-      "step": 158810
-    },
-    {
-      "epoch": 0.6416528965687205,
-      "grad_norm": 906.1963500976562,
-      "learning_rate": 1.7667343943561344e-05,
-      "loss": 63.7832,
-      "step": 158820
-    },
-    {
-      "epoch": 0.6416932978340881,
-      "grad_norm": 335.9945373535156,
-      "learning_rate": 1.7664006880803563e-05,
-      "loss": 51.2343,
-      "step": 158830
-    },
-    {
-      "epoch": 0.6417336990994558,
-      "grad_norm": 1291.095947265625,
-      "learning_rate": 1.7660669961064748e-05,
-      "loss": 59.827,
-      "step": 158840
-    },
-    {
-      "epoch": 0.6417741003648234,
-      "grad_norm": 578.92236328125,
-      "learning_rate": 1.7657333184409936e-05,
-      "loss": 48.6957,
-      "step": 158850
-    },
-    {
-      "epoch": 0.641814501630191,
-      "grad_norm": 1438.8035888671875,
-      "learning_rate": 1.7653996550904205e-05,
-      "loss": 59.5564,
-      "step": 158860
-    },
-    {
-      "epoch": 0.6418549028955587,
-      "grad_norm": 251.0753936767578,
-      "learning_rate": 1.765066006061259e-05,
-      "loss": 51.8155,
-      "step": 158870
-    },
-    {
-      "epoch": 0.6418953041609263,
-      "grad_norm": 525.5086059570312,
-      "learning_rate": 1.7647323713600138e-05,
-      "loss": 51.6161,
-      "step": 158880
-    },
-    {
-      "epoch": 0.641935705426294,
-      "grad_norm": 950.3296508789062,
-      "learning_rate": 1.7643987509931903e-05,
-      "loss": 49.2463,
-      "step": 158890
-    },
-    {
-      "epoch": 0.6419761066916616,
-      "grad_norm": 400.7654724121094,
-      "learning_rate": 1.7640651449672913e-05,
-      "loss": 48.2175,
-      "step": 158900
-    },
-    {
-      "epoch": 0.6420165079570292,
-      "grad_norm": 494.3016662597656,
-      "learning_rate": 1.7637315532888216e-05,
-      "loss": 60.7076,
-      "step": 158910
-    },
-    {
-      "epoch": 0.6420569092223969,
-      "grad_norm": 913.7971801757812,
-      "learning_rate": 1.7633979759642844e-05,
-      "loss": 38.4652,
-      "step": 158920
-    },
-    {
-      "epoch": 0.6420973104877645,
-      "grad_norm": 495.9393310546875,
-      "learning_rate": 1.7630644130001818e-05,
-      "loss": 47.9746,
-      "step": 158930
-    },
-    {
-      "epoch": 0.6421377117531321,
-      "grad_norm": 903.8270263671875,
-      "learning_rate": 1.7627308644030187e-05,
-      "loss": 76.0501,
-      "step": 158940
-    },
-    {
-      "epoch": 0.6421781130184997,
-      "grad_norm": 1427.69970703125,
-      "learning_rate": 1.7623973301792966e-05,
-      "loss": 37.5955,
-      "step": 158950
-    },
-    {
-      "epoch": 0.6422185142838673,
-      "grad_norm": 700.1619262695312,
-      "learning_rate": 1.7620638103355186e-05,
-      "loss": 50.5823,
-      "step": 158960
-    },
-    {
-      "epoch": 0.642258915549235,
-      "grad_norm": 452.8930969238281,
-      "learning_rate": 1.761730304878187e-05,
-      "loss": 42.1029,
-      "step": 158970
-    },
-    {
-      "epoch": 0.6422993168146026,
-      "grad_norm": 248.73570251464844,
-      "learning_rate": 1.7613968138138026e-05,
-      "loss": 32.7044,
-      "step": 158980
-    },
-    {
-      "epoch": 0.6423397180799703,
-      "grad_norm": 779.9566650390625,
-      "learning_rate": 1.7610633371488682e-05,
-      "loss": 61.1448,
-      "step": 158990
-    },
-    {
-      "epoch": 0.6423801193453379,
-      "grad_norm": 1629.21337890625,
-      "learning_rate": 1.7607298748898842e-05,
-      "loss": 63.5136,
-      "step": 159000
-    },
-    {
-      "epoch": 0.6424205206107055,
-      "grad_norm": 824.9151000976562,
-      "learning_rate": 1.760396427043351e-05,
-      "loss": 42.806,
-      "step": 159010
-    },
-    {
-      "epoch": 0.6424609218760732,
-      "grad_norm": 1080.2861328125,
-      "learning_rate": 1.7600629936157708e-05,
-      "loss": 45.2424,
-      "step": 159020
-    },
-    {
-      "epoch": 0.6425013231414408,
-      "grad_norm": 955.01708984375,
-      "learning_rate": 1.759729574613643e-05,
-      "loss": 57.9693,
-      "step": 159030
-    },
-    {
-      "epoch": 0.6425417244068085,
-      "grad_norm": 337.5832824707031,
-      "learning_rate": 1.759396170043469e-05,
-      "loss": 60.3471,
-      "step": 159040
-    },
-    {
-      "epoch": 0.6425821256721761,
-      "grad_norm": 250.19577026367188,
-      "learning_rate": 1.759062779911748e-05,
-      "loss": 50.2848,
-      "step": 159050
-    },
-    {
-      "epoch": 0.6426225269375437,
-      "grad_norm": 1389.3502197265625,
-      "learning_rate": 1.7587294042249797e-05,
-      "loss": 50.535,
-      "step": 159060
-    },
-    {
-      "epoch": 0.6426629282029113,
-      "grad_norm": 741.0910034179688,
-      "learning_rate": 1.758396042989663e-05,
-      "loss": 46.095,
-      "step": 159070
-    },
-    {
-      "epoch": 0.6427033294682789,
-      "grad_norm": 2326.651123046875,
-      "learning_rate": 1.7580626962122977e-05,
-      "loss": 69.9279,
-      "step": 159080
-    },
-    {
-      "epoch": 0.6427437307336465,
-      "grad_norm": 544.2945556640625,
-      "learning_rate": 1.7577293638993808e-05,
-      "loss": 43.031,
-      "step": 159090
-    },
-    {
-      "epoch": 0.6427841319990142,
-      "grad_norm": 1186.790771484375,
-      "learning_rate": 1.7573960460574133e-05,
-      "loss": 47.7816,
-      "step": 159100
-    },
-    {
-      "epoch": 0.6428245332643818,
-      "grad_norm": 1846.28369140625,
-      "learning_rate": 1.757062742692891e-05,
-      "loss": 67.478,
-      "step": 159110
-    },
-    {
-      "epoch": 0.6428649345297495,
-      "grad_norm": 800.3991088867188,
-      "learning_rate": 1.7567294538123142e-05,
-      "loss": 64.417,
-      "step": 159120
-    },
-    {
-      "epoch": 0.6429053357951171,
-      "grad_norm": 534.356201171875,
-      "learning_rate": 1.7563961794221795e-05,
-      "loss": 32.4049,
-      "step": 159130
-    },
-    {
-      "epoch": 0.6429457370604847,
-      "grad_norm": 1035.6060791015625,
-      "learning_rate": 1.756062919528983e-05,
-      "loss": 65.2042,
-      "step": 159140
-    },
-    {
-      "epoch": 0.6429861383258524,
-      "grad_norm": 1449.9736328125,
-      "learning_rate": 1.755729674139224e-05,
-      "loss": 57.8393,
-      "step": 159150
-    },
-    {
-      "epoch": 0.64302653959122,
-      "grad_norm": 1185.867431640625,
-      "learning_rate": 1.7553964432593976e-05,
-      "loss": 56.2053,
-      "step": 159160
-    },
-    {
-      "epoch": 0.6430669408565877,
-      "grad_norm": 261.984375,
-      "learning_rate": 1.7550632268960003e-05,
-      "loss": 43.3817,
-      "step": 159170
-    },
-    {
-      "epoch": 0.6431073421219553,
-      "grad_norm": 1169.52197265625,
-      "learning_rate": 1.7547300250555304e-05,
-      "loss": 70.408,
-      "step": 159180
-    },
-    {
-      "epoch": 0.643147743387323,
-      "grad_norm": 528.8572387695312,
-      "learning_rate": 1.7543968377444806e-05,
-      "loss": 57.1005,
-      "step": 159190
-    },
-    {
-      "epoch": 0.6431881446526905,
-      "grad_norm": 1181.6226806640625,
-      "learning_rate": 1.7540636649693496e-05,
-      "loss": 48.1173,
-      "step": 159200
-    },
-    {
-      "epoch": 0.6432285459180581,
-      "grad_norm": 453.0842590332031,
-      "learning_rate": 1.7537305067366315e-05,
-      "loss": 49.9454,
-      "step": 159210
-    },
-    {
-      "epoch": 0.6432689471834258,
-      "grad_norm": 920.9298095703125,
-      "learning_rate": 1.75339736305282e-05,
-      "loss": 46.288,
-      "step": 159220
-    },
-    {
-      "epoch": 0.6433093484487934,
-      "grad_norm": 843.010009765625,
-      "learning_rate": 1.753064233924413e-05,
-      "loss": 38.9058,
-      "step": 159230
-    },
-    {
-      "epoch": 0.643349749714161,
-      "grad_norm": 239.32803344726562,
-      "learning_rate": 1.752731119357902e-05,
-      "loss": 68.3454,
-      "step": 159240
-    },
-    {
-      "epoch": 0.6433901509795287,
-      "grad_norm": 506.0323791503906,
-      "learning_rate": 1.7523980193597836e-05,
-      "loss": 40.895,
-      "step": 159250
-    },
-    {
-      "epoch": 0.6434305522448963,
-      "grad_norm": 1487.357177734375,
-      "learning_rate": 1.752064933936551e-05,
-      "loss": 71.0815,
-      "step": 159260
-    },
-    {
-      "epoch": 0.643470953510264,
-      "grad_norm": 1509.1661376953125,
-      "learning_rate": 1.7517318630946957e-05,
-      "loss": 53.0446,
-      "step": 159270
-    },
-    {
-      "epoch": 0.6435113547756316,
-      "grad_norm": 1130.5875244140625,
-      "learning_rate": 1.7513988068407146e-05,
-      "loss": 35.6718,
-      "step": 159280
-    },
-    {
-      "epoch": 0.6435517560409992,
-      "grad_norm": 634.6209106445312,
-      "learning_rate": 1.7510657651810986e-05,
-      "loss": 37.0097,
-      "step": 159290
-    },
-    {
-      "epoch": 0.6435921573063669,
-      "grad_norm": 268.3912048339844,
-      "learning_rate": 1.7507327381223405e-05,
-      "loss": 33.5271,
-      "step": 159300
-    },
-    {
-      "epoch": 0.6436325585717345,
-      "grad_norm": 443.1457214355469,
-      "learning_rate": 1.7503997256709342e-05,
-      "loss": 51.4052,
-      "step": 159310
-    },
-    {
-      "epoch": 0.643672959837102,
-      "grad_norm": 687.546142578125,
-      "learning_rate": 1.750066727833371e-05,
-      "loss": 44.2651,
-      "step": 159320
-    },
-    {
-      "epoch": 0.6437133611024697,
-      "grad_norm": 600.9700317382812,
-      "learning_rate": 1.7497337446161432e-05,
-      "loss": 68.676,
-      "step": 159330
-    },
-    {
-      "epoch": 0.6437537623678373,
-      "grad_norm": 565.4005737304688,
-      "learning_rate": 1.749400776025743e-05,
-      "loss": 70.9337,
-      "step": 159340
-    },
-    {
-      "epoch": 0.643794163633205,
-      "grad_norm": 660.9468383789062,
-      "learning_rate": 1.749067822068659e-05,
-      "loss": 87.7747,
-      "step": 159350
-    },
-    {
-      "epoch": 0.6438345648985726,
-      "grad_norm": 744.2423095703125,
-      "learning_rate": 1.748734882751386e-05,
-      "loss": 51.0859,
-      "step": 159360
-    },
-    {
-      "epoch": 0.6438749661639402,
-      "grad_norm": 1477.296630859375,
-      "learning_rate": 1.7484019580804135e-05,
-      "loss": 85.0133,
-      "step": 159370
-    },
-    {
-      "epoch": 0.6439153674293079,
-      "grad_norm": 585.2022705078125,
-      "learning_rate": 1.7480690480622303e-05,
-      "loss": 33.7616,
-      "step": 159380
-    },
-    {
-      "epoch": 0.6439557686946755,
-      "grad_norm": 181.0767364501953,
-      "learning_rate": 1.7477361527033295e-05,
-      "loss": 51.101,
-      "step": 159390
-    },
-    {
-      "epoch": 0.6439961699600432,
-      "grad_norm": 815.1417236328125,
-      "learning_rate": 1.747403272010199e-05,
-      "loss": 53.4556,
-      "step": 159400
-    },
-    {
-      "epoch": 0.6440365712254108,
-      "grad_norm": 615.5349731445312,
-      "learning_rate": 1.7470704059893296e-05,
-      "loss": 48.4075,
-      "step": 159410
-    },
-    {
-      "epoch": 0.6440769724907784,
-      "grad_norm": 322.2975769042969,
-      "learning_rate": 1.7467375546472105e-05,
-      "loss": 56.5311,
-      "step": 159420
-    },
-    {
-      "epoch": 0.6441173737561461,
-      "grad_norm": 621.2068481445312,
-      "learning_rate": 1.7464047179903296e-05,
-      "loss": 64.1465,
-      "step": 159430
-    },
-    {
-      "epoch": 0.6441577750215137,
-      "grad_norm": 922.0742797851562,
-      "learning_rate": 1.7460718960251772e-05,
-      "loss": 39.1898,
-      "step": 159440
-    },
-    {
-      "epoch": 0.6441981762868813,
-      "grad_norm": 427.25860595703125,
-      "learning_rate": 1.745739088758242e-05,
-      "loss": 27.403,
-      "step": 159450
-    },
-    {
-      "epoch": 0.6442385775522489,
-      "grad_norm": 785.6755981445312,
-      "learning_rate": 1.74540629619601e-05,
-      "loss": 59.5198,
-      "step": 159460
-    },
-    {
-      "epoch": 0.6442789788176165,
-      "grad_norm": 1196.318115234375,
-      "learning_rate": 1.7450735183449723e-05,
-      "loss": 53.6108,
-      "step": 159470
-    },
-    {
-      "epoch": 0.6443193800829842,
-      "grad_norm": 1071.03173828125,
-      "learning_rate": 1.744740755211614e-05,
-      "loss": 46.8555,
-      "step": 159480
-    },
-    {
-      "epoch": 0.6443597813483518,
-      "grad_norm": 506.322998046875,
-      "learning_rate": 1.7444080068024243e-05,
-      "loss": 52.3927,
-      "step": 159490
-    },
-    {
-      "epoch": 0.6444001826137195,
-      "grad_norm": 385.1031799316406,
-      "learning_rate": 1.744075273123889e-05,
-      "loss": 48.9555,
-      "step": 159500
-    },
-    {
-      "epoch": 0.6444405838790871,
-      "grad_norm": 782.7656860351562,
-      "learning_rate": 1.7437425541824952e-05,
-      "loss": 57.9691,
-      "step": 159510
-    },
-    {
-      "epoch": 0.6444809851444547,
-      "grad_norm": 398.4875793457031,
-      "learning_rate": 1.7434098499847306e-05,
-      "loss": 47.1612,
-      "step": 159520
-    },
-    {
-      "epoch": 0.6445213864098224,
-      "grad_norm": 736.70947265625,
-      "learning_rate": 1.74307716053708e-05,
-      "loss": 58.8329,
-      "step": 159530
-    },
-    {
-      "epoch": 0.64456178767519,
-      "grad_norm": 814.1843872070312,
-      "learning_rate": 1.742744485846029e-05,
-      "loss": 48.0108,
-      "step": 159540
-    },
-    {
-      "epoch": 0.6446021889405577,
-      "grad_norm": 578.465576171875,
-      "learning_rate": 1.7424118259180654e-05,
-      "loss": 50.4716,
-      "step": 159550
-    },
-    {
-      "epoch": 0.6446425902059253,
-      "grad_norm": 409.29443359375,
-      "learning_rate": 1.742079180759672e-05,
-      "loss": 31.7956,
-      "step": 159560
-    },
-    {
-      "epoch": 0.6446829914712929,
-      "grad_norm": 604.2058715820312,
-      "learning_rate": 1.7417465503773365e-05,
-      "loss": 69.64,
-      "step": 159570
-    },
-    {
-      "epoch": 0.6447233927366605,
-      "grad_norm": 312.2088928222656,
-      "learning_rate": 1.741413934777542e-05,
-      "loss": 49.1468,
-      "step": 159580
-    },
-    {
-      "epoch": 0.6447637940020281,
-      "grad_norm": 241.517822265625,
-      "learning_rate": 1.7410813339667732e-05,
-      "loss": 28.2745,
-      "step": 159590
-    },
-    {
-      "epoch": 0.6448041952673957,
-      "grad_norm": 911.9425048828125,
-      "learning_rate": 1.7407487479515147e-05,
-      "loss": 68.7837,
-      "step": 159600
-    },
-    {
-      "epoch": 0.6448445965327634,
-      "grad_norm": 568.782958984375,
-      "learning_rate": 1.7404161767382504e-05,
-      "loss": 49.4384,
-      "step": 159610
-    },
-    {
-      "epoch": 0.644884997798131,
-      "grad_norm": 350.11181640625,
-      "learning_rate": 1.7400836203334626e-05,
-      "loss": 41.8597,
-      "step": 159620
-    },
-    {
-      "epoch": 0.6449253990634987,
-      "grad_norm": 634.2789916992188,
-      "learning_rate": 1.739751078743637e-05,
-      "loss": 62.7964,
-      "step": 159630
-    },
-    {
-      "epoch": 0.6449658003288663,
-      "grad_norm": 609.7909545898438,
-      "learning_rate": 1.7394185519752545e-05,
-      "loss": 43.3135,
-      "step": 159640
-    },
-    {
-      "epoch": 0.6450062015942339,
-      "grad_norm": 918.4065551757812,
-      "learning_rate": 1.7390860400348002e-05,
-      "loss": 47.5571,
-      "step": 159650
-    },
-    {
-      "epoch": 0.6450466028596016,
-      "grad_norm": 536.011474609375,
-      "learning_rate": 1.7387535429287548e-05,
-      "loss": 43.0701,
-      "step": 159660
-    },
-    {
-      "epoch": 0.6450870041249692,
-      "grad_norm": 1669.569091796875,
-      "learning_rate": 1.7384210606636007e-05,
-      "loss": 52.9317,
-      "step": 159670
-    },
-    {
-      "epoch": 0.6451274053903369,
-      "grad_norm": 346.0846862792969,
-      "learning_rate": 1.7380885932458206e-05,
-      "loss": 51.7841,
-      "step": 159680
-    },
-    {
-      "epoch": 0.6451678066557045,
-      "grad_norm": 665.9485473632812,
-      "learning_rate": 1.7377561406818958e-05,
-      "loss": 40.3949,
-      "step": 159690
-    },
-    {
-      "epoch": 0.6452082079210721,
-      "grad_norm": 804.7376098632812,
-      "learning_rate": 1.7374237029783062e-05,
-      "loss": 56.1418,
-      "step": 159700
-    },
-    {
-      "epoch": 0.6452486091864397,
-      "grad_norm": 856.9612426757812,
-      "learning_rate": 1.7370912801415357e-05,
-      "loss": 64.3835,
-      "step": 159710
-    },
-    {
-      "epoch": 0.6452890104518073,
-      "grad_norm": 518.4335327148438,
-      "learning_rate": 1.736758872178062e-05,
-      "loss": 58.4394,
-      "step": 159720
-    },
-    {
-      "epoch": 0.645329411717175,
-      "grad_norm": 935.9529418945312,
-      "learning_rate": 1.736426479094368e-05,
-      "loss": 31.8426,
-      "step": 159730
-    },
-    {
-      "epoch": 0.6453698129825426,
-      "grad_norm": 951.4691162109375,
-      "learning_rate": 1.7360941008969327e-05,
-      "loss": 60.8805,
-      "step": 159740
-    },
-    {
-      "epoch": 0.6454102142479102,
-      "grad_norm": 809.8295288085938,
-      "learning_rate": 1.735761737592236e-05,
-      "loss": 45.1077,
-      "step": 159750
-    },
-    {
-      "epoch": 0.6454506155132779,
-      "grad_norm": 653.5640869140625,
-      "learning_rate": 1.7354293891867582e-05,
-      "loss": 61.4744,
-      "step": 159760
-    },
-    {
-      "epoch": 0.6454910167786455,
-      "grad_norm": 555.7434692382812,
-      "learning_rate": 1.7350970556869768e-05,
-      "loss": 44.4319,
-      "step": 159770
-    },
-    {
-      "epoch": 0.6455314180440131,
-      "grad_norm": 519.6600952148438,
-      "learning_rate": 1.7347647370993738e-05,
-      "loss": 62.9967,
-      "step": 159780
-    },
-    {
-      "epoch": 0.6455718193093808,
-      "grad_norm": 571.7482299804688,
-      "learning_rate": 1.7344324334304255e-05,
-      "loss": 31.3623,
-      "step": 159790
-    },
-    {
-      "epoch": 0.6456122205747484,
-      "grad_norm": 1434.1971435546875,
-      "learning_rate": 1.73410014468661e-05,
-      "loss": 68.6211,
-      "step": 159800
-    },
-    {
-      "epoch": 0.6456526218401161,
-      "grad_norm": 634.74951171875,
-      "learning_rate": 1.7337678708744083e-05,
-      "loss": 53.1098,
-      "step": 159810
-    },
-    {
-      "epoch": 0.6456930231054837,
-      "grad_norm": 893.255126953125,
-      "learning_rate": 1.7334356120002957e-05,
-      "loss": 66.1552,
-      "step": 159820
-    },
-    {
-      "epoch": 0.6457334243708513,
-      "grad_norm": 776.960205078125,
-      "learning_rate": 1.7331033680707502e-05,
-      "loss": 43.6166,
-      "step": 159830
-    },
-    {
-      "epoch": 0.6457738256362189,
-      "grad_norm": 1306.64306640625,
-      "learning_rate": 1.7327711390922494e-05,
-      "loss": 56.0734,
-      "step": 159840
-    },
-    {
-      "epoch": 0.6458142269015865,
-      "grad_norm": 947.216552734375,
-      "learning_rate": 1.7324389250712702e-05,
-      "loss": 42.4781,
-      "step": 159850
-    },
-    {
-      "epoch": 0.6458546281669542,
-      "grad_norm": 638.41796875,
-      "learning_rate": 1.73210672601429e-05,
-      "loss": 55.3032,
-      "step": 159860
-    },
-    {
-      "epoch": 0.6458950294323218,
-      "grad_norm": 336.3428649902344,
-      "learning_rate": 1.731774541927784e-05,
-      "loss": 43.4402,
-      "step": 159870
-    },
-    {
-      "epoch": 0.6459354306976894,
-      "grad_norm": 432.4233703613281,
-      "learning_rate": 1.7314423728182283e-05,
-      "loss": 35.9008,
-      "step": 159880
-    },
-    {
-      "epoch": 0.6459758319630571,
-      "grad_norm": 796.058837890625,
-      "learning_rate": 1.7311102186921e-05,
-      "loss": 51.9767,
-      "step": 159890
-    },
-    {
-      "epoch": 0.6460162332284247,
-      "grad_norm": 522.548828125,
-      "learning_rate": 1.730778079555874e-05,
-      "loss": 44.1105,
-      "step": 159900
-    },
-    {
-      "epoch": 0.6460566344937924,
-      "grad_norm": 1422.60205078125,
-      "learning_rate": 1.7304459554160245e-05,
-      "loss": 74.7213,
-      "step": 159910
-    },
-    {
-      "epoch": 0.64609703575916,
-      "grad_norm": 675.3711547851562,
-      "learning_rate": 1.7301138462790278e-05,
-      "loss": 104.2761,
-      "step": 159920
-    },
-    {
-      "epoch": 0.6461374370245276,
-      "grad_norm": 591.744140625,
-      "learning_rate": 1.729781752151358e-05,
-      "loss": 38.1823,
-      "step": 159930
-    },
-    {
-      "epoch": 0.6461778382898953,
-      "grad_norm": 238.55694580078125,
-      "learning_rate": 1.7294496730394895e-05,
-      "loss": 35.8654,
-      "step": 159940
-    },
-    {
-      "epoch": 0.6462182395552629,
-      "grad_norm": 416.1772766113281,
-      "learning_rate": 1.7291176089498967e-05,
-      "loss": 51.6108,
-      "step": 159950
-    },
-    {
-      "epoch": 0.6462586408206304,
-      "grad_norm": 1862.9417724609375,
-      "learning_rate": 1.728785559889052e-05,
-      "loss": 52.4495,
-      "step": 159960
-    },
-    {
-      "epoch": 0.6462990420859981,
-      "grad_norm": 415.4091796875,
-      "learning_rate": 1.7284535258634307e-05,
-      "loss": 53.8081,
-      "step": 159970
-    },
-    {
-      "epoch": 0.6463394433513657,
-      "grad_norm": 613.8626098632812,
-      "learning_rate": 1.7281215068795055e-05,
-      "loss": 55.2499,
-      "step": 159980
-    },
-    {
-      "epoch": 0.6463798446167334,
-      "grad_norm": 326.62451171875,
-      "learning_rate": 1.7277895029437474e-05,
-      "loss": 54.2632,
-      "step": 159990
-    },
-    {
-      "epoch": 0.646420245882101,
-      "grad_norm": 860.3509521484375,
-      "learning_rate": 1.7274575140626318e-05,
-      "loss": 44.3942,
-      "step": 160000
-    },
-    {
-      "epoch": 0.6464606471474686,
-      "grad_norm": 1033.29833984375,
-      "learning_rate": 1.727125540242629e-05,
-      "loss": 52.0802,
-      "step": 160010
-    },
-    {
-      "epoch": 0.6465010484128363,
-      "grad_norm": 697.0635986328125,
-      "learning_rate": 1.726793581490212e-05,
-      "loss": 58.4031,
-      "step": 160020
-    },
-    {
-      "epoch": 0.6465414496782039,
-      "grad_norm": 649.52734375,
-      "learning_rate": 1.7264616378118528e-05,
-      "loss": 52.7141,
-      "step": 160030
-    },
-    {
-      "epoch": 0.6465818509435716,
-      "grad_norm": 911.1539306640625,
-      "learning_rate": 1.7261297092140212e-05,
-      "loss": 47.7209,
-      "step": 160040
-    },
-    {
-      "epoch": 0.6466222522089392,
-      "grad_norm": 635.70556640625,
-      "learning_rate": 1.72579779570319e-05,
-      "loss": 43.9278,
-      "step": 160050
-    },
-    {
-      "epoch": 0.6466626534743068,
-      "grad_norm": 839.5745239257812,
-      "learning_rate": 1.7254658972858293e-05,
-      "loss": 38.8251,
-      "step": 160060
-    },
-    {
-      "epoch": 0.6467030547396745,
-      "grad_norm": 529.5610961914062,
-      "learning_rate": 1.7251340139684086e-05,
-      "loss": 37.6862,
-      "step": 160070
-    },
-    {
-      "epoch": 0.6467434560050421,
-      "grad_norm": 743.9073486328125,
-      "learning_rate": 1.7248021457574004e-05,
-      "loss": 48.4526,
-      "step": 160080
-    },
-    {
-      "epoch": 0.6467838572704097,
-      "grad_norm": 0.0,
-      "learning_rate": 1.7244702926592733e-05,
-      "loss": 47.8844,
-      "step": 160090
-    },
-    {
-      "epoch": 0.6468242585357773,
-      "grad_norm": 200.43478393554688,
-      "learning_rate": 1.7241384546804972e-05,
-      "loss": 46.4571,
-      "step": 160100
-    },
-    {
-      "epoch": 0.6468646598011449,
-      "grad_norm": 921.7332763671875,
-      "learning_rate": 1.7238066318275415e-05,
-      "loss": 60.5268,
-      "step": 160110
-    },
-    {
-      "epoch": 0.6469050610665126,
-      "grad_norm": 918.865478515625,
-      "learning_rate": 1.7234748241068742e-05,
-      "loss": 63.9479,
-      "step": 160120
-    },
-    {
-      "epoch": 0.6469454623318802,
-      "grad_norm": 654.3958740234375,
-      "learning_rate": 1.723143031524966e-05,
-      "loss": 71.1433,
-      "step": 160130
-    },
-    {
-      "epoch": 0.6469858635972479,
-      "grad_norm": 1054.6842041015625,
-      "learning_rate": 1.722811254088284e-05,
-      "loss": 45.9198,
-      "step": 160140
-    },
-    {
-      "epoch": 0.6470262648626155,
-      "grad_norm": 494.39654541015625,
-      "learning_rate": 1.722479491803296e-05,
-      "loss": 38.1067,
-      "step": 160150
-    },
-    {
-      "epoch": 0.6470666661279831,
-      "grad_norm": 522.5104370117188,
-      "learning_rate": 1.7221477446764717e-05,
-      "loss": 47.4557,
-      "step": 160160
-    },
-    {
-      "epoch": 0.6471070673933508,
-      "grad_norm": 751.9249267578125,
-      "learning_rate": 1.7218160127142767e-05,
-      "loss": 54.4638,
-      "step": 160170
-    },
-    {
-      "epoch": 0.6471474686587184,
-      "grad_norm": 520.9002075195312,
-      "learning_rate": 1.7214842959231794e-05,
-      "loss": 58.695,
-      "step": 160180
-    },
-    {
-      "epoch": 0.6471878699240861,
-      "grad_norm": 1485.286865234375,
-      "learning_rate": 1.721152594309647e-05,
-      "loss": 80.7229,
-      "step": 160190
-    },
-    {
-      "epoch": 0.6472282711894537,
-      "grad_norm": 872.4847412109375,
-      "learning_rate": 1.7208209078801454e-05,
-      "loss": 44.0047,
-      "step": 160200
-    },
-    {
-      "epoch": 0.6472686724548213,
-      "grad_norm": 574.897705078125,
-      "learning_rate": 1.7204892366411416e-05,
-      "loss": 30.545,
-      "step": 160210
-    },
-    {
-      "epoch": 0.6473090737201889,
-      "grad_norm": 4002.87939453125,
-      "learning_rate": 1.720157580599101e-05,
-      "loss": 69.5485,
-      "step": 160220
-    },
-    {
-      "epoch": 0.6473494749855565,
-      "grad_norm": 231.5006866455078,
-      "learning_rate": 1.7198259397604904e-05,
-      "loss": 42.8945,
-      "step": 160230
-    },
-    {
-      "epoch": 0.6473898762509241,
-      "grad_norm": 545.1617431640625,
-      "learning_rate": 1.719494314131775e-05,
-      "loss": 46.0572,
-      "step": 160240
-    },
-    {
-      "epoch": 0.6474302775162918,
-      "grad_norm": 659.3037719726562,
-      "learning_rate": 1.7191627037194186e-05,
-      "loss": 51.2325,
-      "step": 160250
-    },
-    {
-      "epoch": 0.6474706787816594,
-      "grad_norm": 855.5885620117188,
-      "learning_rate": 1.718831108529888e-05,
-      "loss": 54.8725,
-      "step": 160260
-    },
-    {
-      "epoch": 0.6475110800470271,
-      "grad_norm": 484.8570861816406,
-      "learning_rate": 1.7184995285696477e-05,
-      "loss": 54.7183,
-      "step": 160270
-    },
-    {
-      "epoch": 0.6475514813123947,
-      "grad_norm": 827.1829223632812,
-      "learning_rate": 1.7181679638451603e-05,
-      "loss": 49.625,
-      "step": 160280
-    },
-    {
-      "epoch": 0.6475918825777623,
-      "grad_norm": 199.2564239501953,
-      "learning_rate": 1.7178364143628923e-05,
-      "loss": 29.7762,
-      "step": 160290
-    },
-    {
-      "epoch": 0.64763228384313,
-      "grad_norm": 830.081298828125,
-      "learning_rate": 1.717504880129304e-05,
-      "loss": 66.7952,
-      "step": 160300
-    },
-    {
-      "epoch": 0.6476726851084976,
-      "grad_norm": 859.8870849609375,
-      "learning_rate": 1.7171733611508627e-05,
-      "loss": 51.3871,
-      "step": 160310
-    },
-    {
-      "epoch": 0.6477130863738653,
-      "grad_norm": 718.7318725585938,
-      "learning_rate": 1.7168418574340298e-05,
-      "loss": 62.161,
-      "step": 160320
-    },
-    {
-      "epoch": 0.6477534876392329,
-      "grad_norm": 733.2501831054688,
-      "learning_rate": 1.716510368985267e-05,
-      "loss": 77.6914,
-      "step": 160330
-    },
-    {
-      "epoch": 0.6477938889046005,
-      "grad_norm": 958.80126953125,
-      "learning_rate": 1.7161788958110388e-05,
-      "loss": 69.0079,
-      "step": 160340
-    },
-    {
-      "epoch": 0.6478342901699681,
-      "grad_norm": 597.50439453125,
-      "learning_rate": 1.7158474379178064e-05,
-      "loss": 65.9391,
-      "step": 160350
-    },
-    {
-      "epoch": 0.6478746914353357,
-      "grad_norm": 956.994873046875,
-      "learning_rate": 1.7155159953120313e-05,
-      "loss": 66.6005,
-      "step": 160360
-    },
-    {
-      "epoch": 0.6479150927007034,
-      "grad_norm": 524.7179565429688,
-      "learning_rate": 1.7151845680001767e-05,
-      "loss": 39.2416,
-      "step": 160370
-    },
-    {
-      "epoch": 0.647955493966071,
-      "grad_norm": 451.7310485839844,
-      "learning_rate": 1.7148531559887018e-05,
-      "loss": 55.5202,
-      "step": 160380
-    },
-    {
-      "epoch": 0.6479958952314386,
-      "grad_norm": 627.1416015625,
-      "learning_rate": 1.71452175928407e-05,
-      "loss": 69.2645,
-      "step": 160390
-    },
-    {
-      "epoch": 0.6480362964968063,
-      "grad_norm": 482.4586486816406,
-      "learning_rate": 1.7141903778927406e-05,
-      "loss": 58.0431,
-      "step": 160400
-    },
-    {
-      "epoch": 0.6480766977621739,
-      "grad_norm": 681.709716796875,
-      "learning_rate": 1.7138590118211733e-05,
-      "loss": 52.02,
-      "step": 160410
-    },
-    {
-      "epoch": 0.6481170990275416,
-      "grad_norm": 361.3819885253906,
-      "learning_rate": 1.7135276610758307e-05,
-      "loss": 53.8131,
-      "step": 160420
-    },
-    {
-      "epoch": 0.6481575002929092,
-      "grad_norm": 607.1292724609375,
-      "learning_rate": 1.713196325663171e-05,
-      "loss": 51.5823,
-      "step": 160430
-    },
-    {
-      "epoch": 0.6481979015582768,
-      "grad_norm": 1490.06298828125,
-      "learning_rate": 1.7128650055896535e-05,
-      "loss": 68.2615,
-      "step": 160440
-    },
-    {
-      "epoch": 0.6482383028236445,
-      "grad_norm": 596.255615234375,
-      "learning_rate": 1.7125337008617386e-05,
-      "loss": 49.4564,
-      "step": 160450
-    },
-    {
-      "epoch": 0.6482787040890121,
-      "grad_norm": 834.3792114257812,
-      "learning_rate": 1.7122024114858835e-05,
-      "loss": 49.1376,
-      "step": 160460
-    },
-    {
