diff --git "a/checkpoint-25000/trainer_state.json" "b/checkpoint-25000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-25000/trainer_state.json"
@@ -0,0 +1,17909 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4257713953037063,
+  "eval_steps": 500,
+  "global_step": 25000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009703085581214826,
+      "grad_norm": 0.5093896998931756,
+      "learning_rate": 5e-06,
+      "loss": 1.7158,
+      "step": 10
+    },
+    {
+      "epoch": 0.0019406171162429653,
+      "grad_norm": 0.5499013987843694,
+      "learning_rate": 1e-05,
+      "loss": 1.6625,
+      "step": 20
+    },
+    {
+      "epoch": 0.002910925674364448,
+      "grad_norm": 0.6551534985295564,
+      "learning_rate": 1.5e-05,
+      "loss": 1.6378,
+      "step": 30
+    },
+    {
+      "epoch": 0.0038812342324859306,
+      "grad_norm": 0.7544385328936806,
+      "learning_rate": 2e-05,
+      "loss": 1.5712,
+      "step": 40
+    },
+    {
+      "epoch": 0.004851542790607413,
+      "grad_norm": 0.6539371672140126,
+      "learning_rate": 2.5e-05,
+      "loss": 1.5217,
+      "step": 50
+    },
+    {
+      "epoch": 0.005821851348728896,
+      "grad_norm": 0.6806376549433046,
+      "learning_rate": 3e-05,
+      "loss": 1.4412,
+      "step": 60
+    },
+    {
+      "epoch": 0.0067921599068503785,
+      "grad_norm": 0.9109066482466388,
+      "learning_rate": 3.5e-05,
+      "loss": 1.2303,
+      "step": 70
+    },
+    {
+      "epoch": 0.007762468464971861,
+      "grad_norm": 0.8941485474193254,
+      "learning_rate": 4e-05,
+      "loss": 1.1185,
+      "step": 80
+    },
+    {
+      "epoch": 0.008732777023093344,
+      "grad_norm": 0.6564948122239989,
+      "learning_rate": 4.5e-05,
+      "loss": 1.0481,
+      "step": 90
+    },
+    {
+      "epoch": 0.009703085581214826,
+      "grad_norm": 0.7749322437088195,
+      "learning_rate": 5e-05,
+      "loss": 0.9965,
+      "step": 100
+    },
+    {
+      "epoch": 0.010673394139336309,
+      "grad_norm": 0.7305812400337627,
+      "learning_rate": 4.998377571549095e-05,
+      "loss": 1.0689,
+      "step": 110
+    },
+    {
+      "epoch": 0.011643702697457792,
+      "grad_norm": 0.7320600025098228,
+      "learning_rate": 4.9967551430981895e-05,
+      "loss": 0.9554,
+      "step": 120
+    },
+    {
+      "epoch": 0.012614011255579274,
+      "grad_norm": 0.7997476828820285,
+      "learning_rate": 4.9951327146472845e-05,
+      "loss": 0.9048,
+      "step": 130
+    },
+    {
+      "epoch": 0.013584319813700757,
+      "grad_norm": 0.9942301355244718,
+      "learning_rate": 4.993510286196379e-05,
+      "loss": 0.9736,
+      "step": 140
+    },
+    {
+      "epoch": 0.01455462837182224,
+      "grad_norm": 0.8990043753777502,
+      "learning_rate": 4.991887857745474e-05,
+      "loss": 0.9447,
+      "step": 150
+    },
+    {
+      "epoch": 0.015524936929943722,
+      "grad_norm": 0.6931013922842434,
+      "learning_rate": 4.990265429294568e-05,
+      "loss": 0.8748,
+      "step": 160
+    },
+    {
+      "epoch": 0.016495245488065205,
+      "grad_norm": 0.8305007689150241,
+      "learning_rate": 4.988643000843663e-05,
+      "loss": 0.9467,
+      "step": 170
+    },
+    {
+      "epoch": 0.017465554046186688,
+      "grad_norm": 0.8993058386814992,
+      "learning_rate": 4.9870205723927573e-05,
+      "loss": 0.9394,
+      "step": 180
+    },
+    {
+      "epoch": 0.01843586260430817,
+      "grad_norm": 0.7486900307747556,
+      "learning_rate": 4.985398143941852e-05,
+      "loss": 0.931,
+      "step": 190
+    },
+    {
+      "epoch": 0.019406171162429653,
+      "grad_norm": 0.8879788388070488,
+      "learning_rate": 4.983775715490947e-05,
+      "loss": 0.9217,
+      "step": 200
+    },
+    {
+      "epoch": 0.020376479720551136,
+      "grad_norm": 0.7351527409279133,
+      "learning_rate": 4.9821532870400416e-05,
+      "loss": 0.8438,
+      "step": 210
+    },
+    {
+      "epoch": 0.021346788278672618,
+      "grad_norm": 0.9651267349255982,
+      "learning_rate": 4.9805308585891366e-05,
+      "loss": 0.8806,
+      "step": 220
+    },
+    {
+      "epoch": 0.0223170968367941,
+      "grad_norm": 0.8358141761944308,
+      "learning_rate": 4.978908430138231e-05,
+      "loss": 0.9075,
+      "step": 230
+    },
+    {
+      "epoch": 0.023287405394915583,
+      "grad_norm": 0.9068073418448239,
+      "learning_rate": 4.977286001687326e-05,
+      "loss": 0.9126,
+      "step": 240
+    },
+    {
+      "epoch": 0.024257713953037066,
+      "grad_norm": 0.8510875679665427,
+      "learning_rate": 4.97566357323642e-05,
+      "loss": 0.8509,
+      "step": 250
+    },
+    {
+      "epoch": 0.02522802251115855,
+      "grad_norm": 0.8371392324883873,
+      "learning_rate": 4.974041144785515e-05,
+      "loss": 0.9018,
+      "step": 260
+    },
+    {
+      "epoch": 0.02619833106928003,
+      "grad_norm": 0.7343155911324605,
+      "learning_rate": 4.9724187163346094e-05,
+      "loss": 0.9251,
+      "step": 270
+    },
+    {
+      "epoch": 0.027168639627401514,
+      "grad_norm": 1.012390858760769,
+      "learning_rate": 4.9707962878837044e-05,
+      "loss": 0.8881,
+      "step": 280
+    },
+    {
+      "epoch": 0.028138948185522997,
+      "grad_norm": 0.8092967868905977,
+      "learning_rate": 4.9691738594327994e-05,
+      "loss": 0.8831,
+      "step": 290
+    },
+    {
+      "epoch": 0.02910925674364448,
+      "grad_norm": 0.9545292755123069,
+      "learning_rate": 4.967551430981894e-05,
+      "loss": 0.8871,
+      "step": 300
+    },
+    {
+      "epoch": 0.030079565301765962,
+      "grad_norm": 0.8451115258584492,
+      "learning_rate": 4.9659290025309887e-05,
+      "loss": 0.8913,
+      "step": 310
+    },
+    {
+      "epoch": 0.031049873859887445,
+      "grad_norm": 0.984138459636415,
+      "learning_rate": 4.964306574080083e-05,
+      "loss": 0.9046,
+      "step": 320
+    },
+    {
+      "epoch": 0.03202018241800893,
+      "grad_norm": 0.9275621393016203,
+      "learning_rate": 4.962684145629178e-05,
+      "loss": 0.88,
+      "step": 330
+    },
+    {
+      "epoch": 0.03299049097613041,
+      "grad_norm": 1.0192818586577115,
+      "learning_rate": 4.961061717178272e-05,
+      "loss": 0.8923,
+      "step": 340
+    },
+    {
+      "epoch": 0.03396079953425189,
+      "grad_norm": 0.9779533854801943,
+      "learning_rate": 4.959439288727367e-05,
+      "loss": 0.868,
+      "step": 350
+    },
+    {
+      "epoch": 0.034931108092373375,
+      "grad_norm": 0.8554421373526296,
+      "learning_rate": 4.957816860276462e-05,
+      "loss": 0.8661,
+      "step": 360
+    },
+    {
+      "epoch": 0.03590141665049486,
+      "grad_norm": 0.7404878078324254,
+      "learning_rate": 4.9561944318255565e-05,
+      "loss": 0.8883,
+      "step": 370
+    },
+    {
+      "epoch": 0.03687172520861634,
+      "grad_norm": 0.9879847832445883,
+      "learning_rate": 4.9545720033746515e-05,
+      "loss": 0.8156,
+      "step": 380
+    },
+    {
+      "epoch": 0.03784203376673782,
+      "grad_norm": 1.0209320862188216,
+      "learning_rate": 4.952949574923746e-05,
+      "loss": 0.8212,
+      "step": 390
+    },
+    {
+      "epoch": 0.038812342324859306,
+      "grad_norm": 1.0823734215432599,
+      "learning_rate": 4.951327146472841e-05,
+      "loss": 0.8619,
+      "step": 400
+    },
+    {
+      "epoch": 0.03978265088298079,
+      "grad_norm": 1.033686091190584,
+      "learning_rate": 4.949704718021935e-05,
+      "loss": 0.818,
+      "step": 410
+    },
+    {
+      "epoch": 0.04075295944110227,
+      "grad_norm": 0.9587642487246072,
+      "learning_rate": 4.94808228957103e-05,
+      "loss": 0.8377,
+      "step": 420
+    },
+    {
+      "epoch": 0.041723267999223754,
+      "grad_norm": 0.8072742668289667,
+      "learning_rate": 4.946459861120124e-05,
+      "loss": 0.8043,
+      "step": 430
+    },
+    {
+      "epoch": 0.042693576557345236,
+      "grad_norm": 1.0090353092499478,
+      "learning_rate": 4.944837432669219e-05,
+      "loss": 0.8829,
+      "step": 440
+    },
+    {
+      "epoch": 0.04366388511546672,
+      "grad_norm": 1.177538600328968,
+      "learning_rate": 4.943215004218314e-05,
+      "loss": 0.859,
+      "step": 450
+    },
+    {
+      "epoch": 0.0446341936735882,
+      "grad_norm": 1.225063127801835,
+      "learning_rate": 4.9415925757674086e-05,
+      "loss": 0.8231,
+      "step": 460
+    },
+    {
+      "epoch": 0.045604502231709684,
+      "grad_norm": 0.8224632705866025,
+      "learning_rate": 4.9399701473165035e-05,
+      "loss": 0.8335,
+      "step": 470
+    },
+    {
+      "epoch": 0.04657481078983117,
+      "grad_norm": 1.153934818941284,
+      "learning_rate": 4.938347718865598e-05,
+      "loss": 0.8046,
+      "step": 480
+    },
+    {
+      "epoch": 0.04754511934795265,
+      "grad_norm": 1.0954567838682858,
+      "learning_rate": 4.936725290414693e-05,
+      "loss": 0.8975,
+      "step": 490
+    },
+    {
+      "epoch": 0.04851542790607413,
+      "grad_norm": 1.2466392758405571,
+      "learning_rate": 4.935102861963788e-05,
+      "loss": 0.8361,
+      "step": 500
+    },
+    {
+      "epoch": 0.049485736464195615,
+      "grad_norm": 0.902356430744448,
+      "learning_rate": 4.933480433512883e-05,
+      "loss": 0.8048,
+      "step": 510
+    },
+    {
+      "epoch": 0.0504560450223171,
+      "grad_norm": 1.01027561407788,
+      "learning_rate": 4.931858005061977e-05,
+      "loss": 0.8094,
+      "step": 520
+    },
+    {
+      "epoch": 0.05142635358043858,
+      "grad_norm": 1.1330239054084805,
+      "learning_rate": 4.930235576611072e-05,
+      "loss": 0.8288,
+      "step": 530
+    },
+    {
+      "epoch": 0.05239666213856006,
+      "grad_norm": 1.19235957479236,
+      "learning_rate": 4.9286131481601664e-05,
+      "loss": 0.8358,
+      "step": 540
+    },
+    {
+      "epoch": 0.053366970696681545,
+      "grad_norm": 1.070882879754836,
+      "learning_rate": 4.926990719709261e-05,
+      "loss": 0.783,
+      "step": 550
+    },
+    {
+      "epoch": 0.05433727925480303,
+      "grad_norm": 0.772829083325615,
+      "learning_rate": 4.925368291258356e-05,
+      "loss": 0.7888,
+      "step": 560
+    },
+    {
+      "epoch": 0.05530758781292451,
+      "grad_norm": 1.0485603174488358,
+      "learning_rate": 4.9237458628074506e-05,
+      "loss": 0.8893,
+      "step": 570
+    },
+    {
+      "epoch": 0.05627789637104599,
+      "grad_norm": 1.1365438528959075,
+      "learning_rate": 4.9221234343565456e-05,
+      "loss": 0.8533,
+      "step": 580
+    },
+    {
+      "epoch": 0.057248204929167476,
+      "grad_norm": 1.1791401079390134,
+      "learning_rate": 4.92050100590564e-05,
+      "loss": 0.8208,
+      "step": 590
+    },
+    {
+      "epoch": 0.05821851348728896,
+      "grad_norm": 1.165054586063552,
+      "learning_rate": 4.918878577454735e-05,
+      "loss": 0.7945,
+      "step": 600
+    },
+    {
+      "epoch": 0.05918882204541044,
+      "grad_norm": 1.0001362775159148,
+      "learning_rate": 4.917256149003829e-05,
+      "loss": 0.8426,
+      "step": 610
+    },
+    {
+      "epoch": 0.060159130603531924,
+      "grad_norm": 1.1955538077863535,
+      "learning_rate": 4.915633720552924e-05,
+      "loss": 0.8346,
+      "step": 620
+    },
+    {
+      "epoch": 0.06112943916165341,
+      "grad_norm": 1.016839980469479,
+      "learning_rate": 4.9140112921020184e-05,
+      "loss": 0.8224,
+      "step": 630
+    },
+    {
+      "epoch": 0.06209974771977489,
+      "grad_norm": 0.9782695489269584,
+      "learning_rate": 4.9123888636511134e-05,
+      "loss": 0.849,
+      "step": 640
+    },
+    {
+      "epoch": 0.06307005627789637,
+      "grad_norm": 1.0768176346339298,
+      "learning_rate": 4.9107664352002084e-05,
+      "loss": 0.8519,
+      "step": 650
+    },
+    {
+      "epoch": 0.06404036483601785,
+      "grad_norm": 1.0636061081219466,
+      "learning_rate": 4.909144006749303e-05,
+      "loss": 0.7887,
+      "step": 660
+    },
+    {
+      "epoch": 0.06501067339413934,
+      "grad_norm": 1.0734895912783853,
+      "learning_rate": 4.907521578298398e-05,
+      "loss": 0.8415,
+      "step": 670
+    },
+    {
+      "epoch": 0.06598098195226082,
+      "grad_norm": 1.0796531470403106,
+      "learning_rate": 4.905899149847492e-05,
+      "loss": 0.7512,
+      "step": 680
+    },
+    {
+      "epoch": 0.0669512905103823,
+      "grad_norm": 0.9361767973637918,
+      "learning_rate": 4.904276721396587e-05,
+      "loss": 0.8482,
+      "step": 690
+    },
+    {
+      "epoch": 0.06792159906850379,
+      "grad_norm": 1.5287022172498188,
+      "learning_rate": 4.902654292945681e-05,
+      "loss": 0.8389,
+      "step": 700
+    },
+    {
+      "epoch": 0.06889190762662527,
+      "grad_norm": 1.129894676066631,
+      "learning_rate": 4.901031864494776e-05,
+      "loss": 0.7629,
+      "step": 710
+    },
+    {
+      "epoch": 0.06986221618474675,
+      "grad_norm": 1.060923634010241,
+      "learning_rate": 4.8994094360438705e-05,
+      "loss": 0.8134,
+      "step": 720
+    },
+    {
+      "epoch": 0.07083252474286823,
+      "grad_norm": 1.121507522572716,
+      "learning_rate": 4.8977870075929655e-05,
+      "loss": 0.8295,
+      "step": 730
+    },
+    {
+      "epoch": 0.07180283330098972,
+      "grad_norm": 1.2375518463265478,
+      "learning_rate": 4.8961645791420605e-05,
+      "loss": 0.8006,
+      "step": 740
+    },
+    {
+      "epoch": 0.0727731418591112,
+      "grad_norm": 1.1957590053535605,
+      "learning_rate": 4.894542150691155e-05,
+      "loss": 0.8185,
+      "step": 750
+    },
+    {
+      "epoch": 0.07374345041723268,
+      "grad_norm": 1.1525450079069435,
+      "learning_rate": 4.89291972224025e-05,
+      "loss": 0.8186,
+      "step": 760
+    },
+    {
+      "epoch": 0.07471375897535416,
+      "grad_norm": 1.2637346291101144,
+      "learning_rate": 4.891297293789344e-05,
+      "loss": 0.8094,
+      "step": 770
+    },
+    {
+      "epoch": 0.07568406753347565,
+      "grad_norm": 1.1118603684784645,
+      "learning_rate": 4.889674865338439e-05,
+      "loss": 0.8075,
+      "step": 780
+    },
+    {
+      "epoch": 0.07665437609159713,
+      "grad_norm": 0.987279065085845,
+      "learning_rate": 4.888052436887533e-05,
+      "loss": 0.8088,
+      "step": 790
+    },
+    {
+      "epoch": 0.07762468464971861,
+      "grad_norm": 1.050050445685608,
+      "learning_rate": 4.886430008436628e-05,
+      "loss": 0.8283,
+      "step": 800
+    },
+    {
+      "epoch": 0.0785949932078401,
+      "grad_norm": 1.1064553828881234,
+      "learning_rate": 4.8848075799857226e-05,
+      "loss": 0.7906,
+      "step": 810
+    },
+    {
+      "epoch": 0.07956530176596158,
+      "grad_norm": 1.0679283923210974,
+      "learning_rate": 4.8831851515348176e-05,
+      "loss": 0.7888,
+      "step": 820
+    },
+    {
+      "epoch": 0.08053561032408306,
+      "grad_norm": 1.325000406776113,
+      "learning_rate": 4.8815627230839126e-05,
+      "loss": 0.8573,
+      "step": 830
+    },
+    {
+      "epoch": 0.08150591888220454,
+      "grad_norm": 1.1430990025005974,
+      "learning_rate": 4.879940294633007e-05,
+      "loss": 0.823,
+      "step": 840
+    },
+    {
+      "epoch": 0.08247622744032602,
+      "grad_norm": 0.9708573314011439,
+      "learning_rate": 4.878317866182102e-05,
+      "loss": 0.7812,
+      "step": 850
+    },
+    {
+      "epoch": 0.08344653599844751,
+      "grad_norm": 1.040516439704035,
+      "learning_rate": 4.876695437731196e-05,
+      "loss": 0.8043,
+      "step": 860
+    },
+    {
+      "epoch": 0.08441684455656899,
+      "grad_norm": 1.4624875951419556,
+      "learning_rate": 4.875073009280291e-05,
+      "loss": 0.8054,
+      "step": 870
+    },
+    {
+      "epoch": 0.08538715311469047,
+      "grad_norm": 1.2973550173157966,
+      "learning_rate": 4.8734505808293854e-05,
+      "loss": 0.7693,
+      "step": 880
+    },
+    {
+      "epoch": 0.08635746167281196,
+      "grad_norm": 1.21645485433679,
+      "learning_rate": 4.8718281523784804e-05,
+      "loss": 0.7402,
+      "step": 890
+    },
+    {
+      "epoch": 0.08732777023093344,
+      "grad_norm": 1.2989905455220712,
+      "learning_rate": 4.870205723927575e-05,
+      "loss": 0.7603,
+      "step": 900
+    },
+    {
+      "epoch": 0.08829807878905492,
+      "grad_norm": 1.1647056182819275,
+      "learning_rate": 4.86858329547667e-05,
+      "loss": 0.7576,
+      "step": 910
+    },
+    {
+      "epoch": 0.0892683873471764,
+      "grad_norm": 1.32821951027441,
+      "learning_rate": 4.8669608670257646e-05,
+      "loss": 0.7593,
+      "step": 920
+    },
+    {
+      "epoch": 0.09023869590529789,
+      "grad_norm": 0.9792995911846096,
+      "learning_rate": 4.865338438574859e-05,
+      "loss": 0.7726,
+      "step": 930
+    },
+    {
+      "epoch": 0.09120900446341937,
+      "grad_norm": 1.1716259071546666,
+      "learning_rate": 4.863716010123954e-05,
+      "loss": 0.7585,
+      "step": 940
+    },
+    {
+      "epoch": 0.09217931302154085,
+      "grad_norm": 1.4206264005961533,
+      "learning_rate": 4.862093581673048e-05,
+      "loss": 0.7922,
+      "step": 950
+    },
+    {
+      "epoch": 0.09314962157966233,
+      "grad_norm": 0.9969780924285648,
+      "learning_rate": 4.860471153222143e-05,
+      "loss": 0.8338,
+      "step": 960
+    },
+    {
+      "epoch": 0.09411993013778382,
+      "grad_norm": 1.2259249945683814,
+      "learning_rate": 4.8588487247712375e-05,
+      "loss": 0.7848,
+      "step": 970
+    },
+    {
+      "epoch": 0.0950902386959053,
+      "grad_norm": 1.304284918297249,
+      "learning_rate": 4.8572262963203325e-05,
+      "loss": 0.799,
+      "step": 980
+    },
+    {
+      "epoch": 0.09606054725402678,
+      "grad_norm": 1.382632512351389,
+      "learning_rate": 4.8556038678694274e-05,
+      "loss": 0.7358,
+      "step": 990
+    },
+    {
+      "epoch": 0.09703085581214826,
+      "grad_norm": 1.1047797334502243,
+      "learning_rate": 4.853981439418522e-05,
+      "loss": 0.7175,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09800116437026975,
+      "grad_norm": 1.298146739590951,
+      "learning_rate": 4.852359010967617e-05,
+      "loss": 0.7694,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09897147292839123,
+      "grad_norm": 1.3449159574026,
+      "learning_rate": 4.850736582516711e-05,
+      "loss": 0.7549,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09994178148651271,
+      "grad_norm": 1.0510958795717098,
+      "learning_rate": 4.849114154065806e-05,
+      "loss": 0.7851,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1009120900446342,
+      "grad_norm": 1.2932499353997113,
+      "learning_rate": 4.8474917256149e-05,
+      "loss": 0.7948,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10188239860275568,
+      "grad_norm": 1.024772482994231,
+      "learning_rate": 4.845869297163995e-05,
+      "loss": 0.7551,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10285270716087716,
+      "grad_norm": 0.9151226850367016,
+      "learning_rate": 4.8442468687130896e-05,
+      "loss": 0.8212,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10382301571899864,
+      "grad_norm": 1.0356064137085648,
+      "learning_rate": 4.8426244402621846e-05,
+      "loss": 0.7815,
+      "step": 1070
+    },
+    {
+      "epoch": 0.10479332427712013,
+      "grad_norm": 1.0454220890712578,
+      "learning_rate": 4.8410020118112795e-05,
+      "loss": 0.7813,
+      "step": 1080
+    },
+    {
+      "epoch": 0.10576363283524161,
+      "grad_norm": 1.0245673186100301,
+      "learning_rate": 4.839379583360374e-05,
+      "loss": 0.7759,
+      "step": 1090
+    },
+    {
+      "epoch": 0.10673394139336309,
+      "grad_norm": 1.1805883290044246,
+      "learning_rate": 4.837757154909469e-05,
+      "loss": 0.8016,
+      "step": 1100
+    },
+    {
+      "epoch": 0.10770424995148457,
+      "grad_norm": 1.305171444801399,
+      "learning_rate": 4.836134726458563e-05,
+      "loss": 0.7603,
+      "step": 1110
+    },
+    {
+      "epoch": 0.10867455850960606,
+      "grad_norm": 1.1334982322569604,
+      "learning_rate": 4.834512298007658e-05,
+      "loss": 0.7818,
+      "step": 1120
+    },
+    {
+      "epoch": 0.10964486706772754,
+      "grad_norm": 1.4897469886835581,
+      "learning_rate": 4.8328898695567524e-05,
+      "loss": 0.7391,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11061517562584902,
+      "grad_norm": 1.12299562877673,
+      "learning_rate": 4.8312674411058474e-05,
+      "loss": 0.8129,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1115854841839705,
+      "grad_norm": 1.2581064725802191,
+      "learning_rate": 4.8296450126549417e-05,
+      "loss": 0.7132,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11255579274209199,
+      "grad_norm": 1.1117033056933057,
+      "learning_rate": 4.8280225842040366e-05,
+      "loss": 0.7247,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11352610130021347,
+      "grad_norm": 1.5072697151276053,
+      "learning_rate": 4.8264001557531316e-05,
+      "loss": 0.7832,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11449640985833495,
+      "grad_norm": 1.3629201153146466,
+      "learning_rate": 4.824777727302226e-05,
+      "loss": 0.7894,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11546671841645643,
+      "grad_norm": 1.3456355263382838,
+      "learning_rate": 4.823155298851321e-05,
+      "loss": 0.8145,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11643702697457792,
+      "grad_norm": 1.3206439343547949,
+      "learning_rate": 4.821532870400415e-05,
+      "loss": 0.7802,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1174073355326994,
+      "grad_norm": 1.0980976013487813,
+      "learning_rate": 4.81991044194951e-05,
+      "loss": 0.801,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11837764409082088,
+      "grad_norm": 1.1370443655089153,
+      "learning_rate": 4.8182880134986045e-05,
+      "loss": 0.8006,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11934795264894237,
+      "grad_norm": 1.4354316240483984,
+      "learning_rate": 4.8166655850476994e-05,
+      "loss": 0.8131,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12031826120706385,
+      "grad_norm": 0.9112243461310537,
+      "learning_rate": 4.815043156596794e-05,
+      "loss": 0.8145,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12128856976518533,
+      "grad_norm": 0.8213091259360801,
+      "learning_rate": 4.813420728145889e-05,
+      "loss": 0.7114,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12225887832330681,
+      "grad_norm": 1.3405078182982422,
+      "learning_rate": 4.811798299694984e-05,
+      "loss": 0.7986,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1232291868814283,
+      "grad_norm": 1.4540711324279514,
+      "learning_rate": 4.810175871244078e-05,
+      "loss": 0.7281,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12419949543954978,
+      "grad_norm": 1.0040222075537582,
+      "learning_rate": 4.8085534427931737e-05,
+      "loss": 0.8048,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12516980399767125,
+      "grad_norm": 1.9264161003194282,
+      "learning_rate": 4.806931014342268e-05,
+      "loss": 0.7512,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12614011255579274,
+      "grad_norm": 1.155430871835739,
+      "learning_rate": 4.805308585891363e-05,
+      "loss": 0.7909,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1271104211139142,
+      "grad_norm": 1.382153914724162,
+      "learning_rate": 4.803686157440457e-05,
+      "loss": 0.7613,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1280807296720357,
+      "grad_norm": 1.1079273142955168,
+      "learning_rate": 4.802063728989552e-05,
+      "loss": 0.7747,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12905103823015718,
+      "grad_norm": 1.4797892743755068,
+      "learning_rate": 4.8004413005386465e-05,
+      "loss": 0.7448,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13002134678827867,
+      "grad_norm": 1.2840858833683126,
+      "learning_rate": 4.7988188720877415e-05,
+      "loss": 0.7567,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13099165534640014,
+      "grad_norm": 1.7212676971427285,
+      "learning_rate": 4.797196443636836e-05,
+      "loss": 0.7743,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13196196390452164,
+      "grad_norm": 1.7283159526025742,
+      "learning_rate": 4.795574015185931e-05,
+      "loss": 0.7696,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1329322724626431,
+      "grad_norm": 1.0739621496825589,
+      "learning_rate": 4.793951586735026e-05,
+      "loss": 0.7408,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1339025810207646,
+      "grad_norm": 1.183199109006057,
+      "learning_rate": 4.79232915828412e-05,
+      "loss": 0.7935,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13487288957888607,
+      "grad_norm": 1.3981562096537532,
+      "learning_rate": 4.790706729833215e-05,
+      "loss": 0.7113,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13584319813700757,
+      "grad_norm": 1.192516841319882,
+      "learning_rate": 4.789084301382309e-05,
+      "loss": 0.7786,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13681350669512904,
+      "grad_norm": 1.9269898255758637,
+      "learning_rate": 4.787461872931404e-05,
+      "loss": 0.7262,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13778381525325054,
+      "grad_norm": 1.1218056549529871,
+      "learning_rate": 4.7858394444804986e-05,
+      "loss": 0.6762,
+      "step": 1420
+    },
+    {
+      "epoch": 0.138754123811372,
+      "grad_norm": 1.3635607352483248,
+      "learning_rate": 4.7842170160295936e-05,
+      "loss": 0.7733,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1397244323694935,
+      "grad_norm": 1.2380674478503626,
+      "learning_rate": 4.782594587578688e-05,
+      "loss": 0.7599,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14069474092761497,
+      "grad_norm": 1.4869366640536255,
+      "learning_rate": 4.780972159127783e-05,
+      "loss": 0.7408,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14166504948573647,
+      "grad_norm": 1.413887084722376,
+      "learning_rate": 4.779349730676878e-05,
+      "loss": 0.7792,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14263535804385793,
+      "grad_norm": 1.1522475316568597,
+      "learning_rate": 4.777727302225972e-05,
+      "loss": 0.7494,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14360566660197943,
+      "grad_norm": 1.5458892686752617,
+      "learning_rate": 4.776104873775067e-05,
+      "loss": 0.7151,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1445759751601009,
+      "grad_norm": 1.8892150029961168,
+      "learning_rate": 4.7744824453241614e-05,
+      "loss": 0.7025,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1455462837182224,
+      "grad_norm": 1.5206314987834826,
+      "learning_rate": 4.7728600168732564e-05,
+      "loss": 0.7748,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14651659227634387,
+      "grad_norm": 1.2234736643633124,
+      "learning_rate": 4.771237588422351e-05,
+      "loss": 0.7321,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14748690083446536,
+      "grad_norm": 1.3582944219822406,