-      "epoch": 0.6483191053543798,
-      "grad_norm": 1242.96337890625,
-      "learning_rate": 1.711871137468549e-05,
-      "loss": 66.0371,
-      "step": 160470
-    },
-    {
-      "epoch": 0.6483595066197473,
-      "grad_norm": 1074.2083740234375,
-      "learning_rate": 1.7115398788161925e-05,
-      "loss": 64.567,
-      "step": 160480
-    },
-    {
-      "epoch": 0.6483999078851149,
-      "grad_norm": 1067.1346435546875,
-      "learning_rate": 1.7112086355352707e-05,
-      "loss": 53.1166,
-      "step": 160490
-    },
-    {
-      "epoch": 0.6484403091504826,
-      "grad_norm": 442.2884826660156,
-      "learning_rate": 1.7108774076322443e-05,
-      "loss": 32.6686,
-      "step": 160500
-    },
-    {
-      "epoch": 0.6484807104158502,
-      "grad_norm": 450.2314758300781,
-      "learning_rate": 1.7105461951135686e-05,
-      "loss": 50.7306,
-      "step": 160510
-    },
-    {
-      "epoch": 0.6485211116812178,
-      "grad_norm": 1504.6929931640625,
-      "learning_rate": 1.7102149979857e-05,
-      "loss": 47.6178,
-      "step": 160520
-    },
-    {
-      "epoch": 0.6485615129465855,
-      "grad_norm": 605.5977783203125,
-      "learning_rate": 1.7098838162550983e-05,
-      "loss": 41.464,
-      "step": 160530
-    },
-    {
-      "epoch": 0.6486019142119531,
-      "grad_norm": 867.1982421875,
-      "learning_rate": 1.7095526499282172e-05,
-      "loss": 46.7762,
-      "step": 160540
-    },
-    {
-      "epoch": 0.6486423154773208,
-      "grad_norm": 447.15863037109375,
-      "learning_rate": 1.7092214990115152e-05,
-      "loss": 32.9571,
-      "step": 160550
-    },
-    {
-      "epoch": 0.6486827167426884,
-      "grad_norm": 679.4548950195312,
-      "learning_rate": 1.708890363511447e-05,
-      "loss": 72.9706,
-      "step": 160560
-    },
-    {
-      "epoch": 0.648723118008056,
-      "grad_norm": 1380.580810546875,
-      "learning_rate": 1.708559243434467e-05,
-      "loss": 62.2892,
-      "step": 160570
-    },
-    {
-      "epoch": 0.6487635192734237,
-      "grad_norm": 668.0836181640625,
-      "learning_rate": 1.7082281387870338e-05,
-      "loss": 44.8698,
-      "step": 160580
-    },
-    {
-      "epoch": 0.6488039205387913,
-      "grad_norm": 675.6842651367188,
-      "learning_rate": 1.7078970495756002e-05,
-      "loss": 41.055,
-      "step": 160590
-    },
-    {
-      "epoch": 0.6488443218041589,
-      "grad_norm": 703.9763793945312,
-      "learning_rate": 1.7075659758066208e-05,
-      "loss": 63.1799,
-      "step": 160600
-    },
-    {
-      "epoch": 0.6488847230695265,
-      "grad_norm": 458.03302001953125,
-      "learning_rate": 1.7072349174865513e-05,
-      "loss": 38.6174,
-      "step": 160610
-    },
-    {
-      "epoch": 0.6489251243348941,
-      "grad_norm": 291.4510803222656,
-      "learning_rate": 1.7069038746218445e-05,
-      "loss": 55.229,
-      "step": 160620
-    },
-    {
-      "epoch": 0.6489655256002618,
-      "grad_norm": 2080.05322265625,
-      "learning_rate": 1.7065728472189563e-05,
-      "loss": 68.5965,
-      "step": 160630
-    },
-    {
-      "epoch": 0.6490059268656294,
-      "grad_norm": 536.862548828125,
-      "learning_rate": 1.7062418352843382e-05,
-      "loss": 67.5582,
-      "step": 160640
-    },
-    {
-      "epoch": 0.649046328130997,
-      "grad_norm": 462.50970458984375,
-      "learning_rate": 1.7059108388244432e-05,
-      "loss": 38.4309,
-      "step": 160650
-    },
-    {
-      "epoch": 0.6490867293963647,
-      "grad_norm": 838.4609985351562,
-      "learning_rate": 1.7055798578457266e-05,
-      "loss": 57.7827,
-      "step": 160660
-    },
-    {
-      "epoch": 0.6491271306617323,
-      "grad_norm": 761.4349365234375,
-      "learning_rate": 1.7052488923546396e-05,
-      "loss": 42.0861,
-      "step": 160670
-    },
-    {
-      "epoch": 0.6491675319271,
-      "grad_norm": 357.4921569824219,
-      "learning_rate": 1.7049179423576334e-05,
-      "loss": 42.084,
-      "step": 160680
-    },
-    {
-      "epoch": 0.6492079331924676,
-      "grad_norm": 565.8134765625,
-      "learning_rate": 1.7045870078611627e-05,
-      "loss": 61.3284,
-      "step": 160690
-    },
-    {
-      "epoch": 0.6492483344578353,
-      "grad_norm": 607.5779418945312,
-      "learning_rate": 1.7042560888716766e-05,
-      "loss": 43.1379,
-      "step": 160700
-    },
-    {
-      "epoch": 0.6492887357232029,
-      "grad_norm": 716.2503662109375,
-      "learning_rate": 1.7039251853956283e-05,
-      "loss": 50.0222,
-      "step": 160710
-    },
-    {
-      "epoch": 0.6493291369885705,
-      "grad_norm": 446.25128173828125,
-      "learning_rate": 1.703594297439469e-05,
-      "loss": 26.8644,
-      "step": 160720
-    },
-    {
-      "epoch": 0.6493695382539381,
-      "grad_norm": 1103.192626953125,
-      "learning_rate": 1.7032634250096473e-05,
-      "loss": 58.0758,
-      "step": 160730
-    },
-    {
-      "epoch": 0.6494099395193057,
-      "grad_norm": 270.75579833984375,
-      "learning_rate": 1.7029325681126167e-05,
-      "loss": 37.3797,
-      "step": 160740
-    },
-    {
-      "epoch": 0.6494503407846733,
-      "grad_norm": 871.2796020507812,
-      "learning_rate": 1.702601726754825e-05,
-      "loss": 58.7828,
-      "step": 160750
-    },
-    {
-      "epoch": 0.649490742050041,
-      "grad_norm": 584.755126953125,
-      "learning_rate": 1.702270900942724e-05,
-      "loss": 68.4647,
-      "step": 160760
-    },
-    {
-      "epoch": 0.6495311433154086,
-      "grad_norm": 1826.71044921875,
-      "learning_rate": 1.7019400906827624e-05,
-      "loss": 58.1163,
-      "step": 160770
-    },
-    {
-      "epoch": 0.6495715445807763,
-      "grad_norm": 397.8055419921875,
-      "learning_rate": 1.7016092959813893e-05,
-      "loss": 75.9175,
-      "step": 160780
-    },
-    {
-      "epoch": 0.6496119458461439,
-      "grad_norm": 535.3204956054688,
-      "learning_rate": 1.7012785168450543e-05,
-      "loss": 89.4188,
-      "step": 160790
-    },
-    {
-      "epoch": 0.6496523471115115,
-      "grad_norm": 877.5353393554688,
-      "learning_rate": 1.7009477532802054e-05,
-      "loss": 78.9825,
-      "step": 160800
-    },
-    {
-      "epoch": 0.6496927483768792,
-      "grad_norm": 1077.1090087890625,
-      "learning_rate": 1.7006170052932916e-05,
-      "loss": 43.3018,
-      "step": 160810
-    },
-    {
-      "epoch": 0.6497331496422468,
-      "grad_norm": 596.0875854492188,
-      "learning_rate": 1.700286272890761e-05,
-      "loss": 41.5323,
-      "step": 160820
-    },
-    {
-      "epoch": 0.6497735509076145,
-      "grad_norm": 3221.248291015625,
-      "learning_rate": 1.6999555560790607e-05,
-      "loss": 52.7061,
-      "step": 160830
-    },
-    {
-      "epoch": 0.6498139521729821,
-      "grad_norm": 436.67425537109375,
-      "learning_rate": 1.6996248548646394e-05,
-      "loss": 59.5454,
-      "step": 160840
-    },
-    {
-      "epoch": 0.6498543534383497,
-      "grad_norm": 418.7576599121094,
-      "learning_rate": 1.6992941692539437e-05,
-      "loss": 37.4969,
-      "step": 160850
-    },
-    {
-      "epoch": 0.6498947547037173,
-      "grad_norm": 775.6710815429688,
-      "learning_rate": 1.6989634992534194e-05,
-      "loss": 33.1508,
-      "step": 160860
-    },
-    {
-      "epoch": 0.6499351559690849,
-      "grad_norm": 513.3265380859375,
-      "learning_rate": 1.6986328448695148e-05,
-      "loss": 62.6621,
-      "step": 160870
-    },
-    {
-      "epoch": 0.6499755572344526,
-      "grad_norm": 1174.877685546875,
-      "learning_rate": 1.6983022061086763e-05,
-      "loss": 75.1632,
-      "step": 160880
-    },
-    {
-      "epoch": 0.6500159584998202,
-      "grad_norm": 117.83036041259766,
-      "learning_rate": 1.6979715829773476e-05,
-      "loss": 28.0209,
-      "step": 160890
-    },
-    {
-      "epoch": 0.6500563597651878,
-      "grad_norm": 667.8468627929688,
-      "learning_rate": 1.6976409754819767e-05,
-      "loss": 41.9234,
-      "step": 160900
-    },
-    {
-      "epoch": 0.6500967610305555,
-      "grad_norm": 1031.58544921875,
-      "learning_rate": 1.6973103836290072e-05,
-      "loss": 70.4715,
-      "step": 160910
-    },
-    {
-      "epoch": 0.6501371622959231,
-      "grad_norm": 372.794677734375,
-      "learning_rate": 1.6969798074248858e-05,
-      "loss": 45.1825,
-      "step": 160920
-    },
-    {
-      "epoch": 0.6501775635612908,
-      "grad_norm": 776.9822387695312,
-      "learning_rate": 1.6966492468760565e-05,
-      "loss": 63.0023,
-      "step": 160930
-    },
-    {
-      "epoch": 0.6502179648266584,
-      "grad_norm": 1337.965576171875,
-      "learning_rate": 1.6963187019889626e-05,
-      "loss": 55.9985,
-      "step": 160940
-    },
-    {
-      "epoch": 0.650258366092026,
-      "grad_norm": 828.7103881835938,
-      "learning_rate": 1.6959881727700508e-05,
-      "loss": 55.6604,
-      "step": 160950
-    },
-    {
-      "epoch": 0.6502987673573937,
-      "grad_norm": 626.9825439453125,
-      "learning_rate": 1.6956576592257635e-05,
-      "loss": 46.8431,
-      "step": 160960
-    },
-    {
-      "epoch": 0.6503391686227613,
-      "grad_norm": 660.419189453125,
-      "learning_rate": 1.6953271613625432e-05,
-      "loss": 68.8267,
-      "step": 160970
-    },
-    {
-      "epoch": 0.650379569888129,
-      "grad_norm": 528.8330078125,
-      "learning_rate": 1.694996679186835e-05,
-      "loss": 77.1253,
-      "step": 160980
-    },
-    {
-      "epoch": 0.6504199711534965,
-      "grad_norm": 540.1735229492188,
-      "learning_rate": 1.69466621270508e-05,
-      "loss": 44.3839,
-      "step": 160990
-    },
-    {
-      "epoch": 0.6504603724188641,
-      "grad_norm": 568.5739135742188,
-      "learning_rate": 1.6943357619237226e-05,
-      "loss": 46.5589,
-      "step": 161000
-    },
-    {
-      "epoch": 0.6505007736842318,
-      "grad_norm": 597.6793212890625,
-      "learning_rate": 1.6940053268492045e-05,
-      "loss": 32.6383,
-      "step": 161010
-    },
-    {
-      "epoch": 0.6505411749495994,
-      "grad_norm": 1134.3443603515625,
-      "learning_rate": 1.693674907487966e-05,
-      "loss": 44.2843,
-      "step": 161020
-    },
-    {
-      "epoch": 0.650581576214967,
-      "grad_norm": 681.8280029296875,
-      "learning_rate": 1.6933445038464518e-05,
-      "loss": 51.1013,
-      "step": 161030
-    },
-    {
-      "epoch": 0.6506219774803347,
-      "grad_norm": 1109.9951171875,
-      "learning_rate": 1.693014115931102e-05,
-      "loss": 73.4727,
-      "step": 161040
-    },
-    {
-      "epoch": 0.6506623787457023,
-      "grad_norm": 895.8487548828125,
-      "learning_rate": 1.6926837437483566e-05,
-      "loss": 84.4225,
-      "step": 161050
-    },
-    {
-      "epoch": 0.65070278001107,
-      "grad_norm": 3003.786865234375,
-      "learning_rate": 1.6923533873046578e-05,
-      "loss": 60.5545,
-      "step": 161060
-    },
-    {
-      "epoch": 0.6507431812764376,
-      "grad_norm": 1717.6845703125,
-      "learning_rate": 1.6920230466064446e-05,
-      "loss": 49.7207,
-      "step": 161070
-    },
-    {
-      "epoch": 0.6507835825418052,
-      "grad_norm": 1025.4805908203125,
-      "learning_rate": 1.6916927216601593e-05,
-      "loss": 57.6173,
-      "step": 161080
-    },
-    {
-      "epoch": 0.6508239838071729,
-      "grad_norm": 583.477783203125,
-      "learning_rate": 1.69136241247224e-05,
-      "loss": 56.1462,
-      "step": 161090
-    },
-    {
-      "epoch": 0.6508643850725405,
-      "grad_norm": 2021.0592041015625,
-      "learning_rate": 1.6910321190491263e-05,
-      "loss": 59.2203,
-      "step": 161100
-    },
-    {
-      "epoch": 0.650904786337908,
-      "grad_norm": 571.0921630859375,
-      "learning_rate": 1.6907018413972586e-05,
-      "loss": 31.7513,
-      "step": 161110
-    },
-    {
-      "epoch": 0.6509451876032757,
-      "grad_norm": 1296.876220703125,
-      "learning_rate": 1.690371579523075e-05,
-      "loss": 49.9373,
-      "step": 161120
-    },
-    {
-      "epoch": 0.6509855888686433,
-      "grad_norm": 481.8075866699219,
-      "learning_rate": 1.6900413334330142e-05,
-      "loss": 46.3444,
-      "step": 161130
-    },
-    {
-      "epoch": 0.651025990134011,
-      "grad_norm": 1343.282958984375,
-      "learning_rate": 1.6897111031335145e-05,
-      "loss": 52.6248,
-      "step": 161140
-    },
-    {
-      "epoch": 0.6510663913993786,
-      "grad_norm": 260.4144592285156,
-      "learning_rate": 1.689380888631014e-05,
-      "loss": 47.9411,
-      "step": 161150
-    },
-    {
-      "epoch": 0.6511067926647462,
-      "grad_norm": 760.5383911132812,
-      "learning_rate": 1.689050689931951e-05,
-      "loss": 46.6919,
-      "step": 161160
-    },
-    {
-      "epoch": 0.6511471939301139,
-      "grad_norm": 795.2639770507812,
-      "learning_rate": 1.688720507042762e-05,
-      "loss": 47.9917,
-      "step": 161170
-    },
-    {
-      "epoch": 0.6511875951954815,
-      "grad_norm": 892.8135375976562,
-      "learning_rate": 1.6883903399698833e-05,
-      "loss": 35.0797,
-      "step": 161180
-    },
-    {
-      "epoch": 0.6512279964608492,
-      "grad_norm": 742.6301879882812,
-      "learning_rate": 1.688060188719754e-05,
-      "loss": 46.3297,
-      "step": 161190
-    },
-    {
-      "epoch": 0.6512683977262168,
-      "grad_norm": 897.2402954101562,
-      "learning_rate": 1.6877300532988094e-05,
-      "loss": 47.0836,
-      "step": 161200
-    },
-    {
-      "epoch": 0.6513087989915844,
-      "grad_norm": 856.124267578125,
-      "learning_rate": 1.6873999337134846e-05,
-      "loss": 69.0124,
-      "step": 161210
-    },
-    {
-      "epoch": 0.6513492002569521,
-      "grad_norm": 929.366943359375,
-      "learning_rate": 1.6870698299702177e-05,
-      "loss": 55.2155,
-      "step": 161220
-    },
-    {
-      "epoch": 0.6513896015223197,
-      "grad_norm": 724.1373291015625,
-      "learning_rate": 1.686739742075442e-05,
-      "loss": 50.1667,
-      "step": 161230
-    },
-    {
-      "epoch": 0.6514300027876873,
-      "grad_norm": 252.8017578125,
-      "learning_rate": 1.686409670035594e-05,
-      "loss": 32.0923,
-      "step": 161240
-    },
-    {
-      "epoch": 0.6514704040530549,
-      "grad_norm": 1198.72900390625,
-      "learning_rate": 1.686079613857109e-05,
-      "loss": 43.797,
-      "step": 161250
-    },
-    {
-      "epoch": 0.6515108053184225,
-      "grad_norm": 369.4762268066406,
-      "learning_rate": 1.6857495735464195e-05,
-      "loss": 58.9313,
-      "step": 161260
-    },
-    {
-      "epoch": 0.6515512065837902,
-      "grad_norm": 251.4295654296875,
-      "learning_rate": 1.6854195491099628e-05,
-      "loss": 42.9671,
-      "step": 161270
-    },
-    {
-      "epoch": 0.6515916078491578,
-      "grad_norm": 441.81585693359375,
-      "learning_rate": 1.68508954055417e-05,
-      "loss": 62.5547,
-      "step": 161280
-    },
-    {
-      "epoch": 0.6516320091145255,
-      "grad_norm": 395.6822509765625,
-      "learning_rate": 1.6847595478854773e-05,
-      "loss": 41.6582,
-      "step": 161290
-    },
-    {
-      "epoch": 0.6516724103798931,
-      "grad_norm": 749.1453247070312,
-      "learning_rate": 1.6844295711103167e-05,
-      "loss": 52.6976,
-      "step": 161300
-    },
-    {
-      "epoch": 0.6517128116452607,
-      "grad_norm": 481.84503173828125,
-      "learning_rate": 1.6840996102351212e-05,
-      "loss": 49.7407,
-      "step": 161310
-    },
-    {
-      "epoch": 0.6517532129106284,
-      "grad_norm": 830.3899536132812,
-      "learning_rate": 1.6837696652663242e-05,
-      "loss": 47.5324,
-      "step": 161320
-    },
-    {
-      "epoch": 0.651793614175996,
-      "grad_norm": 1153.26123046875,
-      "learning_rate": 1.6834397362103577e-05,
-      "loss": 60.7175,
-      "step": 161330
-    },
-    {
-      "epoch": 0.6518340154413637,
-      "grad_norm": 422.6778259277344,
-      "learning_rate": 1.6831098230736532e-05,
-      "loss": 32.2861,
-      "step": 161340
-    },
-    {
-      "epoch": 0.6518744167067313,
-      "grad_norm": 689.973388671875,
-      "learning_rate": 1.6827799258626442e-05,
-      "loss": 47.7611,
-      "step": 161350
-    },
-    {
-      "epoch": 0.6519148179720989,
-      "grad_norm": 537.6976318359375,
-      "learning_rate": 1.6824500445837606e-05,
-      "loss": 54.7357,
-      "step": 161360
-    },
-    {
-      "epoch": 0.6519552192374665,
-      "grad_norm": 1012.6497802734375,
-      "learning_rate": 1.6821201792434345e-05,
-      "loss": 80.4569,
-      "step": 161370
-    },
-    {
-      "epoch": 0.6519956205028341,
-      "grad_norm": 550.6050415039062,
-      "learning_rate": 1.681790329848097e-05,
-      "loss": 48.5143,
-      "step": 161380
-    },
-    {
-      "epoch": 0.6520360217682017,
-      "grad_norm": 618.5077514648438,
-      "learning_rate": 1.681460496404178e-05,
-      "loss": 41.0049,
-      "step": 161390
-    },
-    {
-      "epoch": 0.6520764230335694,
-      "grad_norm": 497.9303894042969,
-      "learning_rate": 1.681130678918108e-05,
-      "loss": 41.7235,
-      "step": 161400
-    },
-    {
-      "epoch": 0.652116824298937,
-      "grad_norm": 1291.5350341796875,
-      "learning_rate": 1.6808008773963173e-05,
-      "loss": 70.4364,
-      "step": 161410
-    },
-    {
-      "epoch": 0.6521572255643047,
-      "grad_norm": 955.0973510742188,
-      "learning_rate": 1.6804710918452342e-05,
-      "loss": 50.173,
-      "step": 161420
-    },
-    {
-      "epoch": 0.6521976268296723,
-      "grad_norm": 647.0445556640625,
-      "learning_rate": 1.6801413222712904e-05,
-      "loss": 71.8404,
-      "step": 161430
-    },
-    {
-      "epoch": 0.65223802809504,
-      "grad_norm": 282.2113952636719,
-      "learning_rate": 1.6798115686809125e-05,
-      "loss": 52.0266,
-      "step": 161440
-    },
-    {
-      "epoch": 0.6522784293604076,
-      "grad_norm": 189.90536499023438,
-      "learning_rate": 1.679481831080531e-05,
-      "loss": 68.1988,
-      "step": 161450
-    },
-    {
-      "epoch": 0.6523188306257752,
-      "grad_norm": 474.63739013671875,
-      "learning_rate": 1.679152109476574e-05,
-      "loss": 65.4844,
-      "step": 161460
-    },
-    {
-      "epoch": 0.6523592318911429,
-      "grad_norm": 912.4492797851562,
-      "learning_rate": 1.6788224038754687e-05,
-      "loss": 70.473,
-      "step": 161470
-    },
-    {
-      "epoch": 0.6523996331565105,
-      "grad_norm": 785.5763549804688,
-      "learning_rate": 1.6784927142836436e-05,
-      "loss": 76.3289,
-      "step": 161480
-    },
-    {
-      "epoch": 0.6524400344218781,
-      "grad_norm": 614.8267211914062,
-      "learning_rate": 1.6781630407075257e-05,
-      "loss": 49.6596,
-      "step": 161490
-    },
-    {
-      "epoch": 0.6524804356872457,
-      "grad_norm": 658.1514282226562,
-      "learning_rate": 1.677833383153542e-05,
-      "loss": 59.6776,
-      "step": 161500
-    },
-    {
-      "epoch": 0.6525208369526133,
-      "grad_norm": 416.6296081542969,
-      "learning_rate": 1.677503741628121e-05,
-      "loss": 48.725,
-      "step": 161510
-    },
-    {
-      "epoch": 0.652561238217981,
-      "grad_norm": 769.060791015625,
-      "learning_rate": 1.6771741161376862e-05,
-      "loss": 43.6589,
-      "step": 161520
-    },
-    {
-      "epoch": 0.6526016394833486,
-      "grad_norm": 1873.2025146484375,
-      "learning_rate": 1.676844506688667e-05,
-      "loss": 76.2816,
-      "step": 161530
-    },
-    {
-      "epoch": 0.6526420407487162,
-      "grad_norm": 318.47247314453125,
-      "learning_rate": 1.6765149132874882e-05,
-      "loss": 39.9869,
-      "step": 161540
-    },
-    {
-      "epoch": 0.6526824420140839,
-      "grad_norm": 203.8809814453125,
-      "learning_rate": 1.6761853359405737e-05,
-      "loss": 39.4577,
-      "step": 161550
-    },
-    {
-      "epoch": 0.6527228432794515,
-      "grad_norm": 635.0053100585938,
-      "learning_rate": 1.6758557746543518e-05,
-      "loss": 61.5434,
-      "step": 161560
-    },
-    {
-      "epoch": 0.6527632445448192,
-      "grad_norm": 848.6380004882812,
-      "learning_rate": 1.675526229435245e-05,
-      "loss": 60.4842,
-      "step": 161570
-    },
-    {
-      "epoch": 0.6528036458101868,
-      "grad_norm": 253.58030700683594,
-      "learning_rate": 1.675196700289679e-05,
-      "loss": 40.0244,
-      "step": 161580
-    },
-    {
-      "epoch": 0.6528440470755544,
-      "grad_norm": 3136.26025390625,
-      "learning_rate": 1.6748671872240785e-05,
-      "loss": 46.3562,
-      "step": 161590
-    },
-    {
-      "epoch": 0.6528844483409221,
-      "grad_norm": 560.5701904296875,
-      "learning_rate": 1.6745376902448656e-05,
-      "loss": 104.8585,
-      "step": 161600
-    },
-    {
-      "epoch": 0.6529248496062897,
-      "grad_norm": 1130.5245361328125,
-      "learning_rate": 1.6742082093584667e-05,
-      "loss": 52.8417,
-      "step": 161610
-    },
-    {
-      "epoch": 0.6529652508716574,
-      "grad_norm": 197.07540893554688,
-      "learning_rate": 1.6738787445713037e-05,
-      "loss": 54.9652,
-      "step": 161620
-    },
-    {
-      "epoch": 0.6530056521370249,
-      "grad_norm": 579.043701171875,
-      "learning_rate": 1.673549295889799e-05,
-      "loss": 37.0909,
-      "step": 161630
-    },
-    {
-      "epoch": 0.6530460534023925,
-      "grad_norm": 486.1392517089844,
-      "learning_rate": 1.6732198633203773e-05,
-      "loss": 55.2478,
-      "step": 161640
-    },
-    {
-      "epoch": 0.6530864546677602,
-      "grad_norm": 735.452392578125,
-      "learning_rate": 1.67289044686946e-05,
-      "loss": 43.7261,
-      "step": 161650
-    },
-    {
-      "epoch": 0.6531268559331278,
-      "grad_norm": 1341.6444091796875,
-      "learning_rate": 1.672561046543469e-05,
-      "loss": 53.2419,
-      "step": 161660
-    },
-    {
-      "epoch": 0.6531672571984954,
-      "grad_norm": 713.711181640625,
-      "learning_rate": 1.6722316623488262e-05,
-      "loss": 35.3926,
-      "step": 161670
-    },
-    {
-      "epoch": 0.6532076584638631,
-      "grad_norm": 1008.7979125976562,
-      "learning_rate": 1.6719022942919527e-05,
-      "loss": 34.2371,
-      "step": 161680
-    },
-    {
-      "epoch": 0.6532480597292307,
-      "grad_norm": 1110.8104248046875,
-      "learning_rate": 1.6715729423792714e-05,
-      "loss": 44.102,
-      "step": 161690
-    },
-    {
-      "epoch": 0.6532884609945984,
-      "grad_norm": 527.839599609375,
-      "learning_rate": 1.671243606617202e-05,
-      "loss": 39.0247,
-      "step": 161700
-    },
-    {
-      "epoch": 0.653328862259966,
-      "grad_norm": 533.70849609375,
-      "learning_rate": 1.6709142870121643e-05,
-      "loss": 53.8311,
-      "step": 161710
-    },
-    {
-      "epoch": 0.6533692635253336,
-      "grad_norm": 165.14747619628906,
-      "learning_rate": 1.6705849835705802e-05,
-      "loss": 41.9429,
-      "step": 161720
-    },
-    {
-      "epoch": 0.6534096647907013,
-      "grad_norm": 762.746337890625,
-      "learning_rate": 1.6702556962988693e-05,
-      "loss": 66.698,
-      "step": 161730
-    },
-    {
-      "epoch": 0.6534500660560689,
-      "grad_norm": 1566.11328125,
-      "learning_rate": 1.6699264252034497e-05,
-      "loss": 66.9996,
-      "step": 161740
-    },
-    {
-      "epoch": 0.6534904673214365,
-      "grad_norm": 1001.0677490234375,
-      "learning_rate": 1.6695971702907426e-05,
-      "loss": 88.9084,
-      "step": 161750
-    },
-    {
-      "epoch": 0.6535308685868041,
-      "grad_norm": 656.9507446289062,
-      "learning_rate": 1.669267931567165e-05,
-      "loss": 42.3577,
-      "step": 161760
-    },
-    {
-      "epoch": 0.6535712698521717,
-      "grad_norm": 1003.6537475585938,
-      "learning_rate": 1.668938709039138e-05,
-      "loss": 62.9774,
-      "step": 161770
-    },
-    {
-      "epoch": 0.6536116711175394,
-      "grad_norm": 879.7156372070312,
-      "learning_rate": 1.6686095027130783e-05,
-      "loss": 50.1465,
-      "step": 161780
-    },
-    {
-      "epoch": 0.653652072382907,
-      "grad_norm": 709.9041137695312,
-      "learning_rate": 1.6682803125954037e-05,
-      "loss": 45.2575,
-      "step": 161790
-    },
-    {
-      "epoch": 0.6536924736482747,
-      "grad_norm": 557.7595825195312,
-      "learning_rate": 1.6679511386925337e-05,
-      "loss": 69.0358,
-      "step": 161800
-    },
-    {
-      "epoch": 0.6537328749136423,
-      "grad_norm": 855.1320190429688,
-      "learning_rate": 1.667621981010884e-05,
-      "loss": 45.8865,
-      "step": 161810
-    },
-    {
-      "epoch": 0.6537732761790099,
-      "grad_norm": 1192.5213623046875,
-      "learning_rate": 1.6672928395568727e-05,
-      "loss": 58.5781,
-      "step": 161820
-    },
-    {
-      "epoch": 0.6538136774443776,
-      "grad_norm": 753.4164428710938,
-      "learning_rate": 1.666963714336916e-05,
-      "loss": 49.3136,
-      "step": 161830
-    },
-    {
-      "epoch": 0.6538540787097452,
-      "grad_norm": 836.4158935546875,
-      "learning_rate": 1.66663460535743e-05,
-      "loss": 38.0138,
-      "step": 161840
-    },
-    {
-      "epoch": 0.6538944799751129,
-      "grad_norm": 480.0657043457031,
-      "learning_rate": 1.6663055126248326e-05,
-      "loss": 43.5118,
-      "step": 161850
-    },
-    {
-      "epoch": 0.6539348812404805,
-      "grad_norm": 971.6677856445312,
-      "learning_rate": 1.6659764361455383e-05,
-      "loss": 46.9971,
-      "step": 161860
-    },
-    {
-      "epoch": 0.6539752825058481,
-      "grad_norm": 603.728271484375,
-      "learning_rate": 1.6656473759259614e-05,
-      "loss": 64.1542,
-      "step": 161870
-    },
-    {
-      "epoch": 0.6540156837712157,
-      "grad_norm": 2384.861328125,
-      "learning_rate": 1.6653183319725195e-05,
-      "loss": 45.0268,
-      "step": 161880
-    },
-    {
-      "epoch": 0.6540560850365833,
-      "grad_norm": 783.9608764648438,
-      "learning_rate": 1.6649893042916258e-05,
-      "loss": 52.3407,
-      "step": 161890
-    },
-    {
-      "epoch": 0.6540964863019509,
-      "grad_norm": 922.7523193359375,
-      "learning_rate": 1.6646602928896963e-05,
-      "loss": 43.5483,
-      "step": 161900
-    },
-    {
-      "epoch": 0.6541368875673186,
-      "grad_norm": 1093.2479248046875,
-      "learning_rate": 1.6643312977731445e-05,
-      "loss": 57.16,
-      "step": 161910
-    },
-    {
-      "epoch": 0.6541772888326862,
-      "grad_norm": 265.1941833496094,
-      "learning_rate": 1.6640023189483835e-05,
-      "loss": 43.1179,
-      "step": 161920
-    },
-    {
-      "epoch": 0.6542176900980539,
-      "grad_norm": 732.4922485351562,
-      "learning_rate": 1.663673356421829e-05,
-      "loss": 38.3712,
-      "step": 161930
-    },
-    {
-      "epoch": 0.6542580913634215,
-      "grad_norm": 2143.16162109375,
-      "learning_rate": 1.6633444101998917e-05,
-      "loss": 56.0343,
-      "step": 161940
-    },
-    {
-      "epoch": 0.6542984926287891,
-      "grad_norm": 746.3717041015625,
-      "learning_rate": 1.663015480288986e-05,
-      "loss": 48.6199,
-      "step": 161950
-    },
-    {
-      "epoch": 0.6543388938941568,
-      "grad_norm": 1015.5712280273438,
-      "learning_rate": 1.662686566695525e-05,
-      "loss": 47.5555,
-      "step": 161960
-    },
-    {
-      "epoch": 0.6543792951595244,
-      "grad_norm": 997.4241333007812,
-      "learning_rate": 1.6623576694259194e-05,
-      "loss": 64.299,
-      "step": 161970
-    },
-    {
-      "epoch": 0.6544196964248921,
-      "grad_norm": 938.6635131835938,
-      "learning_rate": 1.662028788486583e-05,
-      "loss": 50.552,
-      "step": 161980
-    },
-    {
-      "epoch": 0.6544600976902597,
-      "grad_norm": 602.8508911132812,
-      "learning_rate": 1.6616999238839277e-05,
-      "loss": 67.2282,
-      "step": 161990
-    },
-    {
-      "epoch": 0.6545004989556273,
-      "grad_norm": 561.122314453125,
-      "learning_rate": 1.6613710756243626e-05,
-      "loss": 51.4449,
-      "step": 162000
-    },
-    {
-      "epoch": 0.6545409002209949,
-      "grad_norm": 705.6483154296875,
-      "learning_rate": 1.6610422437143007e-05,
-      "loss": 63.1683,
-      "step": 162010
-    },
-    {
-      "epoch": 0.6545813014863625,
-      "grad_norm": 537.274658203125,
-      "learning_rate": 1.6607134281601526e-05,
-      "loss": 49.2148,
-      "step": 162020
-    },
-    {
-      "epoch": 0.6546217027517302,
-      "grad_norm": 213.13555908203125,
-      "learning_rate": 1.6603846289683273e-05,
-      "loss": 95.6635,
-      "step": 162030
-    },
-    {
-      "epoch": 0.6546621040170978,
-      "grad_norm": 749.0859985351562,
-      "learning_rate": 1.660055846145237e-05,
-      "loss": 51.4249,
-      "step": 162040
-    },
-    {
-      "epoch": 0.6547025052824654,
-      "grad_norm": 551.2670288085938,
-      "learning_rate": 1.659727079697289e-05,
-      "loss": 31.3299,
-      "step": 162050
-    },
-    {
-      "epoch": 0.6547429065478331,
-      "grad_norm": 787.1797485351562,
-      "learning_rate": 1.659398329630895e-05,
-      "loss": 44.1481,
-      "step": 162060
-    },
-    {
-      "epoch": 0.6547833078132007,
-      "grad_norm": 611.82958984375,
-      "learning_rate": 1.659069595952464e-05,
-      "loss": 50.7975,
-      "step": 162070
-    },
-    {
-      "epoch": 0.6548237090785684,
-      "grad_norm": 473.4167785644531,
-      "learning_rate": 1.6587408786684033e-05,
-      "loss": 47.7533,
-      "step": 162080
-    },
-    {
-      "epoch": 0.654864110343936,
-      "grad_norm": 938.8263549804688,
-      "learning_rate": 1.6584121777851226e-05,
-      "loss": 37.6045,
-      "step": 162090
-    },
-    {
-      "epoch": 0.6549045116093036,
-      "grad_norm": 722.9107666015625,
-      "learning_rate": 1.65808349330903e-05,
-      "loss": 39.03,
-      "step": 162100
-    },
-    {
-      "epoch": 0.6549449128746713,
-      "grad_norm": 877.0104370117188,
-      "learning_rate": 1.6577548252465325e-05,
-      "loss": 24.0046,
-      "step": 162110
-    },
-    {
-      "epoch": 0.6549853141400389,
-      "grad_norm": 733.263671875,
-      "learning_rate": 1.6574261736040387e-05,
-      "loss": 44.6276,
-      "step": 162120
-    },
-    {
-      "epoch": 0.6550257154054066,
-      "grad_norm": 327.0345764160156,
-      "learning_rate": 1.6570975383879546e-05,
-      "loss": 39.1917,
-      "step": 162130
-    },
-    {
-      "epoch": 0.6550661166707741,
-      "grad_norm": 468.39044189453125,
-      "learning_rate": 1.6567689196046894e-05,
-      "loss": 40.167,
-      "step": 162140
-    },
-    {
-      "epoch": 0.6551065179361417,
-      "grad_norm": 842.412109375,
-      "learning_rate": 1.6564403172606475e-05,
-      "loss": 57.5191,
-      "step": 162150
-    },
-    {
-      "epoch": 0.6551469192015094,
-      "grad_norm": 594.8579711914062,
-      "learning_rate": 1.656111731362236e-05,
-      "loss": 68.3253,
-      "step": 162160
-    },
-    {
-      "epoch": 0.655187320466877,
-      "grad_norm": 553.1251220703125,
-      "learning_rate": 1.6557831619158605e-05,
-      "loss": 41.9115,
-      "step": 162170
-    },
-    {
-      "epoch": 0.6552277217322446,
-      "grad_norm": 2678.964599609375,
-      "learning_rate": 1.6554546089279273e-05,
-      "loss": 68.2033,
-      "step": 162180
-    },
-    {
-      "epoch": 0.6552681229976123,
-      "grad_norm": 812.5203857421875,
-      "learning_rate": 1.6551260724048408e-05,
-      "loss": 39.7395,
-      "step": 162190
-    },
-    {
-      "epoch": 0.6553085242629799,
-      "grad_norm": 776.0007934570312,
-      "learning_rate": 1.6547975523530075e-05,
-      "loss": 46.6357,
-      "step": 162200
-    },
-    {
-      "epoch": 0.6553489255283476,
-      "grad_norm": 824.1729736328125,
-      "learning_rate": 1.6544690487788294e-05,
-      "loss": 62.6703,
-      "step": 162210
-    },
-    {
-      "epoch": 0.6553893267937152,
-      "grad_norm": 855.05029296875,
-      "learning_rate": 1.6541405616887137e-05,
-      "loss": 67.054,
-      "step": 162220
-    },
-    {
-      "epoch": 0.6554297280590828,
-      "grad_norm": 578.1856079101562,
-      "learning_rate": 1.6538120910890634e-05,
-      "loss": 57.6205,
-      "step": 162230
-    },
-    {
-      "epoch": 0.6554701293244505,
-      "grad_norm": 680.5081787109375,
-      "learning_rate": 1.6534836369862806e-05,
-      "loss": 83.3065,
-      "step": 162240
-    },
-    {
-      "epoch": 0.6555105305898181,
-      "grad_norm": 532.9240112304688,
-      "learning_rate": 1.6531551993867717e-05,
-      "loss": 28.7695,
-      "step": 162250
-    },
-    {
-      "epoch": 0.6555509318551858,
-      "grad_norm": 596.7019653320312,
-      "learning_rate": 1.652826778296938e-05,
-      "loss": 44.5479,
-      "step": 162260
-    },
-    {
-      "epoch": 0.6555913331205533,
-      "grad_norm": 502.9052429199219,
-      "learning_rate": 1.6524983737231818e-05,
-      "loss": 48.7583,
-      "step": 162270
-    },
-    {
-      "epoch": 0.6556317343859209,
-      "grad_norm": 494.0071716308594,
-      "learning_rate": 1.6521699856719062e-05,
-      "loss": 50.4662,
-      "step": 162280
-    },
-    {
-      "epoch": 0.6556721356512886,
-      "grad_norm": 1018.645751953125,
-      "learning_rate": 1.651841614149513e-05,
-      "loss": 76.721,
-      "step": 162290
-    },
-    {
-      "epoch": 0.6557125369166562,
-      "grad_norm": 755.8010864257812,
-      "learning_rate": 1.6515132591624048e-05,
-      "loss": 48.0349,
-      "step": 162300
-    },
-    {
-      "epoch": 0.6557529381820238,
-      "grad_norm": 1917.8994140625,
-      "learning_rate": 1.6511849207169826e-05,
-      "loss": 59.3606,
-      "step": 162310
-    },
-    {
-      "epoch": 0.6557933394473915,
-      "grad_norm": 778.1441650390625,
-      "learning_rate": 1.650856598819646e-05,
-      "loss": 40.8655,
-      "step": 162320
-    },
-    {
-      "epoch": 0.6558337407127591,
-      "grad_norm": 385.09332275390625,
-      "learning_rate": 1.6505282934767986e-05,
-      "loss": 69.3711,
-      "step": 162330
-    },
-    {
-      "epoch": 0.6558741419781268,
-      "grad_norm": 2902.443603515625,
-      "learning_rate": 1.650200004694839e-05,
-      "loss": 74.7698,
-      "step": 162340
-    },
-    {
-      "epoch": 0.6559145432434944,
-      "grad_norm": 908.2015991210938,
-      "learning_rate": 1.6498717324801683e-05,
-      "loss": 43.1564,
-      "step": 162350
-    },
-    {
-      "epoch": 0.655954944508862,
-      "grad_norm": 1079.4625244140625,
-      "learning_rate": 1.6495434768391855e-05,
-      "loss": 84.2936,
-      "step": 162360
-    },
-    {
-      "epoch": 0.6559953457742297,
-      "grad_norm": 959.7824096679688,
-      "learning_rate": 1.6492152377782898e-05,
-      "loss": 50.2236,
-      "step": 162370
-    },
-    {
-      "epoch": 0.6560357470395973,
-      "grad_norm": 530.7819213867188,
-      "learning_rate": 1.6488870153038815e-05,
-      "loss": 34.8689,
-      "step": 162380
-    },
-    {
-      "epoch": 0.6560761483049649,
-      "grad_norm": 622.3926391601562,
-      "learning_rate": 1.6485588094223597e-05,
-      "loss": 36.584,
-      "step": 162390
-    },
-    {
-      "epoch": 0.6561165495703325,
-      "grad_norm": 1052.944580078125,
-      "learning_rate": 1.648230620140121e-05,
-      "loss": 51.4176,
-      "step": 162400
-    },
-    {
-      "epoch": 0.6561569508357001,
-      "grad_norm": 1051.1805419921875,
-      "learning_rate": 1.6479024474635656e-05,
-      "loss": 52.1934,
-      "step": 162410
-    },
-    {
-      "epoch": 0.6561973521010678,
-      "grad_norm": 729.2684326171875,
-      "learning_rate": 1.6475742913990906e-05,
-      "loss": 39.7446,
-      "step": 162420
-    },
-    {
-      "epoch": 0.6562377533664354,
-      "grad_norm": 1571.4078369140625,
-      "learning_rate": 1.647246151953094e-05,
-      "loss": 68.3066,
-      "step": 162430
-    },
-    {
-      "epoch": 0.6562781546318031,
-      "grad_norm": 1100.865478515625,
-      "learning_rate": 1.6469180291319723e-05,
-      "loss": 52.9918,
-      "step": 162440
-    },
-    {
-      "epoch": 0.6563185558971707,
-      "grad_norm": 888.1741943359375,
-      "learning_rate": 1.6465899229421223e-05,
-      "loss": 84.3379,
-      "step": 162450
-    },
-    {
-      "epoch": 0.6563589571625383,
-      "grad_norm": 887.6325073242188,
-      "learning_rate": 1.6462618333899422e-05,
-      "loss": 66.6163,
-      "step": 162460
-    },
-    {
-      "epoch": 0.656399358427906,
-      "grad_norm": 631.6770629882812,
-      "learning_rate": 1.645933760481827e-05,
-      "loss": 55.3771,
-      "step": 162470
-    },
-    {
-      "epoch": 0.6564397596932736,
-      "grad_norm": 2138.152587890625,
-      "learning_rate": 1.645605704224172e-05,
-      "loss": 51.0784,
-      "step": 162480
-    },
-    {
-      "epoch": 0.6564801609586413,
-      "grad_norm": 785.101318359375,
-      "learning_rate": 1.6452776646233742e-05,
-      "loss": 40.2111,
-      "step": 162490
-    },
-    {
-      "epoch": 0.6565205622240089,
-      "grad_norm": 830.0779418945312,
-      "learning_rate": 1.6449496416858284e-05,
-      "loss": 56.9483,
-      "step": 162500
-    },
-    {
-      "epoch": 0.6565609634893765,
-      "grad_norm": 341.7655334472656,
-      "learning_rate": 1.644621635417929e-05,
-      "loss": 68.2469,
-      "step": 162510
-    },
-    {
-      "epoch": 0.6566013647547441,
-      "grad_norm": 357.12286376953125,
-      "learning_rate": 1.644293645826072e-05,
-      "loss": 47.8341,
-      "step": 162520
-    },
-    {
-      "epoch": 0.6566417660201117,
-      "grad_norm": 1889.9239501953125,
-      "learning_rate": 1.6439656729166508e-05,
-      "loss": 68.8827,
-      "step": 162530
-    },
-    {
-      "epoch": 0.6566821672854793,
-      "grad_norm": 347.9149475097656,
-      "learning_rate": 1.6436377166960597e-05,
-      "loss": 29.565,
-      "step": 162540
-    },
-    {
-      "epoch": 0.656722568550847,
-      "grad_norm": 763.558837890625,
-      "learning_rate": 1.643309777170692e-05,
-      "loss": 52.3415,
-      "step": 162550
-    },
-    {
-      "epoch": 0.6567629698162146,
-      "grad_norm": 651.407470703125,
-      "learning_rate": 1.6429818543469406e-05,
-      "loss": 53.0294,
-      "step": 162560
-    },
-    {
-      "epoch": 0.6568033710815823,
-      "grad_norm": 874.4526977539062,
-      "learning_rate": 1.6426539482311998e-05,
-      "loss": 56.1281,
-      "step": 162570
-    },
-    {
-      "epoch": 0.6568437723469499,
-      "grad_norm": 1284.8436279296875,
-      "learning_rate": 1.642326058829861e-05,
-      "loss": 61.655,
-      "step": 162580
-    },
-    {
-      "epoch": 0.6568841736123175,
-      "grad_norm": 1306.5478515625,
-      "learning_rate": 1.641998186149318e-05,
-      "loss": 43.2158,
-      "step": 162590
-    },
-    {
-      "epoch": 0.6569245748776852,
-      "grad_norm": 624.8187255859375,
-      "learning_rate": 1.6416703301959622e-05,
-      "loss": 47.5994,
-      "step": 162600
-    },
-    {
-      "epoch": 0.6569649761430528,
-      "grad_norm": 691.6862182617188,
-      "learning_rate": 1.6413424909761846e-05,
-      "loss": 53.6551,
-      "step": 162610
-    },
-    {
-      "epoch": 0.6570053774084205,
-      "grad_norm": 1279.94140625,
-      "learning_rate": 1.6410146684963777e-05,
-      "loss": 87.3749,
-      "step": 162620
-    },
-    {
-      "epoch": 0.6570457786737881,
-      "grad_norm": 1330.4853515625,
-      "learning_rate": 1.6406868627629323e-05,
-      "loss": 91.1492,
-      "step": 162630
-    },
-    {
-      "epoch": 0.6570861799391557,
-      "grad_norm": 769.7970581054688,
-      "learning_rate": 1.6403590737822376e-05,
-      "loss": 61.354,
-      "step": 162640
-    },
-    {
-      "epoch": 0.6571265812045233,
-      "grad_norm": 352.4034423828125,
-      "learning_rate": 1.6400313015606865e-05,
-      "loss": 51.5814,
-      "step": 162650
-    },
-    {
-      "epoch": 0.6571669824698909,
-      "grad_norm": 1086.769775390625,
-      "learning_rate": 1.6397035461046672e-05,
-      "loss": 39.5388,
-      "step": 162660
-    },
-    {
-      "epoch": 0.6572073837352586,
-      "grad_norm": 900.6954345703125,
-      "learning_rate": 1.6393758074205708e-05,
-      "loss": 39.8533,
-      "step": 162670
-    },
-    {
-      "epoch": 0.6572477850006262,
-      "grad_norm": 918.2764892578125,
-      "learning_rate": 1.6390480855147862e-05,
-      "loss": 54.463,
-      "step": 162680
-    },
-    {
-      "epoch": 0.6572881862659938,
-      "grad_norm": 147.70904541015625,
-      "learning_rate": 1.638720380393702e-05,
-      "loss": 66.0242,
-      "step": 162690
-    },
-    {
-      "epoch": 0.6573285875313615,
-      "grad_norm": 796.7276611328125,
-      "learning_rate": 1.6383926920637077e-05,
-      "loss": 50.0915,
-      "step": 162700
-    },
-    {
-      "epoch": 0.6573689887967291,
-      "grad_norm": 727.0776977539062,
-      "learning_rate": 1.6380650205311917e-05,
-      "loss": 82.3581,
-      "step": 162710
-    },
-    {
-      "epoch": 0.6574093900620968,
-      "grad_norm": 623.9905395507812,
-      "learning_rate": 1.637737365802541e-05,
-      "loss": 61.5921,
-      "step": 162720
-    },
-    {
-      "epoch": 0.6574497913274644,
-      "grad_norm": 728.1505126953125,
-      "learning_rate": 1.637409727884145e-05,
-      "loss": 44.927,
-      "step": 162730
-    },
-    {
-      "epoch": 0.657490192592832,
-      "grad_norm": 607.5389404296875,
-      "learning_rate": 1.6370821067823893e-05,
-      "loss": 53.0443,
-      "step": 162740
-    },
-    {
-      "epoch": 0.6575305938581997,
-      "grad_norm": 762.04541015625,
-      "learning_rate": 1.6367545025036636e-05,
-      "loss": 63.8751,
-      "step": 162750
-    },
-    {
-      "epoch": 0.6575709951235673,
-      "grad_norm": 425.23284912109375,
-      "learning_rate": 1.6364269150543532e-05,
-      "loss": 50.8446,
-      "step": 162760
-    },
-    {
-      "epoch": 0.657611396388935,
-      "grad_norm": 722.2022094726562,
-      "learning_rate": 1.636099344440844e-05,
-      "loss": 56.3553,
-      "step": 162770
-    },
-    {
-      "epoch": 0.6576517976543025,
-      "grad_norm": 1616.810791015625,
-      "learning_rate": 1.635771790669523e-05,
-      "loss": 45.0226,
-      "step": 162780
-    },
-    {
-      "epoch": 0.6576921989196701,
-      "grad_norm": 447.7799377441406,
-      "learning_rate": 1.6354442537467757e-05,
-      "loss": 67.5927,
-      "step": 162790
-    },
-    {
-      "epoch": 0.6577326001850378,
-      "grad_norm": 717.3369140625,
-      "learning_rate": 1.635116733678988e-05,
-      "loss": 37.616,
-      "step": 162800
-    },
-    {
-      "epoch": 0.6577730014504054,
-      "grad_norm": 862.1865844726562,
-      "learning_rate": 1.6347892304725454e-05,
-      "loss": 62.0235,
-      "step": 162810
-    },
-    {
-      "epoch": 0.657813402715773,
-      "grad_norm": 460.8524169921875,
-      "learning_rate": 1.634461744133831e-05,
-      "loss": 58.7466,
-      "step": 162820
-    },
-    {
-      "epoch": 0.6578538039811407,
-      "grad_norm": 198.9490509033203,
-      "learning_rate": 1.6341342746692317e-05,
-      "loss": 53.2849,
-      "step": 162830
-    },
-    {
-      "epoch": 0.6578942052465083,
-      "grad_norm": 498.9598693847656,
-      "learning_rate": 1.6338068220851306e-05,
-      "loss": 51.7222,
-      "step": 162840
-    },
-    {
-      "epoch": 0.657934606511876,
-      "grad_norm": 794.1101684570312,
-      "learning_rate": 1.63347938638791e-05,
-      "loss": 49.5184,
-      "step": 162850
-    },
-    {
-      "epoch": 0.6579750077772436,
-      "grad_norm": 499.0705261230469,
-      "learning_rate": 1.633151967583956e-05,
-      "loss": 38.8184,
-      "step": 162860
-    },
-    {
-      "epoch": 0.6580154090426112,
-      "grad_norm": 963.583984375,
-      "learning_rate": 1.6328245656796503e-05,
-      "loss": 60.0164,
-      "step": 162870
-    },
-    {
-      "epoch": 0.6580558103079789,
-      "grad_norm": 996.2311401367188,
-      "learning_rate": 1.6324971806813767e-05,
-      "loss": 44.6584,
-      "step": 162880
-    },
-    {
-      "epoch": 0.6580962115733465,
-      "grad_norm": 632.038330078125,
-      "learning_rate": 1.632169812595517e-05,
-      "loss": 36.6395,
-      "step": 162890
-    },
-    {
-      "epoch": 0.6581366128387142,
-      "grad_norm": 457.18304443359375,
-      "learning_rate": 1.6318424614284524e-05,
-      "loss": 55.2279,
-      "step": 162900
-    },
-    {
-      "epoch": 0.6581770141040817,
-      "grad_norm": 841.80029296875,
-      "learning_rate": 1.6315151271865672e-05,
-      "loss": 61.0696,
-      "step": 162910
-    },
-    {
-      "epoch": 0.6582174153694493,
-      "grad_norm": 1407.9495849609375,
-      "learning_rate": 1.6311878098762417e-05,
-      "loss": 35.4202,
-      "step": 162920
-    },
-    {
-      "epoch": 0.658257816634817,
-      "grad_norm": 456.8638610839844,
-      "learning_rate": 1.6308605095038566e-05,
-      "loss": 56.0512,
-      "step": 162930
-    },
-    {
-      "epoch": 0.6582982179001846,
-      "grad_norm": 626.9701538085938,
-      "learning_rate": 1.6305332260757936e-05,
-      "loss": 43.3455,
-      "step": 162940
-    },
-    {
-      "epoch": 0.6583386191655523,
-      "grad_norm": 555.2235107421875,
-      "learning_rate": 1.6302059595984327e-05,
-      "loss": 42.8999,
-      "step": 162950
-    },
-    {
-      "epoch": 0.6583790204309199,
-      "grad_norm": 300.74713134765625,
-      "learning_rate": 1.629878710078155e-05,
-      "loss": 55.2905,
-      "step": 162960
-    },
-    {
-      "epoch": 0.6584194216962875,
-      "grad_norm": 1130.7760009765625,
-      "learning_rate": 1.6295514775213398e-05,
-      "loss": 41.7524,
-      "step": 162970
-    },
-    {
-      "epoch": 0.6584598229616552,
-      "grad_norm": 816.3352661132812,
-      "learning_rate": 1.629224261934366e-05,
-      "loss": 61.8027,
-      "step": 162980
-    },
-    {
-      "epoch": 0.6585002242270228,
-      "grad_norm": 655.1786499023438,
-      "learning_rate": 1.628897063323614e-05,
-      "loss": 24.371,
-      "step": 162990
-    },
-    {
-      "epoch": 0.6585406254923905,
-      "grad_norm": 2886.60498046875,
-      "learning_rate": 1.6285698816954624e-05,
-      "loss": 69.9516,
-      "step": 163000
-    },
-    {
-      "epoch": 0.6585810267577581,
-      "grad_norm": 1917.0316162109375,
-      "learning_rate": 1.628242717056289e-05,
-      "loss": 67.4653,
-      "step": 163010
-    },
-    {
-      "epoch": 0.6586214280231257,
-      "grad_norm": 971.503662109375,
-      "learning_rate": 1.6279155694124732e-05,
-      "loss": 41.1,
-      "step": 163020
-    },
-    {
-      "epoch": 0.6586618292884933,
-      "grad_norm": 797.81005859375,
-      "learning_rate": 1.6275884387703918e-05,
-      "loss": 52.8087,
-      "step": 163030
-    },
-    {
-      "epoch": 0.6587022305538609,
-      "grad_norm": 1382.03955078125,
-      "learning_rate": 1.6272613251364237e-05,
-      "loss": 49.5148,
-      "step": 163040
-    },
-    {
-      "epoch": 0.6587426318192285,
-      "grad_norm": 612.2030639648438,
-      "learning_rate": 1.6269342285169453e-05,
-      "loss": 30.3748,
-      "step": 163050
-    },
-    {
-      "epoch": 0.6587830330845962,
-      "grad_norm": 528.9710693359375,
-      "learning_rate": 1.6266071489183327e-05,
-      "loss": 63.2863,
-      "step": 163060
-    },
-    {
-      "epoch": 0.6588234343499638,
-      "grad_norm": 707.3026123046875,
-      "learning_rate": 1.626280086346964e-05,
-      "loss": 64.9225,
-      "step": 163070
-    },
-    {
-      "epoch": 0.6588638356153315,
-      "grad_norm": 1027.89892578125,
-      "learning_rate": 1.6259530408092154e-05,
-      "loss": 57.8081,
-      "step": 163080
-    },
-    {
-      "epoch": 0.6589042368806991,
-      "grad_norm": 432.06866455078125,
-      "learning_rate": 1.625626012311461e-05,
-      "loss": 47.481,
-      "step": 163090
-    },
-    {
-      "epoch": 0.6589446381460667,
-      "grad_norm": 501.8659362792969,
-      "learning_rate": 1.6252990008600784e-05,
-      "loss": 54.4411,
-      "step": 163100
-    },
-    {
-      "epoch": 0.6589850394114344,
-      "grad_norm": 1181.19677734375,
-      "learning_rate": 1.6249720064614417e-05,
-      "loss": 70.2707,
-      "step": 163110
-    },
-    {
-      "epoch": 0.659025440676802,
-      "grad_norm": 874.8756103515625,
-      "learning_rate": 1.6246450291219266e-05,
-      "loss": 75.9436,
-      "step": 163120
-    },
-    {
-      "epoch": 0.6590658419421697,
-      "grad_norm": 452.441162109375,
-      "learning_rate": 1.624318068847907e-05,
-      "loss": 31.4531,
-      "step": 163130
-    },
-    {
-      "epoch": 0.6591062432075373,
-      "grad_norm": 625.8572998046875,
-      "learning_rate": 1.623991125645758e-05,
-      "loss": 34.9913,
-      "step": 163140
-    },
-    {
-      "epoch": 0.6591466444729049,
-      "grad_norm": 631.9246215820312,
-      "learning_rate": 1.623664199521853e-05,
-      "loss": 41.2059,
-      "step": 163150
-    },
-    {
-      "epoch": 0.6591870457382725,
-      "grad_norm": 444.54461669921875,
-      "learning_rate": 1.6233372904825656e-05,
-      "loss": 48.9469,
-      "step": 163160
-    },
-    {
-      "epoch": 0.6592274470036401,
-      "grad_norm": 860.9658813476562,
-      "learning_rate": 1.623010398534268e-05,
-      "loss": 38.0498,
-      "step": 163170
-    },
-    {
-      "epoch": 0.6592678482690078,
-      "grad_norm": 545.662109375,
-      "learning_rate": 1.6226835236833354e-05,
-      "loss": 44.3886,
-      "step": 163180
-    },
-    {
-      "epoch": 0.6593082495343754,
-      "grad_norm": 326.8611145019531,
-      "learning_rate": 1.622356665936138e-05,
-      "loss": 69.7939,
-      "step": 163190
-    },
-    {
-      "epoch": 0.659348650799743,
-      "grad_norm": 728.9844360351562,
-      "learning_rate": 1.6220298252990502e-05,
-      "loss": 62.5377,
-      "step": 163200
-    },
-    {
-      "epoch": 0.6593890520651107,
-      "grad_norm": 428.49652099609375,
-      "learning_rate": 1.621703001778443e-05,
-      "loss": 37.8547,
-      "step": 163210
-    },
-    {
-      "epoch": 0.6594294533304783,
-      "grad_norm": 528.7867431640625,
-      "learning_rate": 1.621376195380688e-05,
-      "loss": 72.5068,
-      "step": 163220
-    },
-    {
-      "epoch": 0.659469854595846,
-      "grad_norm": 865.8905639648438,
-      "learning_rate": 1.6210494061121562e-05,
-      "loss": 52.1617,
-      "step": 163230
-    },
-    {
-      "epoch": 0.6595102558612136,
-      "grad_norm": 629.0953979492188,
-      "learning_rate": 1.620722633979219e-05,
-      "loss": 59.1401,
-      "step": 163240
-    },
-    {
-      "epoch": 0.6595506571265812,
-      "grad_norm": 538.1380615234375,
-      "learning_rate": 1.6203958789882456e-05,
-      "loss": 49.7746,
-      "step": 163250
-    },
-    {
-      "epoch": 0.6595910583919489,
-      "grad_norm": 967.8890380859375,
-      "learning_rate": 1.6200691411456084e-05,
-      "loss": 55.0573,
-      "step": 163260
-    },
-    {
-      "epoch": 0.6596314596573165,
-      "grad_norm": 1064.715576171875,
-      "learning_rate": 1.6197424204576757e-05,
-      "loss": 52.5899,
-      "step": 163270
-    },
-    {
-      "epoch": 0.6596718609226842,
-      "grad_norm": 542.53466796875,
-      "learning_rate": 1.6194157169308182e-05,
-      "loss": 55.8167,
-      "step": 163280
-    },
-    {
-      "epoch": 0.6597122621880517,
-      "grad_norm": 432.77374267578125,
-      "learning_rate": 1.619089030571405e-05,
-      "loss": 37.0786,
-      "step": 163290
-    },
-    {
-      "epoch": 0.6597526634534193,
-      "grad_norm": 581.6807250976562,
-      "learning_rate": 1.6187623613858038e-05,
-      "loss": 44.2656,
-      "step": 163300
-    },
-    {
-      "epoch": 0.659793064718787,
-      "grad_norm": 0.0,
-      "learning_rate": 1.6184357093803847e-05,
-      "loss": 45.494,
-      "step": 163310
-    },
-    {
-      "epoch": 0.6598334659841546,
-      "grad_norm": 466.3706359863281,
-      "learning_rate": 1.6181090745615147e-05,
-      "loss": 32.6534,
-      "step": 163320
-    },
-    {
-      "epoch": 0.6598738672495222,
-      "grad_norm": 1236.4456787109375,
-      "learning_rate": 1.617782456935563e-05,
-      "loss": 50.6887,
-      "step": 163330
-    },
-    {
-      "epoch": 0.6599142685148899,
-      "grad_norm": 404.6134948730469,
-      "learning_rate": 1.6174558565088965e-05,
-      "loss": 58.2112,
-      "step": 163340
-    },
-    {
-      "epoch": 0.6599546697802575,
-      "grad_norm": 314.0501708984375,
-      "learning_rate": 1.6171292732878812e-05,
-      "loss": 18.9486,
-      "step": 163350
-    },
-    {
-      "epoch": 0.6599950710456252,
-      "grad_norm": 915.3804931640625,
-      "learning_rate": 1.6168027072788867e-05,
-      "loss": 55.3569,
-      "step": 163360
-    },
-    {
-      "epoch": 0.6600354723109928,
-      "grad_norm": 1853.2662353515625,
-      "learning_rate": 1.616476158488278e-05,
-      "loss": 68.6005,
-      "step": 163370
-    },
-    {
-      "epoch": 0.6600758735763604,
-      "grad_norm": 1049.0845947265625,
-      "learning_rate": 1.6161496269224208e-05,
-      "loss": 55.2401,
-      "step": 163380
-    },
-    {
-      "epoch": 0.6601162748417281,
-      "grad_norm": 969.3112182617188,
-      "learning_rate": 1.6158231125876823e-05,
-      "loss": 47.6416,
-      "step": 163390
-    },
-    {
-      "epoch": 0.6601566761070957,
-      "grad_norm": 595.2536010742188,
-      "learning_rate": 1.6154966154904265e-05,
-      "loss": 52.834,
-      "step": 163400
-    },
-    {
-      "epoch": 0.6601970773724634,
-      "grad_norm": 7205.93701171875,
-      "learning_rate": 1.6151701356370203e-05,
-      "loss": 108.3844,
-      "step": 163410
-    },
-    {
-      "epoch": 0.6602374786378309,
-      "grad_norm": 1252.0986328125,
-      "learning_rate": 1.614843673033828e-05,
-      "loss": 53.7415,
-      "step": 163420
-    },
-    {
-      "epoch": 0.6602778799031985,
-      "grad_norm": 825.3316650390625,
-      "learning_rate": 1.614517227687213e-05,
-      "loss": 38.9342,
-      "step": 163430
-    },
-    {
-      "epoch": 0.6603182811685662,
-      "grad_norm": 818.31689453125,
-      "learning_rate": 1.6141907996035415e-05,
-      "loss": 49.7294,
-      "step": 163440
-    },
-    {
-      "epoch": 0.6603586824339338,
-      "grad_norm": 455.1985168457031,
-      "learning_rate": 1.6138643887891763e-05,
-      "loss": 52.1195,
-      "step": 163450
-    },
-    {
-      "epoch": 0.6603990836993014,
-      "grad_norm": 499.15533447265625,
-      "learning_rate": 1.613537995250481e-05,
-      "loss": 62.1534,
-      "step": 163460
-    },
-    {
-      "epoch": 0.6604394849646691,
-      "grad_norm": 266.7299499511719,
-      "learning_rate": 1.6132116189938185e-05,
-      "loss": 43.204,
-      "step": 163470
-    },
-    {
-      "epoch": 0.6604798862300367,
-      "grad_norm": 527.0379028320312,
-      "learning_rate": 1.612885260025552e-05,
-      "loss": 49.2214,
-      "step": 163480
-    },
-    {
-      "epoch": 0.6605202874954044,
-      "grad_norm": 542.5224609375,
-      "learning_rate": 1.6125589183520445e-05,
-      "loss": 40.1502,
-      "step": 163490
-    },
-    {
-      "epoch": 0.660560688760772,
-      "grad_norm": 489.45208740234375,
-      "learning_rate": 1.612232593979658e-05,
-      "loss": 54.8088,
-      "step": 163500
-    },
-    {
-      "epoch": 0.6606010900261396,
-      "grad_norm": 346.20880126953125,
-      "learning_rate": 1.611906286914753e-05,
-      "loss": 25.9498,
-      "step": 163510
-    },
-    {
-      "epoch": 0.6606414912915073,
-      "grad_norm": 526.3746337890625,
-      "learning_rate": 1.611579997163693e-05,
-      "loss": 41.5305,
-      "step": 163520
-    },
-    {
-      "epoch": 0.6606818925568749,
-      "grad_norm": 589.9901123046875,
-      "learning_rate": 1.611253724732839e-05,
-      "loss": 66.6984,
-      "step": 163530
-    },
-    {
-      "epoch": 0.6607222938222426,
-      "grad_norm": 324.912841796875,
-      "learning_rate": 1.6109274696285495e-05,
-      "loss": 54.046,
-      "step": 163540
-    },
-    {
-      "epoch": 0.6607626950876101,
-      "grad_norm": 948.199462890625,
-      "learning_rate": 1.6106012318571877e-05,
-      "loss": 43.5441,
-      "step": 163550
-    },
-    {
-      "epoch": 0.6608030963529777,
-      "grad_norm": 554.431884765625,
-      "learning_rate": 1.610275011425113e-05,
-      "loss": 42.9645,
-      "step": 163560
-    },
-    {
-      "epoch": 0.6608434976183454,
-      "grad_norm": 974.4641723632812,
-      "learning_rate": 1.6099488083386847e-05,
-      "loss": 59.781,
-      "step": 163570
-    },
-    {
-      "epoch": 0.660883898883713,
-      "grad_norm": 727.3328247070312,
-      "learning_rate": 1.6096226226042632e-05,
-      "loss": 56.2628,
-      "step": 163580
-    },
-    {
-      "epoch": 0.6609243001490807,
-      "grad_norm": 553.086669921875,
-      "learning_rate": 1.6092964542282058e-05,
-      "loss": 46.6205,
-      "step": 163590
-    },
-    {
-      "epoch": 0.6609647014144483,
-      "grad_norm": 425.8204345703125,
-      "learning_rate": 1.6089703032168733e-05,
-      "loss": 53.4261,
-      "step": 163600
-    },
-    {
-      "epoch": 0.6610051026798159,
-      "grad_norm": 550.3936157226562,
-      "learning_rate": 1.608644169576624e-05,
-      "loss": 39.1422,
-      "step": 163610
-    },
-    {
-      "epoch": 0.6610455039451836,
-      "grad_norm": 986.0363159179688,
-      "learning_rate": 1.6083180533138143e-05,
-      "loss": 65.5904,
-      "step": 163620
-    },
-    {
-      "epoch": 0.6610859052105512,
-      "grad_norm": 0.0,
-      "learning_rate": 1.6079919544348045e-05,
-      "loss": 39.4999,
-      "step": 163630
-    },
-    {
-      "epoch": 0.6611263064759189,
-      "grad_norm": 246.7885284423828,
-      "learning_rate": 1.60766587294595e-05,
-      "loss": 62.9314,
-      "step": 163640
-    },
-    {
-      "epoch": 0.6611667077412865,
-      "grad_norm": 728.0462036132812,
-      "learning_rate": 1.607339808853609e-05,
-      "loss": 55.5237,
-      "step": 163650
-    },
-    {
-      "epoch": 0.6612071090066541,
-      "grad_norm": 824.4031982421875,
-      "learning_rate": 1.607013762164138e-05,
-      "loss": 44.8297,
-      "step": 163660
-    },
-    {
-      "epoch": 0.6612475102720217,
-      "grad_norm": 863.5546875,
-      "learning_rate": 1.606687732883893e-05,
-      "loss": 37.9698,
-      "step": 163670
-    },
-    {
-      "epoch": 0.6612879115373893,
-      "grad_norm": 554.3681640625,
-      "learning_rate": 1.606361721019231e-05,
-      "loss": 55.0306,
-      "step": 163680
-    },
-    {
-      "epoch": 0.661328312802757,
-      "grad_norm": 908.3670043945312,
-      "learning_rate": 1.6060357265765073e-05,
-      "loss": 46.0023,
-      "step": 163690
-    },
-    {
-      "epoch": 0.6613687140681246,
-      "grad_norm": 517.4747924804688,
-      "learning_rate": 1.6057097495620767e-05,
-      "loss": 33.2749,
-      "step": 163700
-    },
-    {
-      "epoch": 0.6614091153334922,
-      "grad_norm": 377.3349914550781,
-      "learning_rate": 1.605383789982296e-05,
-      "loss": 49.5776,
-      "step": 163710
-    },
-    {
-      "epoch": 0.6614495165988599,
-      "grad_norm": 910.6222534179688,
-      "learning_rate": 1.605057847843518e-05,
-      "loss": 60.544,
-      "step": 163720
-    },
-    {
-      "epoch": 0.6614899178642275,
-      "grad_norm": 997.0488891601562,
-      "learning_rate": 1.604731923152099e-05,
-      "loss": 41.9783,
-      "step": 163730
-    },
-    {
-      "epoch": 0.6615303191295951,
-      "grad_norm": 610.9208374023438,
-      "learning_rate": 1.604406015914392e-05,
-      "loss": 37.3855,
-      "step": 163740
-    },
-    {
-      "epoch": 0.6615707203949628,
-      "grad_norm": 587.4974365234375,
-      "learning_rate": 1.6040801261367493e-05,
-      "loss": 44.2066,
-      "step": 163750
-    },
-    {
-      "epoch": 0.6616111216603304,
-      "grad_norm": 286.9742431640625,
-      "learning_rate": 1.6037542538255274e-05,
-      "loss": 41.9279,
-      "step": 163760
-    },
-    {
-      "epoch": 0.6616515229256981,