+      "learning_rate": 4.7696151599714456e-05,
+      "loss": 0.7524,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14845720939258683,
+      "grad_norm": 1.3505468356868415,
+      "learning_rate": 4.76799273152054e-05,
+      "loss": 0.8032,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14942751795070833,
+      "grad_norm": 1.161565772908295,
+      "learning_rate": 4.766370303069635e-05,
+      "loss": 0.7449,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1503978265088298,
+      "grad_norm": 1.5018575379494306,
+      "learning_rate": 4.76474787461873e-05,
+      "loss": 0.779,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1513681350669513,
+      "grad_norm": 1.2744144878497465,
+      "learning_rate": 4.763125446167824e-05,
+      "loss": 0.7396,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15233844362507276,
+      "grad_norm": 1.2326350258332728,
+      "learning_rate": 4.761503017716919e-05,
+      "loss": 0.7249,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15330875218319426,
+      "grad_norm": 1.2982918445496714,
+      "learning_rate": 4.7598805892660135e-05,
+      "loss": 0.7453,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15427906074131573,
+      "grad_norm": 1.4792372825641715,
+      "learning_rate": 4.7582581608151085e-05,
+      "loss": 0.7573,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15524936929943722,
+      "grad_norm": 1.2935035055138504,
+      "learning_rate": 4.756635732364203e-05,
+      "loss": 0.7125,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1562196778575587,
+      "grad_norm": 1.2527694639896991,
+      "learning_rate": 4.755013303913298e-05,
+      "loss": 0.7406,
+      "step": 1610
+    },
+    {
+      "epoch": 0.1571899864156802,
+      "grad_norm": 1.0724672098454868,
+      "learning_rate": 4.753390875462392e-05,
+      "loss": 0.7108,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15816029497380166,
+      "grad_norm": 1.1696947872465324,
+      "learning_rate": 4.751768447011487e-05,
+      "loss": 0.7636,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15913060353192315,
+      "grad_norm": 1.216857763890884,
+      "learning_rate": 4.750146018560582e-05,
+      "loss": 0.7916,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16010091209004462,
+      "grad_norm": 1.1371281502973842,
+      "learning_rate": 4.748523590109676e-05,
+      "loss": 0.7663,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16107122064816612,
+      "grad_norm": 1.2599433230373354,
+      "learning_rate": 4.746901161658771e-05,
+      "loss": 0.7279,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1620415292062876,
+      "grad_norm": 1.4580521921521419,
+      "learning_rate": 4.7452787332078656e-05,
+      "loss": 0.7241,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16301183776440908,
+      "grad_norm": 1.2227652156436173,
+      "learning_rate": 4.7436563047569605e-05,
+      "loss": 0.7082,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16398214632253055,
+      "grad_norm": 1.3578429437561153,
+      "learning_rate": 4.742033876306055e-05,
+      "loss": 0.6972,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16495245488065205,
+      "grad_norm": 1.2809095001474842,
+      "learning_rate": 4.74041144785515e-05,
+      "loss": 0.7267,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16592276343877352,
+      "grad_norm": 1.1017366555347645,
+      "learning_rate": 4.738789019404245e-05,
+      "loss": 0.7243,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16689307199689501,
+      "grad_norm": 1.2852463688715783,
+      "learning_rate": 4.737166590953339e-05,
+      "loss": 0.7276,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16786338055501648,
+      "grad_norm": 1.3099425645156408,
+      "learning_rate": 4.735544162502434e-05,
+      "loss": 0.7583,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16883368911313798,
+      "grad_norm": 1.5220953005112245,
+      "learning_rate": 4.7339217340515284e-05,
+      "loss": 0.7651,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16980399767125945,
+      "grad_norm": 1.3251019142596,
+      "learning_rate": 4.7322993056006233e-05,
+      "loss": 0.7428,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17077430622938095,
+      "grad_norm": 1.3275994643711895,
+      "learning_rate": 4.7306768771497176e-05,
+      "loss": 0.7552,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17174461478750241,
+      "grad_norm": 1.1363294732621385,
+      "learning_rate": 4.7290544486988126e-05,
+      "loss": 0.7441,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1727149233456239,
+      "grad_norm": 1.6604556501118164,
+      "learning_rate": 4.727432020247907e-05,
+      "loss": 0.7404,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17368523190374538,
+      "grad_norm": 1.4971063178125654,
+      "learning_rate": 4.725809591797002e-05,
+      "loss": 0.707,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17465554046186688,
+      "grad_norm": 1.296038406932857,
+      "learning_rate": 4.724187163346097e-05,
+      "loss": 0.7226,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17562584901998834,
+      "grad_norm": 0.9900558600646284,
+      "learning_rate": 4.722564734895191e-05,
+      "loss": 0.7107,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17659615757810984,
+      "grad_norm": 0.9589095684989648,
+      "learning_rate": 4.720942306444286e-05,
+      "loss": 0.6789,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1775664661362313,
+      "grad_norm": 1.300487413115222,
+      "learning_rate": 4.7193198779933805e-05,
+      "loss": 0.7656,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1785367746943528,
+      "grad_norm": 1.4678054839136885,
+      "learning_rate": 4.7176974495424754e-05,
+      "loss": 0.7559,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17950708325247428,
+      "grad_norm": 1.1487384302132937,
+      "learning_rate": 4.71607502109157e-05,
+      "loss": 0.7218,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18047739181059577,
+      "grad_norm": 1.5013237178369594,
+      "learning_rate": 4.714452592640665e-05,
+      "loss": 0.6618,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18144770036871724,
+      "grad_norm": 1.3544109774626758,
+      "learning_rate": 4.712830164189759e-05,
+      "loss": 0.7348,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18241800892683874,
+      "grad_norm": 1.7098997535129123,
+      "learning_rate": 4.711207735738854e-05,
+      "loss": 0.6721,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1833883174849602,
+      "grad_norm": 1.439742442692076,
+      "learning_rate": 4.709585307287949e-05,
+      "loss": 0.6823,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1843586260430817,
+      "grad_norm": 1.1240799236385792,
+      "learning_rate": 4.707962878837043e-05,
+      "loss": 0.7337,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18532893460120317,
+      "grad_norm": 1.611378134043144,
+      "learning_rate": 4.706340450386138e-05,
+      "loss": 0.7348,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18629924315932467,
+      "grad_norm": 1.15451740477432,
+      "learning_rate": 4.7047180219352325e-05,
+      "loss": 0.6945,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18726955171744614,
+      "grad_norm": 1.5743486470192276,
+      "learning_rate": 4.7030955934843275e-05,
+      "loss": 0.8062,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18823986027556763,
+      "grad_norm": 1.0248040084556684,
+      "learning_rate": 4.701473165033422e-05,
+      "loss": 0.7155,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1892101688336891,
+      "grad_norm": 1.0310414805904136,
+      "learning_rate": 4.699850736582517e-05,
+      "loss": 0.6776,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1901804773918106,
+      "grad_norm": 1.0785156193356158,
+      "learning_rate": 4.698228308131611e-05,
+      "loss": 0.7092,
+      "step": 1960
+    },
+    {
+      "epoch": 0.19115078594993207,
+      "grad_norm": 1.4378941602091937,
+      "learning_rate": 4.696605879680706e-05,
+      "loss": 0.7286,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19212109450805356,
+      "grad_norm": 1.1595135486156671,
+      "learning_rate": 4.694983451229801e-05,
+      "loss": 0.7479,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19309140306617503,
+      "grad_norm": 1.69348514333408,
+      "learning_rate": 4.6933610227788953e-05,
+      "loss": 0.7263,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19406171162429653,
+      "grad_norm": 1.502953634133657,
+      "learning_rate": 4.69173859432799e-05,
+      "loss": 0.7655,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19406171162429653,
+      "eval_loss": 0.7587813138961792,
+      "eval_runtime": 2477.8973,
+      "eval_samples_per_second": 0.723,
+      "eval_steps_per_second": 0.362,
+      "step": 2000
+    },
+    {
+      "epoch": 0.195032020182418,
+      "grad_norm": 1.4070334559445785,
+      "learning_rate": 4.6901161658770846e-05,
+      "loss": 0.7475,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1960023287405395,
+      "grad_norm": 0.8673128363877267,
+      "learning_rate": 4.6884937374261796e-05,
+      "loss": 0.6747,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19697263729866096,
+      "grad_norm": 1.428849127809278,
+      "learning_rate": 4.686871308975274e-05,
+      "loss": 0.7666,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19794294585678246,
+      "grad_norm": 1.2749540514563555,
+      "learning_rate": 4.685248880524369e-05,
+      "loss": 0.7346,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19891325441490393,
+      "grad_norm": 1.2400883720583105,
+      "learning_rate": 4.683626452073464e-05,
+      "loss": 0.7012,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19988356297302542,
+      "grad_norm": 1.4035260703119048,
+      "learning_rate": 4.682004023622559e-05,
+      "loss": 0.6697,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2008538715311469,
+      "grad_norm": 1.9046063428505313,
+      "learning_rate": 4.680381595171653e-05,
+      "loss": 0.7126,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2018241800892684,
+      "grad_norm": 1.3281602578294986,
+      "learning_rate": 4.678759166720748e-05,
+      "loss": 0.6895,
+      "step": 2080
+    },
+    {
+      "epoch": 0.20279448864738986,
+      "grad_norm": 1.271644663424638,
+      "learning_rate": 4.677136738269843e-05,
+      "loss": 0.7354,
+      "step": 2090
+    },
+    {
+      "epoch": 0.20376479720551136,
+      "grad_norm": 1.2206696245686643,
+      "learning_rate": 4.6755143098189374e-05,
+      "loss": 0.7252,
+      "step": 2100
+    },
+    {
+      "epoch": 0.20473510576363282,
+      "grad_norm": 1.3032653564716208,
+      "learning_rate": 4.6738918813680324e-05,
+      "loss": 0.683,
+      "step": 2110
+    },
+    {
+      "epoch": 0.20570541432175432,
+      "grad_norm": 1.3212954807490243,
+      "learning_rate": 4.6722694529171267e-05,
+      "loss": 0.737,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2066757228798758,
+      "grad_norm": 1.2586128849417775,
+      "learning_rate": 4.6706470244662216e-05,
+      "loss": 0.6841,
+      "step": 2130
+    },
+    {
+      "epoch": 0.20764603143799729,
+      "grad_norm": 1.3491929764497603,
+      "learning_rate": 4.669024596015316e-05,
+      "loss": 0.7637,
+      "step": 2140
+    },
+    {
+      "epoch": 0.20861633999611875,
+      "grad_norm": 1.5081978528458062,
+      "learning_rate": 4.667402167564411e-05,
+      "loss": 0.713,
+      "step": 2150
+    },
+    {
+      "epoch": 0.20958664855424025,
+      "grad_norm": 1.5788893940918114,
+      "learning_rate": 4.665779739113505e-05,
+      "loss": 0.6337,
+      "step": 2160
+    },
+    {
+      "epoch": 0.21055695711236172,
+      "grad_norm": 1.422646038947752,
+      "learning_rate": 4.6641573106626e-05,
+      "loss": 0.6653,
+      "step": 2170
+    },
+    {
+      "epoch": 0.21152726567048322,
+      "grad_norm": 1.3417475771584986,
+      "learning_rate": 4.662534882211695e-05,
+      "loss": 0.7246,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21249757422860469,
+      "grad_norm": 1.0925391942237144,
+      "learning_rate": 4.6609124537607895e-05,
+      "loss": 0.7228,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21346788278672618,
+      "grad_norm": 1.8433218311262853,
+      "learning_rate": 4.6592900253098844e-05,
+      "loss": 0.7507,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21443819134484765,
+      "grad_norm": 1.792431274692439,
+      "learning_rate": 4.657667596858979e-05,
+      "loss": 0.6949,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21540849990296915,
+      "grad_norm": 1.3809251069016177,
+      "learning_rate": 4.656045168408074e-05,
+      "loss": 0.7252,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21637880846109062,
+      "grad_norm": 1.3641658734062512,
+      "learning_rate": 4.654422739957168e-05,
+      "loss": 0.7518,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2173491170192121,
+      "grad_norm": 1.452315608698053,
+      "learning_rate": 4.652800311506263e-05,
+      "loss": 0.6684,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21831942557733358,
+      "grad_norm": 1.1444356316783801,
+      "learning_rate": 4.651177883055357e-05,
+      "loss": 0.6714,
+      "step": 2250
+    },
+    {
+      "epoch": 0.21928973413545508,
+      "grad_norm": 1.2977358748240138,
+      "learning_rate": 4.649555454604452e-05,
+      "loss": 0.7129,
+      "step": 2260
+    },
+    {
+      "epoch": 0.22026004269357655,
+      "grad_norm": 1.328329389241565,
+      "learning_rate": 4.647933026153547e-05,
+      "loss": 0.6292,
+      "step": 2270
+    },
+    {
+      "epoch": 0.22123035125169804,
+      "grad_norm": 1.2461104595186587,
+      "learning_rate": 4.6463105977026415e-05,
+      "loss": 0.7372,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2222006598098195,
+      "grad_norm": 1.3008437677725404,
+      "learning_rate": 4.6446881692517365e-05,
+      "loss": 0.6503,
+      "step": 2290
+    },
+    {
+      "epoch": 0.223170968367941,
+      "grad_norm": 1.3630765232741517,
+      "learning_rate": 4.643065740800831e-05,
+      "loss": 0.657,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22414127692606248,
+      "grad_norm": 0.9600325037717949,
+      "learning_rate": 4.641443312349926e-05,
+      "loss": 0.7117,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22511158548418397,
+      "grad_norm": 1.733755330685857,
+      "learning_rate": 4.63982088389902e-05,
+      "loss": 0.7594,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22608189404230544,
+      "grad_norm": 1.722193618002385,
+      "learning_rate": 4.638198455448115e-05,
+      "loss": 0.6555,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22705220260042694,
+      "grad_norm": 1.2647254702280388,
+      "learning_rate": 4.63657602699721e-05,
+      "loss": 0.6996,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2280225111585484,
+      "grad_norm": 1.2801225623311008,
+      "learning_rate": 4.6349535985463043e-05,
+      "loss": 0.7038,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2289928197166699,
+      "grad_norm": 1.214420395864136,
+      "learning_rate": 4.633331170095399e-05,
+      "loss": 0.6617,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22996312827479137,
+      "grad_norm": 1.1970250608654163,
+      "learning_rate": 4.6317087416444936e-05,
+      "loss": 0.7505,
+      "step": 2370
+    },
+    {
+      "epoch": 0.23093343683291287,
+      "grad_norm": 1.318541272085749,
+      "learning_rate": 4.6300863131935886e-05,
+      "loss": 0.703,
+      "step": 2380
+    },
+    {
+      "epoch": 0.23190374539103434,
+      "grad_norm": 1.0733620278703964,
+      "learning_rate": 4.628463884742683e-05,
+      "loss": 0.7076,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23287405394915583,
+      "grad_norm": 1.320066160024292,
+      "learning_rate": 4.626841456291778e-05,
+      "loss": 0.6845,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2338443625072773,
+      "grad_norm": 1.3916335472579557,
+      "learning_rate": 4.625219027840872e-05,
+      "loss": 0.7213,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2348146710653988,
+      "grad_norm": 1.6506684734262902,
+      "learning_rate": 4.623596599389967e-05,
+      "loss": 0.6995,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23578497962352027,
+      "grad_norm": 0.8976988021024955,
+      "learning_rate": 4.621974170939062e-05,
+      "loss": 0.6994,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23675528818164177,
+      "grad_norm": 1.7783475330175254,
+      "learning_rate": 4.6203517424881564e-05,
+      "loss": 0.7475,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23772559673976323,
+      "grad_norm": 1.3329956259541478,
+      "learning_rate": 4.6187293140372514e-05,
+      "loss": 0.6991,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23869590529788473,
+      "grad_norm": 1.3282617232306233,
+      "learning_rate": 4.617106885586346e-05,
+      "loss": 0.7328,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2396662138560062,
+      "grad_norm": 1.1394548446385124,
+      "learning_rate": 4.615484457135441e-05,
+      "loss": 0.7092,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2406365224141277,
+      "grad_norm": 1.3304671581253036,
+      "learning_rate": 4.613862028684535e-05,
+      "loss": 0.6765,
+      "step": 2480
+    },
+    {
+      "epoch": 0.24160683097224916,
+      "grad_norm": 1.5514738877871737,
+      "learning_rate": 4.61223960023363e-05,
+      "loss": 0.6558,
+      "step": 2490
+    },
+    {
+      "epoch": 0.24257713953037066,
+      "grad_norm": 1.3668799500092241,
+      "learning_rate": 4.610617171782724e-05,
+      "loss": 0.7106,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24257713953037066,
+      "eval_loss": 0.7454198598861694,
+      "eval_runtime": 2468.3109,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24354744808849213,
+      "grad_norm": 1.27051542074557,
+      "learning_rate": 4.608994743331819e-05,
+      "loss": 0.7037,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24451775664661363,
+      "grad_norm": 1.4333699367072212,
+      "learning_rate": 4.607372314880914e-05,
+      "loss": 0.7054,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2454880652047351,
+      "grad_norm": 1.3383562796673616,
+      "learning_rate": 4.6057498864300085e-05,
+      "loss": 0.7674,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2464583737628566,
+      "grad_norm": 1.5405408532446832,
+      "learning_rate": 4.6041274579791035e-05,
+      "loss": 0.7541,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24742868232097806,
+      "grad_norm": 1.4592658544748531,
+      "learning_rate": 4.602505029528198e-05,
+      "loss": 0.7286,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24839899087909956,
+      "grad_norm": 1.4060719508768202,
+      "learning_rate": 4.600882601077293e-05,
+      "loss": 0.7722,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24936929943722103,
+      "grad_norm": 1.2688352607364943,
+      "learning_rate": 4.599260172626387e-05,
+      "loss": 0.6851,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2503396079953425,
+      "grad_norm": 1.1798367463314652,
+      "learning_rate": 4.597637744175482e-05,
+      "loss": 0.6897,
+      "step": 2580
+    },
+    {
+      "epoch": 0.251309916553464,
+      "grad_norm": 1.4210256244438406,
+      "learning_rate": 4.5960153157245763e-05,
+      "loss": 0.7005,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2522802251115855,
+      "grad_norm": 1.3304574917634664,
+      "learning_rate": 4.594392887273671e-05,
+      "loss": 0.6878,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25325053366970696,
+      "grad_norm": 1.4281359203156077,
+      "learning_rate": 4.592770458822766e-05,
+      "loss": 0.7519,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2542208422278284,
+      "grad_norm": 1.351420571440429,
+      "learning_rate": 4.5911480303718606e-05,
+      "loss": 0.6939,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25519115078594995,
+      "grad_norm": 1.3618970623647955,
+      "learning_rate": 4.5895256019209556e-05,
+      "loss": 0.669,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2561614593440714,
+      "grad_norm": 1.1008259683717303,
+      "learning_rate": 4.58790317347005e-05,
+      "loss": 0.6331,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2571317679021929,
+      "grad_norm": 1.690823489066079,
+      "learning_rate": 4.586280745019145e-05,
+      "loss": 0.6571,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25810207646031436,
+      "grad_norm": 1.2328191346656623,
+      "learning_rate": 4.584658316568239e-05,
+      "loss": 0.6712,
+      "step": 2660
+    },
+    {
+      "epoch": 0.2590723850184359,
+      "grad_norm": 1.1997509506925832,
+      "learning_rate": 4.583035888117334e-05,
+      "loss": 0.6998,
+      "step": 2670
+    },
+    {
+      "epoch": 0.26004269357655735,
+      "grad_norm": 1.3726212075390893,
+      "learning_rate": 4.5814134596664284e-05,
+      "loss": 0.6577,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2610130021346788,
+      "grad_norm": 1.4778299478224584,
+      "learning_rate": 4.5797910312155234e-05,
+      "loss": 0.7213,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2619833106928003,
+      "grad_norm": 1.2065482843241282,
+      "learning_rate": 4.5781686027646184e-05,
+      "loss": 0.6504,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2629536192509218,
+      "grad_norm": 1.6950271620933635,
+      "learning_rate": 4.576546174313713e-05,
+      "loss": 0.6938,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2639239278090433,
+      "grad_norm": 1.049429219350124,
+      "learning_rate": 4.5749237458628077e-05,
+      "loss": 0.7235,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26489423636716475,
+      "grad_norm": 1.2856525880832654,
+      "learning_rate": 4.573301317411902e-05,
+      "loss": 0.769,
+      "step": 2730
+    },
+    {
+      "epoch": 0.2658645449252862,
+      "grad_norm": 1.1413973985217811,
+      "learning_rate": 4.571678888960997e-05,
+      "loss": 0.6542,
+      "step": 2740
+    },
+    {
+      "epoch": 0.26683485348340774,
+      "grad_norm": 1.2963652895204112,
+      "learning_rate": 4.570056460510091e-05,
+      "loss": 0.6988,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2678051620415292,
+      "grad_norm": 1.200192916715494,
+      "learning_rate": 4.568434032059186e-05,
+      "loss": 0.662,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2687754705996507,
+      "grad_norm": 1.2418786623251215,
+      "learning_rate": 4.566811603608281e-05,
+      "loss": 0.6615,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26974577915777215,
+      "grad_norm": 1.2920018198618357,
+      "learning_rate": 4.5651891751573755e-05,
+      "loss": 0.683,
+      "step": 2780
+    },
+    {
+      "epoch": 0.27071608771589367,
+      "grad_norm": 1.721771869410589,
+      "learning_rate": 4.5635667467064705e-05,
+      "loss": 0.6742,
+      "step": 2790
+    },
+    {
+      "epoch": 0.27168639627401514,
+      "grad_norm": 1.595428896858632,
+      "learning_rate": 4.561944318255565e-05,
+      "loss": 0.6961,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2726567048321366,
+      "grad_norm": 1.5674989639835977,
+      "learning_rate": 4.56032188980466e-05,
+      "loss": 0.6246,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2736270133902581,
+      "grad_norm": 1.333761320008167,
+      "learning_rate": 4.558699461353754e-05,
+      "loss": 0.7142,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2745973219483796,
+      "grad_norm": 1.175942959739195,
+      "learning_rate": 4.557077032902849e-05,
+      "loss": 0.7335,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27556763050650107,
+      "grad_norm": 1.0433896859463463,
+      "learning_rate": 4.555454604451944e-05,
+      "loss": 0.7523,
+      "step": 2840
+    },
+    {
+      "epoch": 0.27653793906462254,
+      "grad_norm": 1.4484346047275096,
+      "learning_rate": 4.553832176001039e-05,
+      "loss": 0.6576,
+      "step": 2850
+    },
+    {
+      "epoch": 0.277508247622744,
+      "grad_norm": 1.922226387784083,
+      "learning_rate": 4.552209747550133e-05,
+      "loss": 0.6797,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27847855618086553,
+      "grad_norm": 1.4433287579337053,
+      "learning_rate": 4.550587319099228e-05,
+      "loss": 0.6787,
+      "step": 2870
+    },
+    {
+      "epoch": 0.279448864738987,
+      "grad_norm": 1.4997180997318538,
+      "learning_rate": 4.5489648906483225e-05,
+      "loss": 0.7091,
+      "step": 2880
+    },
+    {
+      "epoch": 0.28041917329710847,
+      "grad_norm": 1.347204680869145,
+      "learning_rate": 4.5473424621974175e-05,
+      "loss": 0.725,
+      "step": 2890
+    },
+    {
+      "epoch": 0.28138948185522994,
+      "grad_norm": 1.3046706448190106,
+      "learning_rate": 4.5457200337465125e-05,
+      "loss": 0.6669,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28235979041335146,
+      "grad_norm": 1.1467739204887912,
+      "learning_rate": 4.544097605295607e-05,
+      "loss": 0.756,
+      "step": 2910
+    },
+    {
+      "epoch": 0.28333009897147293,
+      "grad_norm": 1.162771827035537,
+      "learning_rate": 4.542475176844702e-05,
+      "loss": 0.7147,
+      "step": 2920
+    },
+    {
+      "epoch": 0.2843004075295944,
+      "grad_norm": 1.6563795609214405,
+      "learning_rate": 4.540852748393796e-05,
+      "loss": 0.6845,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28527071608771587,
+      "grad_norm": 1.6728193237645246,
+      "learning_rate": 4.539230319942891e-05,
+      "loss": 0.673,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2862410246458374,
+      "grad_norm": 1.194612705131068,
+      "learning_rate": 4.5376078914919854e-05,
+      "loss": 0.6942,
+      "step": 2950
+    },
+    {
+      "epoch": 0.28721133320395886,
+      "grad_norm": 1.439958876835649,
+      "learning_rate": 4.53598546304108e-05,
+      "loss": 0.6809,
+      "step": 2960
+    },
+    {
+      "epoch": 0.28818164176208033,
+      "grad_norm": 1.5873335965261735,
+      "learning_rate": 4.534363034590175e-05,
+      "loss": 0.6877,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2891519503202018,
+      "grad_norm": 1.0997513483559662,
+      "learning_rate": 4.5327406061392696e-05,
+      "loss": 0.665,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2901222588783233,
+      "grad_norm": 1.3952744081716912,
+      "learning_rate": 4.5311181776883646e-05,
+      "loss": 0.7524,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2910925674364448,
+      "grad_norm": 1.1613111987177211,
+      "learning_rate": 4.529495749237459e-05,
+      "loss": 0.6659,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2910925674364448,
+      "eval_loss": 0.7351760268211365,
+      "eval_runtime": 2466.7648,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29206287599456626,
+      "grad_norm": 1.404418707672653,
+      "learning_rate": 4.527873320786554e-05,
+      "loss": 0.6635,
+      "step": 3010
+    },
+    {
+      "epoch": 0.29303318455268773,