-      "grad_norm": 296.52197265625,
-      "learning_rate": 1.6034283989870773e-05,
-      "loss": 36.0075,
-      "step": 163770
-    },
-    {
-      "epoch": 0.6616919241910657,
-      "grad_norm": 228.6723175048828,
-      "learning_rate": 1.603102561627751e-05,
-      "loss": 43.1621,
-      "step": 163780
-    },
-    {
-      "epoch": 0.6617323254564333,
-      "grad_norm": 981.3435668945312,
-      "learning_rate": 1.6027767417539036e-05,
-      "loss": 55.8184,
-      "step": 163790
-    },
-    {
-      "epoch": 0.6617727267218009,
-      "grad_norm": 515.3609008789062,
-      "learning_rate": 1.6024509393718844e-05,
-      "loss": 50.0376,
-      "step": 163800
-    },
-    {
-      "epoch": 0.6618131279871685,
-      "grad_norm": 612.1990966796875,
-      "learning_rate": 1.6021251544880467e-05,
-      "loss": 88.1613,
-      "step": 163810
-    },
-    {
-      "epoch": 0.6618535292525362,
-      "grad_norm": 493.67962646484375,
-      "learning_rate": 1.6017993871087418e-05,
-      "loss": 30.4908,
-      "step": 163820
-    },
-    {
-      "epoch": 0.6618939305179038,
-      "grad_norm": 489.2311706542969,
-      "learning_rate": 1.6014736372403198e-05,
-      "loss": 55.9566,
-      "step": 163830
-    },
-    {
-      "epoch": 0.6619343317832714,
-      "grad_norm": 676.2103271484375,
-      "learning_rate": 1.6011479048891324e-05,
-      "loss": 73.5741,
-      "step": 163840
-    },
-    {
-      "epoch": 0.6619747330486391,
-      "grad_norm": 909.5650634765625,
-      "learning_rate": 1.600822190061528e-05,
-      "loss": 58.0165,
-      "step": 163850
-    },
-    {
-      "epoch": 0.6620151343140067,
-      "grad_norm": 576.3464965820312,
-      "learning_rate": 1.6004964927638593e-05,
-      "loss": 97.8101,
-      "step": 163860
-    },
-    {
-      "epoch": 0.6620555355793744,
-      "grad_norm": 921.4266357421875,
-      "learning_rate": 1.6001708130024746e-05,
-      "loss": 58.1987,
-      "step": 163870
-    },
-    {
-      "epoch": 0.662095936844742,
-      "grad_norm": 220.7132568359375,
-      "learning_rate": 1.5998451507837216e-05,
-      "loss": 44.0783,
-      "step": 163880
-    },
-    {
-      "epoch": 0.6621363381101096,
-      "grad_norm": 395.4250793457031,
-      "learning_rate": 1.5995195061139524e-05,
-      "loss": 37.5821,
-      "step": 163890
-    },
-    {
-      "epoch": 0.6621767393754773,
-      "grad_norm": 566.3154907226562,
-      "learning_rate": 1.5991938789995137e-05,
-      "loss": 42.0082,
-      "step": 163900
-    },
-    {
-      "epoch": 0.6622171406408449,
-      "grad_norm": 600.4891357421875,
-      "learning_rate": 1.5988682694467537e-05,
-      "loss": 31.8918,
-      "step": 163910
-    },
-    {
-      "epoch": 0.6622575419062126,
-      "grad_norm": 985.5350341796875,
-      "learning_rate": 1.598542677462021e-05,
-      "loss": 36.2068,
-      "step": 163920
-    },
-    {
-      "epoch": 0.6622979431715801,
-      "grad_norm": 311.7273864746094,
-      "learning_rate": 1.5982171030516623e-05,
-      "loss": 64.975,
-      "step": 163930
-    },
-    {
-      "epoch": 0.6623383444369477,
-      "grad_norm": 513.2449340820312,
-      "learning_rate": 1.5978915462220263e-05,
-      "loss": 44.5756,
-      "step": 163940
-    },
-    {
-      "epoch": 0.6623787457023154,
-      "grad_norm": 645.991943359375,
-      "learning_rate": 1.597566006979459e-05,
-      "loss": 55.4562,
-      "step": 163950
-    },
-    {
-      "epoch": 0.662419146967683,
-      "grad_norm": 2931.17724609375,
-      "learning_rate": 1.5972404853303062e-05,
-      "loss": 60.1816,
-      "step": 163960
-    },
-    {
-      "epoch": 0.6624595482330506,
-      "grad_norm": 712.2529907226562,
-      "learning_rate": 1.5969149812809157e-05,
-      "loss": 38.7537,
-      "step": 163970
-    },
-    {
-      "epoch": 0.6624999494984183,
-      "grad_norm": 560.7094116210938,
-      "learning_rate": 1.5965894948376326e-05,
-      "loss": 32.3941,
-      "step": 163980
-    },
-    {
-      "epoch": 0.6625403507637859,
-      "grad_norm": 1348.53857421875,
-      "learning_rate": 1.5962640260068017e-05,
-      "loss": 50.6592,
-      "step": 163990
-    },
-    {
-      "epoch": 0.6625807520291536,
-      "grad_norm": 534.3580932617188,
-      "learning_rate": 1.5959385747947698e-05,
-      "loss": 67.778,
-      "step": 164000
-    },
-    {
-      "epoch": 0.6626211532945212,
-      "grad_norm": 838.6098022460938,
-      "learning_rate": 1.5956131412078794e-05,
-      "loss": 40.7363,
-      "step": 164010
-    },
-    {
-      "epoch": 0.6626615545598888,
-      "grad_norm": 510.424072265625,
-      "learning_rate": 1.595287725252478e-05,
-      "loss": 34.0819,
-      "step": 164020
-    },
-    {
-      "epoch": 0.6627019558252565,
-      "grad_norm": 616.5304565429688,
-      "learning_rate": 1.5949623269349078e-05,
-      "loss": 49.7259,
-      "step": 164030
-    },
-    {
-      "epoch": 0.6627423570906241,
-      "grad_norm": 1309.3828125,
-      "learning_rate": 1.5946369462615117e-05,
-      "loss": 38.566,
-      "step": 164040
-    },
-    {
-      "epoch": 0.6627827583559918,
-      "grad_norm": 338.8128662109375,
-      "learning_rate": 1.594311583238636e-05,
-      "loss": 48.4212,
-      "step": 164050
-    },
-    {
-      "epoch": 0.6628231596213593,
-      "grad_norm": 530.7924194335938,
-      "learning_rate": 1.593986237872622e-05,
-      "loss": 54.5951,
-      "step": 164060
-    },
-    {
-      "epoch": 0.6628635608867269,
-      "grad_norm": 545.570556640625,
-      "learning_rate": 1.593660910169812e-05,
-      "loss": 55.6956,
-      "step": 164070
-    },
-    {
-      "epoch": 0.6629039621520946,
-      "grad_norm": 484.4242248535156,
-      "learning_rate": 1.59333560013655e-05,
-      "loss": 52.3372,
-      "step": 164080
-    },
-    {
-      "epoch": 0.6629443634174622,
-      "grad_norm": 496.9941101074219,
-      "learning_rate": 1.5930103077791763e-05,
-      "loss": 61.0413,
-      "step": 164090
-    },
-    {
-      "epoch": 0.6629847646828299,
-      "grad_norm": 516.7117309570312,
-      "learning_rate": 1.5926850331040345e-05,
-      "loss": 53.589,
-      "step": 164100
-    },
-    {
-      "epoch": 0.6630251659481975,
-      "grad_norm": 843.0413818359375,
-      "learning_rate": 1.592359776117465e-05,
-      "loss": 65.3876,
-      "step": 164110
-    },
-    {
-      "epoch": 0.6630655672135651,
-      "grad_norm": 1167.8038330078125,
-      "learning_rate": 1.5920345368258084e-05,
-      "loss": 38.0708,
-      "step": 164120
-    },
-    {
-      "epoch": 0.6631059684789328,
-      "grad_norm": 349.5180358886719,
-      "learning_rate": 1.5917093152354062e-05,
-      "loss": 51.9246,
-      "step": 164130
-    },
-    {
-      "epoch": 0.6631463697443004,
-      "grad_norm": 799.0294799804688,
-      "learning_rate": 1.5913841113525992e-05,
-      "loss": 43.2969,
-      "step": 164140
-    },
-    {
-      "epoch": 0.663186771009668,
-      "grad_norm": 468.5010986328125,
-      "learning_rate": 1.5910589251837257e-05,
-      "loss": 41.2818,
-      "step": 164150
-    },
-    {
-      "epoch": 0.6632271722750357,
-      "grad_norm": 636.5210571289062,
-      "learning_rate": 1.5907337567351264e-05,
-      "loss": 62.2918,
-      "step": 164160
-    },
-    {
-      "epoch": 0.6632675735404033,
-      "grad_norm": 900.3253784179688,
-      "learning_rate": 1.5904086060131412e-05,
-      "loss": 62.7353,
-      "step": 164170
-    },
-    {
-      "epoch": 0.6633079748057709,
-      "grad_norm": 743.2527465820312,
-      "learning_rate": 1.5900834730241088e-05,
-      "loss": 68.0121,
-      "step": 164180
-    },
-    {
-      "epoch": 0.6633483760711385,
-      "grad_norm": 1014.6226196289062,
-      "learning_rate": 1.5897583577743676e-05,
-      "loss": 83.6025,
-      "step": 164190
-    },
-    {
-      "epoch": 0.6633887773365061,
-      "grad_norm": 551.89697265625,
-      "learning_rate": 1.5894332602702545e-05,
-      "loss": 43.9586,
-      "step": 164200
-    },
-    {
-      "epoch": 0.6634291786018738,
-      "grad_norm": 629.16162109375,
-      "learning_rate": 1.58910818051811e-05,
-      "loss": 44.4142,
-      "step": 164210
-    },
-    {
-      "epoch": 0.6634695798672414,
-      "grad_norm": 652.8489990234375,
-      "learning_rate": 1.5887831185242703e-05,
-      "loss": 77.5553,
-      "step": 164220
-    },
-    {
-      "epoch": 0.6635099811326091,
-      "grad_norm": 487.5631408691406,
-      "learning_rate": 1.5884580742950723e-05,
-      "loss": 41.1645,
-      "step": 164230
-    },
-    {
-      "epoch": 0.6635503823979767,
-      "grad_norm": 501.0959777832031,
-      "learning_rate": 1.588133047836854e-05,
-      "loss": 35.929,
-      "step": 164240
-    },
-    {
-      "epoch": 0.6635907836633443,
-      "grad_norm": 371.644287109375,
-      "learning_rate": 1.5878080391559508e-05,
-      "loss": 67.765,
-      "step": 164250
-    },
-    {
-      "epoch": 0.663631184928712,
-      "grad_norm": 693.4502563476562,
-      "learning_rate": 1.5874830482587e-05,
-      "loss": 56.9889,
-      "step": 164260
-    },
-    {
-      "epoch": 0.6636715861940796,
-      "grad_norm": 373.0396728515625,
-      "learning_rate": 1.5871580751514374e-05,
-      "loss": 55.6719,
-      "step": 164270
-    },
-    {
-      "epoch": 0.6637119874594473,
-      "grad_norm": 658.143798828125,
-      "learning_rate": 1.586833119840497e-05,
-      "loss": 55.5467,
-      "step": 164280
-    },
-    {
-      "epoch": 0.6637523887248149,
-      "grad_norm": 361.97808837890625,
-      "learning_rate": 1.586508182332216e-05,
-      "loss": 45.0967,
-      "step": 164290
-    },
-    {
-      "epoch": 0.6637927899901825,
-      "grad_norm": 669.2239990234375,
-      "learning_rate": 1.5861832626329282e-05,
-      "loss": 28.2219,
-      "step": 164300
-    },
-    {
-      "epoch": 0.6638331912555501,
-      "grad_norm": 783.1472778320312,
-      "learning_rate": 1.585858360748967e-05,
-      "loss": 64.7959,
-      "step": 164310
-    },
-    {
-      "epoch": 0.6638735925209177,
-      "grad_norm": 620.7758178710938,
-      "learning_rate": 1.585533476686669e-05,
-      "loss": 46.4694,
-      "step": 164320
-    },
-    {
-      "epoch": 0.6639139937862854,
-      "grad_norm": 761.25146484375,
-      "learning_rate": 1.585208610452366e-05,
-      "loss": 48.2226,
-      "step": 164330
-    },
-    {
-      "epoch": 0.663954395051653,
-      "grad_norm": 925.231689453125,
-      "learning_rate": 1.5848837620523927e-05,
-      "loss": 48.1655,
-      "step": 164340
-    },
-    {
-      "epoch": 0.6639947963170206,
-      "grad_norm": 602.0315551757812,
-      "learning_rate": 1.5845589314930813e-05,
-      "loss": 51.8964,
-      "step": 164350
-    },
-    {
-      "epoch": 0.6640351975823883,
-      "grad_norm": 693.58544921875,
-      "learning_rate": 1.584234118780764e-05,
-      "loss": 46.2405,
-      "step": 164360
-    },
-    {
-      "epoch": 0.6640755988477559,
-      "grad_norm": 705.6416015625,
-      "learning_rate": 1.5839093239217754e-05,
-      "loss": 55.715,
-      "step": 164370
-    },
-    {
-      "epoch": 0.6641160001131236,
-      "grad_norm": 921.3760375976562,
-      "learning_rate": 1.5835845469224447e-05,
-      "loss": 36.3578,
-      "step": 164380
-    },
-    {
-      "epoch": 0.6641564013784912,
-      "grad_norm": 596.5466918945312,
-      "learning_rate": 1.5832597877891066e-05,
-      "loss": 56.3782,
-      "step": 164390
-    },
-    {
-      "epoch": 0.6641968026438588,
-      "grad_norm": 265.7430725097656,
-      "learning_rate": 1.58293504652809e-05,
-      "loss": 37.4067,
-      "step": 164400
-    },
-    {
-      "epoch": 0.6642372039092265,
-      "grad_norm": 1141.7506103515625,
-      "learning_rate": 1.582610323145727e-05,
-      "loss": 50.2969,
-      "step": 164410
-    },
-    {
-      "epoch": 0.6642776051745941,
-      "grad_norm": 658.0392456054688,
-      "learning_rate": 1.5822856176483482e-05,
-      "loss": 58.6385,
-      "step": 164420
-    },
-    {
-      "epoch": 0.6643180064399618,
-      "grad_norm": 676.5816040039062,
-      "learning_rate": 1.5819609300422835e-05,
-      "loss": 46.8305,
-      "step": 164430
-    },
-    {
-      "epoch": 0.6643584077053293,
-      "grad_norm": 1031.92236328125,
-      "learning_rate": 1.581636260333863e-05,
-      "loss": 34.6241,
-      "step": 164440
-    },
-    {
-      "epoch": 0.6643988089706969,
-      "grad_norm": 756.4705200195312,
-      "learning_rate": 1.581311608529417e-05,
-      "loss": 51.8875,
-      "step": 164450
-    },
-    {
-      "epoch": 0.6644392102360646,
-      "grad_norm": 296.3645935058594,
-      "learning_rate": 1.5809869746352732e-05,
-      "loss": 52.9606,
-      "step": 164460
-    },
-    {
-      "epoch": 0.6644796115014322,
-      "grad_norm": 570.0635375976562,
-      "learning_rate": 1.5806623586577622e-05,
-      "loss": 36.714,
-      "step": 164470
-    },
-    {
-      "epoch": 0.6645200127667998,
-      "grad_norm": 973.0076293945312,
-      "learning_rate": 1.580337760603212e-05,
-      "loss": 43.9121,
-      "step": 164480
-    },
-    {
-      "epoch": 0.6645604140321675,
-      "grad_norm": 760.7753295898438,
-      "learning_rate": 1.5800131804779504e-05,
-      "loss": 24.1093,
-      "step": 164490
-    },
-    {
-      "epoch": 0.6646008152975351,
-      "grad_norm": 1284.3609619140625,
-      "learning_rate": 1.5796886182883053e-05,
-      "loss": 36.6988,
-      "step": 164500
-    },
-    {
-      "epoch": 0.6646412165629028,
-      "grad_norm": 517.6578369140625,
-      "learning_rate": 1.5793640740406045e-05,
-      "loss": 46.5038,
-      "step": 164510
-    },
-    {
-      "epoch": 0.6646816178282704,
-      "grad_norm": 853.7466430664062,
-      "learning_rate": 1.579039547741175e-05,
-      "loss": 55.7709,
-      "step": 164520
-    },
-    {
-      "epoch": 0.664722019093638,
-      "grad_norm": 606.7926635742188,
-      "learning_rate": 1.578715039396344e-05,
-      "loss": 44.6481,
-      "step": 164530
-    },
-    {
-      "epoch": 0.6647624203590057,
-      "grad_norm": 484.31829833984375,
-      "learning_rate": 1.5783905490124366e-05,
-      "loss": 51.9917,
-      "step": 164540
-    },
-    {
-      "epoch": 0.6648028216243733,
-      "grad_norm": 770.72314453125,
-      "learning_rate": 1.578066076595781e-05,
-      "loss": 38.291,
-      "step": 164550
-    },
-    {
-      "epoch": 0.664843222889741,
-      "grad_norm": 862.612060546875,
-      "learning_rate": 1.577741622152702e-05,
-      "loss": 54.8366,
-      "step": 164560
-    },
-    {
-      "epoch": 0.6648836241551085,
-      "grad_norm": 548.0719604492188,
-      "learning_rate": 1.5774171856895236e-05,
-      "loss": 50.1739,
-      "step": 164570
-    },
-    {
-      "epoch": 0.6649240254204761,
-      "grad_norm": 953.9413452148438,
-      "learning_rate": 1.5770927672125735e-05,
-      "loss": 39.8906,
-      "step": 164580
-    },
-    {
-      "epoch": 0.6649644266858438,
-      "grad_norm": 555.2219848632812,
-      "learning_rate": 1.5767683667281746e-05,
-      "loss": 42.6911,
-      "step": 164590
-    },
-    {
-      "epoch": 0.6650048279512114,
-      "grad_norm": 567.25244140625,
-      "learning_rate": 1.5764439842426515e-05,
-      "loss": 45.8761,
-      "step": 164600
-    },
-    {
-      "epoch": 0.665045229216579,
-      "grad_norm": 525.5807495117188,
-      "learning_rate": 1.576119619762329e-05,
-      "loss": 42.2324,
-      "step": 164610
-    },
-    {
-      "epoch": 0.6650856304819467,
-      "grad_norm": 766.9520263671875,
-      "learning_rate": 1.5757952732935288e-05,
-      "loss": 53.3623,
-      "step": 164620
-    },
-    {
-      "epoch": 0.6651260317473143,
-      "grad_norm": 859.6406860351562,
-      "learning_rate": 1.575470944842577e-05,
-      "loss": 61.6819,
-      "step": 164630
-    },
-    {
-      "epoch": 0.665166433012682,
-      "grad_norm": 1102.3270263671875,
-      "learning_rate": 1.5751466344157947e-05,
-      "loss": 52.0058,
-      "step": 164640
-    },
-    {
-      "epoch": 0.6652068342780496,
-      "grad_norm": 599.8565673828125,
-      "learning_rate": 1.574822342019504e-05,
-      "loss": 47.8022,
-      "step": 164650
-    },
-    {
-      "epoch": 0.6652472355434172,
-      "grad_norm": 1298.8262939453125,
-      "learning_rate": 1.574498067660029e-05,
-      "loss": 60.007,
-      "step": 164660
-    },
-    {
-      "epoch": 0.6652876368087849,
-      "grad_norm": 507.6510925292969,
-      "learning_rate": 1.574173811343691e-05,
-      "loss": 54.002,
-      "step": 164670
-    },
-    {
-      "epoch": 0.6653280380741525,
-      "grad_norm": 820.8052978515625,
-      "learning_rate": 1.5738495730768105e-05,
-      "loss": 51.6573,
-      "step": 164680
-    },
-    {
-      "epoch": 0.6653684393395202,
-      "grad_norm": 242.71102905273438,
-      "learning_rate": 1.5735253528657098e-05,
-      "loss": 29.9105,
-      "step": 164690
-    },
-    {
-      "epoch": 0.6654088406048877,
-      "grad_norm": 1396.0880126953125,
-      "learning_rate": 1.5732011507167084e-05,
-      "loss": 51.9466,
-      "step": 164700
-    },
-    {
-      "epoch": 0.6654492418702553,
-      "grad_norm": 748.735595703125,
-      "learning_rate": 1.5728769666361287e-05,
-      "loss": 45.9762,
-      "step": 164710
-    },
-    {
-      "epoch": 0.665489643135623,
-      "grad_norm": 501.557373046875,
-      "learning_rate": 1.5725528006302898e-05,
-      "loss": 51.5674,
-      "step": 164720
-    },
-    {
-      "epoch": 0.6655300444009906,
-      "grad_norm": 1090.5556640625,
-      "learning_rate": 1.5722286527055106e-05,
-      "loss": 64.3017,
-      "step": 164730
-    },
-    {
-      "epoch": 0.6655704456663583,
-      "grad_norm": 632.6434326171875,
-      "learning_rate": 1.5719045228681125e-05,
-      "loss": 86.9166,
-      "step": 164740
-    },
-    {
-      "epoch": 0.6656108469317259,
-      "grad_norm": 746.149658203125,
-      "learning_rate": 1.5715804111244137e-05,
-      "loss": 45.3344,
-      "step": 164750
-    },
-    {
-      "epoch": 0.6656512481970935,
-      "grad_norm": 922.2538452148438,
-      "learning_rate": 1.5712563174807317e-05,
-      "loss": 45.3737,
-      "step": 164760
-    },
-    {
-      "epoch": 0.6656916494624612,
-      "grad_norm": 435.47711181640625,
-      "learning_rate": 1.570932241943387e-05,
-      "loss": 44.8212,
-      "step": 164770
-    },
-    {
-      "epoch": 0.6657320507278288,
-      "grad_norm": 261.5279541015625,
-      "learning_rate": 1.5706081845186954e-05,
-      "loss": 40.834,
-      "step": 164780
-    },
-    {
-      "epoch": 0.6657724519931965,
-      "grad_norm": 903.6303100585938,
-      "learning_rate": 1.5702841452129765e-05,
-      "loss": 57.0551,
-      "step": 164790
-    },
-    {
-      "epoch": 0.6658128532585641,
-      "grad_norm": 539.7006225585938,
-      "learning_rate": 1.5699601240325474e-05,
-      "loss": 64.2796,
-      "step": 164800
-    },
-    {
-      "epoch": 0.6658532545239317,
-      "grad_norm": 385.1630859375,
-      "learning_rate": 1.5696361209837228e-05,
-      "loss": 49.2995,
-      "step": 164810
-    },
-    {
-      "epoch": 0.6658936557892993,
-      "grad_norm": 1175.4351806640625,
-      "learning_rate": 1.5693121360728223e-05,
-      "loss": 57.2194,
-      "step": 164820
-    },
-    {
-      "epoch": 0.6659340570546669,
-      "grad_norm": 844.3092041015625,
-      "learning_rate": 1.5689881693061607e-05,
-      "loss": 47.4218,
-      "step": 164830
-    },
-    {
-      "epoch": 0.6659744583200345,
-      "grad_norm": 1089.0220947265625,
-      "learning_rate": 1.5686642206900538e-05,
-      "loss": 53.9814,
-      "step": 164840
-    },
-    {
-      "epoch": 0.6660148595854022,
-      "grad_norm": 1410.1602783203125,
-      "learning_rate": 1.5683402902308175e-05,
-      "loss": 59.7386,
-      "step": 164850
-    },
-    {
-      "epoch": 0.6660552608507698,
-      "grad_norm": 522.8424682617188,
-      "learning_rate": 1.5680163779347667e-05,
-      "loss": 47.6538,
-      "step": 164860
-    },
-    {
-      "epoch": 0.6660956621161375,
-      "grad_norm": 1108.32080078125,
-      "learning_rate": 1.5676924838082172e-05,
-      "loss": 47.7382,
-      "step": 164870
-    },
-    {
-      "epoch": 0.6661360633815051,
-      "grad_norm": 781.7514038085938,
-      "learning_rate": 1.5673686078574822e-05,
-      "loss": 52.2223,
-      "step": 164880
-    },
-    {
-      "epoch": 0.6661764646468727,
-      "grad_norm": 787.5927124023438,
-      "learning_rate": 1.5670447500888756e-05,
-      "loss": 49.3363,
-      "step": 164890
-    },
-    {
-      "epoch": 0.6662168659122404,
-      "grad_norm": 2937.76708984375,
-      "learning_rate": 1.5667209105087132e-05,
-      "loss": 80.4566,
-      "step": 164900
-    },
-    {
-      "epoch": 0.666257267177608,
-      "grad_norm": 777.5316772460938,
-      "learning_rate": 1.566397089123306e-05,
-      "loss": 51.269,
-      "step": 164910
-    },
-    {
-      "epoch": 0.6662976684429757,
-      "grad_norm": 330.6843566894531,
-      "learning_rate": 1.5660732859389686e-05,
-      "loss": 41.2739,
-      "step": 164920
-    },
-    {
-      "epoch": 0.6663380697083433,
-      "grad_norm": 474.75457763671875,
-      "learning_rate": 1.565749500962014e-05,
-      "loss": 63.9933,
-      "step": 164930
-    },
-    {
-      "epoch": 0.666378470973711,
-      "grad_norm": 295.5137634277344,
-      "learning_rate": 1.5654257341987532e-05,
-      "loss": 51.5275,
-      "step": 164940
-    },
-    {
-      "epoch": 0.6664188722390785,
-      "grad_norm": 1012.8373413085938,
-      "learning_rate": 1.5651019856554995e-05,
-      "loss": 75.4704,
-      "step": 164950
-    },
-    {
-      "epoch": 0.6664592735044461,
-      "grad_norm": 1193.5340576171875,
-      "learning_rate": 1.5647782553385636e-05,
-      "loss": 55.0925,
-      "step": 164960
-    },
-    {
-      "epoch": 0.6664996747698138,
-      "grad_norm": 672.0003662109375,
-      "learning_rate": 1.564454543254256e-05,
-      "loss": 41.2695,
-      "step": 164970
-    },
-    {
-      "epoch": 0.6665400760351814,
-      "grad_norm": 1096.6192626953125,
-      "learning_rate": 1.56413084940889e-05,
-      "loss": 80.5824,
-      "step": 164980
-    },
-    {
-      "epoch": 0.666580477300549,
-      "grad_norm": 518.763671875,
-      "learning_rate": 1.5638071738087738e-05,
-      "loss": 42.7408,
-      "step": 164990
-    },
-    {
-      "epoch": 0.6666208785659167,
-      "grad_norm": 463.8653564453125,
-      "learning_rate": 1.56348351646022e-05,
-      "loss": 52.6907,
-      "step": 165000
-    },
-    {
-      "epoch": 0.6666612798312843,
-      "grad_norm": 1809.75732421875,
-      "learning_rate": 1.563159877369537e-05,
-      "loss": 63.8859,
-      "step": 165010
-    },
-    {
-      "epoch": 0.666701681096652,
-      "grad_norm": 705.6831665039062,
-      "learning_rate": 1.562836256543034e-05,
-      "loss": 52.4492,
-      "step": 165020
-    },
-    {
-      "epoch": 0.6667420823620196,
-      "grad_norm": 1125.3067626953125,
-      "learning_rate": 1.562512653987021e-05,
-      "loss": 37.5355,
-      "step": 165030
-    },
-    {
-      "epoch": 0.6667824836273872,
-      "grad_norm": 869.19482421875,
-      "learning_rate": 1.562189069707807e-05,
-      "loss": 40.3621,
-      "step": 165040
-    },
-    {
-      "epoch": 0.6668228848927549,
-      "grad_norm": 617.2032470703125,
-      "learning_rate": 1.561865503711698e-05,
-      "loss": 48.8869,
-      "step": 165050
-    },
-    {
-      "epoch": 0.6668632861581225,
-      "grad_norm": 846.593017578125,
-      "learning_rate": 1.5615419560050058e-05,
-      "loss": 69.1997,
-      "step": 165060
-    },
-    {
-      "epoch": 0.6669036874234902,
-      "grad_norm": 470.2166748046875,
-      "learning_rate": 1.5612184265940348e-05,
-      "loss": 51.828,
-      "step": 165070
-    },
-    {
-      "epoch": 0.6669440886888577,
-      "grad_norm": 505.6989440917969,
-      "learning_rate": 1.560894915485095e-05,
-      "loss": 75.3402,
-      "step": 165080
-    },
-    {
-      "epoch": 0.6669844899542253,
-      "grad_norm": 548.412841796875,
-      "learning_rate": 1.5605714226844924e-05,
-      "loss": 62.3138,
-      "step": 165090
-    },
-    {
-      "epoch": 0.667024891219593,
-      "grad_norm": 516.51123046875,
-      "learning_rate": 1.5602479481985333e-05,
-      "loss": 71.4453,
-      "step": 165100
-    },
-    {
-      "epoch": 0.6670652924849606,
-      "grad_norm": 919.295166015625,
-      "learning_rate": 1.5599244920335245e-05,
-      "loss": 48.0811,
-      "step": 165110
-    },
-    {
-      "epoch": 0.6671056937503282,
-      "grad_norm": 540.0143432617188,
-      "learning_rate": 1.5596010541957712e-05,
-      "loss": 39.6037,
-      "step": 165120
-    },
-    {
-      "epoch": 0.6671460950156959,
-      "grad_norm": 554.9569702148438,
-      "learning_rate": 1.5592776346915796e-05,
-      "loss": 56.8077,
-      "step": 165130
-    },
-    {
-      "epoch": 0.6671864962810635,
-      "grad_norm": 352.08392333984375,
-      "learning_rate": 1.5589542335272556e-05,
-      "loss": 41.5959,
-      "step": 165140
-    },
-    {
-      "epoch": 0.6672268975464312,
-      "grad_norm": 666.3216552734375,
-      "learning_rate": 1.5586308507091018e-05,
-      "loss": 54.632,
-      "step": 165150
-    },
-    {
-      "epoch": 0.6672672988117988,
-      "grad_norm": 372.5094299316406,
-      "learning_rate": 1.5583074862434255e-05,
-      "loss": 41.4742,
-      "step": 165160
-    },
-    {
-      "epoch": 0.6673077000771664,
-      "grad_norm": 464.4438781738281,
-      "learning_rate": 1.5579841401365293e-05,
-      "loss": 68.3877,
-      "step": 165170
-    },
-    {
-      "epoch": 0.6673481013425341,
-      "grad_norm": 325.2723388671875,
-      "learning_rate": 1.5576608123947166e-05,
-      "loss": 44.0116,
-      "step": 165180
-    },
-    {
-      "epoch": 0.6673885026079017,
-      "grad_norm": 538.7552490234375,
-      "learning_rate": 1.5573375030242922e-05,
-      "loss": 43.1703,
-      "step": 165190
-    },
-    {
-      "epoch": 0.6674289038732694,
-      "grad_norm": 629.1891479492188,
-      "learning_rate": 1.557014212031559e-05,
-      "loss": 37.3895,
-      "step": 165200
-    },
-    {
-      "epoch": 0.6674693051386369,
-      "grad_norm": 873.6672973632812,
-      "learning_rate": 1.5566909394228178e-05,
-      "loss": 42.7094,
-      "step": 165210
-    },
-    {
-      "epoch": 0.6675097064040045,
-      "grad_norm": 1655.4644775390625,
-      "learning_rate": 1.556367685204374e-05,
-      "loss": 46.7371,
-      "step": 165220
-    },
-    {
-      "epoch": 0.6675501076693722,
-      "grad_norm": 1335.8275146484375,
-      "learning_rate": 1.556044449382526e-05,
-      "loss": 66.9424,
-      "step": 165230
-    },
-    {
-      "epoch": 0.6675905089347398,
-      "grad_norm": 473.89788818359375,
-      "learning_rate": 1.555721231963579e-05,
-      "loss": 44.3112,
-      "step": 165240
-    },
-    {
-      "epoch": 0.6676309102001075,
-      "grad_norm": 938.0970458984375,
-      "learning_rate": 1.5553980329538326e-05,
-      "loss": 43.0849,
-      "step": 165250
-    },
-    {
-      "epoch": 0.6676713114654751,
-      "grad_norm": 714.7005004882812,
-      "learning_rate": 1.555074852359587e-05,
-      "loss": 42.4088,
-      "step": 165260
-    },
-    {
-      "epoch": 0.6677117127308427,
-      "grad_norm": 954.3150634765625,
-      "learning_rate": 1.5547516901871446e-05,
-      "loss": 48.4536,
-      "step": 165270
-    },
-    {
-      "epoch": 0.6677521139962104,
-      "grad_norm": 474.6979064941406,
-      "learning_rate": 1.5544285464428045e-05,
-      "loss": 46.0709,
-      "step": 165280
-    },
-    {
-      "epoch": 0.667792515261578,
-      "grad_norm": 850.7158813476562,
-      "learning_rate": 1.5541054211328663e-05,
-      "loss": 61.7394,
-      "step": 165290
-    },
-    {
-      "epoch": 0.6678329165269457,
-      "grad_norm": 373.7199401855469,
-      "learning_rate": 1.5537823142636305e-05,
-      "loss": 52.8841,
-      "step": 165300
-    },
-    {
-      "epoch": 0.6678733177923133,
-      "grad_norm": 1184.4080810546875,
-      "learning_rate": 1.5534592258413943e-05,
-      "loss": 55.5253,
-      "step": 165310
-    },
-    {
-      "epoch": 0.6679137190576809,
-      "grad_norm": 632.0458984375,
-      "learning_rate": 1.5531361558724587e-05,
-      "loss": 42.4319,
-      "step": 165320
-    },
-    {
-      "epoch": 0.6679541203230486,
-      "grad_norm": 733.4188842773438,
-      "learning_rate": 1.5528131043631215e-05,
-      "loss": 58.4967,
-      "step": 165330
-    },
-    {
-      "epoch": 0.6679945215884161,
-      "grad_norm": 483.6113586425781,
-      "learning_rate": 1.55249007131968e-05,
-      "loss": 46.9925,
-      "step": 165340
-    },
-    {
-      "epoch": 0.6680349228537837,
-      "grad_norm": 1163.3182373046875,
-      "learning_rate": 1.5521670567484325e-05,
-      "loss": 51.8809,
-      "step": 165350
-    },
-    {
-      "epoch": 0.6680753241191514,
-      "grad_norm": 1143.9605712890625,
-      "learning_rate": 1.5518440606556766e-05,
-      "loss": 72.9315,
-      "step": 165360
-    },
-    {
-      "epoch": 0.668115725384519,
-      "grad_norm": 1212.5926513671875,
-      "learning_rate": 1.5515210830477083e-05,
-      "loss": 47.2623,
-      "step": 165370
-    },
-    {
-      "epoch": 0.6681561266498867,
-      "grad_norm": 1403.037109375,
-      "learning_rate": 1.5511981239308256e-05,
-      "loss": 50.8124,
-      "step": 165380
-    },
-    {
-      "epoch": 0.6681965279152543,
-      "grad_norm": 3793.583984375,
-      "learning_rate": 1.5508751833113223e-05,
-      "loss": 54.1592,
-      "step": 165390
-    },
-    {
-      "epoch": 0.6682369291806219,
-      "grad_norm": 926.212890625,
-      "learning_rate": 1.5505522611954975e-05,
-      "loss": 57.2393,
-      "step": 165400
-    },
-    {
-      "epoch": 0.6682773304459896,
-      "grad_norm": 1016.1986694335938,
-      "learning_rate": 1.550229357589645e-05,
-      "loss": 58.6371,
-      "step": 165410
-    },
-    {
-      "epoch": 0.6683177317113572,
-      "grad_norm": 705.27001953125,
-      "learning_rate": 1.5499064725000592e-05,
-      "loss": 35.6693,
-      "step": 165420
-    },
-    {
-      "epoch": 0.6683581329767249,
-      "grad_norm": 657.5211181640625,
-      "learning_rate": 1.549583605933037e-05,
-      "loss": 38.7974,
-      "step": 165430
-    },
-    {
-      "epoch": 0.6683985342420925,
-      "grad_norm": 1164.2186279296875,
-      "learning_rate": 1.549260757894871e-05,
-      "loss": 34.9751,
-      "step": 165440
-    },
-    {
-      "epoch": 0.6684389355074601,
-      "grad_norm": 1214.388427734375,
-      "learning_rate": 1.5489379283918566e-05,
-      "loss": 63.4225,
-      "step": 165450
-    },
-    {
-      "epoch": 0.6684793367728277,
-      "grad_norm": 491.2007751464844,
-      "learning_rate": 1.548615117430286e-05,
-      "loss": 49.4402,
-      "step": 165460
-    },
-    {
-      "epoch": 0.6685197380381953,
-      "grad_norm": 851.6419677734375,
-      "learning_rate": 1.5482923250164537e-05,
-      "loss": 64.906,
-      "step": 165470
-    },
-    {
-      "epoch": 0.668560139303563,
-      "grad_norm": 1071.079345703125,
-      "learning_rate": 1.5479695511566534e-05,
-      "loss": 50.622,
-      "step": 165480
-    },
-    {
-      "epoch": 0.6686005405689306,
-      "grad_norm": 417.3287658691406,
-      "learning_rate": 1.5476467958571767e-05,
-      "loss": 49.5059,
-      "step": 165490
-    },
-    {
-      "epoch": 0.6686409418342982,
-      "grad_norm": 607.785400390625,
-      "learning_rate": 1.547324059124315e-05,
-      "loss": 34.5434,
-      "step": 165500
-    },
-    {
-      "epoch": 0.6686813430996659,
-      "grad_norm": 614.81591796875,
-      "learning_rate": 1.547001340964362e-05,
-      "loss": 39.9333,
-      "step": 165510
-    },
-    {
-      "epoch": 0.6687217443650335,
-      "grad_norm": 649.9052734375,
-      "learning_rate": 1.5466786413836077e-05,
-      "loss": 62.8206,
-      "step": 165520
-    },
-    {
-      "epoch": 0.6687621456304012,
-      "grad_norm": 387.1318359375,
-      "learning_rate": 1.546355960388345e-05,
-      "loss": 37.9586,
-      "step": 165530
-    },
-    {
-      "epoch": 0.6688025468957688,
-      "grad_norm": 216.71209716796875,
-      "learning_rate": 1.5460332979848634e-05,
-      "loss": 36.5199,
-      "step": 165540
-    },
-    {
-      "epoch": 0.6688429481611364,
-      "grad_norm": 913.2258911132812,
-      "learning_rate": 1.5457106541794543e-05,
-      "loss": 52.0842,
-      "step": 165550
-    },
-    {
-      "epoch": 0.6688833494265041,
-      "grad_norm": 1081.210205078125,
-      "learning_rate": 1.5453880289784066e-05,
-      "loss": 55.7764,
-      "step": 165560
-    },
-    {
-      "epoch": 0.6689237506918717,
-      "grad_norm": 1004.1875,
-      "learning_rate": 1.5450654223880117e-05,
-      "loss": 32.5717,
-      "step": 165570
-    },
-    {
-      "epoch": 0.6689641519572394,
-      "grad_norm": 1424.0938720703125,
-      "learning_rate": 1.5447428344145563e-05,
-      "loss": 67.9024,
-      "step": 165580
-    },
-    {
-      "epoch": 0.6690045532226069,
-      "grad_norm": 898.0243530273438,
-      "learning_rate": 1.544420265064333e-05,
-      "loss": 67.8531,
-      "step": 165590
-    },
-    {
-      "epoch": 0.6690449544879745,
-      "grad_norm": 912.253662109375,
-      "learning_rate": 1.544097714343627e-05,
-      "loss": 43.7589,
-      "step": 165600
-    },
-    {
-      "epoch": 0.6690853557533422,
-      "grad_norm": 743.2623901367188,
-      "learning_rate": 1.5437751822587294e-05,
-      "loss": 46.5631,
-      "step": 165610
-    },
-    {
-      "epoch": 0.6691257570187098,
-      "grad_norm": 476.3370056152344,
-      "learning_rate": 1.5434526688159267e-05,
-      "loss": 51.2162,
-      "step": 165620
-    },
-    {
-      "epoch": 0.6691661582840774,
-      "grad_norm": 856.005859375,
-      "learning_rate": 1.5431301740215064e-05,
-      "loss": 51.07,
-      "step": 165630
-    },
-    {
-      "epoch": 0.6692065595494451,
-      "grad_norm": 809.520751953125,
-      "learning_rate": 1.5428076978817562e-05,
-      "loss": 37.5981,
-      "step": 165640
-    },
-    {
-      "epoch": 0.6692469608148127,
-      "grad_norm": 729.1454467773438,
-      "learning_rate": 1.5424852404029634e-05,
-      "loss": 44.3347,
-      "step": 165650
-    },
-    {
-      "epoch": 0.6692873620801804,
-      "grad_norm": 524.4779663085938,
-      "learning_rate": 1.542162801591412e-05,
-      "loss": 74.8679,
-      "step": 165660
-    },
-    {
-      "epoch": 0.669327763345548,
-      "grad_norm": 556.4620971679688,
-      "learning_rate": 1.5418403814533912e-05,
-      "loss": 53.1046,
-      "step": 165670
-    },
-    {
-      "epoch": 0.6693681646109156,
-      "grad_norm": 354.8873596191406,
-      "learning_rate": 1.5415179799951844e-05,
-      "loss": 51.9128,
-      "step": 165680
-    },
-    {
-      "epoch": 0.6694085658762833,
-      "grad_norm": 507.58636474609375,
-      "learning_rate": 1.5411955972230794e-05,
-      "loss": 43.3807,
-      "step": 165690
-    },
-    {
-      "epoch": 0.6694489671416509,
-      "grad_norm": 573.6036987304688,
-      "learning_rate": 1.5408732331433595e-05,
-      "loss": 59.5482,
-      "step": 165700
-    },
-    {
-      "epoch": 0.6694893684070186,
-      "grad_norm": 363.93328857421875,
-      "learning_rate": 1.5405508877623094e-05,
-      "loss": 49.7304,
-      "step": 165710
-    },
-    {
-      "epoch": 0.6695297696723861,
-      "grad_norm": 994.6358032226562,
-      "learning_rate": 1.5402285610862142e-05,
-      "loss": 41.309,
-      "step": 165720
-    },
-    {
-      "epoch": 0.6695701709377537,
-      "grad_norm": 0.0,
-      "learning_rate": 1.539906253121357e-05,
-      "loss": 43.8188,
-      "step": 165730
-    },
-    {
-      "epoch": 0.6696105722031214,
-      "grad_norm": 914.8114013671875,
-      "learning_rate": 1.5395839638740213e-05,
-      "loss": 52.398,
-      "step": 165740
-    },
-    {
-      "epoch": 0.669650973468489,
-      "grad_norm": 894.1461181640625,
-      "learning_rate": 1.539261693350491e-05,
-      "loss": 48.5215,
-      "step": 165750
-    },
-    {
-      "epoch": 0.6696913747338566,
-      "grad_norm": 1071.9180908203125,
-      "learning_rate": 1.538939441557048e-05,
-      "loss": 42.8886,
-      "step": 165760
-    },
-    {
-      "epoch": 0.6697317759992243,
-      "grad_norm": 801.8084106445312,
-      "learning_rate": 1.5386172084999765e-05,
-      "loss": 57.6356,
-      "step": 165770
-    },
-    {
-      "epoch": 0.6697721772645919,
-      "grad_norm": 1373.5555419921875,
-      "learning_rate": 1.5382949941855574e-05,
-      "loss": 67.3987,
-      "step": 165780
-    },
-    {
-      "epoch": 0.6698125785299596,
-      "grad_norm": 693.4388427734375,
-      "learning_rate": 1.5379727986200716e-05,
-      "loss": 76.9919,
-      "step": 165790
-    },
-    {
-      "epoch": 0.6698529797953272,
-      "grad_norm": 586.3527221679688,
-      "learning_rate": 1.5376506218098015e-05,
-      "loss": 77.7751,
-      "step": 165800
-    },
-    {
-      "epoch": 0.6698933810606948,
-      "grad_norm": 481.37579345703125,
-      "learning_rate": 1.537328463761029e-05,
-      "loss": 44.1322,
-      "step": 165810
-    },
-    {
-      "epoch": 0.6699337823260625,
-      "grad_norm": 631.0782470703125,
-      "learning_rate": 1.5370063244800327e-05,
-      "loss": 60.4053,
-      "step": 165820
-    },
-    {
-      "epoch": 0.6699741835914301,
-      "grad_norm": 374.3154602050781,
-      "learning_rate": 1.5366842039730946e-05,
-      "loss": 58.5419,
-      "step": 165830
-    },
-    {
-      "epoch": 0.6700145848567978,
-      "grad_norm": 457.1123046875,
-      "learning_rate": 1.5363621022464924e-05,
-      "loss": 32.6104,
-      "step": 165840
-    },
-    {
-      "epoch": 0.6700549861221653,
-      "grad_norm": 947.1151123046875,
-      "learning_rate": 1.536040019306509e-05,
-      "loss": 45.5976,
-      "step": 165850
-    },
-    {
-      "epoch": 0.6700953873875329,
-      "grad_norm": 882.1751708984375,
-      "learning_rate": 1.535717955159421e-05,
-      "loss": 41.449,
-      "step": 165860
-    },
-    {
-      "epoch": 0.6701357886529006,
-      "grad_norm": 547.3237915039062,
-      "learning_rate": 1.5353959098115072e-05,
-      "loss": 41.6607,
-      "step": 165870
-    },
-    {
-      "epoch": 0.6701761899182682,
-      "grad_norm": 774.8670654296875,
-      "learning_rate": 1.535073883269048e-05,
-      "loss": 40.8859,
-      "step": 165880
-    },
-    {
-      "epoch": 0.6702165911836359,
-      "grad_norm": 1032.724365234375,
-      "learning_rate": 1.534751875538319e-05,
-      "loss": 63.1162,
-      "step": 165890
-    },
-    {
-      "epoch": 0.6702569924490035,
-      "grad_norm": 846.3161010742188,
-      "learning_rate": 1.5344298866256e-05,
-      "loss": 63.9277,
-      "step": 165900
-    },
-    {
-      "epoch": 0.6702973937143711,
-      "grad_norm": 950.7833862304688,
-      "learning_rate": 1.534107916537168e-05,
-      "loss": 54.6691,
-      "step": 165910
-    },
-    {
-      "epoch": 0.6703377949797388,
-      "grad_norm": 385.0533752441406,
-      "learning_rate": 1.533785965279298e-05,
-      "loss": 82.8562,
-      "step": 165920
-    },
-    {
-      "epoch": 0.6703781962451064,
-      "grad_norm": 101.061279296875,
-      "learning_rate": 1.5334640328582688e-05,
-      "loss": 45.9366,
-      "step": 165930
-    },
-    {
-      "epoch": 0.6704185975104741,
-      "grad_norm": 896.1965942382812,
-      "learning_rate": 1.5331421192803565e-05,
-      "loss": 44.7952,
-      "step": 165940
-    },
-    {
-      "epoch": 0.6704589987758417,
-      "grad_norm": 1131.50634765625,
-      "learning_rate": 1.5328202245518347e-05,
-      "loss": 51.4988,
-      "step": 165950
-    },
-    {
-      "epoch": 0.6704994000412093,
-      "grad_norm": 504.31109619140625,
-      "learning_rate": 1.5324983486789818e-05,
-      "loss": 51.8349,
-      "step": 165960
-    },
-    {
-      "epoch": 0.670539801306577,
-      "grad_norm": 683.3836059570312,
-      "learning_rate": 1.532176491668071e-05,
-      "loss": 42.7658,
-      "step": 165970
-    },
-    {
-      "epoch": 0.6705802025719445,
-      "grad_norm": 0.0,
-      "learning_rate": 1.5318546535253785e-05,
-      "loss": 49.9037,
-      "step": 165980
-    },
-    {
-      "epoch": 0.6706206038373121,
-      "grad_norm": 881.3528442382812,
-      "learning_rate": 1.531532834257178e-05,
-      "loss": 79.0532,
-      "step": 165990
-    },
-    {
-      "epoch": 0.6706610051026798,
-      "grad_norm": 662.7510986328125,
-      "learning_rate": 1.5312110338697426e-05,
-      "loss": 49.7831,
-      "step": 166000
-    },
-    {
-      "epoch": 0.6707014063680474,
-      "grad_norm": 586.3132934570312,
-      "learning_rate": 1.5308892523693478e-05,
-      "loss": 50.3877,
-      "step": 166010
-    },
-    {
-      "epoch": 0.6707418076334151,
-      "grad_norm": 1294.632568359375,
-      "learning_rate": 1.5305674897622658e-05,
-      "loss": 48.3198,
-      "step": 166020
-    },
-    {
-      "epoch": 0.6707822088987827,
-      "grad_norm": 0.0,
-      "learning_rate": 1.5302457460547687e-05,
-      "loss": 29.5519,
-      "step": 166030
-    },
-    {
-      "epoch": 0.6708226101641503,
-      "grad_norm": 799.5355834960938,
-      "learning_rate": 1.5299240212531314e-05,
-      "loss": 54.3518,
-      "step": 166040
-    },
-    {
-      "epoch": 0.670863011429518,
-      "grad_norm": 284.1693420410156,
-      "learning_rate": 1.5296023153636235e-05,
-      "loss": 50.3665,
-      "step": 166050
-    },
-    {
-      "epoch": 0.6709034126948856,
-      "grad_norm": 651.7603759765625,
-      "learning_rate": 1.5292806283925193e-05,
-      "loss": 41.6884,
-      "step": 166060
-    },
-    {
-      "epoch": 0.6709438139602533,
-      "grad_norm": 915.2215576171875,
-      "learning_rate": 1.5289589603460885e-05,
-      "loss": 41.2101,
-      "step": 166070
-    },
-    {
-      "epoch": 0.6709842152256209,
-      "grad_norm": 562.0250854492188,
-      "learning_rate": 1.5286373112306018e-05,
-      "loss": 83.8193,
-      "step": 166080
-    },
-    {
-      "epoch": 0.6710246164909885,
-      "grad_norm": 884.758544921875,
-      "learning_rate": 1.528315681052332e-05,
-      "loss": 38.1881,
-      "step": 166090
-    },
-    {
-      "epoch": 0.6710650177563561,
-      "grad_norm": 298.85906982421875,
-      "learning_rate": 1.527994069817548e-05,
-      "loss": 61.6959,
-      "step": 166100
-    },
-    {
-      "epoch": 0.6711054190217237,
-      "grad_norm": 886.3592529296875,
-      "learning_rate": 1.5276724775325192e-05,
-      "loss": 60.5207,
-      "step": 166110
-    },
-    {
-      "epoch": 0.6711458202870914,
-      "grad_norm": 724.6257934570312,
-      "learning_rate": 1.5273509042035172e-05,
-      "loss": 49.5858,
-      "step": 166120
-    },
-    {
-      "epoch": 0.671186221552459,
-      "grad_norm": 1337.0560302734375,
-      "learning_rate": 1.5270293498368093e-05,
-      "loss": 44.4996,
-      "step": 166130
-    },
-    {
-      "epoch": 0.6712266228178266,
-      "grad_norm": 970.5552368164062,
-      "learning_rate": 1.5267078144386654e-05,
-      "loss": 34.8261,
-      "step": 166140
-    },
-    {
-      "epoch": 0.6712670240831943,
-      "grad_norm": 754.7394409179688,
-      "learning_rate": 1.526386298015354e-05,
-      "loss": 48.2782,
-      "step": 166150
-    },
-    {
-      "epoch": 0.6713074253485619,
-      "grad_norm": 591.5711669921875,
-      "learning_rate": 1.5260648005731427e-05,
-      "loss": 43.9917,
-      "step": 166160
-    },
-    {
-      "epoch": 0.6713478266139296,
-      "grad_norm": 557.9755859375,
-      "learning_rate": 1.5257433221182999e-05,
-      "loss": 44.6998,
-      "step": 166170
-    },
-    {
-      "epoch": 0.6713882278792972,
-      "grad_norm": 1885.062744140625,
-      "learning_rate": 1.5254218626570926e-05,
-      "loss": 45.9201,
-      "step": 166180
-    },
-    {
-      "epoch": 0.6714286291446648,
-      "grad_norm": 581.98974609375,
-      "learning_rate": 1.5251004221957865e-05,
-      "loss": 57.501,
-      "step": 166190
-    },
-    {
-      "epoch": 0.6714690304100325,
-      "grad_norm": 1093.8936767578125,
-      "learning_rate": 1.524779000740651e-05,
-      "loss": 60.8076,
-      "step": 166200
-    },
-    {
-      "epoch": 0.6715094316754001,
-      "grad_norm": 2516.6455078125,
-      "learning_rate": 1.5244575982979497e-05,
-      "loss": 56.7106,
-      "step": 166210
-    },
-    {
-      "epoch": 0.6715498329407678,
-      "grad_norm": 470.2278137207031,
-      "learning_rate": 1.5241362148739513e-05,
-      "loss": 58.4495,
-      "step": 166220
-    },
-    {
-      "epoch": 0.6715902342061353,
-      "grad_norm": 861.3046875,
-      "learning_rate": 1.5238148504749195e-05,
-      "loss": 43.9791,
-      "step": 166230
-    },
-    {
-      "epoch": 0.6716306354715029,
-      "grad_norm": 960.9512329101562,
-      "learning_rate": 1.5234935051071192e-05,
-      "loss": 48.8351,
-      "step": 166240
-    },
-    {
-      "epoch": 0.6716710367368706,
-      "grad_norm": 472.22882080078125,
-      "learning_rate": 1.523172178776816e-05,
-      "loss": 56.9353,
-      "step": 166250
-    },
-    {
-      "epoch": 0.6717114380022382,
-      "grad_norm": 870.235595703125,
-      "learning_rate": 1.5228508714902745e-05,
-      "loss": 65.5524,
-      "step": 166260
-    },
-    {
-      "epoch": 0.6717518392676058,
-      "grad_norm": 618.8545532226562,
-      "learning_rate": 1.5225295832537574e-05,
-      "loss": 43.7606,
-      "step": 166270
-    },
-    {
-      "epoch": 0.6717922405329735,
-      "grad_norm": 359.6185607910156,
-      "learning_rate": 1.52220831407353e-05,
-      "loss": 59.9645,
-      "step": 166280
-    },
-    {
-      "epoch": 0.6718326417983411,
-      "grad_norm": 848.9765625,
-      "learning_rate": 1.5218870639558536e-05,
-      "loss": 47.4422,
-      "step": 166290
-    },
-    {
-      "epoch": 0.6718730430637088,
-      "grad_norm": 493.81597900390625,
-      "learning_rate": 1.521565832906994e-05,
-      "loss": 42.8739,
-      "step": 166300
-    },
-    {
-      "epoch": 0.6719134443290764,
-      "grad_norm": 546.1114501953125,
-      "learning_rate": 1.521244620933212e-05,
-      "loss": 50.4388,
-      "step": 166310
-    },
-    {
-      "epoch": 0.671953845594444,
-      "grad_norm": 850.0634155273438,
-      "learning_rate": 1.5209234280407697e-05,
-      "loss": 39.2991,
-      "step": 166320
-    },
-    {
-      "epoch": 0.6719942468598117,
-      "grad_norm": 476.33441162109375,
-      "learning_rate": 1.5206022542359297e-05,
-      "loss": 43.6006,
-      "step": 166330
-    },
-    {
-      "epoch": 0.6720346481251793,
-      "grad_norm": 1327.771484375,
-      "learning_rate": 1.5202810995249527e-05,
-      "loss": 67.9057,
-      "step": 166340
-    },
-    {
-      "epoch": 0.672075049390547,
-      "grad_norm": 998.8424072265625,
-      "learning_rate": 1.5199599639140994e-05,
-      "loss": 48.7459,
-      "step": 166350
-    },
-    {
-      "epoch": 0.6721154506559145,
-      "grad_norm": 401.8807067871094,
-      "learning_rate": 1.5196388474096319e-05,
-      "loss": 31.3531,
-      "step": 166360
-    },
-    {
-      "epoch": 0.6721558519212821,
-      "grad_norm": 647.2686157226562,
-      "learning_rate": 1.5193177500178091e-05,
-      "loss": 53.1445,
-      "step": 166370
-    },
-    {
-      "epoch": 0.6721962531866498,
-      "grad_norm": 4298.70654296875,
-      "learning_rate": 1.5189966717448923e-05,
-      "loss": 134.2395,
-      "step": 166380
-    },
-    {
-      "epoch": 0.6722366544520174,
-      "grad_norm": 331.1536865234375,
-      "learning_rate": 1.5186756125971407e-05,
-      "loss": 37.2909,
-      "step": 166390
-    },
-    {
-      "epoch": 0.672277055717385,
-      "grad_norm": 532.593505859375,
-      "learning_rate": 1.5183545725808127e-05,
-      "loss": 41.4082,
-      "step": 166400
-    },
-    {
-      "epoch": 0.6723174569827527,
-      "grad_norm": 53.21999740600586,
-      "learning_rate": 1.5180335517021682e-05,
-      "loss": 46.6067,
-      "step": 166410
-    },
-    {
-      "epoch": 0.6723578582481203,
-      "grad_norm": 404.0959777832031,
-      "learning_rate": 1.5177125499674638e-05,
-      "loss": 52.84,
-      "step": 166420
-    },
-    {
-      "epoch": 0.672398259513488,
-      "grad_norm": 858.9264526367188,
-      "learning_rate": 1.5173915673829604e-05,
-      "loss": 57.7387,
-      "step": 166430
-    },
-    {
-      "epoch": 0.6724386607788556,
-      "grad_norm": 446.1383972167969,
-      "learning_rate": 1.5170706039549142e-05,
-      "loss": 56.045,
-      "step": 166440
-    },
-    {
-      "epoch": 0.6724790620442233,
-      "grad_norm": 436.2591247558594,
-      "learning_rate": 1.5167496596895814e-05,
-      "loss": 55.9357,
-      "step": 166450
-    },
-    {
-      "epoch": 0.6725194633095909,
-      "grad_norm": 609.2459106445312,
-      "learning_rate": 1.5164287345932216e-05,
-      "loss": 41.1153,
-      "step": 166460
-    },
-    {
-      "epoch": 0.6725598645749585,
-      "grad_norm": 379.0639343261719,
-      "learning_rate": 1.5161078286720898e-05,
-      "loss": 32.7601,
-      "step": 166470
-    },
-    {
-      "epoch": 0.6726002658403262,
-      "grad_norm": 392.2806396484375,
-      "learning_rate": 1.515786941932441e-05,
-      "loss": 30.7783,
-      "step": 166480
-    },
-    {
-      "epoch": 0.6726406671056937,
-      "grad_norm": 710.4434814453125,
-      "learning_rate": 1.515466074380534e-05,
-      "loss": 42.4171,
-      "step": 166490
-    },
-    {
-      "epoch": 0.6726810683710613,
-      "grad_norm": 921.6600952148438,
-      "learning_rate": 1.5151452260226224e-05,
-      "loss": 43.7955,
-      "step": 166500
-    },
-    {
-      "epoch": 0.672721469636429,
-      "grad_norm": 666.1463012695312,
-      "learning_rate": 1.5148243968649617e-05,
-      "loss": 44.6316,
-      "step": 166510
-    },
-    {
-      "epoch": 0.6727618709017966,
-      "grad_norm": 567.9951171875,
-      "learning_rate": 1.5145035869138067e-05,
-      "loss": 51.5773,
-      "step": 166520
-    },
-    {
-      "epoch": 0.6728022721671643,
-      "grad_norm": 778.463134765625,
-      "learning_rate": 1.5141827961754107e-05,
-      "loss": 65.0105,
-      "step": 166530
-    },
-    {
-      "epoch": 0.6728426734325319,
-      "grad_norm": 744.7365112304688,
-      "learning_rate": 1.5138620246560296e-05,
-      "loss": 47.7923,
-      "step": 166540
-    },
-    {
-      "epoch": 0.6728830746978995,
-      "grad_norm": 750.9429931640625,
-      "learning_rate": 1.5135412723619158e-05,
-      "loss": 55.9812,
-      "step": 166550
-    },
-    {
-      "epoch": 0.6729234759632672,
-      "grad_norm": 1654.34765625,
-      "learning_rate": 1.513220539299322e-05,
-      "loss": 49.7286,
-      "step": 166560
-    },
-    {
-      "epoch": 0.6729638772286348,
-      "grad_norm": 596.2041625976562,
-      "learning_rate": 1.512899825474503e-05,
-      "loss": 35.4972,
-      "step": 166570
-    },
-    {
-      "epoch": 0.6730042784940025,
-      "grad_norm": 929.0473022460938,
-      "learning_rate": 1.5125791308937092e-05,
-      "loss": 41.1769,
-      "step": 166580
-    },
-    {
-      "epoch": 0.6730446797593701,
-      "grad_norm": 456.9891662597656,
-      "learning_rate": 1.5122584555631944e-05,
-      "loss": 51.7045,
-      "step": 166590
-    },
-    {
-      "epoch": 0.6730850810247377,
-      "grad_norm": 326.9583435058594,
-      "learning_rate": 1.5119377994892094e-05,
-      "loss": 45.5811,
-      "step": 166600
-    },
-    {
-      "epoch": 0.6731254822901054,
-      "grad_norm": 681.9387817382812,
-      "learning_rate": 1.5116171626780049e-05,
-      "loss": 60.2909,
-      "step": 166610
-    },
-    {
-      "epoch": 0.6731658835554729,
-      "grad_norm": 288.6805419921875,
-      "learning_rate": 1.5112965451358335e-05,
-      "loss": 45.1375,
-      "step": 166620
-    },
-    {
-      "epoch": 0.6732062848208406,
-      "grad_norm": 375.949951171875,
-      "learning_rate": 1.5109759468689449e-05,
-      "loss": 31.1756,
-      "step": 166630
-    },
-    {
-      "epoch": 0.6732466860862082,
-      "grad_norm": 271.1240234375,
-      "learning_rate": 1.5106553678835884e-05,
-      "loss": 19.8992,
-      "step": 166640
-    },
-    {
-      "epoch": 0.6732870873515758,
-      "grad_norm": 294.1463928222656,
-      "learning_rate": 1.5103348081860159e-05,
-      "loss": 50.6929,
-      "step": 166650
-    },
-    {
-      "epoch": 0.6733274886169435,
-      "grad_norm": 755.0875244140625,
-      "learning_rate": 1.5100142677824753e-05,
-      "loss": 44.0782,
-      "step": 166660
-    },
-    {
-      "epoch": 0.6733678898823111,
-      "grad_norm": 622.8639526367188,
-      "learning_rate": 1.5096937466792169e-05,
-      "loss": 54.3434,
-      "step": 166670
-    },
-    {
-      "epoch": 0.6734082911476788,
-      "grad_norm": 942.9237060546875,
-      "learning_rate": 1.5093732448824888e-05,
-      "loss": 58.158,
-      "step": 166680
-    },
-    {
-      "epoch": 0.6734486924130464,
-      "grad_norm": 509.4794616699219,
-      "learning_rate": 1.5090527623985379e-05,
-      "loss": 59.3872,
-      "step": 166690
-    },
-    {
-      "epoch": 0.673489093678414,
-      "grad_norm": 747.8128662109375,
-      "learning_rate": 1.5087322992336147e-05,
-      "loss": 49.9323,
-      "step": 166700
-    },
-    {
-      "epoch": 0.6735294949437817,
-      "grad_norm": 1549.0496826171875,
-      "learning_rate": 1.5084118553939659e-05,
-      "loss": 41.1457,
-      "step": 166710
-    },
-    {
-      "epoch": 0.6735698962091493,
-      "grad_norm": 930.0858154296875,
-      "learning_rate": 1.5080914308858374e-05,
-      "loss": 82.8616,
-      "step": 166720
-    },
-    {
-      "epoch": 0.673610297474517,
-      "grad_norm": 2211.052734375,
-      "learning_rate": 1.5077710257154782e-05,
-      "loss": 81.3171,
-      "step": 166730
-    },
-    {
-      "epoch": 0.6736506987398845,
-      "grad_norm": 1658.010498046875,
-      "learning_rate": 1.5074506398891328e-05,
-      "loss": 54.205,
-      "step": 166740
-    },
-    {
-      "epoch": 0.6736911000052521,
-      "grad_norm": 371.7207336425781,
-      "learning_rate": 1.5071302734130489e-05,
-      "loss": 67.5018,
-      "step": 166750
-    },
-    {
-      "epoch": 0.6737315012706198,
-      "grad_norm": 1372.85546875,
-      "learning_rate": 1.506809926293471e-05,
-      "loss": 54.2526,
-      "step": 166760
-    },
-    {
-      "epoch": 0.6737719025359874,
-      "grad_norm": 307.7871398925781,
-      "learning_rate": 1.506489598536645e-05,
-      "loss": 42.5487,
-      "step": 166770
-    },
-    {
-      "epoch": 0.673812303801355,
-      "grad_norm": 225.70396423339844,
-      "learning_rate": 1.5061692901488162e-05,
-      "loss": 53.6776,
-      "step": 166780
-    },
-    {
-      "epoch": 0.6738527050667227,
-      "grad_norm": 2758.591064453125,
-      "learning_rate": 1.5058490011362286e-05,
-      "loss": 50.1884,
-      "step": 166790
-    },
-    {
-      "epoch": 0.6738931063320903,
-      "grad_norm": 283.5628967285156,
-      "learning_rate": 1.505528731505126e-05,
-      "loss": 72.4942,
-      "step": 166800
-    },
-    {
-      "epoch": 0.673933507597458,
-      "grad_norm": 885.5732421875,
-      "learning_rate": 1.5052084812617533e-05,
-      "loss": 32.8235,
-      "step": 166810
-    },
-    {
-      "epoch": 0.6739739088628256,
-      "grad_norm": 592.6634521484375,
-      "learning_rate": 1.5048882504123529e-05,
-      "loss": 26.64,
-      "step": 166820
-    },
-    {
-      "epoch": 0.6740143101281932,
-      "grad_norm": 448.51971435546875,
-      "learning_rate": 1.5045680389631686e-05,
-      "loss": 59.0059,
-      "step": 166830
-    },
-    {
-      "epoch": 0.6740547113935609,
-      "grad_norm": 760.6209106445312,
-      "learning_rate": 1.5042478469204435e-05,
-      "loss": 45.268,
-      "step": 166840
-    },
-    {
-      "epoch": 0.6740951126589285,
-      "grad_norm": 541.5360107421875,
-      "learning_rate": 1.5039276742904185e-05,
-      "loss": 62.0476,
-      "step": 166850
-    },
-    {
-      "epoch": 0.6741355139242962,
-      "grad_norm": 1311.8394775390625,
-      "learning_rate": 1.5036075210793368e-05,
-      "loss": 86.5055,
-      "step": 166860
-    },
-    {
-      "epoch": 0.6741759151896637,
-      "grad_norm": 871.006591796875,
-      "learning_rate": 1.5032873872934394e-05,
-      "loss": 66.7284,
-      "step": 166870
-    },
-    {
-      "epoch": 0.6742163164550313,
-      "grad_norm": 750.0230712890625,
-      "learning_rate": 1.5029672729389669e-05,
-      "loss": 34.7836,
-      "step": 166880
-    },
-    {
-      "epoch": 0.674256717720399,
-      "grad_norm": 1440.93994140625,
-      "learning_rate": 1.5026471780221612e-05,
-      "loss": 44.9238,
-      "step": 166890
-    },
-    {
-      "epoch": 0.6742971189857666,
-      "grad_norm": 331.9103698730469,
-      "learning_rate": 1.5023271025492618e-05,
-      "loss": 45.4969,
-      "step": 166900
-    },
-    {
-      "epoch": 0.6743375202511342,
-      "grad_norm": 1073.5389404296875,
-      "learning_rate": 1.5020070465265098e-05,
-      "loss": 53.6601,
-      "step": 166910
-    },
-    {
-      "epoch": 0.6743779215165019,
-      "grad_norm": 854.6211547851562,
-      "learning_rate": 1.5016870099601444e-05,
-      "loss": 55.5576,
-      "step": 166920
-    },
-    {
-      "epoch": 0.6744183227818695,
-      "grad_norm": 1300.90478515625,
-      "learning_rate": 1.501366992856404e-05,
-      "loss": 40.2496,
-      "step": 166930
-    },
-    {
-      "epoch": 0.6744587240472372,
-      "grad_norm": 401.09375,
-      "learning_rate": 1.501046995221529e-05,
-      "loss": 31.3694,
-      "step": 166940
-    },
-    {
-      "epoch": 0.6744991253126048,
-      "grad_norm": 386.5203857421875,
-      "learning_rate": 1.500727017061756e-05,
-      "loss": 47.9808,
-      "step": 166950
-    },
-    {
-      "epoch": 0.6745395265779724,
-      "grad_norm": 535.3987426757812,
-      "learning_rate": 1.5004070583833251e-05,
-      "loss": 62.2983,
-      "step": 166960
-    },
-    {
-      "epoch": 0.6745799278433401,
-      "grad_norm": 1989.7027587890625,
-      "learning_rate": 1.5000871191924731e-05,
-      "loss": 70.537,
-      "step": 166970
-    },
-    {
-      "epoch": 0.6746203291087077,
-      "grad_norm": 396.22125244140625,
-      "learning_rate": 1.4997671994954371e-05,
-      "loss": 51.7699,
-      "step": 166980
-    },
-    {
-      "epoch": 0.6746607303740754,
-      "grad_norm": 453.8977355957031,
-      "learning_rate": 1.499447299298455e-05,
-      "loss": 32.1538,
-      "step": 166990
-    },
-    {
-      "epoch": 0.6747011316394429,
-      "grad_norm": 654.4794311523438,
-      "learning_rate": 1.4991274186077632e-05,
-      "loss": 49.3818,
-      "step": 167000
-    },
-    {
-      "epoch": 0.6747415329048105,
-      "grad_norm": 682.2568359375,
-      "learning_rate": 1.4988075574295968e-05,
-      "loss": 68.9815,
-      "step": 167010
-    },
-    {
-      "epoch": 0.6747819341701782,
-      "grad_norm": 1162.35791015625,
-      "learning_rate": 1.4984877157701932e-05,
-      "loss": 72.4098,
-      "step": 167020
-    },
-    {
-      "epoch": 0.6748223354355458,
-      "grad_norm": 1012.3003540039062,
-      "learning_rate": 1.4981678936357863e-05,
-      "loss": 44.8387,
-      "step": 167030
-    },
-    {
-      "epoch": 0.6748627367009135,
-      "grad_norm": 378.43292236328125,
-      "learning_rate": 1.4978480910326132e-05,
-      "loss": 74.7357,
-      "step": 167040
-    },
-    {
-      "epoch": 0.6749031379662811,
-      "grad_norm": 393.8338317871094,
-      "learning_rate": 1.4975283079669072e-05,
-      "loss": 31.4734,
-      "step": 167050
-    },
-    {
-      "epoch": 0.6749435392316487,
-      "grad_norm": 662.2154541015625,
-      "learning_rate": 1.4972085444449018e-05,
-      "loss": 33.0518,
-      "step": 167060
-    },
-    {
-      "epoch": 0.6749839404970164,
-      "grad_norm": 748.7392578125,
-      "learning_rate": 1.4968888004728338e-05,