+      "grad_norm": 1.7333713648105467,
+      "learning_rate": 4.526250892335648e-05,
+      "loss": 0.6918,
+      "step": 3020
+    },
+    {
+      "epoch": 0.29400349311080926,
+      "grad_norm": 1.325947337933012,
+      "learning_rate": 4.524628463884743e-05,
+      "loss": 0.708,
+      "step": 3030
+    },
+    {
+      "epoch": 0.2949738016689307,
+      "grad_norm": 1.420774851884038,
+      "learning_rate": 4.5230060354338374e-05,
+      "loss": 0.7173,
+      "step": 3040
+    },
+    {
+      "epoch": 0.2959441102270522,
+      "grad_norm": 1.1753762610267322,
+      "learning_rate": 4.5213836069829324e-05,
+      "loss": 0.6654,
+      "step": 3050
+    },
+    {
+      "epoch": 0.29691441878517366,
+      "grad_norm": 1.526767780243654,
+      "learning_rate": 4.5197611785320274e-05,
+      "loss": 0.6321,
+      "step": 3060
+    },
+    {
+      "epoch": 0.2978847273432952,
+      "grad_norm": 1.8280158534376985,
+      "learning_rate": 4.518138750081122e-05,
+      "loss": 0.6647,
+      "step": 3070
+    },
+    {
+      "epoch": 0.29885503590141665,
+      "grad_norm": 1.4181228966831005,
+      "learning_rate": 4.516516321630217e-05,
+      "loss": 0.7338,
+      "step": 3080
+    },
+    {
+      "epoch": 0.2998253444595381,
+      "grad_norm": 1.7077549961994072,
+      "learning_rate": 4.514893893179311e-05,
+      "loss": 0.6272,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3007956530176596,
+      "grad_norm": 1.518857601222952,
+      "learning_rate": 4.513271464728406e-05,
+      "loss": 0.6184,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3017659615757811,
+      "grad_norm": 1.5688964933772704,
+      "learning_rate": 4.5116490362775e-05,
+      "loss": 0.6904,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3027362701339026,
+      "grad_norm": 1.346013000056015,
+      "learning_rate": 4.510026607826595e-05,
+      "loss": 0.6591,
+      "step": 3120
+    },
+    {
+      "epoch": 0.30370657869202405,
+      "grad_norm": 1.7289926750048026,
+      "learning_rate": 4.5084041793756895e-05,
+      "loss": 0.602,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3046768872501455,
+      "grad_norm": 1.634884511910698,
+      "learning_rate": 4.5067817509247845e-05,
+      "loss": 0.6344,
+      "step": 3140
+    },
+    {
+      "epoch": 0.30564719580826705,
+      "grad_norm": 1.2072013322253554,
+      "learning_rate": 4.5051593224738795e-05,
+      "loss": 0.6441,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3066175043663885,
+      "grad_norm": 1.4877523069726029,
+      "learning_rate": 4.503536894022974e-05,
+      "loss": 0.6711,
+      "step": 3160
+    },
+    {
+      "epoch": 0.30758781292451,
+      "grad_norm": 1.3820572558751547,
+      "learning_rate": 4.501914465572069e-05,
+      "loss": 0.7005,
+      "step": 3170
+    },
+    {
+      "epoch": 0.30855812148263145,
+      "grad_norm": 1.4545570501775118,
+      "learning_rate": 4.500292037121163e-05,
+      "loss": 0.6796,
+      "step": 3180
+    },
+    {
+      "epoch": 0.309528430040753,
+      "grad_norm": 1.4415846087886384,
+      "learning_rate": 4.498669608670258e-05,
+      "loss": 0.6753,
+      "step": 3190
+    },
+    {
+      "epoch": 0.31049873859887445,
+      "grad_norm": 1.253477394104618,
+      "learning_rate": 4.497047180219352e-05,
+      "loss": 0.745,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3114690471569959,
+      "grad_norm": 1.6938495814472803,
+      "learning_rate": 4.495424751768447e-05,
+      "loss": 0.6656,
+      "step": 3210
+    },
+    {
+      "epoch": 0.3124393557151174,
+      "grad_norm": 1.7041386856543572,
+      "learning_rate": 4.4938023233175416e-05,
+      "loss": 0.678,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3134096642732389,
+      "grad_norm": 1.945535890749437,
+      "learning_rate": 4.4921798948666366e-05,
+      "loss": 0.6741,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3143799728313604,
+      "grad_norm": 1.504845552046309,
+      "learning_rate": 4.4905574664157316e-05,
+      "loss": 0.7055,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31535028138948185,
+      "grad_norm": 1.5218102037928898,
+      "learning_rate": 4.488935037964826e-05,
+      "loss": 0.6664,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3163205899476033,
+      "grad_norm": 1.281958434229701,
+      "learning_rate": 4.487312609513921e-05,
+      "loss": 0.689,
+      "step": 3260
+    },
+    {
+      "epoch": 0.31729089850572484,
+      "grad_norm": 1.4017870385811553,
+      "learning_rate": 4.485690181063015e-05,
+      "loss": 0.656,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3182612070638463,
+      "grad_norm": 1.1312164184452325,
+      "learning_rate": 4.48406775261211e-05,
+      "loss": 0.6956,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3192315156219678,
+      "grad_norm": 1.5553483087810633,
+      "learning_rate": 4.4824453241612044e-05,
+      "loss": 0.627,
+      "step": 3290
+    },
+    {
+      "epoch": 0.32020182418008925,
+      "grad_norm": 1.5414421667820604,
+      "learning_rate": 4.4808228957102994e-05,
+      "loss": 0.6567,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32117213273821077,
+      "grad_norm": 1.3882171728699897,
+      "learning_rate": 4.479200467259394e-05,
+      "loss": 0.6478,
+      "step": 3310
+    },
+    {
+      "epoch": 0.32214244129633224,
+      "grad_norm": 1.2870563704453035,
+      "learning_rate": 4.477578038808489e-05,
+      "loss": 0.6258,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3231127498544537,
+      "grad_norm": 1.2176939882656843,
+      "learning_rate": 4.4759556103575836e-05,
+      "loss": 0.6831,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3240830584125752,
+      "grad_norm": 1.5859704946797195,
+      "learning_rate": 4.474333181906678e-05,
+      "loss": 0.6875,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3250533669706967,
+      "grad_norm": 1.3048860685179964,
+      "learning_rate": 4.472710753455773e-05,
+      "loss": 0.6797,
+      "step": 3350
+    },
+    {
+      "epoch": 0.32602367552881817,
+      "grad_norm": 1.4244729025826581,
+      "learning_rate": 4.471088325004867e-05,
+      "loss": 0.6671,
+      "step": 3360
+    },
+    {
+      "epoch": 0.32699398408693964,
+      "grad_norm": 1.664993378309651,
+      "learning_rate": 4.469465896553962e-05,
+      "loss": 0.6946,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3279642926450611,
+      "grad_norm": 1.4374586076882605,
+      "learning_rate": 4.4678434681030565e-05,
+      "loss": 0.7152,
+      "step": 3380
+    },
+    {
+      "epoch": 0.32893460120318263,
+      "grad_norm": 1.183729655148727,
+      "learning_rate": 4.4662210396521515e-05,
+      "loss": 0.6695,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3299049097613041,
+      "grad_norm": 1.3106066103858482,
+      "learning_rate": 4.4645986112012464e-05,
+      "loss": 0.681,
+      "step": 3400
+    },
+    {
+      "epoch": 0.33087521831942557,
+      "grad_norm": 1.9834642707734547,
+      "learning_rate": 4.462976182750341e-05,
+      "loss": 0.6544,
+      "step": 3410
+    },
+    {
+      "epoch": 0.33184552687754704,
+      "grad_norm": 1.2898353030549436,
+      "learning_rate": 4.461353754299436e-05,
+      "loss": 0.6516,
+      "step": 3420
+    },
+    {
+      "epoch": 0.33281583543566856,
+      "grad_norm": 1.3210516235415775,
+      "learning_rate": 4.45973132584853e-05,
+      "loss": 0.6684,
+      "step": 3430
+    },
+    {
+      "epoch": 0.33378614399379003,
+      "grad_norm": 1.4442553376708276,
+      "learning_rate": 4.458108897397625e-05,
+      "loss": 0.6673,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3347564525519115,
+      "grad_norm": 1.1500209299804536,
+      "learning_rate": 4.456486468946719e-05,
+      "loss": 0.7158,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33572676111003297,
+      "grad_norm": 1.31538694061074,
+      "learning_rate": 4.454864040495814e-05,
+      "loss": 0.647,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3366970696681545,
+      "grad_norm": 1.4081971729681526,
+      "learning_rate": 4.4532416120449086e-05,
+      "loss": 0.6964,
+      "step": 3470
+    },
+    {
+      "epoch": 0.33766737822627596,
+      "grad_norm": 1.4794888005976117,
+      "learning_rate": 4.4516191835940036e-05,
+      "loss": 0.6644,
+      "step": 3480
+    },
+    {
+      "epoch": 0.33863768678439743,
+      "grad_norm": 1.3524205715043236,
+      "learning_rate": 4.4499967551430985e-05,
+      "loss": 0.6421,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3396079953425189,
+      "grad_norm": 1.5105858350763188,
+      "learning_rate": 4.448374326692193e-05,
+      "loss": 0.7105,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3396079953425189,
+      "eval_loss": 0.7246462106704712,
+      "eval_runtime": 2471.0932,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3405783039006404,
+      "grad_norm": 1.4583137688880468,
+      "learning_rate": 4.446751898241288e-05,
+      "loss": 0.6087,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3415486124587619,
+      "grad_norm": 1.4130179418528255,
+      "learning_rate": 4.445129469790382e-05,
+      "loss": 0.6628,
+      "step": 3520
+    },
+    {
+      "epoch": 0.34251892101688336,
+      "grad_norm": 1.3225389415201874,
+      "learning_rate": 4.443507041339477e-05,
+      "loss": 0.625,
+      "step": 3530
+    },
+    {
+      "epoch": 0.34348922957500483,
+      "grad_norm": 1.8587881703474012,
+      "learning_rate": 4.4418846128885714e-05,
+      "loss": 0.6416,
+      "step": 3540
+    },
+    {
+      "epoch": 0.34445953813312635,
+      "grad_norm": 1.2742394540775415,
+      "learning_rate": 4.4402621844376664e-05,
+      "loss": 0.632,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3454298466912478,
+      "grad_norm": 1.64437983962982,
+      "learning_rate": 4.4386397559867607e-05,
+      "loss": 0.6383,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3464001552493693,
+      "grad_norm": 1.3231966805125737,
+      "learning_rate": 4.4370173275358556e-05,
+      "loss": 0.6841,
+      "step": 3570
+    },
+    {
+      "epoch": 0.34737046380749076,
+      "grad_norm": 1.167387762307211,
+      "learning_rate": 4.4353948990849506e-05,
+      "loss": 0.616,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3483407723656123,
+      "grad_norm": 1.4336867980215322,
+      "learning_rate": 4.433772470634045e-05,
+      "loss": 0.6195,
+      "step": 3590
+    },
+    {
+      "epoch": 0.34931108092373375,
+      "grad_norm": 1.403759520909343,
+      "learning_rate": 4.43215004218314e-05,
+      "loss": 0.6787,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3502813894818552,
+      "grad_norm": 1.4695150471417817,
+      "learning_rate": 4.430527613732234e-05,
+      "loss": 0.6433,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3512516980399767,
+      "grad_norm": 1.3322730792588582,
+      "learning_rate": 4.42890518528133e-05,
+      "loss": 0.6827,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3522220065980982,
+      "grad_norm": 1.2541946759332971,
+      "learning_rate": 4.427282756830424e-05,
+      "loss": 0.6767,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3531923151562197,
+      "grad_norm": 1.3220529304919946,
+      "learning_rate": 4.425660328379519e-05,
+      "loss": 0.6785,
+      "step": 3640
+    },
+    {
+      "epoch": 0.35416262371434115,
+      "grad_norm": 1.8159199916167459,
+      "learning_rate": 4.4240378999286134e-05,
+      "loss": 0.6504,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3551329322724626,
+      "grad_norm": 1.5513977697025123,
+      "learning_rate": 4.4224154714777084e-05,
+      "loss": 0.668,
+      "step": 3660
+    },
+    {
+      "epoch": 0.35610324083058414,
+      "grad_norm": 1.9383083203795937,
+      "learning_rate": 4.420793043026803e-05,
+      "loss": 0.6314,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3570735493887056,
+      "grad_norm": 1.5611559659864904,
+      "learning_rate": 4.419170614575898e-05,
+      "loss": 0.6617,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3580438579468271,
+      "grad_norm": 1.2579691148448051,
+      "learning_rate": 4.4175481861249927e-05,
+      "loss": 0.6641,
+      "step": 3690
+    },
+    {
+      "epoch": 0.35901416650494855,
+      "grad_norm": 1.8250727183362423,
+      "learning_rate": 4.415925757674087e-05,
+      "loss": 0.628,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3599844750630701,
+      "grad_norm": 1.8663848290613063,
+      "learning_rate": 4.414303329223182e-05,
+      "loss": 0.6819,
+      "step": 3710
+    },
+    {
+      "epoch": 0.36095478362119154,
+      "grad_norm": 0.998691604867035,
+      "learning_rate": 4.412680900772276e-05,
+      "loss": 0.6603,
+      "step": 3720
+    },
+    {
+      "epoch": 0.361925092179313,
+      "grad_norm": 1.5186010030355703,
+      "learning_rate": 4.411058472321371e-05,
+      "loss": 0.6608,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3628954007374345,
+      "grad_norm": 1.3444105870233574,
+      "learning_rate": 4.4094360438704655e-05,
+      "loss": 0.6684,
+      "step": 3740
+    },
+    {
+      "epoch": 0.363865709295556,
+      "grad_norm": 1.8433609494425311,
+      "learning_rate": 4.4078136154195605e-05,
+      "loss": 0.7007,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3648360178536775,
+      "grad_norm": 1.3736456316572478,
+      "learning_rate": 4.406191186968655e-05,
+      "loss": 0.6691,
+      "step": 3760
+    },
+    {
+      "epoch": 0.36580632641179894,
+      "grad_norm": 1.3845972262001316,
+      "learning_rate": 4.40456875851775e-05,
+      "loss": 0.699,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3667766349699204,
+      "grad_norm": 1.3202439744871353,
+      "learning_rate": 4.402946330066845e-05,
+      "loss": 0.6463,
+      "step": 3780
+    },
+    {
+      "epoch": 0.36774694352804194,
+      "grad_norm": 1.527206134833558,
+      "learning_rate": 4.401323901615939e-05,
+      "loss": 0.6261,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3687172520861634,
+      "grad_norm": 1.85165256113485,
+      "learning_rate": 4.399701473165034e-05,
+      "loss": 0.6268,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3696875606442849,
+      "grad_norm": 1.7774545505998887,
+      "learning_rate": 4.398079044714128e-05,
+      "loss": 0.674,
+      "step": 3810
+    },
+    {
+      "epoch": 0.37065786920240634,
+      "grad_norm": 1.1302938960409563,
+      "learning_rate": 4.396456616263223e-05,
+      "loss": 0.6814,
+      "step": 3820
+    },
+    {
+      "epoch": 0.37162817776052787,
+      "grad_norm": 1.5633030889510193,
+      "learning_rate": 4.3948341878123176e-05,
+      "loss": 0.6821,
+      "step": 3830
+    },
+    {
+      "epoch": 0.37259848631864934,
+      "grad_norm": 1.4226623590722947,
+      "learning_rate": 4.3932117593614126e-05,
+      "loss": 0.6721,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3735687948767708,
+      "grad_norm": 1.3788997600836588,
+      "learning_rate": 4.391589330910507e-05,
+      "loss": 0.6484,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3745391034348923,
+      "grad_norm": 1.9387617060516118,
+      "learning_rate": 4.389966902459602e-05,
+      "loss": 0.7059,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3755094119930138,
+      "grad_norm": 1.4159508230128068,
+      "learning_rate": 4.388344474008697e-05,
+      "loss": 0.6763,
+      "step": 3870
+    },
+    {
+      "epoch": 0.37647972055113527,
+      "grad_norm": 1.4395070409428377,
+      "learning_rate": 4.386722045557791e-05,
+      "loss": 0.6441,
+      "step": 3880
+    },
+    {
+      "epoch": 0.37745002910925674,
+      "grad_norm": 1.5259821305785304,
+      "learning_rate": 4.385099617106886e-05,
+      "loss": 0.7342,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3784203376673782,
+      "grad_norm": 1.58232361711394,
+      "learning_rate": 4.3834771886559804e-05,
+      "loss": 0.7058,
+      "step": 3900
+    },
+    {
+      "epoch": 0.37939064622549973,
+      "grad_norm": 1.2758386103028025,
+      "learning_rate": 4.3818547602050754e-05,
+      "loss": 0.6116,
+      "step": 3910
+    },
+    {
+      "epoch": 0.3803609547836212,
+      "grad_norm": 1.354723087565567,
+      "learning_rate": 4.38023233175417e-05,
+      "loss": 0.6709,
+      "step": 3920
+    },
+    {
+      "epoch": 0.38133126334174267,
+      "grad_norm": 1.5790529961766175,
+      "learning_rate": 4.3786099033032646e-05,
+      "loss": 0.7028,
+      "step": 3930
+    },
+    {
+      "epoch": 0.38230157189986413,
+      "grad_norm": 1.9609521386475108,
+      "learning_rate": 4.376987474852359e-05,
+      "loss": 0.5906,
+      "step": 3940
+    },
+    {
+      "epoch": 0.38327188045798566,
+      "grad_norm": 1.3349118770454718,
+      "learning_rate": 4.375365046401454e-05,
+      "loss": 0.6623,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3842421890161071,
+      "grad_norm": 1.2937167531350466,
+      "learning_rate": 4.373742617950549e-05,
+      "loss": 0.6369,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3852124975742286,
+      "grad_norm": 1.3419705920519438,
+      "learning_rate": 4.372120189499643e-05,
+      "loss": 0.7277,
+      "step": 3970
+    },
+    {
+      "epoch": 0.38618280613235006,
+      "grad_norm": 1.7073814830995901,
+      "learning_rate": 4.370497761048738e-05,
+      "loss": 0.6241,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3871531146904716,
+      "grad_norm": 1.373271354057727,
+      "learning_rate": 4.3688753325978325e-05,
+      "loss": 0.6481,
+      "step": 3990
+    },
+    {
+      "epoch": 0.38812342324859306,
+      "grad_norm": 1.7949929197599008,
+      "learning_rate": 4.3672529041469275e-05,
+      "loss": 0.6125,
+      "step": 4000
+    },
+    {
+      "epoch": 0.38812342324859306,
+      "eval_loss": 0.7137264609336853,
+      "eval_runtime": 2466.144,
+      "eval_samples_per_second": 0.727,
+      "eval_steps_per_second": 0.363,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3890937318067145,
+      "grad_norm": 0.8329579782452979,
+      "learning_rate": 4.365630475696022e-05,
+      "loss": 0.6341,
+      "step": 4010
+    },
+    {
+      "epoch": 0.390064040364836,
+      "grad_norm": 1.2617089929475087,
+      "learning_rate": 4.364008047245117e-05,
+      "loss": 0.671,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3910343489229575,
+      "grad_norm": 1.5577127482655397,
+      "learning_rate": 4.362385618794212e-05,
+      "loss": 0.6793,
+      "step": 4030
+    },
+    {
+      "epoch": 0.392004657481079,
+      "grad_norm": 1.6386238000694935,
+      "learning_rate": 4.360763190343306e-05,
+      "loss": 0.7014,
+      "step": 4040
+    },
+    {
+      "epoch": 0.39297496603920046,
+      "grad_norm": 1.5648938337659175,
+      "learning_rate": 4.359140761892401e-05,
+      "loss": 0.6867,
+      "step": 4050
+    },
+    {
+      "epoch": 0.3939452745973219,
+      "grad_norm": 1.6294675174253543,
+      "learning_rate": 4.357518333441495e-05,
+      "loss": 0.6706,
+      "step": 4060
+    },
+    {
+      "epoch": 0.39491558315544345,
+      "grad_norm": 1.549167727126926,
+      "learning_rate": 4.35589590499059e-05,
+      "loss": 0.6536,
+      "step": 4070
+    },
+    {
+      "epoch": 0.3958858917135649,
+      "grad_norm": 1.4182199112027882,
+      "learning_rate": 4.3542734765396846e-05,
+      "loss": 0.679,
+      "step": 4080
+    },
+    {
+      "epoch": 0.3968562002716864,
+      "grad_norm": 1.2821446945633657,
+      "learning_rate": 4.3526510480887795e-05,
+      "loss": 0.6643,
+      "step": 4090
+    },
+    {
+      "epoch": 0.39782650882980786,
+      "grad_norm": 1.6858376405816184,
+      "learning_rate": 4.351028619637874e-05,
+      "loss": 0.6394,
+      "step": 4100
+    },
+    {
+      "epoch": 0.3987968173879294,
+      "grad_norm": 1.4610732414303427,
+      "learning_rate": 4.349406191186969e-05,
+      "loss": 0.7123,
+      "step": 4110
+    },
+    {
+      "epoch": 0.39976712594605085,
+      "grad_norm": 1.504769761142886,
+      "learning_rate": 4.347783762736064e-05,
+      "loss": 0.6071,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4007374345041723,
+      "grad_norm": 1.6912171907525997,
+      "learning_rate": 4.346161334285158e-05,
+      "loss": 0.5923,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4017077430622938,
+      "grad_norm": 1.5783940240743402,
+      "learning_rate": 4.344538905834253e-05,
+      "loss": 0.5902,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4026780516204153,
+      "grad_norm": 1.300258602942144,
+      "learning_rate": 4.3429164773833474e-05,
+      "loss": 0.668,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4036483601785368,
+      "grad_norm": 1.3879301457682354,
+      "learning_rate": 4.3412940489324423e-05,
+      "loss": 0.6578,
+      "step": 4160
+    },
+    {
+      "epoch": 0.40461866873665825,
+      "grad_norm": 1.0418247331643762,
+      "learning_rate": 4.3396716204815366e-05,
+      "loss": 0.6539,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4055889772947797,
+      "grad_norm": 1.817053798243336,
+      "learning_rate": 4.3380491920306316e-05,
+      "loss": 0.6296,
+      "step": 4180
+    },
+    {
+      "epoch": 0.40655928585290124,
+      "grad_norm": 1.8784845030864437,
+      "learning_rate": 4.336426763579726e-05,
+      "loss": 0.6435,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4075295944110227,
+      "grad_norm": 1.3173177735824406,
+      "learning_rate": 4.334804335128821e-05,
+      "loss": 0.6467,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4084999029691442,
+      "grad_norm": 1.6609920860725593,
+      "learning_rate": 4.333181906677916e-05,
+      "loss": 0.6506,
+      "step": 4210
+    },
+    {
+      "epoch": 0.40947021152726565,
+      "grad_norm": 1.3422418231507067,
+      "learning_rate": 4.33155947822701e-05,
+      "loss": 0.6794,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4104405200853872,
+      "grad_norm": 1.6530218294736676,
+      "learning_rate": 4.329937049776105e-05,
+      "loss": 0.6458,
+      "step": 4230
+    },
+    {
+      "epoch": 0.41141082864350864,
+      "grad_norm": 1.7371351925998049,
+      "learning_rate": 4.3283146213251995e-05,
+      "loss": 0.6085,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4123811372016301,
+      "grad_norm": 1.1986220099049816,
+      "learning_rate": 4.3266921928742944e-05,
+      "loss": 0.6763,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4133514457597516,
+      "grad_norm": 1.4910668696454408,
+      "learning_rate": 4.325069764423389e-05,
+      "loss": 0.6943,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4143217543178731,
+      "grad_norm": 1.605222234480949,
+      "learning_rate": 4.323447335972484e-05,
+      "loss": 0.5683,
+      "step": 4270
+    },
+    {
+      "epoch": 0.41529206287599457,
+      "grad_norm": 1.6517210814862113,
+      "learning_rate": 4.321824907521578e-05,
+      "loss": 0.6121,
+      "step": 4280
+    },
+    {
+      "epoch": 0.41626237143411604,
+      "grad_norm": 1.5780093620130797,
+      "learning_rate": 4.320202479070673e-05,
+      "loss": 0.634,
+      "step": 4290
+    },
+    {
+      "epoch": 0.4172326799922375,
+      "grad_norm": 1.4948523984666717,
+      "learning_rate": 4.318580050619768e-05,
+      "loss": 0.6736,
+      "step": 4300
+    },
+    {
+      "epoch": 0.41820298855035903,
+      "grad_norm": 1.412186726690487,
+      "learning_rate": 4.316957622168862e-05,
+      "loss": 0.6645,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4191732971084805,
+      "grad_norm": 1.3014470286002153,
+      "learning_rate": 4.315335193717957e-05,
+      "loss": 0.611,
+      "step": 4320
+    },
+    {
+      "epoch": 0.42014360566660197,
+      "grad_norm": 1.5892208566989257,
+      "learning_rate": 4.3137127652670515e-05,
+      "loss": 0.6545,
+      "step": 4330
+    },
+    {
+      "epoch": 0.42111391422472344,
+      "grad_norm": 1.912773548887141,
+      "learning_rate": 4.3120903368161465e-05,
+      "loss": 0.6454,
+      "step": 4340
+    },
+    {
+      "epoch": 0.42208422278284496,
+      "grad_norm": 1.7286922382345113,
+      "learning_rate": 4.310467908365241e-05,
+      "loss": 0.6591,
+      "step": 4350
+    },
+    {
+      "epoch": 0.42305453134096643,
+      "grad_norm": 1.8839412216123284,
+      "learning_rate": 4.308845479914336e-05,
+      "loss": 0.5875,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4240248398990879,
+      "grad_norm": 1.433166987175659,
+      "learning_rate": 4.30722305146343e-05,
+      "loss": 0.6405,
+      "step": 4370
+    },
+    {
+      "epoch": 0.42499514845720937,
+      "grad_norm": 1.493313820047122,
+      "learning_rate": 4.305600623012525e-05,
+      "loss": 0.6421,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4259654570153309,
+      "grad_norm": 1.470156638554737,
+      "learning_rate": 4.30397819456162e-05,
+      "loss": 0.6212,
+      "step": 4390
+    },
+    {
+      "epoch": 0.42693576557345236,
+      "grad_norm": 1.801577389486602,
+      "learning_rate": 4.3023557661107143e-05,
+      "loss": 0.589,
+      "step": 4400
+    },
+    {
+      "epoch": 0.42790607413157383,
+      "grad_norm": 1.6174671178388904,
+      "learning_rate": 4.30073333765981e-05,
+      "loss": 0.6612,
+      "step": 4410
+    },
+    {
+      "epoch": 0.4288763826896953,
+      "grad_norm": 1.5169174455759806,
+      "learning_rate": 4.299110909208904e-05,
+      "loss": 0.6418,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4298466912478168,
+      "grad_norm": 1.7159303308076814,
+      "learning_rate": 4.297488480757999e-05,
+      "loss": 0.6363,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4308169998059383,
+      "grad_norm": 1.9158386115366701,
+      "learning_rate": 4.2958660523070936e-05,
+      "loss": 0.6639,
+      "step": 4440
+    },
+    {
+      "epoch": 0.43178730836405976,
+      "grad_norm": 1.6614034382324376,
+      "learning_rate": 4.2942436238561885e-05,
+      "loss": 0.6546,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43275761692218123,
+      "grad_norm": 1.528595189219001,
+      "learning_rate": 4.292621195405283e-05,
+      "loss": 0.6151,
+      "step": 4460
+    },
+    {
+      "epoch": 0.43372792548030276,
+      "grad_norm": 1.297393909600355,
+      "learning_rate": 4.290998766954378e-05,
+      "loss": 0.6142,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4346982340384242,
+      "grad_norm": 1.6025277242190177,
+      "learning_rate": 4.289376338503472e-05,
+      "loss": 0.6415,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4356685425965457,
+      "grad_norm": 1.550877285078,
+      "learning_rate": 4.287753910052567e-05,
+      "loss": 0.6896,
+      "step": 4490
+    },
+    {
+      "epoch": 0.43663885115466716,
+      "grad_norm": 1.8134887382719538,
+      "learning_rate": 4.286131481601662e-05,
+      "loss": 0.674,
+      "step": 4500
+    },
+    {
+      "epoch": 0.43663885115466716,
+      "eval_loss": 0.70569908618927,
+      "eval_runtime": 2470.7726,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4376091597127887,
+      "grad_norm": 1.5715320462855416,
+      "learning_rate": 4.2845090531507564e-05,
+      "loss": 0.6107,
+      "step": 4510
+    },
+    {