-      "loss": 44.0402,
-      "step": 167070
-    },
-    {
-      "epoch": 0.675024341762384,
-      "grad_norm": 1040.579833984375,
-      "learning_rate": 1.4965690760569346e-05,
-      "loss": 54.7236,
-      "step": 167080
-    },
-    {
-      "epoch": 0.6750647430277517,
-      "grad_norm": 2075.803955078125,
-      "learning_rate": 1.4962493712034373e-05,
-      "loss": 70.8729,
-      "step": 167090
-    },
-    {
-      "epoch": 0.6751051442931193,
-      "grad_norm": 787.3817138671875,
-      "learning_rate": 1.4959296859185754e-05,
-      "loss": 44.2748,
-      "step": 167100
-    },
-    {
-      "epoch": 0.6751455455584869,
-      "grad_norm": 816.3740844726562,
-      "learning_rate": 1.4956100202085809e-05,
-      "loss": 52.7223,
-      "step": 167110
-    },
-    {
-      "epoch": 0.6751859468238546,
-      "grad_norm": 268.7018127441406,
-      "learning_rate": 1.4952903740796873e-05,
-      "loss": 39.8699,
-      "step": 167120
-    },
-    {
-      "epoch": 0.6752263480892221,
-      "grad_norm": 341.5850524902344,
-      "learning_rate": 1.4949707475381247e-05,
-      "loss": 61.8581,
-      "step": 167130
-    },
-    {
-      "epoch": 0.6752667493545897,
-      "grad_norm": 373.7789611816406,
-      "learning_rate": 1.4946511405901236e-05,
-      "loss": 43.0861,
-      "step": 167140
-    },
-    {
-      "epoch": 0.6753071506199574,
-      "grad_norm": 1607.9439697265625,
-      "learning_rate": 1.4943315532419177e-05,
-      "loss": 41.1233,
-      "step": 167150
-    },
-    {
-      "epoch": 0.675347551885325,
-      "grad_norm": 417.06597900390625,
-      "learning_rate": 1.4940119854997354e-05,
-      "loss": 28.2353,
-      "step": 167160
-    },
-    {
-      "epoch": 0.6753879531506927,
-      "grad_norm": 493.3361511230469,
-      "learning_rate": 1.4936924373698066e-05,
-      "loss": 52.0371,
-      "step": 167170
-    },
-    {
-      "epoch": 0.6754283544160603,
-      "grad_norm": 429.33819580078125,
-      "learning_rate": 1.4933729088583626e-05,
-      "loss": 54.8033,
-      "step": 167180
-    },
-    {
-      "epoch": 0.675468755681428,
-      "grad_norm": 731.7124633789062,
-      "learning_rate": 1.4930533999716317e-05,
-      "loss": 52.8938,
-      "step": 167190
-    },
-    {
-      "epoch": 0.6755091569467956,
-      "grad_norm": 807.2184448242188,
-      "learning_rate": 1.4927339107158437e-05,
-      "loss": 68.7912,
-      "step": 167200
-    },
-    {
-      "epoch": 0.6755495582121632,
-      "grad_norm": 594.211669921875,
-      "learning_rate": 1.4924144410972265e-05,
-      "loss": 51.8672,
-      "step": 167210
-    },
-    {
-      "epoch": 0.6755899594775309,
-      "grad_norm": 1601.8602294921875,
-      "learning_rate": 1.4920949911220078e-05,
-      "loss": 61.9253,
-      "step": 167220
-    },
-    {
-      "epoch": 0.6756303607428985,
-      "grad_norm": 216.67095947265625,
-      "learning_rate": 1.4917755607964168e-05,
-      "loss": 43.7638,
-      "step": 167230
-    },
-    {
-      "epoch": 0.6756707620082661,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4914561501266805e-05,
-      "loss": 41.52,
-      "step": 167240
-    },
-    {
-      "epoch": 0.6757111632736337,
-      "grad_norm": 675.5355224609375,
-      "learning_rate": 1.4911367591190248e-05,
-      "loss": 38.7212,
-      "step": 167250
-    },
-    {
-      "epoch": 0.6757515645390013,
-      "grad_norm": 790.0083618164062,
-      "learning_rate": 1.4908173877796783e-05,
-      "loss": 49.3089,
-      "step": 167260
-    },
-    {
-      "epoch": 0.675791965804369,
-      "grad_norm": 1060.16796875,
-      "learning_rate": 1.490498036114866e-05,
-      "loss": 71.76,
-      "step": 167270
-    },
-    {
-      "epoch": 0.6758323670697366,
-      "grad_norm": 1394.681640625,
-      "learning_rate": 1.4901787041308146e-05,
-      "loss": 43.9718,
-      "step": 167280
-    },
-    {
-      "epoch": 0.6758727683351042,
-      "grad_norm": 860.03466796875,
-      "learning_rate": 1.4898593918337494e-05,
-      "loss": 42.2888,
-      "step": 167290
-    },
-    {
-      "epoch": 0.6759131696004719,
-      "grad_norm": 1046.5,
-      "learning_rate": 1.4895400992298942e-05,
-      "loss": 40.4044,
-      "step": 167300
-    },
-    {
-      "epoch": 0.6759535708658395,
-      "grad_norm": 387.24810791015625,
-      "learning_rate": 1.4892208263254762e-05,
-      "loss": 54.466,
-      "step": 167310
-    },
-    {
-      "epoch": 0.6759939721312072,
-      "grad_norm": 700.537353515625,
-      "learning_rate": 1.4889015731267186e-05,
-      "loss": 39.7297,
-      "step": 167320
-    },
-    {
-      "epoch": 0.6760343733965748,
-      "grad_norm": 465.908935546875,
-      "learning_rate": 1.4885823396398441e-05,
-      "loss": 43.5209,
-      "step": 167330
-    },
-    {
-      "epoch": 0.6760747746619424,
-      "grad_norm": 840.498291015625,
-      "learning_rate": 1.4882631258710788e-05,
-      "loss": 35.8289,
-      "step": 167340
-    },
-    {
-      "epoch": 0.6761151759273101,
-      "grad_norm": 1104.419189453125,
-      "learning_rate": 1.4879439318266442e-05,
-      "loss": 65.706,
-      "step": 167350
-    },
-    {
-      "epoch": 0.6761555771926777,
-      "grad_norm": 435.9366760253906,
-      "learning_rate": 1.4876247575127641e-05,
-      "loss": 63.7672,
-      "step": 167360
-    },
-    {
-      "epoch": 0.6761959784580454,
-      "grad_norm": 1329.9017333984375,
-      "learning_rate": 1.4873056029356608e-05,
-      "loss": 76.5193,
-      "step": 167370
-    },
-    {
-      "epoch": 0.6762363797234129,
-      "grad_norm": 357.02081298828125,
-      "learning_rate": 1.486986468101555e-05,
-      "loss": 55.5462,
-      "step": 167380
-    },
-    {
-      "epoch": 0.6762767809887805,
-      "grad_norm": 468.8736267089844,
-      "learning_rate": 1.4866673530166703e-05,
-      "loss": 48.5541,
-      "step": 167390
-    },
-    {
-      "epoch": 0.6763171822541482,
-      "grad_norm": 670.1433715820312,
-      "learning_rate": 1.4863482576872275e-05,
-      "loss": 55.1696,
-      "step": 167400
-    },
-    {
-      "epoch": 0.6763575835195158,
-      "grad_norm": 289.0069580078125,
-      "learning_rate": 1.486029182119446e-05,
-      "loss": 44.4765,
-      "step": 167410
-    },
-    {
-      "epoch": 0.6763979847848834,
-      "grad_norm": 803.2169799804688,
-      "learning_rate": 1.4857101263195491e-05,
-      "loss": 38.6828,
-      "step": 167420
-    },
-    {
-      "epoch": 0.6764383860502511,
-      "grad_norm": 852.9284057617188,
-      "learning_rate": 1.4853910902937548e-05,
-      "loss": 36.3954,
-      "step": 167430
-    },
-    {
-      "epoch": 0.6764787873156187,
-      "grad_norm": 875.1856689453125,
-      "learning_rate": 1.485072074048284e-05,
-      "loss": 57.8835,
-      "step": 167440
-    },
-    {
-      "epoch": 0.6765191885809864,
-      "grad_norm": 692.950927734375,
-      "learning_rate": 1.4847530775893554e-05,
-      "loss": 102.5225,
-      "step": 167450
-    },
-    {
-      "epoch": 0.676559589846354,
-      "grad_norm": 897.9561767578125,
-      "learning_rate": 1.4844341009231876e-05,
-      "loss": 65.091,
-      "step": 167460
-    },
-    {
-      "epoch": 0.6765999911117216,
-      "grad_norm": 800.497802734375,
-      "learning_rate": 1.4841151440560009e-05,
-      "loss": 29.9811,
-      "step": 167470
-    },
-    {
-      "epoch": 0.6766403923770893,
-      "grad_norm": 734.5826416015625,
-      "learning_rate": 1.4837962069940114e-05,
-      "loss": 57.3711,
-      "step": 167480
-    },
-    {
-      "epoch": 0.6766807936424569,
-      "grad_norm": 756.9467163085938,
-      "learning_rate": 1.4834772897434388e-05,
-      "loss": 48.2266,
-      "step": 167490
-    },
-    {
-      "epoch": 0.6767211949078246,
-      "grad_norm": 449.9790344238281,
-      "learning_rate": 1.4831583923104999e-05,
-      "loss": 62.687,
-      "step": 167500
-    },
-    {
-      "epoch": 0.6767615961731921,
-      "grad_norm": 1147.2373046875,
-      "learning_rate": 1.4828395147014106e-05,
-      "loss": 46.7687,
-      "step": 167510
-    },
-    {
-      "epoch": 0.6768019974385597,
-      "grad_norm": 546.6746826171875,
-      "learning_rate": 1.4825206569223899e-05,
-      "loss": 51.0683,
-      "step": 167520
-    },
-    {
-      "epoch": 0.6768423987039274,
-      "grad_norm": 871.5731201171875,
-      "learning_rate": 1.4822018189796525e-05,
-      "loss": 49.3357,
-      "step": 167530
-    },
-    {
-      "epoch": 0.676882799969295,
-      "grad_norm": 690.7603759765625,
-      "learning_rate": 1.4818830008794143e-05,
-      "loss": 40.3889,
-      "step": 167540
-    },
-    {
-      "epoch": 0.6769232012346627,
-      "grad_norm": 792.1572265625,
-      "learning_rate": 1.4815642026278913e-05,
-      "loss": 46.8522,
-      "step": 167550
-    },
-    {
-      "epoch": 0.6769636025000303,
-      "grad_norm": 448.63714599609375,
-      "learning_rate": 1.4812454242312979e-05,
-      "loss": 35.1668,
-      "step": 167560
-    },
-    {
-      "epoch": 0.6770040037653979,
-      "grad_norm": 594.5196533203125,
-      "learning_rate": 1.4809266656958504e-05,
-      "loss": 57.1279,
-      "step": 167570
-    },
-    {
-      "epoch": 0.6770444050307656,
-      "grad_norm": 346.8206481933594,
-      "learning_rate": 1.4806079270277623e-05,
-      "loss": 41.4994,
-      "step": 167580
-    },
-    {
-      "epoch": 0.6770848062961332,
-      "grad_norm": 435.75714111328125,
-      "learning_rate": 1.4802892082332461e-05,
-      "loss": 37.8179,
-      "step": 167590
-    },
-    {
-      "epoch": 0.6771252075615009,
-      "grad_norm": 921.9247436523438,
-      "learning_rate": 1.4799705093185181e-05,
-      "loss": 74.8439,
-      "step": 167600
-    },
-    {
-      "epoch": 0.6771656088268685,
-      "grad_norm": 453.77178955078125,
-      "learning_rate": 1.47965183028979e-05,
-      "loss": 51.5178,
-      "step": 167610
-    },
-    {
-      "epoch": 0.6772060100922361,
-      "grad_norm": 590.6558837890625,
-      "learning_rate": 1.4793331711532744e-05,
-      "loss": 48.1793,
-      "step": 167620
-    },
-    {
-      "epoch": 0.6772464113576038,
-      "grad_norm": 971.6336059570312,
-      "learning_rate": 1.4790145319151846e-05,
-      "loss": 48.0082,
-      "step": 167630
-    },
-    {
-      "epoch": 0.6772868126229713,
-      "grad_norm": 933.4710693359375,
-      "learning_rate": 1.4786959125817312e-05,
-      "loss": 42.4729,
-      "step": 167640
-    },
-    {
-      "epoch": 0.6773272138883389,
-      "grad_norm": 1059.8695068359375,
-      "learning_rate": 1.4783773131591278e-05,
-      "loss": 49.6168,
-      "step": 167650
-    },
-    {
-      "epoch": 0.6773676151537066,
-      "grad_norm": 796.0709228515625,
-      "learning_rate": 1.4780587336535844e-05,
-      "loss": 46.6943,
-      "step": 167660
-    },
-    {
-      "epoch": 0.6774080164190742,
-      "grad_norm": 361.644775390625,
-      "learning_rate": 1.4777401740713112e-05,
-      "loss": 36.3145,
-      "step": 167670
-    },
-    {
-      "epoch": 0.6774484176844419,
-      "grad_norm": 745.7164916992188,
-      "learning_rate": 1.4774216344185205e-05,
-      "loss": 55.0647,
-      "step": 167680
-    },
-    {
-      "epoch": 0.6774888189498095,
-      "grad_norm": 143.7425079345703,
-      "learning_rate": 1.477103114701422e-05,
-      "loss": 43.7669,
-      "step": 167690
-    },
-    {
-      "epoch": 0.6775292202151771,
-      "grad_norm": 663.1755981445312,
-      "learning_rate": 1.4767846149262237e-05,
-      "loss": 46.7611,
-      "step": 167700
-    },
-    {
-      "epoch": 0.6775696214805448,
-      "grad_norm": 362.8015441894531,
-      "learning_rate": 1.476466135099137e-05,
-      "loss": 39.5467,
-      "step": 167710
-    },
-    {
-      "epoch": 0.6776100227459124,
-      "grad_norm": 1011.8582153320312,
-      "learning_rate": 1.476147675226369e-05,
-      "loss": 52.9852,
-      "step": 167720
-    },
-    {
-      "epoch": 0.6776504240112801,
-      "grad_norm": 1140.3988037109375,
-      "learning_rate": 1.47582923531413e-05,
-      "loss": 68.2397,
-      "step": 167730
-    },
-    {
-      "epoch": 0.6776908252766477,
-      "grad_norm": 1081.307373046875,
-      "learning_rate": 1.4755108153686275e-05,
-      "loss": 67.8517,
-      "step": 167740
-    },
-    {
-      "epoch": 0.6777312265420153,
-      "grad_norm": 416.0245361328125,
-      "learning_rate": 1.475192415396068e-05,
-      "loss": 51.0669,
-      "step": 167750
-    },
-    {
-      "epoch": 0.677771627807383,
-      "grad_norm": 911.7064819335938,
-      "learning_rate": 1.474874035402661e-05,
-      "loss": 58.926,
-      "step": 167760
-    },
-    {
-      "epoch": 0.6778120290727505,
-      "grad_norm": 510.8608093261719,
-      "learning_rate": 1.4745556753946125e-05,
-      "loss": 64.8826,
-      "step": 167770
-    },
-    {
-      "epoch": 0.6778524303381182,
-      "grad_norm": 1057.5894775390625,
-      "learning_rate": 1.4742373353781285e-05,
-      "loss": 57.3264,
-      "step": 167780
-    },
-    {
-      "epoch": 0.6778928316034858,
-      "grad_norm": 416.422119140625,
-      "learning_rate": 1.4739190153594157e-05,
-      "loss": 27.1778,
-      "step": 167790
-    },
-    {
-      "epoch": 0.6779332328688534,
-      "grad_norm": 879.224365234375,
-      "learning_rate": 1.4736007153446801e-05,
-      "loss": 59.9546,
-      "step": 167800
-    },
-    {
-      "epoch": 0.6779736341342211,
-      "grad_norm": 1042.2764892578125,
-      "learning_rate": 1.4732824353401273e-05,
-      "loss": 60.1474,
-      "step": 167810
-    },
-    {
-      "epoch": 0.6780140353995887,
-      "grad_norm": 612.871826171875,
-      "learning_rate": 1.4729641753519618e-05,
-      "loss": 50.6535,
-      "step": 167820
-    },
-    {
-      "epoch": 0.6780544366649564,
-      "grad_norm": 663.3670654296875,
-      "learning_rate": 1.472645935386388e-05,
-      "loss": 46.683,
-      "step": 167830
-    },
-    {
-      "epoch": 0.678094837930324,
-      "grad_norm": 787.3726196289062,
-      "learning_rate": 1.4723277154496111e-05,
-      "loss": 56.9814,
-      "step": 167840
-    },
-    {
-      "epoch": 0.6781352391956916,
-      "grad_norm": 1040.9354248046875,
-      "learning_rate": 1.472009515547835e-05,
-      "loss": 60.3822,
-      "step": 167850
-    },
-    {
-      "epoch": 0.6781756404610593,
-      "grad_norm": 708.2203979492188,
-      "learning_rate": 1.4716913356872614e-05,
-      "loss": 59.3495,
-      "step": 167860
-    },
-    {
-      "epoch": 0.6782160417264269,
-      "grad_norm": 677.75048828125,
-      "learning_rate": 1.4713731758740956e-05,
-      "loss": 34.9022,
-      "step": 167870
-    },
-    {
-      "epoch": 0.6782564429917946,
-      "grad_norm": 1900.8795166015625,
-      "learning_rate": 1.4710550361145386e-05,
-      "loss": 39.3471,
-      "step": 167880
-    },
-    {
-      "epoch": 0.6782968442571621,
-      "grad_norm": 246.22775268554688,
-      "learning_rate": 1.470736916414794e-05,
-      "loss": 30.3667,
-      "step": 167890
-    },
-    {
-      "epoch": 0.6783372455225297,
-      "grad_norm": 410.8074035644531,
-      "learning_rate": 1.4704188167810635e-05,
-      "loss": 43.8699,
-      "step": 167900
-    },
-    {
-      "epoch": 0.6783776467878974,
-      "grad_norm": 580.0462646484375,
-      "learning_rate": 1.4701007372195469e-05,
-      "loss": 61.4522,
-      "step": 167910
-    },
-    {
-      "epoch": 0.678418048053265,
-      "grad_norm": 951.4273681640625,
-      "learning_rate": 1.4697826777364477e-05,
-      "loss": 83.201,
-      "step": 167920
-    },
-    {
-      "epoch": 0.6784584493186326,
-      "grad_norm": 613.5919189453125,
-      "learning_rate": 1.4694646383379657e-05,
-      "loss": 55.9393,
-      "step": 167930
-    },
-    {
-      "epoch": 0.6784988505840003,
-      "grad_norm": 1002.1031494140625,
-      "learning_rate": 1.4691466190303e-05,
-      "loss": 31.6006,
-      "step": 167940
-    },
-    {
-      "epoch": 0.6785392518493679,
-      "grad_norm": 1075.826904296875,
-      "learning_rate": 1.4688286198196524e-05,
-      "loss": 50.4907,
-      "step": 167950
-    },
-    {
-      "epoch": 0.6785796531147356,
-      "grad_norm": 880.4722290039062,
-      "learning_rate": 1.4685106407122218e-05,
-      "loss": 30.88,
-      "step": 167960
-    },
-    {
-      "epoch": 0.6786200543801032,
-      "grad_norm": 406.95733642578125,
-      "learning_rate": 1.4681926817142071e-05,
-      "loss": 49.4461,
-      "step": 167970
-    },
-    {
-      "epoch": 0.6786604556454708,
-      "grad_norm": 1066.5130615234375,
-      "learning_rate": 1.4678747428318079e-05,
-      "loss": 73.1735,
-      "step": 167980
-    },
-    {
-      "epoch": 0.6787008569108385,
-      "grad_norm": 76.85258483886719,
-      "learning_rate": 1.4675568240712206e-05,
-      "loss": 38.3394,
-      "step": 167990
-    },
-    {
-      "epoch": 0.6787412581762061,
-      "grad_norm": 2073.328125,
-      "learning_rate": 1.467238925438646e-05,
-      "loss": 76.5933,
-      "step": 168000
-    },
-    {
-      "epoch": 0.6787816594415738,
-      "grad_norm": 911.6731567382812,
-      "learning_rate": 1.4669210469402789e-05,
-      "loss": 74.6866,
-      "step": 168010
-    },
-    {
-      "epoch": 0.6788220607069413,
-      "grad_norm": 502.549072265625,
-      "learning_rate": 1.466603188582319e-05,
-      "loss": 36.4497,
-      "step": 168020
-    },
-    {
-      "epoch": 0.6788624619723089,
-      "grad_norm": 222.66180419921875,
-      "learning_rate": 1.4662853503709617e-05,
-      "loss": 78.5684,
-      "step": 168030
-    },
-    {
-      "epoch": 0.6789028632376766,
-      "grad_norm": 624.6636352539062,
-      "learning_rate": 1.4659675323124036e-05,
-      "loss": 48.2924,
-      "step": 168040
-    },
-    {
-      "epoch": 0.6789432645030442,
-      "grad_norm": 322.1091003417969,
-      "learning_rate": 1.4656497344128412e-05,
-      "loss": 40.4815,
-      "step": 168050
-    },
-    {
-      "epoch": 0.6789836657684118,
-      "grad_norm": 1130.6031494140625,
-      "learning_rate": 1.4653319566784696e-05,
-      "loss": 69.0325,
-      "step": 168060
-    },
-    {
-      "epoch": 0.6790240670337795,
-      "grad_norm": 454.5316467285156,
-      "learning_rate": 1.4650141991154832e-05,
-      "loss": 52.8761,
-      "step": 168070
-    },
-    {
-      "epoch": 0.6790644682991471,
-      "grad_norm": 980.3739624023438,
-      "learning_rate": 1.464696461730079e-05,
-      "loss": 50.6851,
-      "step": 168080
-    },
-    {
-      "epoch": 0.6791048695645148,
-      "grad_norm": 1091.614990234375,
-      "learning_rate": 1.464378744528449e-05,
-      "loss": 65.8454,
-      "step": 168090
-    },
-    {
-      "epoch": 0.6791452708298824,
-      "grad_norm": 1084.71337890625,
-      "learning_rate": 1.4640610475167898e-05,
-      "loss": 78.1262,
-      "step": 168100
-    },
-    {
-      "epoch": 0.67918567209525,
-      "grad_norm": 667.9674682617188,
-      "learning_rate": 1.4637433707012938e-05,
-      "loss": 45.5906,
-      "step": 168110
-    },
-    {
-      "epoch": 0.6792260733606177,
-      "grad_norm": 946.9363403320312,
-      "learning_rate": 1.4634257140881536e-05,
-      "loss": 54.872,
-      "step": 168120
-    },
-    {
-      "epoch": 0.6792664746259853,
-      "grad_norm": 640.0570068359375,
-      "learning_rate": 1.4631080776835629e-05,
-      "loss": 41.9037,
-      "step": 168130
-    },
-    {
-      "epoch": 0.679306875891353,
-      "grad_norm": 352.1270446777344,
-      "learning_rate": 1.4627904614937143e-05,
-      "loss": 54.49,
-      "step": 168140
-    },
-    {
-      "epoch": 0.6793472771567205,
-      "grad_norm": 1415.2705078125,
-      "learning_rate": 1.4624728655247995e-05,
-      "loss": 43.0648,
-      "step": 168150
-    },
-    {
-      "epoch": 0.6793876784220881,
-      "grad_norm": 572.370849609375,
-      "learning_rate": 1.462155289783011e-05,
-      "loss": 42.223,
-      "step": 168160
-    },
-    {
-      "epoch": 0.6794280796874558,
-      "grad_norm": 970.8530883789062,
-      "learning_rate": 1.4618377342745381e-05,
-      "loss": 57.9264,
-      "step": 168170
-    },
-    {
-      "epoch": 0.6794684809528234,
-      "grad_norm": 1352.0662841796875,
-      "learning_rate": 1.461520199005574e-05,
-      "loss": 48.2247,
-      "step": 168180
-    },
-    {
-      "epoch": 0.6795088822181911,
-      "grad_norm": 375.2728576660156,
-      "learning_rate": 1.4612026839823084e-05,
-      "loss": 43.6525,
-      "step": 168190
-    },
-    {
-      "epoch": 0.6795492834835587,
-      "grad_norm": 560.6547241210938,
-      "learning_rate": 1.4608851892109304e-05,
-      "loss": 39.3269,
-      "step": 168200
-    },
-    {
-      "epoch": 0.6795896847489263,
-      "grad_norm": 558.0577392578125,
-      "learning_rate": 1.4605677146976315e-05,
-      "loss": 51.4779,
-      "step": 168210
-    },
-    {
-      "epoch": 0.679630086014294,
-      "grad_norm": 431.2486572265625,
-      "learning_rate": 1.4602502604486001e-05,
-      "loss": 53.5672,
-      "step": 168220
-    },
-    {
-      "epoch": 0.6796704872796616,
-      "grad_norm": 616.7715454101562,
-      "learning_rate": 1.4599328264700247e-05,
-      "loss": 36.0276,
-      "step": 168230
-    },
-    {
-      "epoch": 0.6797108885450293,
-      "grad_norm": 1291.5537109375,
-      "learning_rate": 1.4596154127680947e-05,
-      "loss": 43.0771,
-      "step": 168240
-    },
-    {
-      "epoch": 0.6797512898103969,
-      "grad_norm": 1298.318115234375,
-      "learning_rate": 1.4592980193489975e-05,
-      "loss": 53.2581,
-      "step": 168250
-    },
-    {
-      "epoch": 0.6797916910757645,
-      "grad_norm": 845.0426635742188,
-      "learning_rate": 1.458980646218921e-05,
-      "loss": 50.5599,
-      "step": 168260
-    },
-    {
-      "epoch": 0.6798320923411322,
-      "grad_norm": 592.9050903320312,
-      "learning_rate": 1.458663293384053e-05,
-      "loss": 60.6831,
-      "step": 168270
-    },
-    {
-      "epoch": 0.6798724936064997,
-      "grad_norm": 488.923095703125,
-      "learning_rate": 1.4583459608505801e-05,
-      "loss": 43.6718,
-      "step": 168280
-    },
-    {
-      "epoch": 0.6799128948718673,
-      "grad_norm": 390.5216369628906,
-      "learning_rate": 1.45802864862469e-05,
-      "loss": 46.2525,
-      "step": 168290
-    },
-    {
-      "epoch": 0.679953296137235,
-      "grad_norm": 662.7907104492188,
-      "learning_rate": 1.4577113567125669e-05,
-      "loss": 53.4482,
-      "step": 168300
-    },
-    {
-      "epoch": 0.6799936974026026,
-      "grad_norm": 282.6667785644531,
-      "learning_rate": 1.4573940851203974e-05,
-      "loss": 38.2593,
-      "step": 168310
-    },
-    {
-      "epoch": 0.6800340986679703,
-      "grad_norm": 614.1632080078125,
-      "learning_rate": 1.4570768338543672e-05,
-      "loss": 43.634,
-      "step": 168320
-    },
-    {
-      "epoch": 0.6800744999333379,
-      "grad_norm": 550.2304077148438,
-      "learning_rate": 1.4567596029206607e-05,
-      "loss": 68.1899,
-      "step": 168330
-    },
-    {
-      "epoch": 0.6801149011987055,
-      "grad_norm": 692.3876953125,
-      "learning_rate": 1.456442392325463e-05,
-      "loss": 66.2251,
-      "step": 168340
-    },
-    {
-      "epoch": 0.6801553024640732,
-      "grad_norm": 389.5172424316406,
-      "learning_rate": 1.4561252020749591e-05,
-      "loss": 56.9403,
-      "step": 168350
-    },
-    {
-      "epoch": 0.6801957037294408,
-      "grad_norm": 726.8670043945312,
-      "learning_rate": 1.455808032175331e-05,
-      "loss": 59.1227,
-      "step": 168360
-    },
-    {
-      "epoch": 0.6802361049948085,
-      "grad_norm": 1539.648193359375,
-      "learning_rate": 1.4554908826327625e-05,
-      "loss": 50.6336,
-      "step": 168370
-    },
-    {
-      "epoch": 0.6802765062601761,
-      "grad_norm": 463.2121276855469,
-      "learning_rate": 1.4551737534534383e-05,
-      "loss": 36.2807,
-      "step": 168380
-    },
-    {
-      "epoch": 0.6803169075255437,
-      "grad_norm": 535.5940551757812,
-      "learning_rate": 1.4548566446435378e-05,
-      "loss": 48.5567,
-      "step": 168390
-    },
-    {
-      "epoch": 0.6803573087909114,
-      "grad_norm": 514.473388671875,
-      "learning_rate": 1.4545395562092468e-05,
-      "loss": 26.0429,
-      "step": 168400
-    },
-    {
-      "epoch": 0.6803977100562789,
-      "grad_norm": 1582.9013671875,
-      "learning_rate": 1.4542224881567434e-05,
-      "loss": 32.9378,
-      "step": 168410
-    },
-    {
-      "epoch": 0.6804381113216466,
-      "grad_norm": 2122.857177734375,
-      "learning_rate": 1.453905440492213e-05,
-      "loss": 49.2891,
-      "step": 168420
-    },
-    {
-      "epoch": 0.6804785125870142,
-      "grad_norm": 377.3238220214844,
-      "learning_rate": 1.4535884132218342e-05,
-      "loss": 43.1437,
-      "step": 168430
-    },
-    {
-      "epoch": 0.6805189138523818,
-      "grad_norm": 919.3121948242188,
-      "learning_rate": 1.4532714063517871e-05,
-      "loss": 78.3128,
-      "step": 168440
-    },
-    {
-      "epoch": 0.6805593151177495,
-      "grad_norm": 673.9483642578125,
-      "learning_rate": 1.4529544198882544e-05,
-      "loss": 75.3447,
-      "step": 168450
-    },
-    {
-      "epoch": 0.6805997163831171,
-      "grad_norm": 832.7345581054688,
-      "learning_rate": 1.4526374538374132e-05,
-      "loss": 59.8041,
-      "step": 168460
-    },
-    {
-      "epoch": 0.6806401176484848,
-      "grad_norm": 1186.2255859375,
-      "learning_rate": 1.4523205082054442e-05,
-      "loss": 52.9427,
-      "step": 168470
-    },
-    {
-      "epoch": 0.6806805189138524,
-      "grad_norm": 488.5611572265625,
-      "learning_rate": 1.452003582998526e-05,
-      "loss": 36.2347,
-      "step": 168480
-    },
-    {
-      "epoch": 0.68072092017922,
-      "grad_norm": 1167.7965087890625,
-      "learning_rate": 1.4516866782228378e-05,
-      "loss": 41.4017,
-      "step": 168490
-    },
-    {
-      "epoch": 0.6807613214445877,
-      "grad_norm": 2220.746337890625,
-      "learning_rate": 1.4513697938845572e-05,
-      "loss": 52.8377,
-      "step": 168500
-    },
-    {
-      "epoch": 0.6808017227099553,
-      "grad_norm": 1385.3912353515625,
-      "learning_rate": 1.4510529299898634e-05,
-      "loss": 49.8378,
-      "step": 168510
-    },
-    {
-      "epoch": 0.680842123975323,
-      "grad_norm": 735.9927978515625,
-      "learning_rate": 1.4507360865449319e-05,
-      "loss": 50.0806,
-      "step": 168520
-    },
-    {
-      "epoch": 0.6808825252406905,
-      "grad_norm": 548.2672729492188,
-      "learning_rate": 1.4504192635559406e-05,
-      "loss": 60.145,
-      "step": 168530
-    },
-    {
-      "epoch": 0.6809229265060581,
-      "grad_norm": 212.44097900390625,
-      "learning_rate": 1.4501024610290658e-05,
-      "loss": 42.0871,
-      "step": 168540
-    },
-    {
-      "epoch": 0.6809633277714258,
-      "grad_norm": 632.0999755859375,
-      "learning_rate": 1.4497856789704844e-05,
-      "loss": 48.5092,
-      "step": 168550
-    },
-    {
-      "epoch": 0.6810037290367934,
-      "grad_norm": 508.2789611816406,
-      "learning_rate": 1.4494689173863726e-05,
-      "loss": 44.463,
-      "step": 168560
-    },
-    {
-      "epoch": 0.681044130302161,
-      "grad_norm": 684.1690063476562,
-      "learning_rate": 1.4491521762829034e-05,
-      "loss": 47.4132,
-      "step": 168570
-    },
-    {
-      "epoch": 0.6810845315675287,
-      "grad_norm": 519.2853393554688,
-      "learning_rate": 1.4488354556662554e-05,
-      "loss": 48.8163,
-      "step": 168580
-    },
-    {
-      "epoch": 0.6811249328328963,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4485187555426005e-05,
-      "loss": 36.66,
-      "step": 168590
-    },
-    {
-      "epoch": 0.681165334098264,
-      "grad_norm": 1602.7708740234375,
-      "learning_rate": 1.4482020759181135e-05,
-      "loss": 54.6894,
-      "step": 168600
-    },
-    {
-      "epoch": 0.6812057353636316,
-      "grad_norm": 423.5799255371094,
-      "learning_rate": 1.4478854167989687e-05,
-      "loss": 50.2314,
-      "step": 168610
-    },
-    {
-      "epoch": 0.6812461366289992,
-      "grad_norm": 447.10101318359375,
-      "learning_rate": 1.4475687781913394e-05,
-      "loss": 47.8248,
-      "step": 168620
-    },
-    {
-      "epoch": 0.6812865378943669,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4472521601013995e-05,
-      "loss": 40.6153,
-      "step": 168630
-    },
-    {
-      "epoch": 0.6813269391597345,
-      "grad_norm": 577.5380859375,
-      "learning_rate": 1.4469355625353198e-05,
-      "loss": 48.3852,
-      "step": 168640
-    },
-    {
-      "epoch": 0.6813673404251022,
-      "grad_norm": 1142.417724609375,
-      "learning_rate": 1.4466189854992735e-05,
-      "loss": 52.0295,
-      "step": 168650
-    },
-    {
-      "epoch": 0.6814077416904697,
-      "grad_norm": 444.90838623046875,
-      "learning_rate": 1.4463024289994322e-05,
-      "loss": 56.2271,
-      "step": 168660
-    },
-    {
-      "epoch": 0.6814481429558373,
-      "grad_norm": 688.0332641601562,
-      "learning_rate": 1.4459858930419689e-05,
-      "loss": 56.1599,
-      "step": 168670
-    },
-    {
-      "epoch": 0.681488544221205,
-      "grad_norm": 634.889404296875,
-      "learning_rate": 1.445669377633051e-05,
-      "loss": 51.8216,
-      "step": 168680
-    },
-    {
-      "epoch": 0.6815289454865726,
-      "grad_norm": 318.47491455078125,
-      "learning_rate": 1.4453528827788531e-05,
-      "loss": 38.8029,
-      "step": 168690
-    },
-    {
-      "epoch": 0.6815693467519403,
-      "grad_norm": 428.7381286621094,
-      "learning_rate": 1.4450364084855433e-05,
-      "loss": 33.6186,
-      "step": 168700
-    },
-    {
-      "epoch": 0.6816097480173079,
-      "grad_norm": 1699.4671630859375,
-      "learning_rate": 1.4447199547592916e-05,
-      "loss": 77.7904,
-      "step": 168710
-    },
-    {
-      "epoch": 0.6816501492826755,
-      "grad_norm": 631.262451171875,
-      "learning_rate": 1.4444035216062684e-05,
-      "loss": 83.7363,
-      "step": 168720
-    },
-    {
-      "epoch": 0.6816905505480432,
-      "grad_norm": 686.1653442382812,
-      "learning_rate": 1.4440871090326404e-05,
-      "loss": 37.8094,
-      "step": 168730
-    },
-    {
-      "epoch": 0.6817309518134108,
-      "grad_norm": 547.116943359375,
-      "learning_rate": 1.4437707170445797e-05,
-      "loss": 51.7289,
-      "step": 168740
-    },
-    {
-      "epoch": 0.6817713530787785,
-      "grad_norm": 458.6316223144531,
-      "learning_rate": 1.443454345648252e-05,
-      "loss": 45.8365,
-      "step": 168750
-    },
-    {
-      "epoch": 0.6818117543441461,
-      "grad_norm": 436.25347900390625,
-      "learning_rate": 1.4431379948498253e-05,
-      "loss": 42.5683,
-      "step": 168760
-    },
-    {
-      "epoch": 0.6818521556095137,
-      "grad_norm": 293.4028015136719,
-      "learning_rate": 1.4428216646554676e-05,
-      "loss": 45.4803,
-      "step": 168770
-    },
-    {
-      "epoch": 0.6818925568748814,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4425053550713458e-05,
-      "loss": 41.6429,
-      "step": 168780
-    },
-    {
-      "epoch": 0.6819329581402489,
-      "grad_norm": 1097.0491943359375,
-      "learning_rate": 1.4421890661036275e-05,
-      "loss": 54.7557,
-      "step": 168790
-    },
-    {
-      "epoch": 0.6819733594056165,
-      "grad_norm": 2632.930419921875,
-      "learning_rate": 1.4418727977584774e-05,
-      "loss": 50.7967,
-      "step": 168800
-    },
-    {
-      "epoch": 0.6820137606709842,
-      "grad_norm": 806.8290405273438,
-      "learning_rate": 1.4415565500420613e-05,
-      "loss": 49.4605,
-      "step": 168810
-    },
-    {
-      "epoch": 0.6820541619363518,
-      "grad_norm": 1056.539794921875,
-      "learning_rate": 1.4412403229605454e-05,
-      "loss": 55.0219,
-      "step": 168820
-    },
-    {
-      "epoch": 0.6820945632017195,
-      "grad_norm": 464.64453125,
-      "learning_rate": 1.4409241165200954e-05,
-      "loss": 62.6499,
-      "step": 168830
-    },
-    {
-      "epoch": 0.6821349644670871,
-      "grad_norm": 978.5242919921875,
-      "learning_rate": 1.4406079307268734e-05,
-      "loss": 45.2906,
-      "step": 168840
-    },
-    {
-      "epoch": 0.6821753657324547,
-      "grad_norm": 529.989990234375,
-      "learning_rate": 1.4402917655870466e-05,
-      "loss": 33.3289,
-      "step": 168850
-    },
-    {
-      "epoch": 0.6822157669978224,
-      "grad_norm": 556.6727294921875,
-      "learning_rate": 1.4399756211067766e-05,
-      "loss": 35.7993,
-      "step": 168860
-    },
-    {
-      "epoch": 0.68225616826319,
-      "grad_norm": 943.91357421875,
-      "learning_rate": 1.4396594972922278e-05,
-      "loss": 42.8627,
-      "step": 168870
-    },
-    {
-      "epoch": 0.6822965695285577,
-      "grad_norm": 813.277099609375,
-      "learning_rate": 1.4393433941495637e-05,
-      "loss": 48.5582,
-      "step": 168880
-    },
-    {
-      "epoch": 0.6823369707939253,
-      "grad_norm": 795.2565307617188,
-      "learning_rate": 1.4390273116849445e-05,
-      "loss": 47.6735,
-      "step": 168890
-    },
-    {
-      "epoch": 0.6823773720592929,
-      "grad_norm": 1392.483642578125,
-      "learning_rate": 1.438711249904536e-05,
-      "loss": 55.0963,
-      "step": 168900
-    },
-    {
-      "epoch": 0.6824177733246606,
-      "grad_norm": 1057.4271240234375,
-      "learning_rate": 1.438395208814497e-05,
-      "loss": 63.7595,
-      "step": 168910
-    },
-    {
-      "epoch": 0.6824581745900281,
-      "grad_norm": 1027.0804443359375,
-      "learning_rate": 1.43807918842099e-05,
-      "loss": 47.1839,
-      "step": 168920
-    },
-    {
-      "epoch": 0.6824985758553958,
-      "grad_norm": 1541.439453125,
-      "learning_rate": 1.437763188730176e-05,
-      "loss": 56.8558,
-      "step": 168930
-    },
-    {
-      "epoch": 0.6825389771207634,
-      "grad_norm": 744.2750854492188,
-      "learning_rate": 1.4374472097482155e-05,
-      "loss": 48.9892,
-      "step": 168940
-    },
-    {
-      "epoch": 0.682579378386131,
-      "grad_norm": 217.60009765625,
-      "learning_rate": 1.4371312514812685e-05,
-      "loss": 49.317,
-      "step": 168950
-    },
-    {
-      "epoch": 0.6826197796514987,
-      "grad_norm": 938.9303588867188,
-      "learning_rate": 1.4368153139354962e-05,
-      "loss": 36.7838,
-      "step": 168960
-    },
-    {
-      "epoch": 0.6826601809168663,
-      "grad_norm": 616.41748046875,
-      "learning_rate": 1.4364993971170553e-05,
-      "loss": 47.57,
-      "step": 168970
-    },
-    {
-      "epoch": 0.682700582182234,
-      "grad_norm": 393.9440002441406,
-      "learning_rate": 1.4361835010321067e-05,
-      "loss": 59.8542,
-      "step": 168980
-    },
-    {
-      "epoch": 0.6827409834476016,
-      "grad_norm": 1040.7830810546875,
-      "learning_rate": 1.435867625686808e-05,
-      "loss": 46.2593,
-      "step": 168990
-    },
-    {
-      "epoch": 0.6827813847129692,
-      "grad_norm": 1211.5755615234375,
-      "learning_rate": 1.4355517710873184e-05,
-      "loss": 63.6152,
-      "step": 169000
-    },
-    {
-      "epoch": 0.6828217859783369,
-      "grad_norm": 734.5034790039062,
-      "learning_rate": 1.4352359372397955e-05,
-      "loss": 48.3726,
-      "step": 169010
-    },
-    {
-      "epoch": 0.6828621872437045,
-      "grad_norm": 933.177001953125,
-      "learning_rate": 1.4349201241503943e-05,
-      "loss": 72.7957,
-      "step": 169020
-    },
-    {
-      "epoch": 0.6829025885090722,
-      "grad_norm": 532.6458129882812,
-      "learning_rate": 1.4346043318252756e-05,
-      "loss": 47.9236,
-      "step": 169030
-    },
-    {
-      "epoch": 0.6829429897744398,
-      "grad_norm": 795.7175903320312,
-      "learning_rate": 1.434288560270593e-05,
-      "loss": 43.2592,
-      "step": 169040
-    },
-    {
-      "epoch": 0.6829833910398073,
-      "grad_norm": 717.64697265625,
-      "learning_rate": 1.4339728094925037e-05,
-      "loss": 62.2802,
-      "step": 169050
-    },
-    {
-      "epoch": 0.683023792305175,
-      "grad_norm": 1106.865234375,
-      "learning_rate": 1.4336570794971643e-05,
-      "loss": 62.9854,
-      "step": 169060
-    },
-    {
-      "epoch": 0.6830641935705426,
-      "grad_norm": 452.23504638671875,
-      "learning_rate": 1.4333413702907267e-05,
-      "loss": 54.1656,
-      "step": 169070
-    },
-    {
-      "epoch": 0.6831045948359102,
-      "grad_norm": 1227.8212890625,
-      "learning_rate": 1.4330256818793508e-05,
-      "loss": 49.5026,
-      "step": 169080
-    },
-    {
-      "epoch": 0.6831449961012779,
-      "grad_norm": 637.03466796875,
-      "learning_rate": 1.4327100142691874e-05,
-      "loss": 41.3022,
-      "step": 169090
-    },
-    {
-      "epoch": 0.6831853973666455,
-      "grad_norm": 551.9191284179688,
-      "learning_rate": 1.4323943674663914e-05,
-      "loss": 50.8832,
-      "step": 169100
-    },
-    {
-      "epoch": 0.6832257986320132,
-      "grad_norm": 1289.005859375,
-      "learning_rate": 1.432078741477117e-05,
-      "loss": 45.6479,
-      "step": 169110
-    },
-    {
-      "epoch": 0.6832661998973808,
-      "grad_norm": 875.027099609375,
-      "learning_rate": 1.4317631363075184e-05,
-      "loss": 51.9001,
-      "step": 169120
-    },
-    {
-      "epoch": 0.6833066011627484,
-      "grad_norm": 655.9733276367188,
-      "learning_rate": 1.4314475519637466e-05,
-      "loss": 65.7755,
-      "step": 169130
-    },
-    {
-      "epoch": 0.6833470024281161,
-      "grad_norm": 248.5944061279297,
-      "learning_rate": 1.4311319884519547e-05,
-      "loss": 59.2009,
-      "step": 169140
-    },
-    {
-      "epoch": 0.6833874036934837,
-      "grad_norm": 251.61972045898438,
-      "learning_rate": 1.4308164457782952e-05,
-      "loss": 36.2274,
-      "step": 169150
-    },
-    {
-      "epoch": 0.6834278049588514,
-      "grad_norm": 244.47169494628906,
-      "learning_rate": 1.4305009239489192e-05,
-      "loss": 40.4156,
-      "step": 169160
-    },
-    {
-      "epoch": 0.6834682062242189,
-      "grad_norm": 425.4827575683594,
-      "learning_rate": 1.4301854229699796e-05,
-      "loss": 34.5485,
-      "step": 169170
-    },
-    {
-      "epoch": 0.6835086074895865,
-      "grad_norm": 579.6253662109375,
-      "learning_rate": 1.4298699428476236e-05,
-      "loss": 35.4702,
-      "step": 169180
-    },
-    {
-      "epoch": 0.6835490087549542,
-      "grad_norm": 1097.84716796875,
-      "learning_rate": 1.4295544835880065e-05,
-      "loss": 56.0326,
-      "step": 169190
-    },
-    {
-      "epoch": 0.6835894100203218,
-      "grad_norm": 1029.4461669921875,
-      "learning_rate": 1.4292390451972745e-05,
-      "loss": 36.3401,
-      "step": 169200
-    },
-    {
-      "epoch": 0.6836298112856894,
-      "grad_norm": 428.86077880859375,
-      "learning_rate": 1.4289236276815787e-05,
-      "loss": 59.5827,
-      "step": 169210
-    },
-    {
-      "epoch": 0.6836702125510571,
-      "grad_norm": 933.2636108398438,
-      "learning_rate": 1.4286082310470692e-05,
-      "loss": 47.7912,
-      "step": 169220
-    },
-    {
-      "epoch": 0.6837106138164247,
-      "grad_norm": 396.597412109375,
-      "learning_rate": 1.428292855299892e-05,
-      "loss": 49.1154,
-      "step": 169230
-    },
-    {
-      "epoch": 0.6837510150817924,
-      "grad_norm": 856.8510131835938,
-      "learning_rate": 1.427977500446199e-05,
-      "loss": 77.9437,
-      "step": 169240
-    },
-    {
-      "epoch": 0.68379141634716,
-      "grad_norm": 612.7291259765625,
-      "learning_rate": 1.4276621664921357e-05,
-      "loss": 50.0896,
-      "step": 169250
-    },
-    {
-      "epoch": 0.6838318176125276,
-      "grad_norm": 544.8532104492188,
-      "learning_rate": 1.4273468534438505e-05,
-      "loss": 55.9285,
-      "step": 169260
-    },
-    {
-      "epoch": 0.6838722188778953,
-      "grad_norm": 466.5942077636719,
-      "learning_rate": 1.4270315613074906e-05,
-      "loss": 46.574,
-      "step": 169270
-    },
-    {
-      "epoch": 0.6839126201432629,
-      "grad_norm": 1225.802978515625,
-      "learning_rate": 1.426716290089204e-05,
-      "loss": 32.1837,
-      "step": 169280
-    },
-    {
-      "epoch": 0.6839530214086306,
-      "grad_norm": 1749.95556640625,
-      "learning_rate": 1.4264010397951335e-05,
-      "loss": 49.6508,
-      "step": 169290
-    },
-    {
-      "epoch": 0.6839934226739981,
-      "grad_norm": 439.6357727050781,
-      "learning_rate": 1.4260858104314297e-05,
-      "loss": 46.1547,
-      "step": 169300
-    },
-    {
-      "epoch": 0.6840338239393657,
-      "grad_norm": 879.0947875976562,
-      "learning_rate": 1.425770602004235e-05,
-      "loss": 46.7996,
-      "step": 169310
-    },
-    {
-      "epoch": 0.6840742252047334,
-      "grad_norm": 616.0336303710938,
-      "learning_rate": 1.4254554145196953e-05,
-      "loss": 51.1238,
-      "step": 169320
-    },
-    {
-      "epoch": 0.684114626470101,
-      "grad_norm": 2811.0615234375,
-      "learning_rate": 1.4251402479839564e-05,
-      "loss": 37.678,
-      "step": 169330
-    },
-    {
-      "epoch": 0.6841550277354687,
-      "grad_norm": 554.9786987304688,
-      "learning_rate": 1.42482510240316e-05,
-      "loss": 49.2597,
-      "step": 169340
-    },
-    {
-      "epoch": 0.6841954290008363,
-      "grad_norm": 344.1712341308594,
-      "learning_rate": 1.4245099777834539e-05,
-      "loss": 71.6305,
-      "step": 169350
-    },
-    {
-      "epoch": 0.6842358302662039,
-      "grad_norm": 464.71051025390625,
-      "learning_rate": 1.4241948741309782e-05,
-      "loss": 54.1099,
-      "step": 169360
-    },
-    {
-      "epoch": 0.6842762315315716,
-      "grad_norm": 1372.66064453125,
-      "learning_rate": 1.4238797914518776e-05,
-      "loss": 56.0686,
-      "step": 169370
-    },
-    {
-      "epoch": 0.6843166327969392,
-      "grad_norm": 504.1880798339844,
-      "learning_rate": 1.4235647297522942e-05,
-      "loss": 44.2797,
-      "step": 169380
-    },
-    {
-      "epoch": 0.6843570340623069,
-      "grad_norm": 579.6629638671875,
-      "learning_rate": 1.4232496890383706e-05,
-      "loss": 97.3262,
-      "step": 169390
-    },
-    {
-      "epoch": 0.6843974353276745,
-      "grad_norm": 648.5736694335938,
-      "learning_rate": 1.42293466931625e-05,
-      "loss": 63.3319,
-      "step": 169400
-    },
-    {
-      "epoch": 0.6844378365930421,
-      "grad_norm": 450.7187194824219,
-      "learning_rate": 1.4226196705920708e-05,
-      "loss": 73.6804,
-      "step": 169410
-    },
-    {
-      "epoch": 0.6844782378584098,
-      "grad_norm": 456.9446716308594,
-      "learning_rate": 1.4223046928719763e-05,
-      "loss": 48.2182,
-      "step": 169420
-    },
-    {
-      "epoch": 0.6845186391237773,
-      "grad_norm": 623.4013671875,
-      "learning_rate": 1.4219897361621065e-05,
-      "loss": 72.1421,
-      "step": 169430
-    },
-    {
-      "epoch": 0.684559040389145,
-      "grad_norm": 421.9169006347656,
-      "learning_rate": 1.421674800468603e-05,
-      "loss": 28.4017,
-      "step": 169440
-    },
-    {
-      "epoch": 0.6845994416545126,
-      "grad_norm": 623.2809448242188,
-      "learning_rate": 1.4213598857976024e-05,
-      "loss": 45.7994,
-      "step": 169450
-    },
-    {
-      "epoch": 0.6846398429198802,
-      "grad_norm": 431.9721984863281,
-      "learning_rate": 1.421044992155248e-05,
-      "loss": 35.0073,
-      "step": 169460
-    },
-    {
-      "epoch": 0.6846802441852479,
-      "grad_norm": 415.11993408203125,
-      "learning_rate": 1.4207301195476757e-05,
-      "loss": 52.3375,
-      "step": 169470
-    },
-    {
-      "epoch": 0.6847206454506155,
-      "grad_norm": 1272.1280517578125,
-      "learning_rate": 1.4204152679810258e-05,
-      "loss": 51.1304,
-      "step": 169480
-    },
-    {
-      "epoch": 0.6847610467159831,
-      "grad_norm": 571.7431030273438,
-      "learning_rate": 1.4201004374614369e-05,
-      "loss": 75.1111,
-      "step": 169490
-    },
-    {
-      "epoch": 0.6848014479813508,
-      "grad_norm": 519.0687255859375,
-      "learning_rate": 1.4197856279950438e-05,
-      "loss": 25.3569,
-      "step": 169500
-    },
-    {
-      "epoch": 0.6848418492467184,
-      "grad_norm": 544.158203125,
-      "learning_rate": 1.4194708395879886e-05,
-      "loss": 63.012,
-      "step": 169510
-    },
-    {
-      "epoch": 0.6848822505120861,
-      "grad_norm": 387.5862121582031,
-      "learning_rate": 1.4191560722464032e-05,
-      "loss": 34.8872,
-      "step": 169520
-    },
-    {
-      "epoch": 0.6849226517774537,
-      "grad_norm": 666.9616088867188,
-      "learning_rate": 1.4188413259764285e-05,
-      "loss": 42.9067,
-      "step": 169530
-    },
-    {
-      "epoch": 0.6849630530428213,
-      "grad_norm": 0.0,
-      "learning_rate": 1.418526600784198e-05,
-      "loss": 37.349,
-      "step": 169540
-    },
-    {
-      "epoch": 0.685003454308189,
-      "grad_norm": 418.4957580566406,
-      "learning_rate": 1.4182118966758481e-05,
-      "loss": 42.9753,
-      "step": 169550
-    },
-    {
-      "epoch": 0.6850438555735565,
-      "grad_norm": 254.7620849609375,
-      "learning_rate": 1.4178972136575153e-05,
-      "loss": 42.3305,
-      "step": 169560
-    },
-    {
-      "epoch": 0.6850842568389242,
-      "grad_norm": 1166.2672119140625,
-      "learning_rate": 1.4175825517353325e-05,
-      "loss": 51.9849,
-      "step": 169570
-    },
-    {
-      "epoch": 0.6851246581042918,
-      "grad_norm": 608.7311401367188,
-      "learning_rate": 1.4172679109154349e-05,
-      "loss": 38.9989,
-      "step": 169580
-    },
-    {
-      "epoch": 0.6851650593696594,
-      "grad_norm": 354.56768798828125,
-      "learning_rate": 1.4169532912039568e-05,
-      "loss": 40.5726,
-      "step": 169590
-    },
-    {
-      "epoch": 0.6852054606350271,
-      "grad_norm": 324.2098693847656,
-      "learning_rate": 1.4166386926070322e-05,
-      "loss": 46.7577,
-      "step": 169600
-    },
-    {
-      "epoch": 0.6852458619003947,
-      "grad_norm": 484.006591796875,
-      "learning_rate": 1.4163241151307937e-05,
-      "loss": 40.9271,
-      "step": 169610
-    },
-    {
-      "epoch": 0.6852862631657624,
-      "grad_norm": 694.5880737304688,
-      "learning_rate": 1.4160095587813757e-05,
-      "loss": 45.7319,
-      "step": 169620
-    },
-    {
-      "epoch": 0.68532666443113,
-      "grad_norm": 252.2886962890625,
-      "learning_rate": 1.4156950235649074e-05,
-      "loss": 63.6754,
-      "step": 169630
-    },
-    {
-      "epoch": 0.6853670656964976,
-      "grad_norm": 233.86050415039062,
-      "learning_rate": 1.4153805094875248e-05,
-      "loss": 56.9261,
-      "step": 169640
-    },
-    {
-      "epoch": 0.6854074669618653,
-      "grad_norm": 710.7440795898438,
-      "learning_rate": 1.4150660165553564e-05,
-      "loss": 52.6019,
-      "step": 169650
-    },
-    {
-      "epoch": 0.6854478682272329,
-      "grad_norm": 1143.4447021484375,
-      "learning_rate": 1.4147515447745349e-05,
-      "loss": 56.2893,
-      "step": 169660
-    },
-    {
-      "epoch": 0.6854882694926006,
-      "grad_norm": 590.0914306640625,
-      "learning_rate": 1.4144370941511915e-05,
-      "loss": 57.5833,
-      "step": 169670
-    },
-    {
-      "epoch": 0.6855286707579682,
-      "grad_norm": 617.8530883789062,
-      "learning_rate": 1.4141226646914541e-05,
-      "loss": 49.1148,
-      "step": 169680
-    },
-    {
-      "epoch": 0.6855690720233357,
-      "grad_norm": 1288.4437255859375,
-      "learning_rate": 1.413808256401456e-05,
-      "loss": 56.3085,
-      "step": 169690
-    },
-    {
-      "epoch": 0.6856094732887034,
-      "grad_norm": 1510.7669677734375,
-      "learning_rate": 1.4134938692873245e-05,
-      "loss": 50.2129,
-      "step": 169700
-    },
-    {
-      "epoch": 0.685649874554071,
-      "grad_norm": 490.9001159667969,
-      "learning_rate": 1.4131795033551892e-05,
-      "loss": 53.7312,
-      "step": 169710
-    },
-    {
-      "epoch": 0.6856902758194386,
-      "grad_norm": 1059.6722412109375,
-      "learning_rate": 1.412865158611179e-05,
-      "loss": 50.0878,
-      "step": 169720
-    },
-    {
-      "epoch": 0.6857306770848063,
-      "grad_norm": 717.5321655273438,
-      "learning_rate": 1.4125508350614236e-05,
-      "loss": 41.077,
-      "step": 169730
-    },
-    {
-      "epoch": 0.6857710783501739,
-      "grad_norm": 368.7514343261719,
-      "learning_rate": 1.412236532712048e-05,
-      "loss": 59.4592,
-      "step": 169740
-    },
-    {
-      "epoch": 0.6858114796155416,
-      "grad_norm": 1239.5657958984375,
-      "learning_rate": 1.4119222515691816e-05,
-      "loss": 45.6636,
-      "step": 169750
-    },
-    {
-      "epoch": 0.6858518808809092,
-      "grad_norm": 589.4757690429688,
-      "learning_rate": 1.4116079916389507e-05,
-      "loss": 54.7634,
-      "step": 169760
-    },
-    {
-      "epoch": 0.6858922821462768,
-      "grad_norm": 566.4857177734375,
-      "learning_rate": 1.4112937529274828e-05,
-      "loss": 40.5211,
-      "step": 169770
-    },
-    {
-      "epoch": 0.6859326834116445,
-      "grad_norm": 991.4258422851562,
-      "learning_rate": 1.4109795354409044e-05,
-      "loss": 36.051,
-      "step": 169780
-    },
-    {
-      "epoch": 0.6859730846770121,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4106653391853386e-05,
-      "loss": 43.4424,
-      "step": 169790
-    },
-    {
-      "epoch": 0.6860134859423798,
-      "grad_norm": 680.9237670898438,
-      "learning_rate": 1.4103511641669152e-05,
-      "loss": 52.9324,
-      "step": 169800
-    },
-    {
-      "epoch": 0.6860538872077473,
-      "grad_norm": 287.92718505859375,
-      "learning_rate": 1.4100370103917554e-05,
-      "loss": 30.8766,
-      "step": 169810
-    },
-    {
-      "epoch": 0.6860942884731149,
-      "grad_norm": 353.46734619140625,
-      "learning_rate": 1.4097228778659854e-05,
-      "loss": 31.8396,
-      "step": 169820
-    },
-    {
-      "epoch": 0.6861346897384826,
-      "grad_norm": 1042.081298828125,
-      "learning_rate": 1.4094087665957304e-05,
-      "loss": 67.0164,
-      "step": 169830
-    },
-    {
-      "epoch": 0.6861750910038502,
-      "grad_norm": 489.08367919921875,
-      "learning_rate": 1.4090946765871104e-05,
-      "loss": 60.9808,
-      "step": 169840
-    },
-    {
-      "epoch": 0.6862154922692179,
-      "grad_norm": 874.5740356445312,
-      "learning_rate": 1.408780607846254e-05,
-      "loss": 50.7961,
-      "step": 169850
-    },
-    {
-      "epoch": 0.6862558935345855,
-      "grad_norm": 740.3602905273438,
-      "learning_rate": 1.4084665603792802e-05,
-      "loss": 42.8049,
-      "step": 169860
-    },
-    {
-      "epoch": 0.6862962947999531,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4081525341923127e-05,
-      "loss": 42.2607,
-      "step": 169870
-    },
-    {
-      "epoch": 0.6863366960653208,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4078385292914736e-05,
-      "loss": 40.3975,
-      "step": 169880
-    },
-    {
-      "epoch": 0.6863770973306884,
-      "grad_norm": 1021.9735717773438,
-      "learning_rate": 1.4075245456828857e-05,
-      "loss": 42.3641,
-      "step": 169890
-    },
-    {
-      "epoch": 0.686417498596056,
-      "grad_norm": 699.235595703125,
-      "learning_rate": 1.4072105833726684e-05,
-      "loss": 50.5739,
-      "step": 169900
-    },
-    {
-      "epoch": 0.6864578998614237,
-      "grad_norm": 343.7049560546875,
-      "learning_rate": 1.4068966423669433e-05,
-      "loss": 37.3967,
-      "step": 169910
-    },
-    {
-      "epoch": 0.6864983011267913,
-      "grad_norm": 680.474853515625,
-      "learning_rate": 1.406582722671831e-05,
-      "loss": 49.679,
-      "step": 169920
-    },
-    {
-      "epoch": 0.686538702392159,
-      "grad_norm": 1111.1092529296875,
-      "learning_rate": 1.406268824293451e-05,
-      "loss": 57.0753,
-      "step": 169930
-    },
-    {
-      "epoch": 0.6865791036575265,
-      "grad_norm": 523.4867553710938,
-      "learning_rate": 1.4059549472379247e-05,
-      "loss": 44.4473,
-      "step": 169940
-    },
-    {
-      "epoch": 0.6866195049228941,
-      "grad_norm": 776.1953125,
-      "learning_rate": 1.405641091511368e-05,
-      "loss": 53.1783,
-      "step": 169950
-    },
-    {
-      "epoch": 0.6866599061882618,
-      "grad_norm": 712.0708618164062,
-      "learning_rate": 1.4053272571199036e-05,
-      "loss": 95.6826,
-      "step": 169960
-    },
-    {
-      "epoch": 0.6867003074536294,
-      "grad_norm": 309.5819396972656,
-      "learning_rate": 1.405013444069647e-05,
-      "loss": 54.1302,
-      "step": 169970
-    },
-    {
-      "epoch": 0.6867407087189971,
-      "grad_norm": 682.5097045898438,
-      "learning_rate": 1.4046996523667166e-05,
-      "loss": 63.7726,
-      "step": 169980
-    },
-    {
-      "epoch": 0.6867811099843647,
-      "grad_norm": 0.0,
-      "learning_rate": 1.4043858820172309e-05,
-      "loss": 42.362,
-      "step": 169990
-    },
-    {
-      "epoch": 0.6868215112497323,
-      "grad_norm": 384.0200500488281,
-      "learning_rate": 1.4040721330273062e-05,
-      "loss": 49.3784,
-      "step": 170000
-    },
-    {
-      "epoch": 0.6868619125151,
-      "grad_norm": 550.7702026367188,
-      "learning_rate": 1.4037584054030606e-05,
-      "loss": 50.9029,
-      "step": 170010
-    },
-    {
-      "epoch": 0.6869023137804676,
-      "grad_norm": 1652.8741455078125,
-      "learning_rate": 1.4034446991506083e-05,
-      "loss": 54.4098,
-      "step": 170020
-    },
-    {
-      "epoch": 0.6869427150458353,
-      "grad_norm": 1165.6163330078125,
-      "learning_rate": 1.4031310142760662e-05,
-      "loss": 43.6158,
-      "step": 170030
-    },
-    {
-      "epoch": 0.6869831163112029,
-      "grad_norm": 1255.7396240234375,
-      "learning_rate": 1.4028173507855493e-05,
-      "loss": 56.6457,
-      "step": 170040
-    },
-    {
-      "epoch": 0.6870235175765705,
-      "grad_norm": 379.06463623046875,
-      "learning_rate": 1.4025037086851733e-05,
-      "loss": 51.2328,
-      "step": 170050
-    },
-    {
-      "epoch": 0.6870639188419382,
-      "grad_norm": 446.58673095703125,
-      "learning_rate": 1.4021900879810526e-05,
-      "loss": 44.2981,
-      "step": 170060
-    },
-    {
-      "epoch": 0.6871043201073057,
-      "grad_norm": 889.757080078125,
-      "learning_rate": 1.4018764886793023e-05,
-      "loss": 51.626,
-      "step": 170070
-    },
-    {
-      "epoch": 0.6871447213726734,
-      "grad_norm": 544.6858520507812,
-      "learning_rate": 1.401562910786034e-05,
-      "loss": 42.926,
-      "step": 170080
-    },
-    {
-      "epoch": 0.687185122638041,
-      "grad_norm": 575.81884765625,
-      "learning_rate": 1.4012493543073623e-05,
-      "loss": 46.7048,
-      "step": 170090
-    },
-    {
-      "epoch": 0.6872255239034086,
-      "grad_norm": 798.9896850585938,
-      "learning_rate": 1.4009358192494016e-05,
-      "loss": 43.1169,
-      "step": 170100
-    },
-    {
-      "epoch": 0.6872659251687763,
-      "grad_norm": 314.7311096191406,
-      "learning_rate": 1.4006223056182604e-05,
-      "loss": 46.7615,
-      "step": 170110
-    },
-    {
-      "epoch": 0.6873063264341439,
-      "grad_norm": 587.7617797851562,
-      "learning_rate": 1.4003088134200553e-05,
-      "loss": 41.9043,
-      "step": 170120
-    },
-    {
-      "epoch": 0.6873467276995116,
-      "grad_norm": 593.9027099609375,
-      "learning_rate": 1.3999953426608941e-05,
-      "loss": 52.3961,
-      "step": 170130
-    },
-    {
-      "epoch": 0.6873871289648792,
-      "grad_norm": 1898.5985107421875,
-      "learning_rate": 1.3996818933468925e-05,
-      "loss": 67.1244,
-      "step": 170140
-    },
-    {
-      "epoch": 0.6874275302302468,
-      "grad_norm": 993.4326171875,
-      "learning_rate": 1.3993684654841574e-05,
-      "loss": 46.3975,
-      "step": 170150
-    },
-    {
-      "epoch": 0.6874679314956145,
-      "grad_norm": 220.27293395996094,
-      "learning_rate": 1.399055059078801e-05,
-      "loss": 40.9544,
-      "step": 170160
-    },
-    {
-      "epoch": 0.6875083327609821,
-      "grad_norm": 702.5592651367188,
-      "learning_rate": 1.3987416741369336e-05,
-      "loss": 38.6786,
-      "step": 170170
-    },
-    {
-      "epoch": 0.6875487340263498,
-      "grad_norm": 484.0182800292969,
-      "learning_rate": 1.3984283106646636e-05,
-      "loss": 79.0217,
-      "step": 170180
-    },
-    {
-      "epoch": 0.6875891352917174,
-      "grad_norm": 1016.0274658203125,
-      "learning_rate": 1.3981149686681008e-05,
-      "loss": 60.7541,
-      "step": 170190
-    },
-    {
-      "epoch": 0.6876295365570849,
-      "grad_norm": 781.2263793945312,
-      "learning_rate": 1.397801648153354e-05,
-      "loss": 40.7296,
-      "step": 170200
-    },
-    {
-      "epoch": 0.6876699378224526,
-      "grad_norm": 935.5984497070312,
-      "learning_rate": 1.3974883491265312e-05,
-      "loss": 60.1782,
-      "step": 170210
-    },
-    {
-      "epoch": 0.6877103390878202,
-      "grad_norm": 721.454345703125,
-      "learning_rate": 1.397175071593741e-05,
-      "loss": 62.3025,
-      "step": 170220
-    },
-    {
-      "epoch": 0.6877507403531878,
-      "grad_norm": 550.3572998046875,
-      "learning_rate": 1.3968618155610913e-05,
-      "loss": 50.8357,
-      "step": 170230
-    },
-    {
-      "epoch": 0.6877911416185555,
-      "grad_norm": 467.0953674316406,
-      "learning_rate": 1.3965485810346874e-05,
-      "loss": 48.71,
-      "step": 170240
-    },
-    {
-      "epoch": 0.6878315428839231,
-      "grad_norm": 613.0184326171875,
-      "learning_rate": 1.3962353680206373e-05,
-      "loss": 58.2839,
-      "step": 170250
-    },
-    {
-      "epoch": 0.6878719441492908,
-      "grad_norm": 8435.68359375,
-      "learning_rate": 1.3959221765250469e-05,
-      "loss": 84.0615,
-      "step": 170260
-    },
-    {
-      "epoch": 0.6879123454146584,
-      "grad_norm": 1174.869384765625,
-      "learning_rate": 1.3956090065540217e-05,
-      "loss": 50.7304,
-      "step": 170270
-    },
-    {
-      "epoch": 0.687952746680026,
-      "grad_norm": 838.7512817382812,
-      "learning_rate": 1.3952958581136688e-05,
-      "loss": 43.5792,
-      "step": 170280
-    },
-    {
-      "epoch": 0.6879931479453937,
-      "grad_norm": 585.5516357421875,
-      "learning_rate": 1.39498273121009e-05,
-      "loss": 46.7663,
-      "step": 170290
-    },
-    {
-      "epoch": 0.6880335492107613,
-      "grad_norm": 595.8783569335938,
-      "learning_rate": 1.3946696258493936e-05,
-      "loss": 29.3378,
-      "step": 170300
-    },
-    {
-      "epoch": 0.688073950476129,
-      "grad_norm": 642.0225219726562,
-      "learning_rate": 1.3943565420376808e-05,
-      "loss": 30.4124,
-      "step": 170310
-    },
-    {
-      "epoch": 0.6881143517414965,
-      "grad_norm": 515.0293579101562,
-      "learning_rate": 1.3940434797810568e-05,
-      "loss": 29.4962,
-      "step": 170320
-    },
-    {
-      "epoch": 0.6881547530068641,
-      "grad_norm": 521.1478271484375,
-      "learning_rate": 1.3937304390856242e-05,
-      "loss": 47.4596,
-      "step": 170330
-    },
-    {
-      "epoch": 0.6881951542722318,
-      "grad_norm": 691.7684936523438,
-      "learning_rate": 1.3934174199574872e-05,
-      "loss": 65.3109,
-      "step": 170340
-    },
-    {
-      "epoch": 0.6882355555375994,
-      "grad_norm": 650.1378784179688,
-      "learning_rate": 1.3931044224027468e-05,
-      "loss": 50.7096,
-      "step": 170350
-    },
-    {
-      "epoch": 0.688275956802967,
-      "grad_norm": 544.4131469726562,
-      "learning_rate": 1.392791446427505e-05,
-      "loss": 40.2387,
-      "step": 170360
-    },
-    {
-      "epoch": 0.6883163580683347,
-      "grad_norm": 1095.565185546875,
-      "learning_rate": 1.3924784920378642e-05,
-      "loss": 86.7689,
-      "step": 170370