+      "epoch": 0.43857946827091016,
+      "grad_norm": 1.659219569116577,
+      "learning_rate": 4.2828866246998514e-05,
+      "loss": 0.554,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4395497768290316,
+      "grad_norm": 1.5014148593583845,
+      "learning_rate": 4.2812641962489457e-05,
+      "loss": 0.6541,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4405200853871531,
+      "grad_norm": 1.2080187388338057,
+      "learning_rate": 4.2796417677980406e-05,
+      "loss": 0.6704,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4414903939452746,
+      "grad_norm": 1.4344070923299042,
+      "learning_rate": 4.278019339347135e-05,
+      "loss": 0.637,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4424607025033961,
+      "grad_norm": 1.9816113300875973,
+      "learning_rate": 4.27639691089623e-05,
+      "loss": 0.6036,
+      "step": 4560
+    },
+    {
+      "epoch": 0.44343101106151755,
+      "grad_norm": 1.270379951275499,
+      "learning_rate": 4.274774482445324e-05,
+      "loss": 0.6481,
+      "step": 4570
+    },
+    {
+      "epoch": 0.444401319619639,
+      "grad_norm": 1.559280795643455,
+      "learning_rate": 4.273152053994419e-05,
+      "loss": 0.62,
+      "step": 4580
+    },
+    {
+      "epoch": 0.44537162817776055,
+      "grad_norm": 1.8377481053949813,
+      "learning_rate": 4.271529625543514e-05,
+      "loss": 0.6662,
+      "step": 4590
+    },
+    {
+      "epoch": 0.446341936735882,
+      "grad_norm": 1.3952819132659193,
+      "learning_rate": 4.2699071970926085e-05,
+      "loss": 0.6969,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4473122452940035,
+      "grad_norm": 1.5967451654603113,
+      "learning_rate": 4.2682847686417034e-05,
+      "loss": 0.6604,
+      "step": 4610
+    },
+    {
+      "epoch": 0.44828255385212495,
+      "grad_norm": 1.5849329398639342,
+      "learning_rate": 4.266662340190798e-05,
+      "loss": 0.6589,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4492528624102465,
+      "grad_norm": 1.7238645892594557,
+      "learning_rate": 4.265039911739893e-05,
+      "loss": 0.6658,
+      "step": 4630
+    },
+    {
+      "epoch": 0.45022317096836795,
+      "grad_norm": 1.7849098949361206,
+      "learning_rate": 4.263417483288987e-05,
+      "loss": 0.6507,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4511934795264894,
+      "grad_norm": 1.3539873989398281,
+      "learning_rate": 4.261795054838082e-05,
+      "loss": 0.631,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4521637880846109,
+      "grad_norm": 1.6177606598072791,
+      "learning_rate": 4.260172626387176e-05,
+      "loss": 0.7667,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4531340966427324,
+      "grad_norm": 1.7683798129102917,
+      "learning_rate": 4.258550197936271e-05,
+      "loss": 0.6182,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4541044052008539,
+      "grad_norm": 1.5536790285936453,
+      "learning_rate": 4.256927769485366e-05,
+      "loss": 0.6515,
+      "step": 4680
+    },
+    {
+      "epoch": 0.45507471375897535,
+      "grad_norm": 1.4626963492189242,
+      "learning_rate": 4.2553053410344605e-05,
+      "loss": 0.5595,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4560450223170968,
+      "grad_norm": 1.8455989589758681,
+      "learning_rate": 4.2536829125835555e-05,
+      "loss": 0.6382,
+      "step": 4700
+    },
+    {
+      "epoch": 0.45701533087521834,
+      "grad_norm": 1.8260482347716946,
+      "learning_rate": 4.25206048413265e-05,
+      "loss": 0.6708,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4579856394333398,
+      "grad_norm": 1.506588655076192,
+      "learning_rate": 4.250438055681745e-05,
+      "loss": 0.6138,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4589559479914613,
+      "grad_norm": 1.7415606928937182,
+      "learning_rate": 4.248815627230839e-05,
+      "loss": 0.6688,
+      "step": 4730
+    },
+    {
+      "epoch": 0.45992625654958275,
+      "grad_norm": 1.2860177177533143,
+      "learning_rate": 4.247193198779934e-05,
+      "loss": 0.6496,
+      "step": 4740
+    },
+    {
+      "epoch": 0.46089656510770427,
+      "grad_norm": 1.419953340190783,
+      "learning_rate": 4.245570770329029e-05,
+      "loss": 0.6399,
+      "step": 4750
+    },
+    {
+      "epoch": 0.46186687366582574,
+      "grad_norm": 1.4197804283366926,
+      "learning_rate": 4.2439483418781234e-05,
+      "loss": 0.572,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4628371822239472,
+      "grad_norm": 1.1761100117238914,
+      "learning_rate": 4.242325913427218e-05,
+      "loss": 0.6099,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4638074907820687,
+      "grad_norm": 1.5767612487613212,
+      "learning_rate": 4.2407034849763126e-05,
+      "loss": 0.6223,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4647777993401902,
+      "grad_norm": 1.8050008733247063,
+      "learning_rate": 4.2390810565254076e-05,
+      "loss": 0.5695,
+      "step": 4790
+    },
+    {
+      "epoch": 0.46574810789831167,
+      "grad_norm": 1.3088703288484584,
+      "learning_rate": 4.237458628074502e-05,
+      "loss": 0.5723,
+      "step": 4800
+    },
+    {
+      "epoch": 0.46671841645643314,
+      "grad_norm": 1.7711140961973422,
+      "learning_rate": 4.235836199623597e-05,
+      "loss": 0.6173,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4676887250145546,
+      "grad_norm": 1.6431929005817145,
+      "learning_rate": 4.234213771172691e-05,
+      "loss": 0.5982,
+      "step": 4820
+    },
+    {
+      "epoch": 0.46865903357267613,
+      "grad_norm": 1.557431489902951,
+      "learning_rate": 4.232591342721786e-05,
+      "loss": 0.6098,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4696293421307976,
+      "grad_norm": 1.4479656995240782,
+      "learning_rate": 4.230968914270881e-05,
+      "loss": 0.5699,
+      "step": 4840
+    },
+    {
+      "epoch": 0.47059965068891907,
+      "grad_norm": 1.6046344258439647,
+      "learning_rate": 4.2293464858199754e-05,
+      "loss": 0.6243,
+      "step": 4850
+    },
+    {
+      "epoch": 0.47156995924704054,
+      "grad_norm": 1.252218532607539,
+      "learning_rate": 4.2277240573690704e-05,
+      "loss": 0.5787,
+      "step": 4860
+    },
+    {
+      "epoch": 0.47254026780516206,
+      "grad_norm": 1.4372595855894126,
+      "learning_rate": 4.226101628918165e-05,
+      "loss": 0.6483,
+      "step": 4870
+    },
+    {
+      "epoch": 0.47351057636328353,
+      "grad_norm": 1.6112811676393963,
+      "learning_rate": 4.22447920046726e-05,
+      "loss": 0.6042,
+      "step": 4880
+    },
+    {
+      "epoch": 0.474480884921405,
+      "grad_norm": 1.6387669351547591,
+      "learning_rate": 4.222856772016354e-05,
+      "loss": 0.6176,
+      "step": 4890
+    },
+    {
+      "epoch": 0.47545119347952647,
+      "grad_norm": 1.4893296879883764,
+      "learning_rate": 4.221234343565449e-05,
+      "loss": 0.5976,
+      "step": 4900
+    },
+    {
+      "epoch": 0.476421502037648,
+      "grad_norm": 1.663980815080282,
+      "learning_rate": 4.219611915114543e-05,
+      "loss": 0.6655,
+      "step": 4910
+    },
+    {
+      "epoch": 0.47739181059576946,
+      "grad_norm": 1.3387695933594599,
+      "learning_rate": 4.217989486663638e-05,
+      "loss": 0.6166,
+      "step": 4920
+    },
+    {
+      "epoch": 0.47836211915389093,
+      "grad_norm": 1.7319521884893254,
+      "learning_rate": 4.216367058212733e-05,
+      "loss": 0.6225,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4793324277120124,
+      "grad_norm": 1.4563922038982844,
+      "learning_rate": 4.2147446297618275e-05,
+      "loss": 0.6351,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4803027362701339,
+      "grad_norm": 1.7473358339447822,
+      "learning_rate": 4.2131222013109225e-05,
+      "loss": 0.6205,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4812730448282554,
+      "grad_norm": 1.4006136660964967,
+      "learning_rate": 4.211499772860017e-05,
+      "loss": 0.6992,
+      "step": 4960
+    },
+    {
+      "epoch": 0.48224335338637686,
+      "grad_norm": 1.7446969484471915,
+      "learning_rate": 4.209877344409112e-05,
+      "loss": 0.6251,
+      "step": 4970
+    },
+    {
+      "epoch": 0.48321366194449833,
+      "grad_norm": 1.3789648220512414,
+      "learning_rate": 4.208254915958206e-05,
+      "loss": 0.5856,
+      "step": 4980
+    },
+    {
+      "epoch": 0.48418397050261985,
+      "grad_norm": 1.77919007957107,
+      "learning_rate": 4.206632487507301e-05,
+      "loss": 0.621,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4851542790607413,
+      "grad_norm": 1.5926084629766706,
+      "learning_rate": 4.2050100590563953e-05,
+      "loss": 0.6542,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4851542790607413,
+      "eval_loss": 0.6975318789482117,
+      "eval_runtime": 2469.0533,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4861245876188628,
+      "grad_norm": 1.7733999980186248,
+      "learning_rate": 4.20338763060549e-05,
+      "loss": 0.6943,
+      "step": 5010
+    },
+    {
+      "epoch": 0.48709489617698426,
+      "grad_norm": 1.7221032223753934,
+      "learning_rate": 4.201765202154585e-05,
+      "loss": 0.6338,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4880652047351058,
+      "grad_norm": 1.7180735514930228,
+      "learning_rate": 4.2001427737036796e-05,
+      "loss": 0.6698,
+      "step": 5030
+    },
+    {
+      "epoch": 0.48903551329322725,
+      "grad_norm": 1.698242629280347,
+      "learning_rate": 4.1985203452527746e-05,
+      "loss": 0.6972,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4900058218513487,
+      "grad_norm": 1.8269714939747912,
+      "learning_rate": 4.196897916801869e-05,
+      "loss": 0.6212,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4909761304094702,
+      "grad_norm": 1.4693827904505679,
+      "learning_rate": 4.195275488350964e-05,
+      "loss": 0.5933,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4919464389675917,
+      "grad_norm": 1.49361517085645,
+      "learning_rate": 4.193653059900058e-05,
+      "loss": 0.5489,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4929167475257132,
+      "grad_norm": 1.5630723710374232,
+      "learning_rate": 4.192030631449153e-05,
+      "loss": 0.6578,
+      "step": 5080
+    },
+    {
+      "epoch": 0.49388705608383465,
+      "grad_norm": 1.6495843346442778,
+      "learning_rate": 4.1904082029982474e-05,
+      "loss": 0.6906,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4948573646419561,
+      "grad_norm": 1.7983678208711196,
+      "learning_rate": 4.1887857745473424e-05,
+      "loss": 0.5486,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49582767320007765,
+      "grad_norm": 2.00722752034129,
+      "learning_rate": 4.1871633460964374e-05,
+      "loss": 0.648,
+      "step": 5110
+    },
+    {
+      "epoch": 0.4967979817581991,
+      "grad_norm": 1.7918897206429327,
+      "learning_rate": 4.185540917645532e-05,
+      "loss": 0.6693,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4977682903163206,
+      "grad_norm": 1.3410097713324785,
+      "learning_rate": 4.1839184891946267e-05,
+      "loss": 0.6065,
+      "step": 5130
+    },
+    {
+      "epoch": 0.49873859887444205,
+      "grad_norm": 1.8485275360772722,
+      "learning_rate": 4.182296060743721e-05,
+      "loss": 0.6802,
+      "step": 5140
+    },
+    {
+      "epoch": 0.4997089074325636,
+      "grad_norm": 1.923253820990962,
+      "learning_rate": 4.180673632292816e-05,
+      "loss": 0.6014,
+      "step": 5150
+    },
+    {
+      "epoch": 0.500679215990685,
+      "grad_norm": 1.325223633752785,
+      "learning_rate": 4.17905120384191e-05,
+      "loss": 0.5787,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5016495245488065,
+      "grad_norm": 1.7937826702440574,
+      "learning_rate": 4.177428775391005e-05,
+      "loss": 0.6197,
+      "step": 5170
+    },
+    {
+      "epoch": 0.502619833106928,
+      "grad_norm": 1.4050667202069218,
+      "learning_rate": 4.1758063469401e-05,
+      "loss": 0.6,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5035901416650495,
+      "grad_norm": 1.5656335055581363,
+      "learning_rate": 4.174183918489195e-05,
+      "loss": 0.6924,
+      "step": 5190
+    },
+    {
+      "epoch": 0.504560450223171,
+      "grad_norm": 1.3325372640233277,
+      "learning_rate": 4.1725614900382895e-05,
+      "loss": 0.6477,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5055307587812925,
+      "grad_norm": 1.5612035852842387,
+      "learning_rate": 4.1709390615873844e-05,
+      "loss": 0.6372,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5065010673394139,
+      "grad_norm": 1.6164770277307943,
+      "learning_rate": 4.1693166331364794e-05,
+      "loss": 0.5822,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5074713758975354,
+      "grad_norm": 1.8957196740567803,
+      "learning_rate": 4.167694204685574e-05,
+      "loss": 0.5695,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5084416844556569,
+      "grad_norm": 1.505057761308431,
+      "learning_rate": 4.166071776234669e-05,
+      "loss": 0.5893,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5094119930137784,
+      "grad_norm": 1.2392435413477463,
+      "learning_rate": 4.164449347783763e-05,
+      "loss": 0.6641,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5103823015718999,
+      "grad_norm": 1.8002682332596205,
+      "learning_rate": 4.162826919332858e-05,
+      "loss": 0.5892,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5113526101300213,
+      "grad_norm": 1.5161558066332184,
+      "learning_rate": 4.161204490881952e-05,
+      "loss": 0.5985,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5123229186881428,
+      "grad_norm": 1.3773663290971068,
+      "learning_rate": 4.159582062431047e-05,
+      "loss": 0.6244,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5132932272462644,
+      "grad_norm": 1.8649060352228473,
+      "learning_rate": 4.1579596339801415e-05,
+      "loss": 0.6144,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5142635358043858,
+      "grad_norm": 1.711468575333758,
+      "learning_rate": 4.1563372055292365e-05,
+      "loss": 0.5889,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5152338443625073,
+      "grad_norm": 1.4531536302097425,
+      "learning_rate": 4.1547147770783315e-05,
+      "loss": 0.6063,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5162041529206287,
+      "grad_norm": 1.6628035885794548,
+      "learning_rate": 4.153092348627426e-05,
+      "loss": 0.5872,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5171744614787502,
+      "grad_norm": 1.43386966167638,
+      "learning_rate": 4.151469920176521e-05,
+      "loss": 0.5836,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5181447700368718,
+      "grad_norm": 1.6338584720484304,
+      "learning_rate": 4.149847491725615e-05,
+      "loss": 0.583,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5191150785949932,
+      "grad_norm": 1.602663089927568,
+      "learning_rate": 4.14822506327471e-05,
+      "loss": 0.6466,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5200853871531147,
+      "grad_norm": 0.993656188560276,
+      "learning_rate": 4.1466026348238044e-05,
+      "loss": 0.5972,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5210556957112362,
+      "grad_norm": 1.5353879720655148,
+      "learning_rate": 4.144980206372899e-05,
+      "loss": 0.6159,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5220260042693576,
+      "grad_norm": 1.4710027502404226,
+      "learning_rate": 4.143357777921994e-05,
+      "loss": 0.6231,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5229963128274792,
+      "grad_norm": 1.3154974116618938,
+      "learning_rate": 4.1417353494710886e-05,
+      "loss": 0.6414,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5239666213856006,
+      "grad_norm": 1.5917256061619933,
+      "learning_rate": 4.1401129210201836e-05,
+      "loss": 0.5574,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5249369299437221,
+      "grad_norm": 1.999701613939348,
+      "learning_rate": 4.138490492569278e-05,
+      "loss": 0.6659,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5259072385018436,
+      "grad_norm": 1.158059375940914,
+      "learning_rate": 4.136868064118373e-05,
+      "loss": 0.5702,
+      "step": 5420
+    },
+    {
+      "epoch": 0.526877547059965,
+      "grad_norm": 1.6503529559993917,
+      "learning_rate": 4.135245635667467e-05,
+      "loss": 0.5866,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5278478556180866,
+      "grad_norm": 1.778595452682844,
+      "learning_rate": 4.133623207216562e-05,
+      "loss": 0.6727,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5288181641762081,
+      "grad_norm": 2.153378806689067,
+      "learning_rate": 4.1320007787656564e-05,
+      "loss": 0.5849,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5297884727343295,
+      "grad_norm": 1.5480145123110607,
+      "learning_rate": 4.1303783503147514e-05,
+      "loss": 0.5985,
+      "step": 5460
+    },
+    {
+      "epoch": 0.530758781292451,
+      "grad_norm": 1.3074515838089584,
+      "learning_rate": 4.1287559218638464e-05,
+      "loss": 0.6567,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5317290898505724,
+      "grad_norm": 1.2634236320868193,
+      "learning_rate": 4.127133493412941e-05,
+      "loss": 0.6744,
+      "step": 5480
+    },
+    {
+      "epoch": 0.532699398408694,
+      "grad_norm": 1.4158920202942755,
+      "learning_rate": 4.125511064962036e-05,
+      "loss": 0.6089,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5336697069668155,
+      "grad_norm": 1.530835103672291,
+      "learning_rate": 4.12388863651113e-05,
+      "loss": 0.6637,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5336697069668155,
+      "eval_loss": 0.6922717094421387,
+      "eval_runtime": 2471.913,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5346400155249369,
+      "grad_norm": 1.0979844633621818,
+      "learning_rate": 4.122266208060225e-05,
+      "loss": 0.6711,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5356103240830584,
+      "grad_norm": 1.9531703151807043,
+      "learning_rate": 4.120643779609319e-05,
+      "loss": 0.634,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5365806326411799,
+      "grad_norm": 1.4345990864009392,
+      "learning_rate": 4.119021351158414e-05,
+      "loss": 0.6289,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5375509411993014,
+      "grad_norm": 1.5002874747132873,
+      "learning_rate": 4.1173989227075085e-05,
+      "loss": 0.6655,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5385212497574229,
+      "grad_norm": 1.4114133378596565,
+      "learning_rate": 4.1157764942566035e-05,
+      "loss": 0.6314,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5394915583155443,
+      "grad_norm": 1.2792208401822267,
+      "learning_rate": 4.1141540658056985e-05,
+      "loss": 0.5622,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5404618668736658,
+      "grad_norm": 1.4436849148969537,
+      "learning_rate": 4.112531637354793e-05,
+      "loss": 0.6758,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5414321754317873,
+      "grad_norm": 1.495469226735889,
+      "learning_rate": 4.110909208903888e-05,
+      "loss": 0.6457,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5424024839899088,
+      "grad_norm": 1.5696729181281173,
+      "learning_rate": 4.109286780452982e-05,
+      "loss": 0.5434,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5433727925480303,
+      "grad_norm": 1.4884099597371836,
+      "learning_rate": 4.107664352002077e-05,
+      "loss": 0.6602,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5443431011061518,
+      "grad_norm": 1.364422329620206,
+      "learning_rate": 4.106041923551171e-05,
+      "loss": 0.6093,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5453134096642732,
+      "grad_norm": 1.710525492450016,
+      "learning_rate": 4.104419495100266e-05,
+      "loss": 0.5733,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5462837182223947,
+      "grad_norm": 1.8645615729328109,
+      "learning_rate": 4.1027970666493606e-05,
+      "loss": 0.6301,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5472540267805162,
+      "grad_norm": 2.1212649730065403,
+      "learning_rate": 4.1011746381984556e-05,
+      "loss": 0.6489,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5482243353386377,
+      "grad_norm": 1.3978835293983003,
+      "learning_rate": 4.0995522097475506e-05,
+      "loss": 0.6651,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5491946438967592,
+      "grad_norm": 1.968855731350402,
+      "learning_rate": 4.097929781296645e-05,
+      "loss": 0.5849,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5501649524548806,
+      "grad_norm": 1.3003479487793523,
+      "learning_rate": 4.09630735284574e-05,
+      "loss": 0.5981,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5511352610130021,
+      "grad_norm": 1.3093675797291962,
+      "learning_rate": 4.094684924394834e-05,
+      "loss": 0.6001,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5521055695711237,
+      "grad_norm": 1.386619668616355,
+      "learning_rate": 4.093062495943929e-05,
+      "loss": 0.587,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5530758781292451,
+      "grad_norm": 2.1925053494451108,
+      "learning_rate": 4.0914400674930234e-05,
+      "loss": 0.6748,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5540461866873666,
+      "grad_norm": 1.401533269192193,
+      "learning_rate": 4.0898176390421184e-05,
+      "loss": 0.5987,
+      "step": 5710
+    },
+    {
+      "epoch": 0.555016495245488,
+      "grad_norm": 1.8417454170154262,
+      "learning_rate": 4.088195210591213e-05,
+      "loss": 0.6169,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5559868038036095,
+      "grad_norm": 1.70211748961807,
+      "learning_rate": 4.086572782140308e-05,
+      "loss": 0.6231,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5569571123617311,
+      "grad_norm": 1.449836249162655,
+      "learning_rate": 4.0849503536894026e-05,
+      "loss": 0.6636,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5579274209198525,
+      "grad_norm": 1.411704552119965,
+      "learning_rate": 4.083327925238497e-05,
+      "loss": 0.6334,
+      "step": 5750
+    },
+    {
+      "epoch": 0.558897729477974,
+      "grad_norm": 1.3373247464479125,
+      "learning_rate": 4.081705496787592e-05,
+      "loss": 0.6127,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5598680380360955,
+      "grad_norm": 1.6669305203635734,
+      "learning_rate": 4.080083068336686e-05,
+      "loss": 0.6419,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5608383465942169,
+      "grad_norm": 1.3599639094085172,
+      "learning_rate": 4.078460639885781e-05,
+      "loss": 0.6838,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5618086551523385,
+      "grad_norm": 1.5754322431175416,
+      "learning_rate": 4.0768382114348755e-05,
+      "loss": 0.5997,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5627789637104599,
+      "grad_norm": 1.667207521590911,
+      "learning_rate": 4.0752157829839705e-05,
+      "loss": 0.6199,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5637492722685814,
+      "grad_norm": 1.5869372089733027,
+      "learning_rate": 4.0735933545330654e-05,
+      "loss": 0.5779,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5647195808267029,
+      "grad_norm": 1.7150097071735784,
+      "learning_rate": 4.07197092608216e-05,
+      "loss": 0.5978,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5656898893848243,
+      "grad_norm": 1.1835353092728575,
+      "learning_rate": 4.070348497631255e-05,
+      "loss": 0.6781,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5666601979429459,
+      "grad_norm": 1.7307613010314937,
+      "learning_rate": 4.068726069180349e-05,
+      "loss": 0.634,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5676305065010674,
+      "grad_norm": 1.6364191249486493,
+      "learning_rate": 4.067103640729444e-05,
+      "loss": 0.6377,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5686008150591888,
+      "grad_norm": 1.4754833764988036,
+      "learning_rate": 4.065481212278538e-05,
+      "loss": 0.6148,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5695711236173103,
+      "grad_norm": 1.8389583610281375,
+      "learning_rate": 4.063858783827633e-05,
+      "loss": 0.6739,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5705414321754317,
+      "grad_norm": 1.4669099461048227,
+      "learning_rate": 4.0622363553767276e-05,
+      "loss": 0.598,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5715117407335533,
+      "grad_norm": 1.609973267529918,
+      "learning_rate": 4.0606139269258226e-05,
+      "loss": 0.5476,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5724820492916748,
+      "grad_norm": 1.6951117185016165,
+      "learning_rate": 4.0589914984749175e-05,
+      "loss": 0.595,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5734523578497962,
+      "grad_norm": 1.494158253886906,
+      "learning_rate": 4.057369070024012e-05,
+      "loss": 0.6886,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5744226664079177,
+      "grad_norm": 1.3036559330975726,
+      "learning_rate": 4.055746641573107e-05,
+      "loss": 0.5986,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5753929749660393,
+      "grad_norm": 1.4737709906961922,
+      "learning_rate": 4.054124213122201e-05,
+      "loss": 0.615,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5763632835241607,
+      "grad_norm": 1.4022175413897375,
+      "learning_rate": 4.052501784671296e-05,
+      "loss": 0.6398,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5773335920822822,
+      "grad_norm": 1.433959705605443,
+      "learning_rate": 4.0508793562203904e-05,
+      "loss": 0.6415,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5783039006404036,
+      "grad_norm": 1.7123129741732308,
+      "learning_rate": 4.0492569277694854e-05,
+      "loss": 0.6216,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5792742091985251,
+      "grad_norm": 2.131052124619936,
+      "learning_rate": 4.04763449931858e-05,
+      "loss": 0.5596,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5802445177566466,
+      "grad_norm": 1.2233223645362516,
+      "learning_rate": 4.046012070867675e-05,
+      "loss": 0.6023,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5812148263147681,
+      "grad_norm": 1.734811390420522,
+      "learning_rate": 4.0443896424167696e-05,
+      "loss": 0.6347,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5821851348728896,
+      "grad_norm": 1.626024609025115,
+      "learning_rate": 4.0427672139658646e-05,
+      "loss": 0.5688,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5821851348728896,
+      "eval_loss": 0.6903010606765747,
+      "eval_runtime": 2473.5711,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.362,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5831554434310111,
+      "grad_norm": 1.3892850877811105,
+      "learning_rate": 4.0411447855149596e-05,
+      "loss": 0.6131,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5841257519891325,
+      "grad_norm": 1.6290044308163973,
+      "learning_rate": 4.039522357064054e-05,
+      "loss": 0.6319,
+      "step": 6020
+    },
+    {
+      "epoch": 0.585096060547254,
+      "grad_norm": 1.7770012777694764,
+      "learning_rate": 4.037899928613149e-05,
+      "loss": 0.5756,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5860663691053755,
+      "grad_norm": 0.9954028008191703,
+      "learning_rate": 4.036277500162243e-05,
+      "loss": 0.6161,
+      "step": 6040
+    },
+    {
+      "epoch": 0.587036677663497,
+      "grad_norm": 1.7893541919562175,
+      "learning_rate": 4.034655071711338e-05,
+      "loss": 0.62,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5880069862216185,
+      "grad_norm": 1.6500512739247042,
+      "learning_rate": 4.0330326432604324e-05,
+      "loss": 0.5928,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5889772947797399,
+      "grad_norm": 1.6568317215206447,
+      "learning_rate": 4.0314102148095274e-05,
+      "loss": 0.6043,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5899476033378614,
+      "grad_norm": 2.006336465212855,
+      "learning_rate": 4.029787786358622e-05,
+      "loss": 0.6283,
+      "step": 6080
+    },
+    {
+      "epoch": 0.590917911895983,
+      "grad_norm": 1.481658511307882,
+      "learning_rate": 4.028165357907717e-05,
+      "loss": 0.6229,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5918882204541044,
+      "grad_norm": 1.7467752898199094,
+      "learning_rate": 4.0265429294568117e-05,
+      "loss": 0.5997,
+      "step": 6100
+    },
+    {
+      "epoch": 0.5928585290122259,
+      "grad_norm": 1.747997366501937,
+      "learning_rate": 4.024920501005906e-05,
+      "loss": 0.6441,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5938288375703473,
+      "grad_norm": 1.6963969244893895,
+      "learning_rate": 4.023298072555001e-05,
+      "loss": 0.6059,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5947991461284688,
+      "grad_norm": 1.335184560881826,
+      "learning_rate": 4.021675644104095e-05,
+      "loss": 0.5627,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5957694546865904,
+      "grad_norm": 1.5256295408049876,
+      "learning_rate": 4.02005321565319e-05,
+      "loss": 0.578,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5967397632447118,
+      "grad_norm": 1.7615333752357474,
+      "learning_rate": 4.0184307872022845e-05,
+      "loss": 0.6046,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5977100718028333,
+      "grad_norm": 1.9321393707932597,
+      "learning_rate": 4.0168083587513795e-05,
+      "loss": 0.615,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5986803803609548,
+      "grad_norm": 1.2267052175289888,
+      "learning_rate": 4.015185930300474e-05,
+      "loss": 0.6238,
+      "step": 6170
+    },
+    {
+      "epoch": 0.5996506889190762,
+      "grad_norm": 1.8092624768121868,
+      "learning_rate": 4.013563501849569e-05,
+      "loss": 0.6514,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6006209974771978,
+      "grad_norm": 1.546108807816704,
+      "learning_rate": 4.011941073398664e-05,
+      "loss": 0.6071,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6015913060353192,
+      "grad_norm": 1.8186789877657976,
+      "learning_rate": 4.010318644947758e-05,
+      "loss": 0.6624,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6025616145934407,
+      "grad_norm": 1.6349446430924426,
+      "learning_rate": 4.008696216496853e-05,
+      "loss": 0.6208,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6035319231515622,
+      "grad_norm": 1.6574782074117176,
+      "learning_rate": 4.007073788045947e-05,
+      "loss": 0.6067,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6045022317096836,
+      "grad_norm": 1.372441926371392,
+      "learning_rate": 4.005451359595042e-05,
+      "loss": 0.5805,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6054725402678052,
+      "grad_norm": 2.1366961584133164,
+      "learning_rate": 4.0038289311441366e-05,
+      "loss": 0.6797,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6064428488259267,
+      "grad_norm": 1.8553936516223448,
+      "learning_rate": 4.0022065026932316e-05,
+      "loss": 0.582,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6074131573840481,
+      "grad_norm": 1.95303083914795,
+      "learning_rate": 4.000584074242326e-05,
+      "loss": 0.6474,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6083834659421696,
+      "grad_norm": 1.6884162400220937,
+      "learning_rate": 3.998961645791421e-05,
+      "loss": 0.6339,
+      "step": 6270
+    },
+    {
+      "epoch": 0.609353774500291,
+      "grad_norm": 1.7825690304753616,
+      "learning_rate": 3.997339217340516e-05,
+      "loss": 0.5568,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6103240830584126,
+      "grad_norm": 1.5564493386693086,
+      "learning_rate": 3.99571678888961e-05,
+      "loss": 0.602,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6112943916165341,
+      "grad_norm": 1.667415447715766,
+      "learning_rate": 3.994094360438705e-05,
+      "loss": 0.5877,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6122647001746555,
+      "grad_norm": 1.6232084554425263,
+      "learning_rate": 3.9924719319877994e-05,
+      "loss": 0.6247,
+      "step": 6310
+    },
+    {
+      "epoch": 0.613235008732777,
+      "grad_norm": 1.7806635141428606,
+      "learning_rate": 3.9908495035368944e-05,
+      "loss": 0.5651,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6142053172908986,
+      "grad_norm": 1.2898375089967073,
+      "learning_rate": 3.989227075085989e-05,
+      "loss": 0.6206,
+      "step": 6330
+    },
+    {
+      "epoch": 0.61517562584902,
+      "grad_norm": 1.762789516395008,
+      "learning_rate": 3.9876046466350836e-05,
+      "loss": 0.5538,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6161459344071415,
+      "grad_norm": 1.8092369800491652,
+      "learning_rate": 3.985982218184178e-05,
+      "loss": 0.6236,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6171162429652629,
+      "grad_norm": 1.8060820825881239,
+      "learning_rate": 3.984359789733273e-05,
+      "loss": 0.5875,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6180865515233844,
+      "grad_norm": 1.7317977897542403,
+      "learning_rate": 3.982737361282368e-05,
+      "loss": 0.594,
+      "step": 6370
+    },
+    {
+      "epoch": 0.619056860081506,
+      "grad_norm": 1.5504516693013288,
+      "learning_rate": 3.981114932831462e-05,
+      "loss": 0.5833,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6200271686396274,
+      "grad_norm": 1.3714460442712506,
+      "learning_rate": 3.979492504380557e-05,
+      "loss": 0.6048,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6209974771977489,
+      "grad_norm": 1.741408754147969,
+      "learning_rate": 3.9778700759296515e-05,
+      "loss": 0.6729,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6219677857558704,
+      "grad_norm": 1.0600787702033703,
+      "learning_rate": 3.9762476474787465e-05,
+      "loss": 0.6934,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6229380943139918,
+      "grad_norm": 1.5722449067246123,
+      "learning_rate": 3.974625219027841e-05,
+      "loss": 0.6043,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6239084028721134,
+      "grad_norm": 1.3176596782485965,
+      "learning_rate": 3.973002790576936e-05,
+      "loss": 0.5898,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6248787114302348,
+      "grad_norm": 1.4177190750303141,
+      "learning_rate": 3.971380362126031e-05,
+      "loss": 0.6146,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6258490199883563,
+      "grad_norm": 1.3993450966845369,
+      "learning_rate": 3.969757933675125e-05,
+      "loss": 0.608,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6268193285464778,
+      "grad_norm": 1.6552784809097676,
+      "learning_rate": 3.96813550522422e-05,
+      "loss": 0.6078,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6277896371045992,
+      "grad_norm": 1.315768136368434,
+      "learning_rate": 3.966513076773314e-05,
+      "loss": 0.6021,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6287599456627208,
+      "grad_norm": 1.431816586824017,
+      "learning_rate": 3.964890648322409e-05,
+      "loss": 0.5571,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6297302542208423,
+      "grad_norm": 1.4950226300857892,
+      "learning_rate": 3.9632682198715036e-05,
+      "loss": 0.6297,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6307005627789637,
+      "grad_norm": 2.0781333722978284,
+      "learning_rate": 3.9616457914205985e-05,
+      "loss": 0.596,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6307005627789637,
+      "eval_loss": 0.6844401955604553,
+      "eval_runtime": 2468.3658,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6316708713370852,
+      "grad_norm": 1.5337482284796442,
+      "learning_rate": 3.960023362969693e-05,
+      "loss": 0.6634,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6326411798952066,
+      "grad_norm": 1.7040234243162709,
+      "learning_rate": 3.958400934518788e-05,
+      "loss": 0.5936,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6336114884533282,
+      "grad_norm": 1.7943789476429368,
+      "learning_rate": 3.956778506067883e-05,
+      "loss": 0.6361,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6345817970114497,
+      "grad_norm": 1.5737933588874193,
+      "learning_rate": 3.955156077616977e-05,
+      "loss": 0.6192,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6355521055695711,
+      "grad_norm": 1.532774889369556,
+      "learning_rate": 3.953533649166072e-05,
+      "loss": 0.5694,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6365224141276926,
+      "grad_norm": 1.6854751291833254,
+      "learning_rate": 3.9519112207151664e-05,
+      "loss": 0.6365,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6374927226858141,
+      "grad_norm": 1.4850693595634958,
+      "learning_rate": 3.9502887922642613e-05,
+      "loss": 0.5966,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6384630312439356,
+      "grad_norm": 1.5032167913978984,
+      "learning_rate": 3.9486663638133556e-05,
+      "loss": 0.6413,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6394333398020571,
+      "grad_norm": 1.4058296063426399,
+      "learning_rate": 3.9470439353624506e-05,
+      "loss": 0.5845,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6404036483601785,
+      "grad_norm": 1.563521067932564,
+      "learning_rate": 3.945421506911545e-05,
+      "loss": 0.6179,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6413739569183,
+      "grad_norm": 1.4036808350972751,
+      "learning_rate": 3.94379907846064e-05,
+      "loss": 0.5942,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6423442654764215,
+      "grad_norm": 1.4964922216668848,
+      "learning_rate": 3.942176650009735e-05,
+      "loss": 0.6516,
+      "step": 6620
+    },
+    {
+      "epoch": 0.643314574034543,
+      "grad_norm": 1.6801427039047954,
+      "learning_rate": 3.940554221558829e-05,
+      "loss": 0.6192,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6442848825926645,
+      "grad_norm": 1.582294622357827,
+      "learning_rate": 3.938931793107924e-05,
+      "loss": 0.6051,
+      "step": 6640
+    },
+    {
+      "epoch": 0.645255191150786,
+      "grad_norm": 1.531859995817891,
+      "learning_rate": 3.9373093646570185e-05,
+      "loss": 0.6407,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6462254997089074,
+      "grad_norm": 1.9088041020547684,
+      "learning_rate": 3.9356869362061134e-05,
+      "loss": 0.6078,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6471958082670289,
+      "grad_norm": 1.3066981837793257,
+      "learning_rate": 3.934064507755208e-05,
+      "loss": 0.5665,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6481661168251504,
+      "grad_norm": 1.313352722546981,
+      "learning_rate": 3.932442079304303e-05,
+      "loss": 0.5201,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6491364253832719,
+      "grad_norm": 1.8145770193834194,
+      "learning_rate": 3.930819650853397e-05,
+      "loss": 0.636,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6501067339413934,
+      "grad_norm": 1.6198734288858008,
+      "learning_rate": 3.929197222402492e-05,
+      "loss": 0.6529,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6510770424995148,
+      "grad_norm": 1.6029042771301745,
+      "learning_rate": 3.927574793951587e-05,
+      "loss": 0.5777,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6520473510576363,
+      "grad_norm": 1.7982640944263595,
+      "learning_rate": 3.925952365500681e-05,
+      "loss": 0.5833,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6530176596157579,
+      "grad_norm": 1.9034497019204402,
+      "learning_rate": 3.924329937049776e-05,
+      "loss": 0.5981,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6539879681738793,
+      "grad_norm": 1.3239435404528657,
+      "learning_rate": 3.9227075085988705e-05,
+      "loss": 0.5861,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6549582767320008,
+      "grad_norm": 1.7333381509447607,
+      "learning_rate": 3.921085080147966e-05,
+      "loss": 0.5696,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6559285852901222,
+      "grad_norm": 1.7408114949745195,
+      "learning_rate": 3.9194626516970605e-05,
+      "loss": 0.5868,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6568988938482437,
+      "grad_norm": 1.4140478484561587,
+      "learning_rate": 3.9178402232461555e-05,
+      "loss": 0.6126,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6578692024063653,
+      "grad_norm": 1.2359753742563322,
+      "learning_rate": 3.91621779479525e-05,
+      "loss": 0.595,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6588395109644867,
+      "grad_norm": 1.4986954153208918,
+      "learning_rate": 3.914595366344345e-05,
+      "loss": 0.6399,
+      "step": 6790
+    },
+    {
+      "epoch": 0.6598098195226082,
+      "grad_norm": 1.60396773259511,
+      "learning_rate": 3.912972937893439e-05,
+      "loss": 0.573,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6607801280807297,
+      "grad_norm": 1.6446241731745532,
+      "learning_rate": 3.911350509442534e-05,
+      "loss": 0.5758,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6617504366388511,
+      "grad_norm": 1.6940327126152295,
+      "learning_rate": 3.909728080991629e-05,
+      "loss": 0.5646,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6627207451969727,
+      "grad_norm": 1.6725911752868794,
+      "learning_rate": 3.908105652540723e-05,
+      "loss": 0.6282,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6636910537550941,
+      "grad_norm": 1.724226678968073,
+      "learning_rate": 3.906483224089818e-05,
+      "loss": 0.6391,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6646613623132156,
+      "grad_norm": 1.5959383703623191,
+      "learning_rate": 3.9048607956389126e-05,
+      "loss": 0.6003,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6656316708713371,
+      "grad_norm": 1.7487148965006274,
+      "learning_rate": 3.9032383671880075e-05,
+      "loss": 0.6019,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6666019794294585,
+      "grad_norm": 2.0088682339975312,
+      "learning_rate": 3.901615938737102e-05,
+      "loss": 0.5846,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6675722879875801,
+      "grad_norm": 1.5646052877933723,
+      "learning_rate": 3.899993510286197e-05,
+      "loss": 0.6222,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6685425965457016,
+      "grad_norm": 1.5750290096746722,
+      "learning_rate": 3.898371081835291e-05,
+      "loss": 0.6018,
+      "step": 6890
+    },
+    {
+      "epoch": 0.669512905103823,
+      "grad_norm": 1.2258053581934112,
+      "learning_rate": 3.896748653384386e-05,
+      "loss": 0.5697,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6704832136619445,
+      "grad_norm": 1.9376975148901512,
+      "learning_rate": 3.895126224933481e-05,
+      "loss": 0.6168,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6714535222200659,
+      "grad_norm": 1.4670771869188424,
+      "learning_rate": 3.8935037964825754e-05,
+      "loss": 0.5964,
+      "step": 6920
+    },
+    {
+      "epoch": 0.6724238307781875,
+      "grad_norm": 1.324084539118919,
+      "learning_rate": 3.8918813680316704e-05,
+      "loss": 0.5346,
+      "step": 6930
+    },
+    {
+      "epoch": 0.673394139336309,
+      "grad_norm": 1.717496207353836,
+      "learning_rate": 3.8902589395807647e-05,
+      "loss": 0.6112,
+      "step": 6940
+    },
+    {
+      "epoch": 0.6743644478944304,
+      "grad_norm": 2.019875964877309,
+      "learning_rate": 3.8886365111298596e-05,
+      "loss": 0.6522,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6753347564525519,
+      "grad_norm": 1.3354957180640497,
+      "learning_rate": 3.887014082678954e-05,
+      "loss": 0.6256,
+      "step": 6960
+    },
+    {
+      "epoch": 0.6763050650106734,
+      "grad_norm": 1.6937952057212555,
+      "learning_rate": 3.885391654228049e-05,
+      "loss": 0.6176,
+      "step": 6970
+    },
+    {
+      "epoch": 0.6772753735687949,
+      "grad_norm": 1.5422166727550382,
+      "learning_rate": 3.883769225777143e-05,
+      "loss": 0.5546,
+      "step": 6980
+    },
+    {
+      "epoch": 0.6782456821269164,
+      "grad_norm": 1.7417491098472007,
+      "learning_rate": 3.882146797326238e-05,
+      "loss": 0.6267,
+      "step": 6990
+    },
+    {
+      "epoch": 0.6792159906850378,
+      "grad_norm": 1.5366921907221318,
+      "learning_rate": 3.880524368875333e-05,
+      "loss": 0.6218,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6792159906850378,
+      "eval_loss": 0.678726851940155,
+      "eval_runtime": 2469.9254,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6801862992431593,
+      "grad_norm": 1.5097589222422163,
+      "learning_rate": 3.8789019404244275e-05,
+      "loss": 0.5925,
+      "step": 7010
+    },
+    {
+      "epoch": 0.6811566078012808,
+      "grad_norm": 2.1865443717654256,
+      "learning_rate": 3.8772795119735224e-05,
+      "loss": 0.6125,
+      "step": 7020
+    },
+    {
+      "epoch": 0.6821269163594023,
+      "grad_norm": 1.3913795509553841,
+      "learning_rate": 3.875657083522617e-05,
+      "loss": 0.6366,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6830972249175238,
+      "grad_norm": 1.314507220249943,
+      "learning_rate": 3.874034655071712e-05,
+      "loss": 0.5609,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6840675334756453,
+      "grad_norm": 1.5985439411272098,
+      "learning_rate": 3.872412226620806e-05,
+      "loss": 0.5946,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6850378420337667,
+      "grad_norm": 1.8453526914168807,
+      "learning_rate": 3.870789798169901e-05,
+      "loss": 0.5389,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6860081505918882,
+      "grad_norm": 1.746744853463753,
+      "learning_rate": 3.869167369718995e-05,
+      "loss": 0.5821,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6869784591500097,
+      "grad_norm": 1.351619112366914,
+      "learning_rate": 3.86754494126809e-05,
+      "loss": 0.6333,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6879487677081312,
+      "grad_norm": 1.6766693580749343,
+      "learning_rate": 3.865922512817185e-05,
+      "loss": 0.5905,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6889190762662527,
+      "grad_norm": 1.6907020127934513,
+      "learning_rate": 3.8643000843662795e-05,
+      "loss": 0.583,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6898893848243741,
+      "grad_norm": 1.8609743061215689,
+      "learning_rate": 3.8626776559153745e-05,
+      "loss": 0.6131,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6908596933824956,
+      "grad_norm": 1.3507375736553595,
+      "learning_rate": 3.861055227464469e-05,
+      "loss": 0.6399,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6918300019406172,
+      "grad_norm": 1.7786145489846785,
+      "learning_rate": 3.859432799013564e-05,
+      "loss": 0.5148,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6928003104987386,
+      "grad_norm": 1.7336491727001198,
+      "learning_rate": 3.857810370562658e-05,
+      "loss": 0.6154,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6937706190568601,
+      "grad_norm": 1.4164103938896966,
+      "learning_rate": 3.856187942111753e-05,
+      "loss": 0.5768,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6947409276149815,
+      "grad_norm": 1.5705135987803418,
+      "learning_rate": 3.854565513660848e-05,
+      "loss": 0.6102,
+      "step": 7160
+    },
+    {
+      "epoch": 0.695711236173103,
+      "grad_norm": 1.4121262983361098,
+      "learning_rate": 3.8529430852099424e-05,
+      "loss": 0.61,
+      "step": 7170
+    },
+    {
+      "epoch": 0.6966815447312246,
+      "grad_norm": 1.8472104212242206,
+      "learning_rate": 3.851320656759037e-05,
+      "loss": 0.5911,
+      "step": 7180
+    },
+    {
+      "epoch": 0.697651853289346,
+      "grad_norm": 1.5344754452736826,
+      "learning_rate": 3.8496982283081316e-05,
+      "loss": 0.606,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6986221618474675,
+      "grad_norm": 1.6606427535357149,
+      "learning_rate": 3.8480757998572266e-05,
+      "loss": 0.6062,
+      "step": 7200
+    },
+    {
+      "epoch": 0.699592470405589,
+      "grad_norm": 1.633183446097436,
+      "learning_rate": 3.846453371406321e-05,
+      "loss": 0.5828,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7005627789637104,
+      "grad_norm": 1.646557901144212,
+      "learning_rate": 3.844830942955416e-05,
+      "loss": 0.5354,
+      "step": 7220
+    },
+    {
+      "epoch": 0.701533087521832,
+      "grad_norm": 1.64918421348089,
+      "learning_rate": 3.84320851450451e-05,
+      "loss": 0.551,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7025033960799534,
+      "grad_norm": 1.6382049443147468,
+      "learning_rate": 3.841586086053605e-05,
+      "loss": 0.5723,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7034737046380749,
+      "grad_norm": 2.1034428309614523,
+      "learning_rate": 3.8399636576027e-05,
+      "loss": 0.6282,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7044440131961964,
+      "grad_norm": 1.4993896668880777,
+      "learning_rate": 3.8383412291517944e-05,
+      "loss": 0.5221,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7054143217543178,
+      "grad_norm": 1.5357479402580956,
+      "learning_rate": 3.8367188007008894e-05,
+      "loss": 0.6176,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7063846303124394,
+      "grad_norm": 1.5289777074279678,
+      "learning_rate": 3.835096372249984e-05,
+      "loss": 0.6045,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7073549388705609,
+      "grad_norm": 1.6907343435757922,
+      "learning_rate": 3.833473943799079e-05,
+      "loss": 0.5778,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7083252474286823,
+      "grad_norm": 1.546188967929013,
+      "learning_rate": 3.831851515348173e-05,
+      "loss": 0.5966,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7092955559868038,
+      "grad_norm": 1.6668813308937025,
+      "learning_rate": 3.830229086897268e-05,
+      "loss": 0.5443,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7102658645449252,
+      "grad_norm": 2.0411822746490444,
+      "learning_rate": 3.828606658446362e-05,
+      "loss": 0.5663,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7112361731030468,
+      "grad_norm": 1.5825597761459882,
+      "learning_rate": 3.826984229995457e-05,
+      "loss": 0.6143,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7122064816611683,
+      "grad_norm": 1.7867885913406227,
+      "learning_rate": 3.825361801544552e-05,
+      "loss": 0.6094,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7131767902192897,
+      "grad_norm": 1.4415937345324663,
+      "learning_rate": 3.8237393730936465e-05,
+      "loss": 0.552,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7141470987774112,
+      "grad_norm": 1.8623916924146842,
+      "learning_rate": 3.8221169446427415e-05,
+      "loss": 0.6035,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7151174073355328,
+      "grad_norm": 1.6112921767840875,
+      "learning_rate": 3.820494516191836e-05,
+      "loss": 0.5957,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7160877158936542,
+      "grad_norm": 1.6295510822010462,
+      "learning_rate": 3.818872087740931e-05,
+      "loss": 0.5906,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7170580244517757,
+      "grad_norm": 1.3787962703695553,
+      "learning_rate": 3.817249659290025e-05,
+      "loss": 0.5884,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7180283330098971,
+      "grad_norm": 1.5293927693421874,
+      "learning_rate": 3.81562723083912e-05,
+      "loss": 0.6277,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7189986415680186,
+      "grad_norm": 1.4147252260121417,
+      "learning_rate": 3.8140048023882143e-05,
+      "loss": 0.5725,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7199689501261402,
+      "grad_norm": 1.7736403425897478,
+      "learning_rate": 3.812382373937309e-05,
+      "loss": 0.533,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7209392586842616,
+      "grad_norm": 1.642532067779172,
+      "learning_rate": 3.810759945486404e-05,
+      "loss": 0.6405,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7219095672423831,
+      "grad_norm": 1.8702737255141961,
+      "learning_rate": 3.8091375170354986e-05,
+      "loss": 0.523,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7228798758005046,
+      "grad_norm": 1.82343375200645,
+      "learning_rate": 3.8075150885845936e-05,
+      "loss": 0.54,
+      "step": 7450
+    },
+    {
+      "epoch": 0.723850184358626,
+      "grad_norm": 1.6212132399265973,
+      "learning_rate": 3.805892660133688e-05,
+      "loss": 0.6016,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7248204929167475,
+      "grad_norm": 1.7707895342391216,
+      "learning_rate": 3.804270231682783e-05,
+      "loss": 0.5724,
+      "step": 7470
+    },
+    {
+      "epoch": 0.725790801474869,
+      "grad_norm": 1.4832754109529855,
+      "learning_rate": 3.802647803231877e-05,
+      "loss": 0.5554,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7267611100329905,
+      "grad_norm": 1.4890517160303556,
+      "learning_rate": 3.801025374780972e-05,
+      "loss": 0.6065,
+      "step": 7490
+    },
+    {
+      "epoch": 0.727731418591112,
+      "grad_norm": 2.047151127095432,
+      "learning_rate": 3.799402946330067e-05,
+      "loss": 0.6299,
+      "step": 7500
+    },
+    {
+      "epoch": 0.727731418591112,
+      "eval_loss": 0.6740881204605103,
+      "eval_runtime": 2470.6314,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7287017271492334,
+      "grad_norm": 1.6234544208509818,
+      "learning_rate": 3.7977805178791614e-05,
+      "loss": 0.6013,
+      "step": 7510
+    },
+    {
+      "epoch": 0.729672035707355,
+      "grad_norm": 1.4452320512475898,
+      "learning_rate": 3.7961580894282564e-05,
+      "loss": 0.5479,
+      "step": 7520
+    },
+    {
+      "epoch": 0.7306423442654765,