-    },
-    {
-      "epoch": 0.6883567593337023,
-      "grad_norm": 509.0849304199219,
-      "learning_rate": 1.3921655592399254e-05,
-      "loss": 52.0616,
-      "step": 170380
-    },
-    {
-      "epoch": 0.68839716059907,
-      "grad_norm": 1857.7972412109375,
-      "learning_rate": 1.3918526480397902e-05,
-      "loss": 45.2592,
-      "step": 170390
-    },
-    {
-      "epoch": 0.6884375618644376,
-      "grad_norm": 533.4299926757812,
-      "learning_rate": 1.3915397584435563e-05,
-      "loss": 49.5395,
-      "step": 170400
-    },
-    {
-      "epoch": 0.6884779631298052,
-      "grad_norm": 1264.0267333984375,
-      "learning_rate": 1.3912268904573277e-05,
-      "loss": 48.0885,
-      "step": 170410
-    },
-    {
-      "epoch": 0.6885183643951729,
-      "grad_norm": 870.8333129882812,
-      "learning_rate": 1.3909140440872004e-05,
-      "loss": 48.8337,
-      "step": 170420
-    },
-    {
-      "epoch": 0.6885587656605405,
-      "grad_norm": 267.7932434082031,
-      "learning_rate": 1.3906012193392748e-05,
-      "loss": 57.9533,
-      "step": 170430
-    },
-    {
-      "epoch": 0.6885991669259082,
-      "grad_norm": 2385.821533203125,
-      "learning_rate": 1.3902884162196508e-05,
-      "loss": 50.6058,
-      "step": 170440
-    },
-    {
-      "epoch": 0.6886395681912757,
-      "grad_norm": 1392.8212890625,
-      "learning_rate": 1.3899756347344234e-05,
-      "loss": 33.6519,
-      "step": 170450
-    },
-    {
-      "epoch": 0.6886799694566433,
-      "grad_norm": 1187.458251953125,
-      "learning_rate": 1.3896628748896945e-05,
-      "loss": 34.4566,
-      "step": 170460
-    },
-    {
-      "epoch": 0.688720370722011,
-      "grad_norm": 783.3659057617188,
-      "learning_rate": 1.3893501366915582e-05,
-      "loss": 47.9276,
-      "step": 170470
-    },
-    {
-      "epoch": 0.6887607719873786,
-      "grad_norm": 535.8975830078125,
-      "learning_rate": 1.3890374201461132e-05,
-      "loss": 36.1128,
-      "step": 170480
-    },
-    {
-      "epoch": 0.6888011732527463,
-      "grad_norm": 462.8658752441406,
-      "learning_rate": 1.3887247252594554e-05,
-      "loss": 70.0482,
-      "step": 170490
-    },
-    {
-      "epoch": 0.6888415745181139,
-      "grad_norm": 1207.6224365234375,
-      "learning_rate": 1.388412052037682e-05,
-      "loss": 50.3268,
-      "step": 170500
-    },
-    {
-      "epoch": 0.6888819757834815,
-      "grad_norm": 885.7490234375,
-      "learning_rate": 1.3880994004868872e-05,
-      "loss": 36.7898,
-      "step": 170510
-    },
-    {
-      "epoch": 0.6889223770488492,
-      "grad_norm": 971.1849975585938,
-      "learning_rate": 1.387786770613167e-05,
-      "loss": 30.5436,
-      "step": 170520
-    },
-    {
-      "epoch": 0.6889627783142168,
-      "grad_norm": 850.4481811523438,
-      "learning_rate": 1.3874741624226162e-05,
-      "loss": 40.6387,
-      "step": 170530
-    },
-    {
-      "epoch": 0.6890031795795845,
-      "grad_norm": 1244.720703125,
-      "learning_rate": 1.3871615759213297e-05,
-      "loss": 50.7725,
-      "step": 170540
-    },
-    {
-      "epoch": 0.6890435808449521,
-      "grad_norm": 444.43341064453125,
-      "learning_rate": 1.386849011115402e-05,
-      "loss": 42.5565,
-      "step": 170550
-    },
-    {
-      "epoch": 0.6890839821103197,
-      "grad_norm": 666.0784301757812,
-      "learning_rate": 1.386536468010924e-05,
-      "loss": 79.6655,
-      "step": 170560
-    },
-    {
-      "epoch": 0.6891243833756874,
-      "grad_norm": 687.4263305664062,
-      "learning_rate": 1.3862239466139926e-05,
-      "loss": 50.3608,
-      "step": 170570
-    },
-    {
-      "epoch": 0.6891647846410549,
-      "grad_norm": 901.666015625,
-      "learning_rate": 1.3859114469306977e-05,
-      "loss": 38.5541,
-      "step": 170580
-    },
-    {
-      "epoch": 0.6892051859064225,
-      "grad_norm": 721.3048095703125,
-      "learning_rate": 1.3855989689671328e-05,
-      "loss": 67.1086,
-      "step": 170590
-    },
-    {
-      "epoch": 0.6892455871717902,
-      "grad_norm": 1656.072998046875,
-      "learning_rate": 1.3852865127293902e-05,
-      "loss": 90.0558,
-      "step": 170600
-    },
-    {
-      "epoch": 0.6892859884371578,
-      "grad_norm": 631.8057861328125,
-      "learning_rate": 1.3849740782235604e-05,
-      "loss": 58.9932,
-      "step": 170610
-    },
-    {
-      "epoch": 0.6893263897025255,
-      "grad_norm": 841.3690185546875,
-      "learning_rate": 1.3846616654557359e-05,
-      "loss": 51.6805,
-      "step": 170620
-    },
-    {
-      "epoch": 0.6893667909678931,
-      "grad_norm": 746.2327270507812,
-      "learning_rate": 1.3843492744320059e-05,
-      "loss": 35.7624,
-      "step": 170630
-    },
-    {
-      "epoch": 0.6894071922332607,
-      "grad_norm": 583.4542236328125,
-      "learning_rate": 1.3840369051584612e-05,
-      "loss": 66.87,
-      "step": 170640
-    },
-    {
-      "epoch": 0.6894475934986284,
-      "grad_norm": 435.5360412597656,
-      "learning_rate": 1.3837245576411911e-05,
-      "loss": 47.8244,
-      "step": 170650
-    },
-    {
-      "epoch": 0.689487994763996,
-      "grad_norm": 462.8669738769531,
-      "learning_rate": 1.3834122318862858e-05,
-      "loss": 81.1973,
-      "step": 170660
-    },
-    {
-      "epoch": 0.6895283960293637,
-      "grad_norm": 496.9759826660156,
-      "learning_rate": 1.3830999278998335e-05,
-      "loss": 37.6189,
-      "step": 170670
-    },
-    {
-      "epoch": 0.6895687972947313,
-      "grad_norm": 1232.743408203125,
-      "learning_rate": 1.3827876456879246e-05,
-      "loss": 61.267,
-      "step": 170680
-    },
-    {
-      "epoch": 0.689609198560099,
-      "grad_norm": 288.8763732910156,
-      "learning_rate": 1.3824753852566447e-05,
-      "loss": 45.0395,
-      "step": 170690
-    },
-    {
-      "epoch": 0.6896495998254666,
-      "grad_norm": 925.8347778320312,
-      "learning_rate": 1.3821631466120821e-05,
-      "loss": 43.501,
-      "step": 170700
-    },
-    {
-      "epoch": 0.6896900010908341,
-      "grad_norm": 400.8254699707031,
-      "learning_rate": 1.381850929760326e-05,
-      "loss": 55.7822,
-      "step": 170710
-    },
-    {
-      "epoch": 0.6897304023562018,
-      "grad_norm": 523.3338623046875,
-      "learning_rate": 1.3815387347074594e-05,
-      "loss": 34.1284,
-      "step": 170720
-    },
-    {
-      "epoch": 0.6897708036215694,
-      "grad_norm": 1366.3890380859375,
-      "learning_rate": 1.3812265614595727e-05,
-      "loss": 48.7812,
-      "step": 170730
-    },
-    {
-      "epoch": 0.689811204886937,
-      "grad_norm": 1213.1644287109375,
-      "learning_rate": 1.3809144100227484e-05,
-      "loss": 46.902,
-      "step": 170740
-    },
-    {
-      "epoch": 0.6898516061523047,
-      "grad_norm": 996.2909545898438,
-      "learning_rate": 1.380602280403076e-05,
-      "loss": 89.2399,
-      "step": 170750
-    },
-    {
-      "epoch": 0.6898920074176723,
-      "grad_norm": 497.2900695800781,
-      "learning_rate": 1.3802901726066375e-05,
-      "loss": 64.0509,
-      "step": 170760
-    },
-    {
-      "epoch": 0.68993240868304,
-      "grad_norm": 583.2584838867188,
-      "learning_rate": 1.3799780866395182e-05,
-      "loss": 36.3958,
-      "step": 170770
-    },
-    {
-      "epoch": 0.6899728099484076,
-      "grad_norm": 0.0,
-      "learning_rate": 1.379666022507804e-05,
-      "loss": 44.8411,
-      "step": 170780
-    },
-    {
-      "epoch": 0.6900132112137752,
-      "grad_norm": 1305.6693115234375,
-      "learning_rate": 1.3793539802175764e-05,
-      "loss": 56.8153,
-      "step": 170790
-    },
-    {
-      "epoch": 0.6900536124791429,
-      "grad_norm": 906.8377685546875,
-      "learning_rate": 1.3790419597749199e-05,
-      "loss": 55.3632,
-      "step": 170800
-    },
-    {
-      "epoch": 0.6900940137445105,
-      "grad_norm": 580.7864990234375,
-      "learning_rate": 1.3787299611859172e-05,
-      "loss": 53.7129,
-      "step": 170810
-    },
-    {
-      "epoch": 0.6901344150098782,
-      "grad_norm": 467.542236328125,
-      "learning_rate": 1.3784179844566515e-05,
-      "loss": 55.4573,
-      "step": 170820
-    },
-    {
-      "epoch": 0.6901748162752458,
-      "grad_norm": 1431.475830078125,
-      "learning_rate": 1.3781060295932046e-05,
-      "loss": 44.2515,
-      "step": 170830
-    },
-    {
-      "epoch": 0.6902152175406133,
-      "grad_norm": 1465.8927001953125,
-      "learning_rate": 1.3777940966016592e-05,
-      "loss": 45.1074,
-      "step": 170840
-    },
-    {
-      "epoch": 0.690255618805981,
-      "grad_norm": 598.7523803710938,
-      "learning_rate": 1.3774821854880948e-05,
-      "loss": 34.8125,
-      "step": 170850
-    },
-    {
-      "epoch": 0.6902960200713486,
-      "grad_norm": 1547.8724365234375,
-      "learning_rate": 1.3771702962585928e-05,
-      "loss": 60.3331,
-      "step": 170860
-    },
-    {
-      "epoch": 0.6903364213367162,
-      "grad_norm": 894.2749633789062,
-      "learning_rate": 1.3768584289192351e-05,
-      "loss": 57.0751,
-      "step": 170870
-    },
-    {
-      "epoch": 0.6903768226020839,
-      "grad_norm": 703.2276611328125,
-      "learning_rate": 1.3765465834760988e-05,
-      "loss": 69.8606,
-      "step": 170880
-    },
-    {
-      "epoch": 0.6904172238674515,
-      "grad_norm": 684.1668701171875,
-      "learning_rate": 1.3762347599352667e-05,
-      "loss": 52.2663,
-      "step": 170890
-    },
-    {
-      "epoch": 0.6904576251328192,
-      "grad_norm": 286.8528747558594,
-      "learning_rate": 1.375922958302815e-05,
-      "loss": 33.0038,
-      "step": 170900
-    },
-    {
-      "epoch": 0.6904980263981868,
-      "grad_norm": 626.8790893554688,
-      "learning_rate": 1.3756111785848258e-05,
-      "loss": 52.7146,
-      "step": 170910
-    },
-    {
-      "epoch": 0.6905384276635544,
-      "grad_norm": 390.28375244140625,
-      "learning_rate": 1.3752994207873743e-05,
-      "loss": 44.1779,
-      "step": 170920
-    },
-    {
-      "epoch": 0.6905788289289221,
-      "grad_norm": 2262.665771484375,
-      "learning_rate": 1.37498768491654e-05,
-      "loss": 85.5712,
-      "step": 170930
-    },
-    {
-      "epoch": 0.6906192301942897,
-      "grad_norm": 681.7750854492188,
-      "learning_rate": 1.3746759709783996e-05,
-      "loss": 43.9216,
-      "step": 170940
-    },
-    {
-      "epoch": 0.6906596314596574,
-      "grad_norm": 703.8005981445312,
-      "learning_rate": 1.3743642789790318e-05,
-      "loss": 71.855,
-      "step": 170950
-    },
-    {
-      "epoch": 0.6907000327250249,
-      "grad_norm": 788.2094116210938,
-      "learning_rate": 1.3740526089245109e-05,
-      "loss": 61.7674,
-      "step": 170960
-    },
-    {
-      "epoch": 0.6907404339903925,
-      "grad_norm": 383.26641845703125,
-      "learning_rate": 1.373740960820914e-05,
-      "loss": 35.7818,
-      "step": 170970
-    },
-    {
-      "epoch": 0.6907808352557602,
-      "grad_norm": 1133.83837890625,
-      "learning_rate": 1.3734293346743168e-05,
-      "loss": 36.8474,
-      "step": 170980
-    },
-    {
-      "epoch": 0.6908212365211278,
-      "grad_norm": 877.5737915039062,
-      "learning_rate": 1.373117730490795e-05,
-      "loss": 42.4732,
-      "step": 170990
-    },
-    {
-      "epoch": 0.6908616377864955,
-      "grad_norm": 639.6044921875,
-      "learning_rate": 1.3728061482764238e-05,
-      "loss": 48.5779,
-      "step": 171000
-    },
-    {
-      "epoch": 0.6909020390518631,
-      "grad_norm": 1117.853759765625,
-      "learning_rate": 1.3724945880372753e-05,
-      "loss": 53.8607,
-      "step": 171010
-    },
-    {
-      "epoch": 0.6909424403172307,
-      "grad_norm": 613.6959228515625,
-      "learning_rate": 1.3721830497794275e-05,
-      "loss": 44.6761,
-      "step": 171020
-    },
-    {
-      "epoch": 0.6909828415825984,
-      "grad_norm": 2479.233642578125,
-      "learning_rate": 1.3718715335089504e-05,
-      "loss": 81.8502,
-      "step": 171030
-    },
-    {
-      "epoch": 0.691023242847966,
-      "grad_norm": 857.9766845703125,
-      "learning_rate": 1.3715600392319186e-05,
-      "loss": 41.1113,
-      "step": 171040
-    },
-    {
-      "epoch": 0.6910636441133337,
-      "grad_norm": 8812.421875,
-      "learning_rate": 1.371248566954406e-05,
-      "loss": 114.4573,
-      "step": 171050
-    },
-    {
-      "epoch": 0.6911040453787013,
-      "grad_norm": 1065.7593994140625,
-      "learning_rate": 1.3709371166824813e-05,
-      "loss": 36.8817,
-      "step": 171060
-    },
-    {
-      "epoch": 0.6911444466440689,
-      "grad_norm": 574.299072265625,
-      "learning_rate": 1.3706256884222213e-05,
-      "loss": 47.0386,
-      "step": 171070
-    },
-    {
-      "epoch": 0.6911848479094366,
-      "grad_norm": 650.17724609375,
-      "learning_rate": 1.3703142821796938e-05,
-      "loss": 47.3403,
-      "step": 171080
-    },
-    {
-      "epoch": 0.6912252491748041,
-      "grad_norm": 359.5917053222656,
-      "learning_rate": 1.3700028979609708e-05,
-      "loss": 34.8721,
-      "step": 171090
-    },
-    {
-      "epoch": 0.6912656504401717,
-      "grad_norm": 823.551025390625,
-      "learning_rate": 1.369691535772123e-05,
-      "loss": 39.6124,
-      "step": 171100
-    },
-    {
-      "epoch": 0.6913060517055394,
-      "grad_norm": 408.6918640136719,
-      "learning_rate": 1.369380195619221e-05,
-      "loss": 70.1454,
-      "step": 171110
-    },
-    {
-      "epoch": 0.691346452970907,
-      "grad_norm": 309.99774169921875,
-      "learning_rate": 1.369068877508335e-05,
-      "loss": 40.091,
-      "step": 171120
-    },
-    {
-      "epoch": 0.6913868542362747,
-      "grad_norm": 988.518798828125,
-      "learning_rate": 1.3687575814455322e-05,
-      "loss": 55.3994,
-      "step": 171130
-    },
-    {
-      "epoch": 0.6914272555016423,
-      "grad_norm": 1013.3809814453125,
-      "learning_rate": 1.368446307436883e-05,
-      "loss": 42.0952,
-      "step": 171140
-    },
-    {
-      "epoch": 0.6914676567670099,
-      "grad_norm": 844.95361328125,
-      "learning_rate": 1.3681350554884559e-05,
-      "loss": 49.0532,
-      "step": 171150
-    },
-    {
-      "epoch": 0.6915080580323776,
-      "grad_norm": 673.6221313476562,
-      "learning_rate": 1.367823825606319e-05,
-      "loss": 56.2636,
-      "step": 171160
-    },
-    {
-      "epoch": 0.6915484592977452,
-      "grad_norm": 1622.76806640625,
-      "learning_rate": 1.3675126177965381e-05,
-      "loss": 50.5701,
-      "step": 171170
-    },
-    {
-      "epoch": 0.6915888605631129,
-      "grad_norm": 507.78253173828125,
-      "learning_rate": 1.3672014320651832e-05,
-      "loss": 52.1529,
-      "step": 171180
-    },
-    {
-      "epoch": 0.6916292618284805,
-      "grad_norm": 2060.35693359375,
-      "learning_rate": 1.3668902684183188e-05,
-      "loss": 86.6979,
-      "step": 171190
-    },
-    {
-      "epoch": 0.6916696630938481,
-      "grad_norm": 681.6329956054688,
-      "learning_rate": 1.366579126862012e-05,
-      "loss": 47.6263,
-      "step": 171200
-    },
-    {
-      "epoch": 0.6917100643592158,
-      "grad_norm": 1296.0411376953125,
-      "learning_rate": 1.36626800740233e-05,
-      "loss": 65.7112,
-      "step": 171210
-    },
-    {
-      "epoch": 0.6917504656245833,
-      "grad_norm": 1287.6270751953125,
-      "learning_rate": 1.3659569100453346e-05,
-      "loss": 47.9453,
-      "step": 171220
-    },
-    {
-      "epoch": 0.691790866889951,
-      "grad_norm": 942.5064086914062,
-      "learning_rate": 1.3656458347970949e-05,
-      "loss": 53.8814,
-      "step": 171230
-    },
-    {
-      "epoch": 0.6918312681553186,
-      "grad_norm": 847.1381225585938,
-      "learning_rate": 1.3653347816636727e-05,
-      "loss": 34.2508,
-      "step": 171240
-    },
-    {
-      "epoch": 0.6918716694206862,
-      "grad_norm": 662.107666015625,
-      "learning_rate": 1.3650237506511331e-05,
-      "loss": 69.0237,
-      "step": 171250
-    },
-    {
-      "epoch": 0.6919120706860539,
-      "grad_norm": 1095.3472900390625,
-      "learning_rate": 1.3647127417655401e-05,
-      "loss": 45.7568,
-      "step": 171260
-    },
-    {
-      "epoch": 0.6919524719514215,
-      "grad_norm": 483.9328308105469,
-      "learning_rate": 1.3644017550129564e-05,
-      "loss": 40.4629,
-      "step": 171270
-    },
-    {
-      "epoch": 0.6919928732167892,
-      "grad_norm": 740.6072387695312,
-      "learning_rate": 1.3640907903994454e-05,
-      "loss": 36.811,
-      "step": 171280
-    },
-    {
-      "epoch": 0.6920332744821568,
-      "grad_norm": 743.8606567382812,
-      "learning_rate": 1.3637798479310698e-05,
-      "loss": 54.7159,
-      "step": 171290
-    },
-    {
-      "epoch": 0.6920736757475244,
-      "grad_norm": 2306.760986328125,
-      "learning_rate": 1.3634689276138904e-05,
-      "loss": 64.2762,
-      "step": 171300
-    },
-    {
-      "epoch": 0.6921140770128921,
-      "grad_norm": 525.9327392578125,
-      "learning_rate": 1.3631580294539692e-05,
-      "loss": 58.8963,
-      "step": 171310
-    },
-    {
-      "epoch": 0.6921544782782597,
-      "grad_norm": 743.9605712890625,
-      "learning_rate": 1.3628471534573686e-05,
-      "loss": 30.9349,
-      "step": 171320
-    },
-    {
-      "epoch": 0.6921948795436274,
-      "grad_norm": 478.309326171875,
-      "learning_rate": 1.3625362996301463e-05,
-      "loss": 75.0899,
-      "step": 171330
-    },
-    {
-      "epoch": 0.692235280808995,
-      "grad_norm": 845.4862060546875,
-      "learning_rate": 1.3622254679783663e-05,
-      "loss": 38.9251,
-      "step": 171340
-    },
-    {
-      "epoch": 0.6922756820743625,
-      "grad_norm": 520.022705078125,
-      "learning_rate": 1.3619146585080845e-05,
-      "loss": 43.2606,
-      "step": 171350
-    },
-    {
-      "epoch": 0.6923160833397302,
-      "grad_norm": 795.574462890625,
-      "learning_rate": 1.3616038712253646e-05,
-      "loss": 50.8741,
-      "step": 171360
-    },
-    {
-      "epoch": 0.6923564846050978,
-      "grad_norm": 481.3109130859375,
-      "learning_rate": 1.3612931061362622e-05,
-      "loss": 63.0619,
-      "step": 171370
-    },
-    {
-      "epoch": 0.6923968858704654,
-      "grad_norm": 702.5056762695312,
-      "learning_rate": 1.3609823632468366e-05,
-      "loss": 57.0607,
-      "step": 171380
-    },
-    {
-      "epoch": 0.6924372871358331,
-      "grad_norm": 493.5089111328125,
-      "learning_rate": 1.3606716425631478e-05,
-      "loss": 42.103,
-      "step": 171390
-    },
-    {
-      "epoch": 0.6924776884012007,
-      "grad_norm": 725.418701171875,
-      "learning_rate": 1.3603609440912507e-05,
-      "loss": 71.7141,
-      "step": 171400
-    },
-    {
-      "epoch": 0.6925180896665684,
-      "grad_norm": 417.903076171875,
-      "learning_rate": 1.3600502678372035e-05,
-      "loss": 21.3099,
-      "step": 171410
-    },
-    {
-      "epoch": 0.692558490931936,
-      "grad_norm": 189.60548400878906,
-      "learning_rate": 1.3597396138070633e-05,
-      "loss": 47.8742,
-      "step": 171420
-    },
-    {
-      "epoch": 0.6925988921973036,
-      "grad_norm": 627.2072143554688,
-      "learning_rate": 1.3594289820068864e-05,
-      "loss": 53.868,
-      "step": 171430
-    },
-    {
-      "epoch": 0.6926392934626713,
-      "grad_norm": 681.5715942382812,
-      "learning_rate": 1.3591183724427286e-05,
-      "loss": 74.8162,
-      "step": 171440
-    },
-    {
-      "epoch": 0.6926796947280389,
-      "grad_norm": 150.87960815429688,
-      "learning_rate": 1.3588077851206468e-05,
-      "loss": 32.7783,
-      "step": 171450
-    },
-    {
-      "epoch": 0.6927200959934066,
-      "grad_norm": 918.0814208984375,
-      "learning_rate": 1.3584972200466937e-05,
-      "loss": 32.5443,
-      "step": 171460
-    },
-    {
-      "epoch": 0.6927604972587742,
-      "grad_norm": 623.6495361328125,
-      "learning_rate": 1.3581866772269247e-05,
-      "loss": 54.2402,
-      "step": 171470
-    },
-    {
-      "epoch": 0.6928008985241417,
-      "grad_norm": 858.6541137695312,
-      "learning_rate": 1.3578761566673954e-05,
-      "loss": 39.7497,
-      "step": 171480
-    },
-    {
-      "epoch": 0.6928412997895094,
-      "grad_norm": 875.178466796875,
-      "learning_rate": 1.3575656583741564e-05,
-      "loss": 37.1591,
-      "step": 171490
-    },
-    {
-      "epoch": 0.692881701054877,
-      "grad_norm": 1105.78466796875,
-      "learning_rate": 1.3572551823532654e-05,
-      "loss": 47.9929,
-      "step": 171500
-    },
-    {
-      "epoch": 0.6929221023202446,
-      "grad_norm": 927.5897827148438,
-      "learning_rate": 1.3569447286107705e-05,
-      "loss": 58.4196,
-      "step": 171510
-    },
-    {
-      "epoch": 0.6929625035856123,
-      "grad_norm": 428.81097412109375,
-      "learning_rate": 1.3566342971527291e-05,
-      "loss": 61.7513,
-      "step": 171520
-    },
-    {
-      "epoch": 0.6930029048509799,
-      "grad_norm": 565.3176879882812,
-      "learning_rate": 1.3563238879851892e-05,
-      "loss": 44.2183,
-      "step": 171530
-    },
-    {
-      "epoch": 0.6930433061163476,
-      "grad_norm": 1194.6748046875,
-      "learning_rate": 1.3560135011142039e-05,
-      "loss": 59.5806,
-      "step": 171540
-    },
-    {
-      "epoch": 0.6930837073817152,
-      "grad_norm": 592.6781005859375,
-      "learning_rate": 1.3557031365458256e-05,
-      "loss": 33.0638,
-      "step": 171550
-    },
-    {
-      "epoch": 0.6931241086470828,
-      "grad_norm": 1740.63720703125,
-      "learning_rate": 1.3553927942861016e-05,
-      "loss": 57.521,
-      "step": 171560
-    },
-    {
-      "epoch": 0.6931645099124505,
-      "grad_norm": 603.8390502929688,
-      "learning_rate": 1.3550824743410866e-05,
-      "loss": 45.9965,
-      "step": 171570
-    },
-    {
-      "epoch": 0.6932049111778181,
-      "grad_norm": 387.3785400390625,
-      "learning_rate": 1.3547721767168272e-05,
-      "loss": 33.6153,
-      "step": 171580
-    },
-    {
-      "epoch": 0.6932453124431858,
-      "grad_norm": 1378.163330078125,
-      "learning_rate": 1.3544619014193737e-05,
-      "loss": 56.0015,
-      "step": 171590
-    },
-    {
-      "epoch": 0.6932857137085533,
-      "grad_norm": 0.0,
-      "learning_rate": 1.3541516484547753e-05,
-      "loss": 42.2162,
-      "step": 171600
-    },
-    {
-      "epoch": 0.6933261149739209,
-      "grad_norm": 476.3367919921875,
-      "learning_rate": 1.3538414178290815e-05,
-      "loss": 47.5764,
-      "step": 171610
-    },
-    {
-      "epoch": 0.6933665162392886,
-      "grad_norm": 599.8919677734375,
-      "learning_rate": 1.3535312095483371e-05,
-      "loss": 46.6292,
-      "step": 171620
-    },
-    {
-      "epoch": 0.6934069175046562,
-      "grad_norm": 845.22802734375,
-      "learning_rate": 1.3532210236185938e-05,
-      "loss": 32.0494,
-      "step": 171630
-    },
-    {
-      "epoch": 0.6934473187700239,
-      "grad_norm": 485.71490478515625,
-      "learning_rate": 1.3529108600458967e-05,
-      "loss": 37.3266,
-      "step": 171640
-    },
-    {
-      "epoch": 0.6934877200353915,
-      "grad_norm": 1181.7938232421875,
-      "learning_rate": 1.3526007188362924e-05,
-      "loss": 43.7579,
-      "step": 171650
-    },
-    {
-      "epoch": 0.6935281213007591,
-      "grad_norm": 727.456787109375,
-      "learning_rate": 1.352290599995829e-05,
-      "loss": 65.2715,
-      "step": 171660
-    },
-    {
-      "epoch": 0.6935685225661268,
-      "grad_norm": 999.767578125,
-      "learning_rate": 1.3519805035305495e-05,
-      "loss": 33.2939,
-      "step": 171670
-    },
-    {
-      "epoch": 0.6936089238314944,
-      "grad_norm": 398.591552734375,
-      "learning_rate": 1.3516704294465027e-05,
-      "loss": 41.9499,
-      "step": 171680
-    },
-    {
-      "epoch": 0.6936493250968621,
-      "grad_norm": 462.4875793457031,
-      "learning_rate": 1.3513603777497313e-05,
-      "loss": 70.5573,
-      "step": 171690
-    },
-    {
-      "epoch": 0.6936897263622297,
-      "grad_norm": 1010.8731079101562,
-      "learning_rate": 1.3510503484462805e-05,
-      "loss": 88.3927,
-      "step": 171700
-    },
-    {
-      "epoch": 0.6937301276275973,
-      "grad_norm": 438.3502197265625,
-      "learning_rate": 1.350740341542195e-05,
-      "loss": 38.3786,
-      "step": 171710
-    },
-    {
-      "epoch": 0.693770528892965,
-      "grad_norm": 757.7011108398438,
-      "learning_rate": 1.3504303570435179e-05,
-      "loss": 40.7736,
-      "step": 171720
-    },
-    {
-      "epoch": 0.6938109301583325,
-      "grad_norm": 503.13299560546875,
-      "learning_rate": 1.350120394956294e-05,
-      "loss": 44.1673,
-      "step": 171730
-    },
-    {
-      "epoch": 0.6938513314237001,
-      "grad_norm": 375.2835388183594,
-      "learning_rate": 1.3498104552865636e-05,
-      "loss": 61.8349,
-      "step": 171740
-    },
-    {
-      "epoch": 0.6938917326890678,
-      "grad_norm": 724.8334350585938,
-      "learning_rate": 1.349500538040371e-05,
-      "loss": 42.6491,
-      "step": 171750
-    },
-    {
-      "epoch": 0.6939321339544354,
-      "grad_norm": 1060.08837890625,
-      "learning_rate": 1.3491906432237577e-05,
-      "loss": 72.335,
-      "step": 171760
-    },
-    {
-      "epoch": 0.6939725352198031,
-      "grad_norm": 1090.9224853515625,
-      "learning_rate": 1.3488807708427662e-05,
-      "loss": 53.5348,
-      "step": 171770
-    },
-    {
-      "epoch": 0.6940129364851707,
-      "grad_norm": 1613.0850830078125,
-      "learning_rate": 1.3485709209034347e-05,
-      "loss": 56.2584,
-      "step": 171780
-    },
-    {
-      "epoch": 0.6940533377505383,
-      "grad_norm": 1278.3367919921875,
-      "learning_rate": 1.3482610934118078e-05,
-      "loss": 70.852,
-      "step": 171790
-    },
-    {
-      "epoch": 0.694093739015906,
-      "grad_norm": 647.9417114257812,
-      "learning_rate": 1.3479512883739232e-05,
-      "loss": 50.4589,
-      "step": 171800
-    },
-    {
-      "epoch": 0.6941341402812736,
-      "grad_norm": 942.9307861328125,
-      "learning_rate": 1.3476415057958209e-05,
-      "loss": 37.4094,
-      "step": 171810
-    },
-    {
-      "epoch": 0.6941745415466413,
-      "grad_norm": 480.3348388671875,
-      "learning_rate": 1.3473317456835421e-05,
-      "loss": 47.3434,
-      "step": 171820
-    },
-    {
-      "epoch": 0.6942149428120089,
-      "grad_norm": 502.8962097167969,
-      "learning_rate": 1.347022008043122e-05,
-      "loss": 65.3622,
-      "step": 171830
-    },
-    {
-      "epoch": 0.6942553440773765,
-      "grad_norm": 536.8987426757812,
-      "learning_rate": 1.3467122928806037e-05,
-      "loss": 54.2566,
-      "step": 171840
-    },
-    {
-      "epoch": 0.6942957453427442,
-      "grad_norm": 370.1306457519531,
-      "learning_rate": 1.3464026002020219e-05,
-      "loss": 46.8355,
-      "step": 171850
-    },
-    {
-      "epoch": 0.6943361466081117,
-      "grad_norm": 531.8828735351562,
-      "learning_rate": 1.346092930013415e-05,
-      "loss": 60.7364,
-      "step": 171860
-    },
-    {
-      "epoch": 0.6943765478734794,
-      "grad_norm": 652.61328125,
-      "learning_rate": 1.345783282320821e-05,
-      "loss": 54.955,
-      "step": 171870
-    },
-    {
-      "epoch": 0.694416949138847,
-      "grad_norm": 455.7646484375,
-      "learning_rate": 1.3454736571302763e-05,
-      "loss": 32.8962,
-      "step": 171880
-    },
-    {
-      "epoch": 0.6944573504042146,
-      "grad_norm": 532.4978637695312,
-      "learning_rate": 1.3451640544478175e-05,
-      "loss": 58.8792,
-      "step": 171890
-    },
-    {
-      "epoch": 0.6944977516695823,
-      "grad_norm": 708.6436767578125,
-      "learning_rate": 1.3448544742794791e-05,
-      "loss": 43.3127,
-      "step": 171900
-    },
-    {
-      "epoch": 0.6945381529349499,
-      "grad_norm": 745.49169921875,
-      "learning_rate": 1.3445449166312977e-05,
-      "loss": 52.9133,
-      "step": 171910
-    },
-    {
-      "epoch": 0.6945785542003176,
-      "grad_norm": 1188.466796875,
-      "learning_rate": 1.3442353815093078e-05,
-      "loss": 57.6101,
-      "step": 171920
-    },
-    {
-      "epoch": 0.6946189554656852,
-      "grad_norm": 650.7738647460938,
-      "learning_rate": 1.3439258689195453e-05,
-      "loss": 50.3102,
-      "step": 171930
-    },
-    {
-      "epoch": 0.6946593567310528,
-      "grad_norm": 368.1375732421875,
-      "learning_rate": 1.3436163788680411e-05,
-      "loss": 52.3002,
-      "step": 171940
-    },
-    {
-      "epoch": 0.6946997579964205,
-      "grad_norm": 0.0,
-      "learning_rate": 1.343306911360833e-05,
-      "loss": 56.2097,
-      "step": 171950
-    },
-    {
-      "epoch": 0.6947401592617881,
-      "grad_norm": 676.087890625,
-      "learning_rate": 1.3429974664039501e-05,
-      "loss": 47.7845,
-      "step": 171960
-    },
-    {
-      "epoch": 0.6947805605271558,
-      "grad_norm": 1077.6400146484375,
-      "learning_rate": 1.3426880440034292e-05,
-      "loss": 69.1431,
-      "step": 171970
-    },
-    {
-      "epoch": 0.6948209617925234,
-      "grad_norm": 658.7960205078125,
-      "learning_rate": 1.3423786441652998e-05,
-      "loss": 49.965,
-      "step": 171980
-    },
-    {
-      "epoch": 0.6948613630578909,
-      "grad_norm": 1153.1220703125,
-      "learning_rate": 1.3420692668955947e-05,
-      "loss": 45.7714,
-      "step": 171990
-    },
-    {
-      "epoch": 0.6949017643232586,
-      "grad_norm": 482.1449279785156,
-      "learning_rate": 1.3417599122003464e-05,
-      "loss": 68.6874,
-      "step": 172000
     }
   ],
   "logging_steps": 10,
-  "max_steps": 250000,
+  "max_steps": 500000,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
-  "save_steps": 2000,
+  "save_steps": 4000,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
@@ -120427,7 +5627,7 @@
     }
   },
   "total_flos": 0.0,
-  "train_batch_size": 4,
+  "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null
 }