+      "grad_norm": 1.2290296761164277,
+      "learning_rate": 3.794535660977351e-05,
+      "loss": 0.5471,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7316126528235979,
+      "grad_norm": 1.567003648013224,
+      "learning_rate": 3.792913232526446e-05,
+      "loss": 0.6092,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7325829613817194,
+      "grad_norm": 1.6011455152210408,
+      "learning_rate": 3.7912908040755406e-05,
+      "loss": 0.629,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7335532699398408,
+      "grad_norm": 2.1218593111989192,
+      "learning_rate": 3.7896683756246356e-05,
+      "loss": 0.5773,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7345235784979623,
+      "grad_norm": 1.8323104555880425,
+      "learning_rate": 3.78804594717373e-05,
+      "loss": 0.618,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7354938870560839,
+      "grad_norm": 1.4152516707860787,
+      "learning_rate": 3.786423518722825e-05,
+      "loss": 0.6062,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7364641956142053,
+      "grad_norm": 1.6284692342846199,
+      "learning_rate": 3.784801090271919e-05,
+      "loss": 0.5671,
+      "step": 7590
+    },
+    {
+      "epoch": 0.7374345041723268,
+      "grad_norm": 1.426043120608999,
+      "learning_rate": 3.783178661821014e-05,
+      "loss": 0.5972,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7384048127304483,
+      "grad_norm": 1.893183882199419,
+      "learning_rate": 3.7815562333701085e-05,
+      "loss": 0.6193,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7393751212885697,
+      "grad_norm": 1.9982181406611617,
+      "learning_rate": 3.7799338049192034e-05,
+      "loss": 0.5532,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7403454298466913,
+      "grad_norm": 1.8557864601006913,
+      "learning_rate": 3.7783113764682984e-05,
+      "loss": 0.5777,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7413157384048127,
+      "grad_norm": 1.8775040642513798,
+      "learning_rate": 3.776688948017393e-05,
+      "loss": 0.6268,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7422860469629342,
+      "grad_norm": 1.758574155734976,
+      "learning_rate": 3.775066519566488e-05,
+      "loss": 0.5662,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7432563555210557,
+      "grad_norm": 2.194684585403217,
+      "learning_rate": 3.773444091115582e-05,
+      "loss": 0.6628,
+      "step": 7660
+    },
+    {
+      "epoch": 0.7442266640791771,
+      "grad_norm": 1.3673983887882022,
+      "learning_rate": 3.771821662664677e-05,
+      "loss": 0.6061,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7451969726372987,
+      "grad_norm": 1.2724099570692131,
+      "learning_rate": 3.770199234213771e-05,
+      "loss": 0.6337,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7461672811954202,
+      "grad_norm": 1.6823171665935568,
+      "learning_rate": 3.768576805762866e-05,
+      "loss": 0.5528,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7471375897535416,
+      "grad_norm": 1.9633396330523931,
+      "learning_rate": 3.7669543773119605e-05,
+      "loss": 0.5808,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7481078983116631,
+      "grad_norm": 1.7782111547784767,
+      "learning_rate": 3.7653319488610555e-05,
+      "loss": 0.5767,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7490782068697845,
+      "grad_norm": 2.3110549543336205,
+      "learning_rate": 3.7637095204101505e-05,
+      "loss": 0.6062,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7500485154279061,
+      "grad_norm": 1.6939270574842946,
+      "learning_rate": 3.762087091959245e-05,
+      "loss": 0.6225,
+      "step": 7730
+    },
+    {
+      "epoch": 0.7510188239860276,
+      "grad_norm": 1.4820399391564874,
+      "learning_rate": 3.76046466350834e-05,
+      "loss": 0.6047,
+      "step": 7740
+    },
+    {
+      "epoch": 0.751989132544149,
+      "grad_norm": 1.5672189262381615,
+      "learning_rate": 3.758842235057434e-05,
+      "loss": 0.5879,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7529594411022705,
+      "grad_norm": 1.6472834744978406,
+      "learning_rate": 3.757219806606529e-05,
+      "loss": 0.5895,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7539297496603921,
+      "grad_norm": 1.1201299864920753,
+      "learning_rate": 3.7555973781556234e-05,
+      "loss": 0.5801,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7549000582185135,
+      "grad_norm": 2.1519831187317107,
+      "learning_rate": 3.753974949704718e-05,
+      "loss": 0.5806,
+      "step": 7780
+    },
+    {
+      "epoch": 0.755870366776635,
+      "grad_norm": 1.6558911877127844,
+      "learning_rate": 3.752352521253813e-05,
+      "loss": 0.6243,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7568406753347564,
+      "grad_norm": 1.56259604588081,
+      "learning_rate": 3.7507300928029076e-05,
+      "loss": 0.5611,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7578109838928779,
+      "grad_norm": 1.7228227668603993,
+      "learning_rate": 3.7491076643520026e-05,
+      "loss": 0.5517,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7587812924509995,
+      "grad_norm": 1.4902669117767953,
+      "learning_rate": 3.747485235901097e-05,
+      "loss": 0.538,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7597516010091209,
+      "grad_norm": 1.782224663749718,
+      "learning_rate": 3.745862807450192e-05,
+      "loss": 0.5753,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7607219095672424,
+      "grad_norm": 1.9997818457050736,
+      "learning_rate": 3.744240378999286e-05,
+      "loss": 0.6007,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7616922181253639,
+      "grad_norm": 1.4094657223176301,
+      "learning_rate": 3.742617950548381e-05,
+      "loss": 0.6338,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7626625266834853,
+      "grad_norm": 1.9288686398933608,
+      "learning_rate": 3.7409955220974754e-05,
+      "loss": 0.5906,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7636328352416069,
+      "grad_norm": 1.7390584642787001,
+      "learning_rate": 3.7393730936465704e-05,
+      "loss": 0.5738,
+      "step": 7870
+    },
+    {
+      "epoch": 0.7646031437997283,
+      "grad_norm": 1.9438974814009289,
+      "learning_rate": 3.7377506651956654e-05,
+      "loss": 0.5121,
+      "step": 7880
+    },
+    {
+      "epoch": 0.7655734523578498,
+      "grad_norm": 1.740068812702714,
+      "learning_rate": 3.73612823674476e-05,
+      "loss": 0.557,
+      "step": 7890
+    },
+    {
+      "epoch": 0.7665437609159713,
+      "grad_norm": 1.7983844321630307,
+      "learning_rate": 3.734505808293855e-05,
+      "loss": 0.5748,
+      "step": 7900
+    },
+    {
+      "epoch": 0.7675140694740927,
+      "grad_norm": 1.5464273991421298,
+      "learning_rate": 3.732883379842949e-05,
+      "loss": 0.547,
+      "step": 7910
+    },
+    {
+      "epoch": 0.7684843780322143,
+      "grad_norm": 1.5357637162840818,
+      "learning_rate": 3.731260951392044e-05,
+      "loss": 0.5628,
+      "step": 7920
+    },
+    {
+      "epoch": 0.7694546865903358,
+      "grad_norm": 1.5933643507985389,
+      "learning_rate": 3.729638522941138e-05,
+      "loss": 0.6377,
+      "step": 7930
+    },
+    {
+      "epoch": 0.7704249951484572,
+      "grad_norm": 1.885102854313203,
+      "learning_rate": 3.728016094490233e-05,
+      "loss": 0.5542,
+      "step": 7940
+    },
+    {
+      "epoch": 0.7713953037065787,
+      "grad_norm": 1.2646336337440816,
+      "learning_rate": 3.7263936660393275e-05,
+      "loss": 0.5909,
+      "step": 7950
+    },
+    {
+      "epoch": 0.7723656122647001,
+      "grad_norm": 1.5489929838962764,
+      "learning_rate": 3.7247712375884225e-05,
+      "loss": 0.6417,
+      "step": 7960
+    },
+    {
+      "epoch": 0.7733359208228217,
+      "grad_norm": 1.5642843810424312,
+      "learning_rate": 3.7231488091375175e-05,
+      "loss": 0.5739,
+      "step": 7970
+    },
+    {
+      "epoch": 0.7743062293809432,
+      "grad_norm": 1.2994344306176584,
+      "learning_rate": 3.721526380686612e-05,
+      "loss": 0.6521,
+      "step": 7980
+    },
+    {
+      "epoch": 0.7752765379390646,
+      "grad_norm": 1.7547352757346097,
+      "learning_rate": 3.719903952235707e-05,
+      "loss": 0.6055,
+      "step": 7990
+    },
+    {
+      "epoch": 0.7762468464971861,
+      "grad_norm": 1.8680201530781706,
+      "learning_rate": 3.718281523784801e-05,
+      "loss": 0.5468,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7762468464971861,
+      "eval_loss": 0.6728888750076294,
+      "eval_runtime": 2470.4642,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7772171550553076,
+      "grad_norm": 1.8353606131921394,
+      "learning_rate": 3.716659095333896e-05,
+      "loss": 0.6019,
+      "step": 8010
+    },
+    {
+      "epoch": 0.778187463613429,
+      "grad_norm": 2.1165186332460624,
+      "learning_rate": 3.71503666688299e-05,
+      "loss": 0.6144,
+      "step": 8020
+    },
+    {
+      "epoch": 0.7791577721715506,
+      "grad_norm": 1.2983357938097837,
+      "learning_rate": 3.713414238432085e-05,
+      "loss": 0.5765,
+      "step": 8030
+    },
+    {
+      "epoch": 0.780128080729672,
+      "grad_norm": 1.6998652219779764,
+      "learning_rate": 3.7117918099811796e-05,
+      "loss": 0.5673,
+      "step": 8040
+    },
+    {
+      "epoch": 0.7810983892877935,
+      "grad_norm": 1.6651384645995448,
+      "learning_rate": 3.7101693815302746e-05,
+      "loss": 0.5813,
+      "step": 8050
+    },
+    {
+      "epoch": 0.782068697845915,
+      "grad_norm": 1.711156688730374,
+      "learning_rate": 3.7085469530793696e-05,
+      "loss": 0.5804,
+      "step": 8060
+    },
+    {
+      "epoch": 0.7830390064040365,
+      "grad_norm": 1.8429084688690953,
+      "learning_rate": 3.706924524628464e-05,
+      "loss": 0.6076,
+      "step": 8070
+    },
+    {
+      "epoch": 0.784009314962158,
+      "grad_norm": 2.029890621987729,
+      "learning_rate": 3.705302096177559e-05,
+      "loss": 0.6248,
+      "step": 8080
+    },
+    {
+      "epoch": 0.7849796235202795,
+      "grad_norm": 2.075042801349425,
+      "learning_rate": 3.703679667726653e-05,
+      "loss": 0.5546,
+      "step": 8090
+    },
+    {
+      "epoch": 0.7859499320784009,
+      "grad_norm": 1.9442644766970887,
+      "learning_rate": 3.702057239275748e-05,
+      "loss": 0.5754,
+      "step": 8100
+    },
+    {
+      "epoch": 0.7869202406365224,
+      "grad_norm": 2.2898169839511167,
+      "learning_rate": 3.7004348108248424e-05,
+      "loss": 0.5479,
+      "step": 8110
+    },
+    {
+      "epoch": 0.7878905491946439,
+      "grad_norm": 1.1971567816899984,
+      "learning_rate": 3.6988123823739374e-05,
+      "loss": 0.5841,
+      "step": 8120
+    },
+    {
+      "epoch": 0.7888608577527654,
+      "grad_norm": 1.6483060799493618,
+      "learning_rate": 3.697189953923032e-05,
+      "loss": 0.5993,
+      "step": 8130
+    },
+    {
+      "epoch": 0.7898311663108869,
+      "grad_norm": 2.1844480030037094,
+      "learning_rate": 3.695567525472127e-05,
+      "loss": 0.5639,
+      "step": 8140
+    },
+    {
+      "epoch": 0.7908014748690083,
+      "grad_norm": 1.9208684488820418,
+      "learning_rate": 3.6939450970212216e-05,
+      "loss": 0.5548,
+      "step": 8150
+    },
+    {
+      "epoch": 0.7917717834271298,
+      "grad_norm": 1.7298769058966599,
+      "learning_rate": 3.692322668570316e-05,
+      "loss": 0.6189,
+      "step": 8160
+    },
+    {
+      "epoch": 0.7927420919852514,
+      "grad_norm": 2.043188779971351,
+      "learning_rate": 3.690700240119411e-05,
+      "loss": 0.6341,
+      "step": 8170
+    },
+    {
+      "epoch": 0.7937124005433728,
+      "grad_norm": 1.4155988967270856,
+      "learning_rate": 3.689077811668505e-05,
+      "loss": 0.5803,
+      "step": 8180
+    },
+    {
+      "epoch": 0.7946827091014943,
+      "grad_norm": 1.876027618508082,
+      "learning_rate": 3.6874553832176e-05,
+      "loss": 0.5469,
+      "step": 8190
+    },
+    {
+      "epoch": 0.7956530176596157,
+      "grad_norm": 1.8526890916404788,
+      "learning_rate": 3.6858329547666945e-05,
+      "loss": 0.5692,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7966233262177372,
+      "grad_norm": 1.454859578114591,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.5861,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7975936347758588,
+      "grad_norm": 1.6396970355230962,
+      "learning_rate": 3.6825880978648844e-05,
+      "loss": 0.5864,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7985639433339802,
+      "grad_norm": 2.035219184211101,
+      "learning_rate": 3.680965669413979e-05,
+      "loss": 0.5761,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7995342518921017,
+      "grad_norm": 1.4771264704118183,
+      "learning_rate": 3.679343240963074e-05,
+      "loss": 0.5901,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8005045604502232,
+      "grad_norm": 1.366640822196709,
+      "learning_rate": 3.677720812512168e-05,
+      "loss": 0.5976,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8014748690083446,
+      "grad_norm": 1.6158534552665804,
+      "learning_rate": 3.676098384061263e-05,
+      "loss": 0.5676,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8024451775664662,
+      "grad_norm": 1.906704908176893,
+      "learning_rate": 3.674475955610357e-05,
+      "loss": 0.6014,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8034154861245876,
+      "grad_norm": 1.72316238215741,
+      "learning_rate": 3.672853527159452e-05,
+      "loss": 0.5341,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8043857946827091,
+      "grad_norm": 1.6228489307580705,
+      "learning_rate": 3.6712310987085466e-05,
+      "loss": 0.6174,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8053561032408306,
+      "grad_norm": 1.6343652614101287,
+      "learning_rate": 3.6696086702576416e-05,
+      "loss": 0.5628,
+      "step": 8300
+    },
+    {
+      "epoch": 0.806326411798952,
+      "grad_norm": 1.4875484854044787,
+      "learning_rate": 3.6679862418067365e-05,
+      "loss": 0.5559,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8072967203570736,
+      "grad_norm": 1.7360867259906014,
+      "learning_rate": 3.6663638133558315e-05,
+      "loss": 0.6122,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8082670289151951,
+      "grad_norm": 1.3760924603009739,
+      "learning_rate": 3.664741384904926e-05,
+      "loss": 0.5369,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8092373374733165,
+      "grad_norm": 2.1326347379272033,
+      "learning_rate": 3.663118956454021e-05,
+      "loss": 0.6026,
+      "step": 8340
+    },
+    {
+      "epoch": 0.810207646031438,
+      "grad_norm": 1.8445530020541556,
+      "learning_rate": 3.661496528003116e-05,
+      "loss": 0.5924,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8111779545895594,
+      "grad_norm": 2.1620839328051153,
+      "learning_rate": 3.65987409955221e-05,
+      "loss": 0.5556,
+      "step": 8360
+    },
+    {
+      "epoch": 0.812148263147681,
+      "grad_norm": 1.91334025126679,
+      "learning_rate": 3.658251671101305e-05,
+      "loss": 0.5787,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8131185717058025,
+      "grad_norm": 1.6197394739350461,
+      "learning_rate": 3.656629242650399e-05,
+      "loss": 0.5364,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8140888802639239,
+      "grad_norm": 1.7496855520727714,
+      "learning_rate": 3.655006814199494e-05,
+      "loss": 0.604,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8150591888220454,
+      "grad_norm": 1.9072833025355298,
+      "learning_rate": 3.6533843857485886e-05,
+      "loss": 0.6127,
+      "step": 8400
+    },
+    {
+      "epoch": 0.816029497380167,
+      "grad_norm": 1.693494475477864,
+      "learning_rate": 3.6517619572976836e-05,
+      "loss": 0.5708,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8169998059382884,
+      "grad_norm": 1.694549481689151,
+      "learning_rate": 3.6501395288467786e-05,
+      "loss": 0.544,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8179701144964099,
+      "grad_norm": 1.77015793724743,
+      "learning_rate": 3.648517100395873e-05,
+      "loss": 0.5973,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8189404230545313,
+      "grad_norm": 1.738049084314835,
+      "learning_rate": 3.646894671944968e-05,
+      "loss": 0.5912,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8199107316126528,
+      "grad_norm": 1.647267019109384,
+      "learning_rate": 3.645272243494062e-05,
+      "loss": 0.6346,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8208810401707743,
+      "grad_norm": 1.5240403667548303,
+      "learning_rate": 3.643649815043157e-05,
+      "loss": 0.6057,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8218513487288958,
+      "grad_norm": 1.9527577485511822,
+      "learning_rate": 3.6420273865922514e-05,
+      "loss": 0.5615,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8228216572870173,
+      "grad_norm": 1.937531929878238,
+      "learning_rate": 3.6404049581413464e-05,
+      "loss": 0.5625,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8237919658451388,
+      "grad_norm": 1.7329982779170732,
+      "learning_rate": 3.638782529690441e-05,
+      "loss": 0.613,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8247622744032602,
+      "grad_norm": 1.880843027338221,
+      "learning_rate": 3.637160101239536e-05,
+      "loss": 0.5869,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8247622744032602,
+      "eval_loss": 0.6701070070266724,
+      "eval_runtime": 2471.9751,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8257325829613817,
+      "grad_norm": 1.2566468431839666,
+      "learning_rate": 3.6355376727886307e-05,
+      "loss": 0.5433,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8267028915195032,
+      "grad_norm": 1.8546070309693423,
+      "learning_rate": 3.633915244337725e-05,
+      "loss": 0.5669,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8276732000776247,
+      "grad_norm": 1.268249173975751,
+      "learning_rate": 3.63229281588682e-05,
+      "loss": 0.5895,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8286435086357462,
+      "grad_norm": 1.7652424871651577,
+      "learning_rate": 3.630670387435914e-05,
+      "loss": 0.5597,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8296138171938676,
+      "grad_norm": 1.5480894985375517,
+      "learning_rate": 3.629047958985009e-05,
+      "loss": 0.5486,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8305841257519891,
+      "grad_norm": 1.750414852898824,
+      "learning_rate": 3.6274255305341035e-05,
+      "loss": 0.584,
+      "step": 8560
+    },
+    {
+      "epoch": 0.8315544343101107,
+      "grad_norm": 1.7102419104322222,
+      "learning_rate": 3.6258031020831985e-05,
+      "loss": 0.5668,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8325247428682321,
+      "grad_norm": 1.5528462400057457,
+      "learning_rate": 3.624180673632293e-05,
+      "loss": 0.5896,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8334950514263536,
+      "grad_norm": 1.4801001203063542,
+      "learning_rate": 3.622558245181388e-05,
+      "loss": 0.5997,
+      "step": 8590
+    },
+    {
+      "epoch": 0.834465359984475,
+      "grad_norm": 1.391432340520497,
+      "learning_rate": 3.620935816730483e-05,
+      "loss": 0.5795,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8354356685425965,
+      "grad_norm": 1.4373875976848383,
+      "learning_rate": 3.619313388279577e-05,
+      "loss": 0.5769,
+      "step": 8610
+    },
+    {
+      "epoch": 0.8364059771007181,
+      "grad_norm": 1.8411413590399401,
+      "learning_rate": 3.617690959828672e-05,
+      "loss": 0.5626,
+      "step": 8620
+    },
+    {
+      "epoch": 0.8373762856588395,
+      "grad_norm": 1.6990663233303336,
+      "learning_rate": 3.616068531377766e-05,
+      "loss": 0.6057,
+      "step": 8630
+    },
+    {
+      "epoch": 0.838346594216961,
+      "grad_norm": 1.9737814861957261,
+      "learning_rate": 3.614446102926861e-05,
+      "loss": 0.564,
+      "step": 8640
+    },
+    {
+      "epoch": 0.8393169027750825,
+      "grad_norm": 1.6163853470752478,
+      "learning_rate": 3.6128236744759556e-05,
+      "loss": 0.5429,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8402872113332039,
+      "grad_norm": 1.5911941575436375,
+      "learning_rate": 3.6112012460250506e-05,
+      "loss": 0.6203,
+      "step": 8660
+    },
+    {
+      "epoch": 0.8412575198913255,
+      "grad_norm": 2.101242238346334,
+      "learning_rate": 3.609578817574145e-05,
+      "loss": 0.5263,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8422278284494469,
+      "grad_norm": 1.671876183792032,
+      "learning_rate": 3.60795638912324e-05,
+      "loss": 0.5699,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8431981370075684,
+      "grad_norm": 1.7729527985320428,
+      "learning_rate": 3.606333960672335e-05,
+      "loss": 0.5525,
+      "step": 8690
+    },
+    {
+      "epoch": 0.8441684455656899,
+      "grad_norm": 1.7198189021299524,
+      "learning_rate": 3.604711532221429e-05,
+      "loss": 0.5789,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8451387541238113,
+      "grad_norm": 1.754160746828024,
+      "learning_rate": 3.603089103770524e-05,
+      "loss": 0.5126,
+      "step": 8710
+    },
+    {
+      "epoch": 0.8461090626819329,
+      "grad_norm": 1.8621321453517432,
+      "learning_rate": 3.6014666753196184e-05,
+      "loss": 0.593,
+      "step": 8720
+    },
+    {
+      "epoch": 0.8470793712400544,
+      "grad_norm": 1.4506925576933114,
+      "learning_rate": 3.5998442468687134e-05,
+      "loss": 0.5489,
+      "step": 8730
+    },
+    {
+      "epoch": 0.8480496797981758,
+      "grad_norm": 1.9426709583278723,
+      "learning_rate": 3.598221818417808e-05,
+      "loss": 0.498,
+      "step": 8740
+    },
+    {
+      "epoch": 0.8490199883562973,
+      "grad_norm": 1.7346244341640757,
+      "learning_rate": 3.5965993899669026e-05,
+      "loss": 0.5583,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8499902969144187,
+      "grad_norm": 1.926381813212122,
+      "learning_rate": 3.594976961515997e-05,
+      "loss": 0.5931,
+      "step": 8760
+    },
+    {
+      "epoch": 0.8509606054725403,
+      "grad_norm": 1.9797943448639521,
+      "learning_rate": 3.593354533065092e-05,
+      "loss": 0.5773,
+      "step": 8770
+    },
+    {
+      "epoch": 0.8519309140306618,
+      "grad_norm": 2.243092907709638,
+      "learning_rate": 3.591732104614187e-05,
+      "loss": 0.5862,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8529012225887832,
+      "grad_norm": 1.6686507496640315,
+      "learning_rate": 3.590109676163281e-05,
+      "loss": 0.5861,
+      "step": 8790
+    },
+    {
+      "epoch": 0.8538715311469047,
+      "grad_norm": 1.767743857045935,
+      "learning_rate": 3.588487247712376e-05,
+      "loss": 0.6008,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8548418397050263,
+      "grad_norm": 1.789840488197533,
+      "learning_rate": 3.5868648192614705e-05,
+      "loss": 0.5578,
+      "step": 8810
+    },
+    {
+      "epoch": 0.8558121482631477,
+      "grad_norm": 1.826043320699609,
+      "learning_rate": 3.5852423908105655e-05,
+      "loss": 0.5568,
+      "step": 8820
+    },
+    {
+      "epoch": 0.8567824568212692,
+      "grad_norm": 1.4847524529624125,
+      "learning_rate": 3.58361996235966e-05,
+      "loss": 0.5984,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8577527653793906,
+      "grad_norm": 1.5302240137475795,
+      "learning_rate": 3.581997533908755e-05,
+      "loss": 0.5619,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8587230739375121,
+      "grad_norm": 1.6025870419513641,
+      "learning_rate": 3.58037510545785e-05,
+      "loss": 0.601,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8596933824956337,
+      "grad_norm": 1.8930896480563768,
+      "learning_rate": 3.578752677006944e-05,
+      "loss": 0.5935,
+      "step": 8860
+    },
+    {
+      "epoch": 0.8606636910537551,
+      "grad_norm": 1.6413896209556986,
+      "learning_rate": 3.577130248556039e-05,
+      "loss": 0.6019,
+      "step": 8870
+    },
+    {
+      "epoch": 0.8616339996118766,
+      "grad_norm": 1.6030981794189743,
+      "learning_rate": 3.575507820105133e-05,
+      "loss": 0.5761,
+      "step": 8880
+    },
+    {
+      "epoch": 0.8626043081699981,
+      "grad_norm": 1.4798866694040977,
+      "learning_rate": 3.573885391654228e-05,
+      "loss": 0.5561,
+      "step": 8890
+    },
+    {
+      "epoch": 0.8635746167281195,
+      "grad_norm": 1.5790323051766768,
+      "learning_rate": 3.5722629632033226e-05,
+      "loss": 0.5986,
+      "step": 8900
+    },
+    {
+      "epoch": 0.864544925286241,
+      "grad_norm": 1.8409121182605548,
+      "learning_rate": 3.5706405347524175e-05,
+      "loss": 0.5878,
+      "step": 8910
+    },
+    {
+      "epoch": 0.8655152338443625,
+      "grad_norm": 1.4956048954650922,
+      "learning_rate": 3.569018106301512e-05,
+      "loss": 0.5548,
+      "step": 8920
+    },
+    {
+      "epoch": 0.866485542402484,
+      "grad_norm": 1.6969221289264929,
+      "learning_rate": 3.567395677850607e-05,
+      "loss": 0.5937,
+      "step": 8930
+    },
+    {
+      "epoch": 0.8674558509606055,
+      "grad_norm": 1.8986285895813184,
+      "learning_rate": 3.565773249399702e-05,
+      "loss": 0.5742,
+      "step": 8940
+    },
+    {
+      "epoch": 0.8684261595187269,
+      "grad_norm": 1.5434045122281053,
+      "learning_rate": 3.564150820948796e-05,
+      "loss": 0.561,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8693964680768484,
+      "grad_norm": 1.69523497453855,
+      "learning_rate": 3.562528392497891e-05,
+      "loss": 0.5386,
+      "step": 8960
+    },
+    {
+      "epoch": 0.87036677663497,
+      "grad_norm": 1.382817285428295,
+      "learning_rate": 3.5609059640469854e-05,
+      "loss": 0.555,
+      "step": 8970
+    },
+    {
+      "epoch": 0.8713370851930914,
+      "grad_norm": 1.7354010988203836,
+      "learning_rate": 3.5592835355960803e-05,
+      "loss": 0.5963,
+      "step": 8980
+    },
+    {
+      "epoch": 0.8723073937512129,
+      "grad_norm": 1.9761497522561469,
+      "learning_rate": 3.5576611071451746e-05,
+      "loss": 0.5859,
+      "step": 8990
+    },
+    {
+      "epoch": 0.8732777023093343,
+      "grad_norm": 1.6264062899083487,
+      "learning_rate": 3.5560386786942696e-05,
+      "loss": 0.5596,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8732777023093343,
+      "eval_loss": 0.664995551109314,
+      "eval_runtime": 2472.6792,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8742480108674558,
+      "grad_norm": 1.7886226308798359,
+      "learning_rate": 3.554416250243364e-05,
+      "loss": 0.5484,
+      "step": 9010
+    },
+    {
+      "epoch": 0.8752183194255774,
+      "grad_norm": 1.5768009291424698,
+      "learning_rate": 3.552793821792459e-05,
+      "loss": 0.5511,
+      "step": 9020
+    },
+    {
+      "epoch": 0.8761886279836988,
+      "grad_norm": 1.5632964481761753,
+      "learning_rate": 3.551171393341554e-05,
+      "loss": 0.5173,
+      "step": 9030
+    },
+    {
+      "epoch": 0.8771589365418203,
+      "grad_norm": 2.0563484372782264,
+      "learning_rate": 3.549548964890648e-05,
+      "loss": 0.5586,
+      "step": 9040
+    },
+    {
+      "epoch": 0.8781292450999418,
+      "grad_norm": 1.9786114879020535,
+      "learning_rate": 3.547926536439743e-05,
+      "loss": 0.5767,
+      "step": 9050
+    },
+    {
+      "epoch": 0.8790995536580632,
+      "grad_norm": 1.2143656544779613,
+      "learning_rate": 3.5463041079888375e-05,
+      "loss": 0.5469,
+      "step": 9060
+    },
+    {
+      "epoch": 0.8800698622161848,
+      "grad_norm": 1.696976506773579,
+      "learning_rate": 3.5446816795379324e-05,
+      "loss": 0.5819,
+      "step": 9070
+    },
+    {
+      "epoch": 0.8810401707743062,
+      "grad_norm": 1.8002454640393335,
+      "learning_rate": 3.543059251087027e-05,
+      "loss": 0.5947,
+      "step": 9080
+    },
+    {
+      "epoch": 0.8820104793324277,
+      "grad_norm": 1.826272101277873,
+      "learning_rate": 3.541436822636122e-05,
+      "loss": 0.632,
+      "step": 9090
+    },
+    {
+      "epoch": 0.8829807878905492,
+      "grad_norm": 1.8609691197855038,
+      "learning_rate": 3.539814394185217e-05,
+      "loss": 0.649,
+      "step": 9100
+    },
+    {
+      "epoch": 0.8839510964486706,
+      "grad_norm": 1.7165708438826977,
+      "learning_rate": 3.5381919657343117e-05,
+      "loss": 0.5746,
+      "step": 9110
+    },
+    {
+      "epoch": 0.8849214050067922,
+      "grad_norm": 2.0390486982174454,
+      "learning_rate": 3.536569537283406e-05,
+      "loss": 0.5834,
+      "step": 9120
+    },
+    {
+      "epoch": 0.8858917135649137,
+      "grad_norm": 1.3107579501174733,
+      "learning_rate": 3.534947108832501e-05,
+      "loss": 0.5605,
+      "step": 9130
+    },
+    {
+      "epoch": 0.8868620221230351,
+      "grad_norm": 1.7750714401302556,
+      "learning_rate": 3.533324680381596e-05,
+      "loss": 0.5778,
+      "step": 9140
+    },
+    {
+      "epoch": 0.8878323306811566,
+      "grad_norm": 1.9002696712889475,
+      "learning_rate": 3.53170225193069e-05,
+      "loss": 0.5928,
+      "step": 9150
+    },
+    {
+      "epoch": 0.888802639239278,
+      "grad_norm": 1.602034501274119,
+      "learning_rate": 3.530079823479785e-05,
+      "loss": 0.5941,
+      "step": 9160
+    },
+    {
+      "epoch": 0.8897729477973996,
+      "grad_norm": 1.5338224625276715,
+      "learning_rate": 3.5284573950288795e-05,
+      "loss": 0.5678,
+      "step": 9170
+    },
+    {
+      "epoch": 0.8907432563555211,
+      "grad_norm": 1.6656771631689804,
+      "learning_rate": 3.5268349665779745e-05,
+      "loss": 0.5621,
+      "step": 9180
+    },
+    {
+      "epoch": 0.8917135649136425,
+      "grad_norm": 1.6734086655118368,
+      "learning_rate": 3.525212538127069e-05,
+      "loss": 0.5836,
+      "step": 9190
+    },
+    {
+      "epoch": 0.892683873471764,
+      "grad_norm": 1.7321566684016627,
+      "learning_rate": 3.523590109676164e-05,
+      "loss": 0.6049,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8936541820298856,
+      "grad_norm": 1.8468990646584322,
+      "learning_rate": 3.521967681225258e-05,
+      "loss": 0.5265,
+      "step": 9210
+    },
+    {
+      "epoch": 0.894624490588007,
+      "grad_norm": 1.8260102068484456,
+      "learning_rate": 3.520345252774353e-05,
+      "loss": 0.6003,
+      "step": 9220
+    },
+    {
+      "epoch": 0.8955947991461285,
+      "grad_norm": 1.5049364704966368,
+      "learning_rate": 3.518722824323448e-05,
+      "loss": 0.6362,
+      "step": 9230
+    },
+    {
+      "epoch": 0.8965651077042499,
+      "grad_norm": 1.8567290478944525,
+      "learning_rate": 3.517100395872542e-05,
+      "loss": 0.5666,
+      "step": 9240
+    },
+    {
+      "epoch": 0.8975354162623714,
+      "grad_norm": 1.83554173576104,
+      "learning_rate": 3.515477967421637e-05,
+      "loss": 0.525,
+      "step": 9250
+    },
+    {
+      "epoch": 0.898505724820493,
+      "grad_norm": 1.756580666977339,
+      "learning_rate": 3.5138555389707316e-05,
+      "loss": 0.5297,
+      "step": 9260
+    },
+    {
+      "epoch": 0.8994760333786144,
+      "grad_norm": 1.3709870685247603,
+      "learning_rate": 3.5122331105198265e-05,
+      "loss": 0.5529,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9004463419367359,
+      "grad_norm": 1.910129116074302,
+      "learning_rate": 3.510610682068921e-05,
+      "loss": 0.5575,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9014166504948574,
+      "grad_norm": 1.457636418218358,
+      "learning_rate": 3.508988253618016e-05,
+      "loss": 0.6219,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9023869590529788,
+      "grad_norm": 1.7400018831329018,
+      "learning_rate": 3.50736582516711e-05,
+      "loss": 0.597,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9033572676111004,
+      "grad_norm": 1.7711972448898297,
+      "learning_rate": 3.505743396716205e-05,
+      "loss": 0.5953,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9043275761692218,
+      "grad_norm": 1.3127391028092956,
+      "learning_rate": 3.5041209682653e-05,
+      "loss": 0.5771,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9052978847273433,
+      "grad_norm": 1.707083390377331,
+      "learning_rate": 3.5024985398143944e-05,
+      "loss": 0.5941,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9062681932854648,
+      "grad_norm": 2.3395444029475425,
+      "learning_rate": 3.5008761113634894e-05,
+      "loss": 0.5249,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9072385018435862,
+      "grad_norm": 2.15851320522422,
+      "learning_rate": 3.4992536829125837e-05,
+      "loss": 0.5754,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9082088104017078,
+      "grad_norm": 1.566455099295683,
+      "learning_rate": 3.4976312544616786e-05,
+      "loss": 0.5982,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9091791189598293,
+      "grad_norm": 1.6447262177778976,
+      "learning_rate": 3.496008826010773e-05,
+      "loss": 0.5819,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9101494275179507,
+      "grad_norm": 1.9281752331049982,
+      "learning_rate": 3.494386397559868e-05,
+      "loss": 0.5789,
+      "step": 9380
+    },
+    {
+      "epoch": 0.9111197360760722,
+      "grad_norm": 1.9237214019216216,
+      "learning_rate": 3.492763969108962e-05,
+      "loss": 0.5773,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9120900446341936,
+      "grad_norm": 1.5099620370471458,
+      "learning_rate": 3.491141540658057e-05,
+      "loss": 0.5954,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9130603531923152,
+      "grad_norm": 1.861214942110368,
+      "learning_rate": 3.489519112207152e-05,
+      "loss": 0.5956,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9140306617504367,
+      "grad_norm": 1.780788157697634,
+      "learning_rate": 3.4878966837562465e-05,
+      "loss": 0.5492,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9150009703085581,
+      "grad_norm": 1.7391155816623414,
+      "learning_rate": 3.4862742553053414e-05,
+      "loss": 0.5984,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9159712788666796,
+      "grad_norm": 2.1229665459752125,
+      "learning_rate": 3.484651826854436e-05,
+      "loss": 0.5924,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9169415874248011,
+      "grad_norm": 2.1306613515387625,
+      "learning_rate": 3.483029398403531e-05,
+      "loss": 0.5148,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9179118959829226,
+      "grad_norm": 1.312566076533743,
+      "learning_rate": 3.481406969952625e-05,
+      "loss": 0.5919,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9188822045410441,
+      "grad_norm": 1.4886826509371758,
+      "learning_rate": 3.47978454150172e-05,
+      "loss": 0.5853,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9198525130991655,
+      "grad_norm": 2.0143652338550098,
+      "learning_rate": 3.478162113050815e-05,
+      "loss": 0.5689,
+      "step": 9480
+    },
+    {
+      "epoch": 0.920822821657287,
+      "grad_norm": 1.1368414203141723,
+      "learning_rate": 3.476539684599909e-05,
+      "loss": 0.5546,
+      "step": 9490
+    },
+    {
+      "epoch": 0.9217931302154085,
+      "grad_norm": 1.6341362563091877,
+      "learning_rate": 3.474917256149004e-05,
+      "loss": 0.5481,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9217931302154085,
+      "eval_loss": 0.6622401475906372,
+      "eval_runtime": 2474.1087,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.362,
+      "step": 9500
+    },
+    {
+      "epoch": 0.92276343877353,
+      "grad_norm": 1.8501035490125046,
+      "learning_rate": 3.4732948276980985e-05,
+      "loss": 0.519,
+      "step": 9510
+    },
+    {
+      "epoch": 0.9237337473316515,
+      "grad_norm": 1.788403250607821,
+      "learning_rate": 3.4716723992471935e-05,
+      "loss": 0.5934,
+      "step": 9520
+    },
+    {
+      "epoch": 0.924704055889773,
+      "grad_norm": 2.0089841746536212,
+      "learning_rate": 3.470049970796288e-05,
+      "loss": 0.5463,
+      "step": 9530
+    },
+    {
+      "epoch": 0.9256743644478944,
+      "grad_norm": 1.4797021959834935,
+      "learning_rate": 3.468427542345383e-05,
+      "loss": 0.538,
+      "step": 9540
+    },
+    {
+      "epoch": 0.9266446730060159,
+      "grad_norm": 1.5018994978955122,
+      "learning_rate": 3.466805113894477e-05,
+      "loss": 0.5859,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9276149815641374,
+      "grad_norm": 1.6445575492561615,
+      "learning_rate": 3.465182685443572e-05,
+      "loss": 0.5335,
+      "step": 9560
+    },
+    {
+      "epoch": 0.9285852901222589,
+      "grad_norm": 1.910218513281919,
+      "learning_rate": 3.463560256992667e-05,
+      "loss": 0.5284,
+      "step": 9570
+    },
+    {
+      "epoch": 0.9295555986803804,
+      "grad_norm": 1.5061058308088753,
+      "learning_rate": 3.4619378285417614e-05,
+      "loss": 0.5332,
+      "step": 9580
+    },
+    {
+      "epoch": 0.9305259072385018,
+      "grad_norm": 1.5869569364806828,
+      "learning_rate": 3.460315400090856e-05,
+      "loss": 0.5591,
+      "step": 9590
+    },
+    {
+      "epoch": 0.9314962157966233,
+      "grad_norm": 1.709001551959916,
+      "learning_rate": 3.4586929716399506e-05,
+      "loss": 0.5499,
+      "step": 9600
+    },
+    {
+      "epoch": 0.9324665243547449,
+      "grad_norm": 1.5648665735772118,
+      "learning_rate": 3.4570705431890456e-05,
+      "loss": 0.5677,
+      "step": 9610
+    },
+    {
+      "epoch": 0.9334368329128663,
+      "grad_norm": 2.297106138114182,
+      "learning_rate": 3.45544811473814e-05,
+      "loss": 0.5687,
+      "step": 9620
+    },
+    {
+      "epoch": 0.9344071414709878,
+      "grad_norm": 1.940344115414216,
+      "learning_rate": 3.453825686287235e-05,
+      "loss": 0.5614,
+      "step": 9630
+    },
+    {
+      "epoch": 0.9353774500291092,
+      "grad_norm": 2.2931746226047336,
+      "learning_rate": 3.452203257836329e-05,
+      "loss": 0.5888,
+      "step": 9640
+    },
+    {
+      "epoch": 0.9363477585872307,
+      "grad_norm": 1.6726396768074983,
+      "learning_rate": 3.450580829385424e-05,
+      "loss": 0.555,
+      "step": 9650
+    },
+    {
+      "epoch": 0.9373180671453523,
+      "grad_norm": 1.5245738077800575,
+      "learning_rate": 3.448958400934519e-05,
+      "loss": 0.5464,
+      "step": 9660
+    },
+    {
+      "epoch": 0.9382883757034737,
+      "grad_norm": 1.3863585051832457,
+      "learning_rate": 3.4473359724836134e-05,
+      "loss": 0.5033,
+      "step": 9670
+    },
+    {
+      "epoch": 0.9392586842615952,
+      "grad_norm": 1.2261387961871664,
+      "learning_rate": 3.4457135440327084e-05,
+      "loss": 0.5229,
+      "step": 9680
+    },
+    {
+      "epoch": 0.9402289928197167,
+      "grad_norm": 1.8933274253957586,
+      "learning_rate": 3.444091115581803e-05,
+      "loss": 0.552,
+      "step": 9690
+    },
+    {
+      "epoch": 0.9411993013778381,
+      "grad_norm": 1.7235797326078635,
+      "learning_rate": 3.442468687130898e-05,
+      "loss": 0.5899,
+      "step": 9700
+    },
+    {
+      "epoch": 0.9421696099359597,
+      "grad_norm": 1.428965938918239,
+      "learning_rate": 3.440846258679992e-05,
+      "loss": 0.5733,
+      "step": 9710
+    },
+    {
+      "epoch": 0.9431399184940811,
+      "grad_norm": 1.4190853376920558,
+      "learning_rate": 3.439223830229087e-05,
+      "loss": 0.5762,
+      "step": 9720
+    },
+    {
+      "epoch": 0.9441102270522026,
+      "grad_norm": 1.569528489090731,
+      "learning_rate": 3.437601401778181e-05,
+      "loss": 0.5535,
+      "step": 9730
+    },
+    {
+      "epoch": 0.9450805356103241,
+      "grad_norm": 1.869896863596011,
+      "learning_rate": 3.435978973327276e-05,
+      "loss": 0.5651,
+      "step": 9740
+    },
+    {
+      "epoch": 0.9460508441684455,
+      "grad_norm": 2.3801435462427785,
+      "learning_rate": 3.434356544876371e-05,
+      "loss": 0.5366,
+      "step": 9750
+    },
+    {
+      "epoch": 0.9470211527265671,
+      "grad_norm": 1.7543924621581104,
+      "learning_rate": 3.4327341164254655e-05,
+      "loss": 0.5407,
+      "step": 9760
+    },
+    {
+      "epoch": 0.9479914612846886,
+      "grad_norm": 1.6645160945117223,
+      "learning_rate": 3.4311116879745605e-05,
+      "loss": 0.5776,
+      "step": 9770
+    },
+    {
+      "epoch": 0.94896176984281,
+      "grad_norm": 1.5226969850196896,
+      "learning_rate": 3.429489259523655e-05,
+      "loss": 0.5189,
+      "step": 9780
+    },
+    {
+      "epoch": 0.9499320784009315,
+      "grad_norm": 1.7957341969322531,
+      "learning_rate": 3.42786683107275e-05,
+      "loss": 0.5771,
+      "step": 9790
+    },
+    {
+      "epoch": 0.9509023869590529,
+      "grad_norm": 1.4922624955935235,
+      "learning_rate": 3.426244402621844e-05,
+      "loss": 0.5339,
+      "step": 9800
+    },
+    {
+      "epoch": 0.9518726955171745,
+      "grad_norm": 1.9179842292383775,
+      "learning_rate": 3.424621974170939e-05,
+      "loss": 0.5408,
+      "step": 9810
+    },
+    {
+      "epoch": 0.952843004075296,
+      "grad_norm": 1.8930907076270356,
+      "learning_rate": 3.4229995457200333e-05,
+      "loss": 0.5622,
+      "step": 9820
+    },
+    {
+      "epoch": 0.9538133126334174,
+      "grad_norm": 1.781854001943581,
+      "learning_rate": 3.421377117269128e-05,
+      "loss": 0.5654,
+      "step": 9830
+    },
+    {
+      "epoch": 0.9547836211915389,
+      "grad_norm": 1.9058249916201926,
+      "learning_rate": 3.419754688818223e-05,
+      "loss": 0.6089,
+      "step": 9840
+    },
+    {
+      "epoch": 0.9557539297496604,
+      "grad_norm": 1.5813398061915347,
+      "learning_rate": 3.4181322603673176e-05,
+      "loss": 0.5444,
+      "step": 9850
+    },
+    {
+      "epoch": 0.9567242383077819,
+      "grad_norm": 1.6480219585268394,
+      "learning_rate": 3.4165098319164126e-05,
+      "loss": 0.5215,
+      "step": 9860
+    },
+    {
+      "epoch": 0.9576945468659034,
+      "grad_norm": 1.6302754198626406,
+      "learning_rate": 3.414887403465507e-05,
+      "loss": 0.5225,
+      "step": 9870
+    },
+    {
+      "epoch": 0.9586648554240248,
+      "grad_norm": 1.9428020773502297,
+      "learning_rate": 3.4132649750146025e-05,
+      "loss": 0.5227,
+      "step": 9880
+    },
+    {
+      "epoch": 0.9596351639821463,
+      "grad_norm": 2.1394044994306376,
+      "learning_rate": 3.411642546563697e-05,
+      "loss": 0.5227,
+      "step": 9890
+    },
+    {
+      "epoch": 0.9606054725402678,
+      "grad_norm": 1.4958041091695313,
+      "learning_rate": 3.410020118112792e-05,
+      "loss": 0.5536,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9615757810983893,
+      "grad_norm": 1.65783511931855,
+      "learning_rate": 3.408397689661886e-05,
+      "loss": 0.6116,
+      "step": 9910
+    },
+    {
+      "epoch": 0.9625460896565108,
+      "grad_norm": 1.2687727005359897,
+      "learning_rate": 3.406775261210981e-05,
+      "loss": 0.582,
+      "step": 9920
+    },
+    {
+      "epoch": 0.9635163982146323,
+      "grad_norm": 1.642309856430725,
+      "learning_rate": 3.4051528327600754e-05,
+      "loss": 0.5156,
+      "step": 9930
+    },
+    {
+      "epoch": 0.9644867067727537,
+      "grad_norm": 1.5540151182331825,
+      "learning_rate": 3.4035304043091704e-05,
+      "loss": 0.514,
+      "step": 9940
+    },
+    {
+      "epoch": 0.9654570153308752,
+      "grad_norm": 1.6334411063744383,
+      "learning_rate": 3.401907975858265e-05,
+      "loss": 0.585,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9664273238889967,
+      "grad_norm": 1.5262322683208274,
+      "learning_rate": 3.4002855474073596e-05,
+      "loss": 0.5493,
+      "step": 9960
+    },
+    {
+      "epoch": 0.9673976324471182,
+      "grad_norm": 2.041216469634578,
+      "learning_rate": 3.3986631189564546e-05,
+      "loss": 0.5701,
+      "step": 9970
+    },
+    {
+      "epoch": 0.9683679410052397,
+      "grad_norm": 1.6826626866998198,
+      "learning_rate": 3.397040690505549e-05,
+      "loss": 0.5522,
+      "step": 9980
+    },
+    {
+      "epoch": 0.9693382495633611,
+      "grad_norm": 1.3784779820091337,
+      "learning_rate": 3.395418262054644e-05,
+      "loss": 0.5401,
+      "step": 9990
+    },
+    {
+      "epoch": 0.9703085581214826,
+      "grad_norm": 2.164538127382688,
+      "learning_rate": 3.393795833603738e-05,
+      "loss": 0.5493,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9703085581214826,
+      "eval_loss": 0.6604536771774292,
+      "eval_runtime": 2471.8323,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9712788666796042,
+      "grad_norm": 1.641886541364288,
+      "learning_rate": 3.392173405152833e-05,
+      "loss": 0.5592,
+      "step": 10010
+    },
+    {
+      "epoch": 0.9722491752377256,
+      "grad_norm": 1.676600577801569,
+      "learning_rate": 3.3905509767019275e-05,
+      "loss": 0.5675,
+      "step": 10020
+    },
+    {
+      "epoch": 0.9732194837958471,
+      "grad_norm": 1.575522175535914,
+      "learning_rate": 3.3889285482510224e-05,
+      "loss": 0.5564,
+      "step": 10030
+    },
+    {
+      "epoch": 0.9741897923539685,
+      "grad_norm": 1.4256611852011571,
+      "learning_rate": 3.3873061198001174e-05,
+      "loss": 0.5654,
+      "step": 10040
+    },
+    {
+      "epoch": 0.97516010091209,
+      "grad_norm": 1.9766555201121603,
+      "learning_rate": 3.385683691349212e-05,
+      "loss": 0.5853,
+      "step": 10050
+    },
+    {
+      "epoch": 0.9761304094702116,
+      "grad_norm": 1.4257493175991105,
+      "learning_rate": 3.384061262898307e-05,
+      "loss": 0.602,
+      "step": 10060
+    },
+    {
+      "epoch": 0.977100718028333,
+      "grad_norm": 1.7942732122646,
+      "learning_rate": 3.382438834447401e-05,
+      "loss": 0.5083,
+      "step": 10070
+    },
+    {
+      "epoch": 0.9780710265864545,
+      "grad_norm": 1.970290799031593,
+      "learning_rate": 3.380816405996496e-05,
+      "loss": 0.5353,
+      "step": 10080
+    },
+    {
+      "epoch": 0.979041335144576,
+      "grad_norm": 2.177029103778447,
+      "learning_rate": 3.37919397754559e-05,
+      "loss": 0.5364,
+      "step": 10090
+    },
+    {
+      "epoch": 0.9800116437026974,
+      "grad_norm": 1.6319237609838204,
+      "learning_rate": 3.377571549094685e-05,
+      "loss": 0.5498,
+      "step": 10100
+    },
+    {
+      "epoch": 0.980981952260819,
+      "grad_norm": 1.9418003047270103,
+      "learning_rate": 3.3759491206437795e-05,
+      "loss": 0.6225,
+      "step": 10110
+    },
+    {
+      "epoch": 0.9819522608189404,
+      "grad_norm": 1.5233382574211756,
+      "learning_rate": 3.3743266921928745e-05,
+      "loss": 0.5488,
+      "step": 10120
+    },
+    {
+      "epoch": 0.9829225693770619,
+      "grad_norm": 1.7034150218696569,
+      "learning_rate": 3.3727042637419695e-05,
+      "loss": 0.5912,
+      "step": 10130
+    },
+    {
+      "epoch": 0.9838928779351834,
+      "grad_norm": 1.9808892841805323,
+      "learning_rate": 3.371081835291064e-05,
+      "loss": 0.5657,
+      "step": 10140
+    },
+    {
+      "epoch": 0.9848631864933048,
+      "grad_norm": 1.9429830219605533,
+      "learning_rate": 3.369459406840159e-05,
+      "loss": 0.563,
+      "step": 10150
+    },
+    {
+      "epoch": 0.9858334950514264,
+      "grad_norm": 1.7823608925927663,
+      "learning_rate": 3.367836978389253e-05,
+      "loss": 0.5315,
+      "step": 10160
+    },
+    {
+      "epoch": 0.9868038036095479,
+      "grad_norm": 1.6317868773777158,
+      "learning_rate": 3.366214549938348e-05,
+      "loss": 0.5663,
+      "step": 10170
+    },
+    {
+      "epoch": 0.9877741121676693,
+      "grad_norm": 1.9950884655573202,
+      "learning_rate": 3.3645921214874424e-05,
+      "loss": 0.5704,
+      "step": 10180
+    },
+    {
+      "epoch": 0.9887444207257908,
+      "grad_norm": 1.6320762630806733,
+      "learning_rate": 3.362969693036537e-05,
+      "loss": 0.5746,
+      "step": 10190
+    },
+    {
+      "epoch": 0.9897147292839122,
+      "grad_norm": 1.685041060441873,
+      "learning_rate": 3.361347264585632e-05,
+      "loss": 0.5917,
+      "step": 10200
+    },
+    {
+      "epoch": 0.9906850378420338,
+      "grad_norm": 1.7530117501023248,
+      "learning_rate": 3.3597248361347266e-05,
+      "loss": 0.5365,
+      "step": 10210
+    },
+    {
+      "epoch": 0.9916553464001553,
+      "grad_norm": 1.9196791146989973,
+      "learning_rate": 3.3581024076838216e-05,
+      "loss": 0.5795,
+      "step": 10220
+    },
+    {
+      "epoch": 0.9926256549582767,
+      "grad_norm": 1.9870737501998446,
+      "learning_rate": 3.356479979232916e-05,
+      "loss": 0.6544,
+      "step": 10230
+    },
+    {
+      "epoch": 0.9935959635163982,
+      "grad_norm": 1.491102870770748,
+      "learning_rate": 3.354857550782011e-05,
+      "loss": 0.5083,
+      "step": 10240
+    },
+    {
+      "epoch": 0.9945662720745198,
+      "grad_norm": 1.5900809359608934,
+      "learning_rate": 3.353235122331105e-05,
+      "loss": 0.5693,
+      "step": 10250
+    },
+    {
+      "epoch": 0.9955365806326412,
+      "grad_norm": 1.6635095304395011,
+      "learning_rate": 3.3516126938802e-05,
+      "loss": 0.5744,
+      "step": 10260
+    },
+    {
+      "epoch": 0.9965068891907627,
+      "grad_norm": 2.0049230325912957,
+      "learning_rate": 3.3499902654292944e-05,
+      "loss": 0.5461,
+      "step": 10270
+    },
+    {
+      "epoch": 0.9974771977488841,
+      "grad_norm": 1.50147581851131,
+      "learning_rate": 3.3483678369783894e-05,
+      "loss": 0.5803,
+      "step": 10280
+    },
+    {
+      "epoch": 0.9984475063070056,
+      "grad_norm": 1.8064338359868768,
+      "learning_rate": 3.3467454085274844e-05,
+      "loss": 0.506,
+      "step": 10290
+    },
+    {
+      "epoch": 0.9994178148651272,
+      "grad_norm": 2.052231872791701,
+      "learning_rate": 3.345122980076579e-05,
+      "loss": 0.5752,
+      "step": 10300
+    },
+    {
+      "epoch": 1.0003881234232486,
+      "grad_norm": 1.7418958102493116,
+      "learning_rate": 3.343500551625674e-05,
+      "loss": 0.6082,
+      "step": 10310
+    },
+    {
+      "epoch": 1.00135843198137,
+      "grad_norm": 1.8975860607542987,
+      "learning_rate": 3.341878123174768e-05,
+      "loss": 0.5524,
+      "step": 10320
+    },
+    {
+      "epoch": 1.0023287405394916,
+      "grad_norm": 1.7076807320811012,
+      "learning_rate": 3.340255694723863e-05,
+      "loss": 0.6075,
+      "step": 10330
+    },
+    {
+      "epoch": 1.003299049097613,
+      "grad_norm": 1.4300451205657956,
+      "learning_rate": 3.338633266272957e-05,
+      "loss": 0.5326,
+      "step": 10340
+    },
+    {
+      "epoch": 1.0042693576557344,
+      "grad_norm": 2.0682797223020777,
+      "learning_rate": 3.337010837822052e-05,
+      "loss": 0.5484,
+      "step": 10350
+    },
+    {
+      "epoch": 1.005239666213856,
+      "grad_norm": 1.536692096590345,
+      "learning_rate": 3.3353884093711465e-05,
+      "loss": 0.546,
+      "step": 10360
+    },
+    {
+      "epoch": 1.0062099747719775,
+      "grad_norm": 1.7848861749442593,
+      "learning_rate": 3.3337659809202415e-05,
+      "loss": 0.5763,
+      "step": 10370
+    },
+    {
+      "epoch": 1.007180283330099,
+      "grad_norm": 1.4925347515246201,
+      "learning_rate": 3.3321435524693365e-05,
+      "loss": 0.5005,
+      "step": 10380
+    },
+    {
+      "epoch": 1.0081505918882205,
+      "grad_norm": 1.4245048085109102,
+      "learning_rate": 3.330521124018431e-05,
+      "loss": 0.5352,
+      "step": 10390
+    },
+    {
+      "epoch": 1.009120900446342,
+      "grad_norm": 1.6528542775713155,
+      "learning_rate": 3.328898695567526e-05,
+      "loss": 0.5794,
+      "step": 10400
+    },
+    {
+      "epoch": 1.0100912090044634,
+      "grad_norm": 1.7391134864648952,
+      "learning_rate": 3.32727626711662e-05,
+      "loss": 0.6133,
+      "step": 10410
+    },
+    {
+      "epoch": 1.011061517562585,
+      "grad_norm": 1.8040782879083466,
+      "learning_rate": 3.325653838665715e-05,
+      "loss": 0.5124,
+      "step": 10420
+    },
+    {
+      "epoch": 1.0120318261207064,
+      "grad_norm": 1.730132568689756,
+      "learning_rate": 3.324031410214809e-05,
+      "loss": 0.4803,
+      "step": 10430
+    },
+    {
+      "epoch": 1.0130021346788278,
+      "grad_norm": 1.8540575748734034,
+      "learning_rate": 3.322408981763904e-05,
+      "loss": 0.5188,
+      "step": 10440
+    },
+    {
+      "epoch": 1.0139724432369492,
+      "grad_norm": 1.7683659307789739,
+      "learning_rate": 3.3207865533129986e-05,
+      "loss": 0.5341,
+      "step": 10450
+    },
+    {
+      "epoch": 1.0149427517950709,
+      "grad_norm": 1.7745239019667731,
+      "learning_rate": 3.3191641248620936e-05,
+      "loss": 0.5858,
+      "step": 10460
+    },
+    {
+      "epoch": 1.0159130603531923,
+      "grad_norm": 1.5664467690196044,
+      "learning_rate": 3.3175416964111886e-05,
+      "loss": 0.6587,
+      "step": 10470
+    },
+    {
+      "epoch": 1.0168833689113137,
+      "grad_norm": 1.92833985335436,
+      "learning_rate": 3.315919267960283e-05,
+      "loss": 0.5532,
+      "step": 10480
+    },
+    {
+      "epoch": 1.0178536774694353,
+      "grad_norm": 1.646924763771934,
+      "learning_rate": 3.314296839509378e-05,
+      "loss": 0.5638,
+      "step": 10490
+    },
+    {
+      "epoch": 1.0188239860275567,
+      "grad_norm": 1.477844051399051,
+      "learning_rate": 3.312674411058472e-05,
+      "loss": 0.5254,
+      "step": 10500
+    },
+    {
+      "epoch": 1.0188239860275567,
+      "eval_loss": 0.6569487452507019,
+      "eval_runtime": 2472.3859,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 10500
+    },
+    {
+      "epoch": 1.0197942945856782,
+      "grad_norm": 1.7732534556121307,
+      "learning_rate": 3.311051982607567e-05,
+      "loss": 0.5098,
+      "step": 10510
+    },
+    {
+      "epoch": 1.0207646031437998,
+      "grad_norm": 1.8592910812190746,
+      "learning_rate": 3.3094295541566614e-05,
+      "loss": 0.5836,
+      "step": 10520
+    },
+    {
+      "epoch": 1.0217349117019212,
+      "grad_norm": 1.7186074858012894,
+      "learning_rate": 3.3078071257057564e-05,
+      "loss": 0.5445,
+      "step": 10530
+    },
+    {
+      "epoch": 1.0227052202600426,
+      "grad_norm": 2.043653022917622,
+      "learning_rate": 3.306184697254851e-05,
+      "loss": 0.5627,
+      "step": 10540
+    },
+    {
+      "epoch": 1.0236755288181643,
+      "grad_norm": 1.7054473164163695,
+      "learning_rate": 3.304562268803946e-05,
+      "loss": 0.6426,
+      "step": 10550
+    },
+    {
+      "epoch": 1.0246458373762857,
+      "grad_norm": 1.5920397197878904,
+      "learning_rate": 3.3029398403530406e-05,
+      "loss": 0.5542,
+      "step": 10560
+    },
+    {
+      "epoch": 1.025616145934407,
+      "grad_norm": 1.6885876369635333,
+      "learning_rate": 3.301317411902135e-05,
+      "loss": 0.5399,
+      "step": 10570
+    },
+    {
+      "epoch": 1.0265864544925287,
+      "grad_norm": 1.712761359550596,
+      "learning_rate": 3.29969498345123e-05,
+      "loss": 0.5609,
+      "step": 10580
+    },
+    {
+      "epoch": 1.0275567630506501,
+      "grad_norm": 1.849118205679511,
+      "learning_rate": 3.298072555000324e-05,
+      "loss": 0.5817,
+      "step": 10590
+    },
+    {
+      "epoch": 1.0285270716087715,
+      "grad_norm": 1.7919320749353798,
+      "learning_rate": 3.296450126549419e-05,
+      "loss": 0.562,
+      "step": 10600
+    },
+    {
+      "epoch": 1.0294973801668932,
+      "grad_norm": 1.9867833035377922,
+      "learning_rate": 3.2948276980985135e-05,
+      "loss": 0.5266,
+      "step": 10610
+    },
+    {
+      "epoch": 1.0304676887250146,
+      "grad_norm": 1.4844422926334193,
+      "learning_rate": 3.2932052696476085e-05,
+      "loss": 0.5989,
+      "step": 10620
+    },
+    {
+      "epoch": 1.031437997283136,
+      "grad_norm": 1.902397349373601,
+      "learning_rate": 3.2915828411967034e-05,
+      "loss": 0.4869,
+      "step": 10630
+    },
+    {
+      "epoch": 1.0324083058412574,
+      "grad_norm": 1.595731562090853,
+      "learning_rate": 3.289960412745798e-05,
+      "loss": 0.5014,
+      "step": 10640
+    },
+    {
+      "epoch": 1.033378614399379,
+      "grad_norm": 1.641533553754366,
+      "learning_rate": 3.288337984294893e-05,
+      "loss": 0.5847,
+      "step": 10650
+    },
+    {
+      "epoch": 1.0343489229575005,
+      "grad_norm": 1.6793350674365874,
+      "learning_rate": 3.286715555843988e-05,
+      "loss": 0.5433,
+      "step": 10660
+    },
+    {
+      "epoch": 1.0353192315156219,
+      "grad_norm": 1.7886777094595252,
+      "learning_rate": 3.285093127393083e-05,
+      "loss": 0.5936,
+      "step": 10670
+    },
+    {
+      "epoch": 1.0362895400737435,
+      "grad_norm": 1.7628097025479255,
+      "learning_rate": 3.283470698942177e-05,
+      "loss": 0.5176,
+      "step": 10680
+    },
+    {
+      "epoch": 1.037259848631865,
+      "grad_norm": 1.7117894064748884,
+      "learning_rate": 3.281848270491272e-05,
+      "loss": 0.5166,
+      "step": 10690
+    },
+    {
+      "epoch": 1.0382301571899863,
+      "grad_norm": 1.8595449574787446,
+      "learning_rate": 3.280225842040366e-05,
+      "loss": 0.5511,
+      "step": 10700
+    },
+    {
+      "epoch": 1.039200465748108,
+      "grad_norm": 1.5833347089731324,
+      "learning_rate": 3.278603413589461e-05,
+      "loss": 0.5808,
+      "step": 10710
+    },
+    {
+      "epoch": 1.0401707743062294,
+      "grad_norm": 1.5168916043350162,
+      "learning_rate": 3.2769809851385555e-05,
+      "loss": 0.5295,
+      "step": 10720
+    },
+    {
+      "epoch": 1.0411410828643508,
+      "grad_norm": 1.9583788202686643,
+      "learning_rate": 3.2753585566876505e-05,
+      "loss": 0.5859,
+      "step": 10730
+    },
+    {
+      "epoch": 1.0421113914224724,
+      "grad_norm": 1.6490551146953607,
+      "learning_rate": 3.273736128236745e-05,
+      "loss": 0.5988,
+      "step": 10740
+    },
+    {
+      "epoch": 1.0430816999805939,
+      "grad_norm": 2.2945646534561734,
+      "learning_rate": 3.27211369978584e-05,
+      "loss": 0.5481,
+      "step": 10750
+    },
+    {
+      "epoch": 1.0440520085387153,
+      "grad_norm": 1.6823894633457166,
+      "learning_rate": 3.270491271334935e-05,
+      "loss": 0.566,
+      "step": 10760
+    },
+    {
+      "epoch": 1.0450223170968367,
+      "grad_norm": 1.9698263256075523,
+      "learning_rate": 3.268868842884029e-05,
+      "loss": 0.582,
+      "step": 10770
+    },
+    {
+      "epoch": 1.0459926256549583,
+      "grad_norm": 1.6140673075775909,
+      "learning_rate": 3.267246414433124e-05,
+      "loss": 0.4717,
+      "step": 10780
+    },
+    {
+      "epoch": 1.0469629342130797,
+      "grad_norm": 1.916378749987929,
+      "learning_rate": 3.265623985982218e-05,
+      "loss": 0.5244,
+      "step": 10790
+    },
+    {
+      "epoch": 1.0479332427712011,
+      "grad_norm": 1.9482061690949193,
+      "learning_rate": 3.264001557531313e-05,
+      "loss": 0.5973,
+      "step": 10800
+    },
+    {
+      "epoch": 1.0489035513293228,
+      "grad_norm": 2.142491022649256,
+      "learning_rate": 3.2623791290804076e-05,
+      "loss": 0.5317,
+      "step": 10810
+    },
+    {
+      "epoch": 1.0498738598874442,
+      "grad_norm": 1.706373160041174,
+      "learning_rate": 3.2607567006295026e-05,
+      "loss": 0.5136,
+      "step": 10820
+    },
+    {
+      "epoch": 1.0508441684455656,
+      "grad_norm": 1.8959475848770957,
+      "learning_rate": 3.2591342721785976e-05,
+      "loss": 0.5655,
+      "step": 10830
+    },
+    {
+      "epoch": 1.0518144770036872,
+      "grad_norm": 1.8921424824926663,
+      "learning_rate": 3.257511843727692e-05,
+      "loss": 0.5164,
+      "step": 10840
+    },
+    {
+      "epoch": 1.0527847855618087,
+      "grad_norm": 2.0234555341020664,
+      "learning_rate": 3.255889415276787e-05,
+      "loss": 0.5335,
+      "step": 10850
+    },
+    {
+      "epoch": 1.05375509411993,
+      "grad_norm": 2.0614821240519947,
+      "learning_rate": 3.254266986825881e-05,
+      "loss": 0.5022,
+      "step": 10860
+    },
+    {
+      "epoch": 1.0547254026780517,
+      "grad_norm": 1.7142183353166638,
+      "learning_rate": 3.252644558374976e-05,
+      "loss": 0.5452,
+      "step": 10870
+    },
+    {
+      "epoch": 1.0556957112361731,
+      "grad_norm": 1.8265079239939517,
+      "learning_rate": 3.2510221299240704e-05,
+      "loss": 0.5302,
+      "step": 10880
+    },
+    {
+      "epoch": 1.0566660197942945,
+      "grad_norm": 1.7953764101996608,
+      "learning_rate": 3.2493997014731654e-05,
+      "loss": 0.5418,
+      "step": 10890
+    },
+    {
+      "epoch": 1.0576363283524162,
+      "grad_norm": 1.8762589557600082,
+      "learning_rate": 3.24777727302226e-05,
+      "loss": 0.5743,
+      "step": 10900
+    },
+    {
+      "epoch": 1.0586066369105376,
+      "grad_norm": 1.4926319666858994,
+      "learning_rate": 3.246154844571355e-05,
+      "loss": 0.5291,
+      "step": 10910
+    },
+    {
+      "epoch": 1.059576945468659,
+      "grad_norm": 1.9278698099129787,
+      "learning_rate": 3.2445324161204497e-05,
+      "loss": 0.5409,
+      "step": 10920
+    },
+    {
+      "epoch": 1.0605472540267806,
+      "grad_norm": 1.8973044877439684,
+      "learning_rate": 3.242909987669544e-05,
+      "loss": 0.5466,
+      "step": 10930
+    },
+    {
+      "epoch": 1.061517562584902,
+      "grad_norm": 2.174563710599117,
+      "learning_rate": 3.241287559218639e-05,
+      "loss": 0.516,
+      "step": 10940
+    },
+    {
+      "epoch": 1.0624878711430235,
+      "grad_norm": 2.1559495316330786,
+      "learning_rate": 3.239665130767733e-05,
+      "loss": 0.5951,
+      "step": 10950
+    },
+    {
+      "epoch": 1.0634581797011449,
+      "grad_norm": 2.224958308598168,
+      "learning_rate": 3.238042702316828e-05,
+      "loss": 0.5216,
+      "step": 10960
+    },
+    {
+      "epoch": 1.0644284882592665,
+      "grad_norm": 1.5388616251824314,
+      "learning_rate": 3.2364202738659225e-05,
+      "loss": 0.551,
+      "step": 10970
+    },
+    {
+      "epoch": 1.065398796817388,
+      "grad_norm": 1.8563910144302744,
+      "learning_rate": 3.2347978454150175e-05,
+      "loss": 0.5325,
+      "step": 10980
+    },
+    {
+      "epoch": 1.0663691053755093,
+      "grad_norm": 1.249386464606806,
+      "learning_rate": 3.233175416964112e-05,
+      "loss": 0.5223,
+      "step": 10990
+    },
+    {
+      "epoch": 1.067339413933631,
+      "grad_norm": 1.7788595735717652,
+      "learning_rate": 3.231552988513207e-05,
+      "loss": 0.5353,
+      "step": 11000
+    },
+    {
+      "epoch": 1.067339413933631,
+      "eval_loss": 0.6541542410850525,
+      "eval_runtime": 2467.7272,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 11000
+    },
+    {
+      "epoch": 1.0683097224917524,
+      "grad_norm": 2.093397486341294,
+      "learning_rate": 3.229930560062302e-05,
+      "loss": 0.5373,
+      "step": 11010
+    },
+    {
+      "epoch": 1.0692800310498738,
+      "grad_norm": 1.6292327190910305,
+      "learning_rate": 3.228308131611396e-05,
+      "loss": 0.5417,
+      "step": 11020
+    },
+    {
+      "epoch": 1.0702503396079954,
+      "grad_norm": 1.6545192933610855,
+      "learning_rate": 3.226685703160491e-05,
+      "loss": 0.5508,
+      "step": 11030
+    },
+    {
+      "epoch": 1.0712206481661168,
+      "grad_norm": 1.8906889663679678,
+      "learning_rate": 3.225063274709585e-05,
+      "loss": 0.5268,
+      "step": 11040
+    },
+    {
+      "epoch": 1.0721909567242383,
+      "grad_norm": 1.392471317144995,
+      "learning_rate": 3.22344084625868e-05,
+      "loss": 0.5226,
+      "step": 11050
+    },
+    {
+      "epoch": 1.0731612652823599,
+      "grad_norm": 1.9171337047147172,
+      "learning_rate": 3.2218184178077746e-05,
+      "loss": 0.5212,
+      "step": 11060
+    },
+    {
+      "epoch": 1.0741315738404813,
+      "grad_norm": 1.4857874915171814,
+      "learning_rate": 3.2201959893568696e-05,
+      "loss": 0.5335,
+      "step": 11070
+    },
+    {
+      "epoch": 1.0751018823986027,
+      "grad_norm": 1.7538116790409093,
+      "learning_rate": 3.218573560905964e-05,
+      "loss": 0.542,
+      "step": 11080
+    },
+    {
+      "epoch": 1.0760721909567241,
+      "grad_norm": 1.6310285074576256,
+      "learning_rate": 3.216951132455059e-05,
+      "loss": 0.5114,
+      "step": 11090
+    },
+    {
+      "epoch": 1.0770424995148458,
+      "grad_norm": 1.6921457584921478,
+      "learning_rate": 3.215328704004154e-05,
+      "loss": 0.4924,
+      "step": 11100
+    },
+    {
+      "epoch": 1.0780128080729672,
+      "grad_norm": 1.7745242342830565,
+      "learning_rate": 3.213706275553248e-05,
+      "loss": 0.5567,
+      "step": 11110
+    },
+    {
+      "epoch": 1.0789831166310886,
+      "grad_norm": 1.69496703066604,
+      "learning_rate": 3.212083847102343e-05,
+      "loss": 0.5237,
+      "step": 11120
+    },
+    {
+      "epoch": 1.0799534251892102,
+      "grad_norm": 1.7029551999976154,
+      "learning_rate": 3.2104614186514374e-05,
+      "loss": 0.5023,
+      "step": 11130
+    },
+    {
+      "epoch": 1.0809237337473316,
+      "grad_norm": 1.71464384875071,
+      "learning_rate": 3.2088389902005324e-05,
+      "loss": 0.5111,
+      "step": 11140
+    },
+    {
+      "epoch": 1.081894042305453,
+      "grad_norm": 1.745797205638,
+      "learning_rate": 3.207216561749627e-05,
+      "loss": 0.5259,
+      "step": 11150
+    },
+    {
+      "epoch": 1.0828643508635747,
+      "grad_norm": 1.8756703389755518,
+      "learning_rate": 3.2055941332987216e-05,
+      "loss": 0.5135,
+      "step": 11160
+    },
+    {
+      "epoch": 1.083834659421696,
+      "grad_norm": 1.7773801380695633,
+      "learning_rate": 3.203971704847816e-05,
+      "loss": 0.606,
+      "step": 11170
+    },
+    {
+      "epoch": 1.0848049679798175,
+      "grad_norm": 1.9149914262933831,
+      "learning_rate": 3.202349276396911e-05,
+      "loss": 0.504,
+      "step": 11180
+    },
+    {
+      "epoch": 1.0857752765379391,
+      "grad_norm": 1.5792891192161234,
+      "learning_rate": 3.200726847946006e-05,
+      "loss": 0.4899,
+      "step": 11190
+    },
+    {
+      "epoch": 1.0867455850960606,
+      "grad_norm": 1.3849113367423667,
+      "learning_rate": 3.1991044194951e-05,
+      "loss": 0.5435,
+      "step": 11200
+    },
+    {
+      "epoch": 1.087715893654182,
+      "grad_norm": 1.836894182259094,
+      "learning_rate": 3.197481991044195e-05,
+      "loss": 0.5417,
+      "step": 11210
+    },
+    {
+      "epoch": 1.0886862022123036,
+      "grad_norm": 2.5716819476436723,
+      "learning_rate": 3.1958595625932895e-05,
+      "loss": 0.5487,
+      "step": 11220
+    },
+    {
+      "epoch": 1.089656510770425,
+      "grad_norm": 1.7054225586630418,
+      "learning_rate": 3.1942371341423845e-05,
+      "loss": 0.5366,
+      "step": 11230
+    },
+    {
+      "epoch": 1.0906268193285464,
+      "grad_norm": 1.9132021835433188,
+      "learning_rate": 3.192614705691479e-05,
+      "loss": 0.5212,
+      "step": 11240
+    },
+    {
+      "epoch": 1.091597127886668,
+      "grad_norm": 1.5687725973259348,
+      "learning_rate": 3.190992277240574e-05,
+      "loss": 0.5013,
+      "step": 11250
+    },
+    {
+      "epoch": 1.0925674364447895,
+      "grad_norm": 1.7910741705827617,
+      "learning_rate": 3.189369848789669e-05,
+      "loss": 0.5359,
+      "step": 11260
+    },
+    {
+      "epoch": 1.093537745002911,
+      "grad_norm": 1.4238221915445326,
+      "learning_rate": 3.187747420338763e-05,
+      "loss": 0.5729,
+      "step": 11270
+    },
+    {
+      "epoch": 1.0945080535610323,
+      "grad_norm": 1.8958035882900321,
+      "learning_rate": 3.186124991887858e-05,
+      "loss": 0.5349,
+      "step": 11280
+    },
+    {
+      "epoch": 1.095478362119154,
+      "grad_norm": 1.5644842460614365,
+      "learning_rate": 3.184502563436952e-05,
+      "loss": 0.5275,
+      "step": 11290
+    },
+    {
+      "epoch": 1.0964486706772754,
+      "grad_norm": 1.6966656078568068,
+      "learning_rate": 3.182880134986047e-05,
+      "loss": 0.5714,
+      "step": 11300
+    },
+    {
+      "epoch": 1.0974189792353968,
+      "grad_norm": 1.5845176878742038,
+      "learning_rate": 3.1812577065351416e-05,
+      "loss": 0.547,
+      "step": 11310
+    },
+    {
+      "epoch": 1.0983892877935184,
+      "grad_norm": 1.9638352416110092,
+      "learning_rate": 3.1796352780842365e-05,
+      "loss": 0.5371,
+      "step": 11320
+    },
+    {
+      "epoch": 1.0993595963516398,
+      "grad_norm": 1.3333867282862815,
+      "learning_rate": 3.178012849633331e-05,
+      "loss": 0.5558,
+      "step": 11330
+    },
+    {
+      "epoch": 1.1003299049097612,
+      "grad_norm": 2.028988375070847,
+      "learning_rate": 3.176390421182426e-05,
+      "loss": 0.5425,
+      "step": 11340
+    },
+    {
+      "epoch": 1.1013002134678829,
+      "grad_norm": 2.0233727263417745,
+      "learning_rate": 3.174767992731521e-05,
+      "loss": 0.5128,
+      "step": 11350
+    },
+    {
+      "epoch": 1.1022705220260043,
+      "grad_norm": 1.6415117808780524,
+      "learning_rate": 3.173145564280615e-05,
+      "loss": 0.476,
+      "step": 11360
+    },
+    {
+      "epoch": 1.1032408305841257,
+      "grad_norm": 2.0488869521619972,
+      "learning_rate": 3.17152313582971e-05,
+      "loss": 0.5221,
+      "step": 11370
+    },
+    {
+      "epoch": 1.1042111391422473,
+      "grad_norm": 2.11079167436828,
+      "learning_rate": 3.1699007073788044e-05,
+      "loss": 0.5566,
+      "step": 11380
+    },
+    {
+      "epoch": 1.1051814477003687,
+      "grad_norm": 2.0421176306313398,
+      "learning_rate": 3.1682782789278993e-05,
+      "loss": 0.5061,
+      "step": 11390
+    },
+    {
+      "epoch": 1.1061517562584902,
+      "grad_norm": 1.9992500086060474,
+      "learning_rate": 3.1666558504769936e-05,
+      "loss": 0.542,
+      "step": 11400
+    },
+    {
+      "epoch": 1.1071220648166116,
+      "grad_norm": 1.872752246092324,
+      "learning_rate": 3.1650334220260886e-05,
+      "loss": 0.5602,
+      "step": 11410
+    },
+    {
+      "epoch": 1.1080923733747332,
+      "grad_norm": 2.0723616338374646,
+      "learning_rate": 3.163410993575183e-05,
+      "loss": 0.5052,
+      "step": 11420
+    },
+    {
+      "epoch": 1.1090626819328546,
+      "grad_norm": 1.7543258598505558,
+      "learning_rate": 3.161788565124278e-05,
+      "loss": 0.5951,
+      "step": 11430
+    },
+    {
+      "epoch": 1.110032990490976,
+      "grad_norm": 1.741111503275317,
+      "learning_rate": 3.160166136673373e-05,
+      "loss": 0.5481,
+      "step": 11440
+    },
+    {
+      "epoch": 1.1110032990490977,
+      "grad_norm": 1.3801479018015765,
+      "learning_rate": 3.158543708222468e-05,
+      "loss": 0.5104,
+      "step": 11450
+    },
+    {
+      "epoch": 1.111973607607219,
+      "grad_norm": 1.9442170634773426,
+      "learning_rate": 3.156921279771563e-05,
+      "loss": 0.5129,
+      "step": 11460
+    },
+    {
+      "epoch": 1.1129439161653405,
+      "grad_norm": 1.9331832081031561,
+      "learning_rate": 3.155298851320657e-05,
+      "loss": 0.5139,
+      "step": 11470
+    },
+    {
+      "epoch": 1.1139142247234621,
+      "grad_norm": 1.5810959938815903,
+      "learning_rate": 3.153676422869752e-05,
+      "loss": 0.603,
+      "step": 11480
+    },
+    {
+      "epoch": 1.1148845332815835,
+      "grad_norm": 1.7478463817804297,
+      "learning_rate": 3.1520539944188464e-05,
+      "loss": 0.5573,
+      "step": 11490
+    },
+    {
+      "epoch": 1.115854841839705,
+      "grad_norm": 1.512993291404137,
+      "learning_rate": 3.1504315659679414e-05,
+      "loss": 0.4983,
+      "step": 11500
+    },
+    {
+      "epoch": 1.115854841839705,
+      "eval_loss": 0.6542506814002991,
+      "eval_runtime": 2471.8207,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 11500
+    },
+    {
+      "epoch": 1.1168251503978266,
+      "grad_norm": 1.9794498366078679,
+      "learning_rate": 3.148809137517036e-05,
+      "loss": 0.5397,
+      "step": 11510
+    },
+    {
+      "epoch": 1.117795458955948,
+      "grad_norm": 1.6989190060539383,
+      "learning_rate": 3.1471867090661307e-05,
+      "loss": 0.4416,
+      "step": 11520
+    },
+    {
+      "epoch": 1.1187657675140694,
+      "grad_norm": 1.8417193394056512,
+      "learning_rate": 3.145564280615225e-05,
+      "loss": 0.5761,
+      "step": 11530
+    },
+    {
+      "epoch": 1.119736076072191,
+      "grad_norm": 1.8784314737872851,
+      "learning_rate": 3.14394185216432e-05,
+      "loss": 0.5179,
+      "step": 11540
+    },
+    {
+      "epoch": 1.1207063846303125,
+      "grad_norm": 1.724566892104882,
+      "learning_rate": 3.142319423713415e-05,
+      "loss": 0.5145,
+      "step": 11550
+    },
+    {
+      "epoch": 1.1216766931884339,
+      "grad_norm": 1.6271566863595974,
+      "learning_rate": 3.140696995262509e-05,
+      "loss": 0.5557,
+      "step": 11560
+    },
+    {
+      "epoch": 1.1226470017465555,
+      "grad_norm": 1.632474409823301,
+      "learning_rate": 3.139074566811604e-05,
+      "loss": 0.5351,
+      "step": 11570
+    },
+    {
+      "epoch": 1.123617310304677,
+      "grad_norm": 1.8885164133175332,
+      "learning_rate": 3.1374521383606985e-05,
+      "loss": 0.5066,
+      "step": 11580
+    },
+    {
+      "epoch": 1.1245876188627983,
+      "grad_norm": 1.86325603833634,
+      "learning_rate": 3.1358297099097935e-05,
+      "loss": 0.5561,
+      "step": 11590
+    },
+    {
+      "epoch": 1.1255579274209198,
+      "grad_norm": 1.8325010805036752,
+      "learning_rate": 3.134207281458888e-05,
+      "loss": 0.5518,
+      "step": 11600
+    },
+    {
+      "epoch": 1.1265282359790414,
+      "grad_norm": 1.7732624170282982,
+      "learning_rate": 3.132584853007983e-05,
+      "loss": 0.5264,
+      "step": 11610
+    },
+    {
+      "epoch": 1.1274985445371628,
+      "grad_norm": 1.6849511508626207,
+      "learning_rate": 3.130962424557077e-05,
+      "loss": 0.4938,
+      "step": 11620
+    },
+    {
+      "epoch": 1.1284688530952842,
+      "grad_norm": 1.7088287322634965,
+      "learning_rate": 3.129339996106172e-05,
+      "loss": 0.5432,
+      "step": 11630
+    },
+    {
+      "epoch": 1.1294391616534059,
+      "grad_norm": 2.4521787808904345,
+      "learning_rate": 3.127717567655267e-05,
+      "loss": 0.5222,
+      "step": 11640
+    },
+    {
+      "epoch": 1.1304094702115273,
+      "grad_norm": 1.8568933343609308,
+      "learning_rate": 3.126095139204361e-05,
+      "loss": 0.5155,
+      "step": 11650
+    },
+    {
+      "epoch": 1.1313797787696487,
+      "grad_norm": 1.6581222418410275,
+      "learning_rate": 3.124472710753456e-05,
+      "loss": 0.5039,
+      "step": 11660
+    },
+    {
+      "epoch": 1.1323500873277703,
+      "grad_norm": 1.8783913354540864,
+      "learning_rate": 3.1228502823025506e-05,
+      "loss": 0.5083,
+      "step": 11670
+    },
+    {
+      "epoch": 1.1333203958858917,
+      "grad_norm": 1.7156898303987873,
+      "learning_rate": 3.1212278538516455e-05,
+      "loss": 0.5184,
+      "step": 11680
+    },
+    {
+      "epoch": 1.1342907044440131,
+      "grad_norm": 2.1185799396739093,
+      "learning_rate": 3.11960542540074e-05,
+      "loss": 0.5252,
+      "step": 11690
+    },
+    {
+      "epoch": 1.1352610130021348,
+      "grad_norm": 1.2978327151961555,
+      "learning_rate": 3.117982996949835e-05,
+      "loss": 0.5548,
+      "step": 11700
+    },
+    {
+      "epoch": 1.1362313215602562,
+      "grad_norm": 1.6276998380365333,
+      "learning_rate": 3.116360568498929e-05,
+      "loss": 0.5412,
+      "step": 11710
+    },
+    {
+      "epoch": 1.1372016301183776,
+      "grad_norm": 1.5330498087853446,
+      "learning_rate": 3.114738140048024e-05,
+      "loss": 0.4811,
+      "step": 11720
+    },
+    {
+      "epoch": 1.138171938676499,
+      "grad_norm": 1.545547307525802,
+      "learning_rate": 3.113115711597119e-05,
+      "loss": 0.4841,
+      "step": 11730
+    },
+    {
+      "epoch": 1.1391422472346207,
+      "grad_norm": 1.8963148998956563,
+      "learning_rate": 3.1114932831462134e-05,
+      "loss": 0.5048,
+      "step": 11740
+    },
+    {
+      "epoch": 1.140112555792742,
+      "grad_norm": 1.927144240645965,
+      "learning_rate": 3.1098708546953084e-05,
+      "loss": 0.5437,
+      "step": 11750
+    },
+    {
+      "epoch": 1.1410828643508635,
+      "grad_norm": 1.322758693894623,
+      "learning_rate": 3.1082484262444027e-05,
+      "loss": 0.5824,
+      "step": 11760
+    },
+    {
+      "epoch": 1.1420531729089851,
+      "grad_norm": 1.9210557222184033,
+      "learning_rate": 3.1066259977934976e-05,
+      "loss": 0.5208,
+      "step": 11770
+    },
+    {
+      "epoch": 1.1430234814671065,
+      "grad_norm": 1.8904898905530507,
+      "learning_rate": 3.105003569342592e-05,
+      "loss": 0.5646,
+      "step": 11780
+    },
+    {
+      "epoch": 1.143993790025228,
+      "grad_norm": 2.2135650729717478,
+      "learning_rate": 3.103381140891687e-05,
+      "loss": 0.5823,
+      "step": 11790
+    },
+    {
+      "epoch": 1.1449640985833496,
+      "grad_norm": 1.8257267748894057,
+      "learning_rate": 3.101758712440781e-05,
+      "loss": 0.4957,
+      "step": 11800
+    },
+    {
+      "epoch": 1.145934407141471,
+      "grad_norm": 1.8399910987928954,
+      "learning_rate": 3.100136283989876e-05,
+      "loss": 0.5589,
+      "step": 11810
+    },
+    {
+      "epoch": 1.1469047156995924,
+      "grad_norm": 1.751016659591738,
+      "learning_rate": 3.098513855538971e-05,
+      "loss": 0.4636,
+      "step": 11820
+    },
+    {
+      "epoch": 1.147875024257714,
+      "grad_norm": 2.3784962245655743,
+      "learning_rate": 3.0968914270880655e-05,
+      "loss": 0.558,
+      "step": 11830
+    },
+    {
+      "epoch": 1.1488453328158355,
+      "grad_norm": 1.7453204083727232,
+      "learning_rate": 3.0952689986371604e-05,
+      "loss": 0.5152,
+      "step": 11840
+    },
+    {
+      "epoch": 1.1498156413739569,
+      "grad_norm": 1.6493432346857455,
+      "learning_rate": 3.093646570186255e-05,
+      "loss": 0.5396,
+      "step": 11850
+    },
+    {
+      "epoch": 1.1507859499320783,
+      "grad_norm": 1.4016634963337722,
+      "learning_rate": 3.09202414173535e-05,
+      "loss": 0.5392,
+      "step": 11860
+    },
+    {
+      "epoch": 1.1517562584902,
+      "grad_norm": 1.5144392427573337,
+      "learning_rate": 3.090401713284444e-05,
+      "loss": 0.5625,
+      "step": 11870
+    },
+    {
+      "epoch": 1.1527265670483213,
+      "grad_norm": 1.9911432006869803,
+      "learning_rate": 3.088779284833539e-05,
+      "loss": 0.5187,
+      "step": 11880
+    },
+    {
+      "epoch": 1.153696875606443,
+      "grad_norm": 1.989669700307941,
+      "learning_rate": 3.087156856382634e-05,
+      "loss": 0.516,
+      "step": 11890
+    },
+    {
+      "epoch": 1.1546671841645644,
+      "grad_norm": 2.0321346469591717,
+      "learning_rate": 3.085534427931728e-05,
+      "loss": 0.5028,
+      "step": 11900
+    },
+    {
+      "epoch": 1.1556374927226858,
+      "grad_norm": 1.6419270337849394,
+      "learning_rate": 3.083911999480823e-05,
+      "loss": 0.5114,
+      "step": 11910
+    },
+    {
+      "epoch": 1.1566078012808072,
+      "grad_norm": 1.8199861134594042,
+      "learning_rate": 3.0822895710299175e-05,
+      "loss": 0.4862,
+      "step": 11920
+    },
+    {
+      "epoch": 1.1575781098389288,
+      "grad_norm": 2.092599945357918,
+      "learning_rate": 3.0806671425790125e-05,
+      "loss": 0.4862,
+      "step": 11930
+    },
+    {
+      "epoch": 1.1585484183970503,
+      "grad_norm": 1.6390208725289623,
+      "learning_rate": 3.079044714128107e-05,
+      "loss": 0.5518,
+      "step": 11940
+    },
+    {
+      "epoch": 1.1595187269551717,
+      "grad_norm": 1.4035901760825538,
+      "learning_rate": 3.077422285677202e-05,
+      "loss": 0.5195,
+      "step": 11950
+    },
+    {
+      "epoch": 1.1604890355132933,
+      "grad_norm": 1.965204556535071,
+      "learning_rate": 3.075799857226296e-05,
+      "loss": 0.5198,
+      "step": 11960
+    },
+    {
+      "epoch": 1.1614593440714147,
+      "grad_norm": 1.797317897425162,
+      "learning_rate": 3.074177428775391e-05,
+      "loss": 0.5223,
+      "step": 11970
+    },
+    {
+      "epoch": 1.1624296526295361,
+      "grad_norm": 2.0562783101788713,
+      "learning_rate": 3.072555000324486e-05,
+      "loss": 0.5207,
+      "step": 11980
+    },
+    {
+      "epoch": 1.1633999611876578,
+      "grad_norm": 1.674272970925229,
+      "learning_rate": 3.0709325718735804e-05,
+      "loss": 0.5034,
+      "step": 11990
+    },
+    {
+      "epoch": 1.1643702697457792,
+      "grad_norm": 1.8613409955734859,
+      "learning_rate": 3.069310143422675e-05,
+      "loss": 0.4921,
+      "step": 12000
+    },
+    {
+      "epoch": 1.1643702697457792,
+      "eval_loss": 0.6506599187850952,
+      "eval_runtime": 2467.2044,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 12000
+    },
+    {
+      "epoch": 1.1653405783039006,
+      "grad_norm": 1.561512098916604,
+      "learning_rate": 3.0676877149717696e-05,
+      "loss": 0.5497,
+      "step": 12010
+    },
+    {
+      "epoch": 1.1663108868620222,
+      "grad_norm": 1.865230398957055,
+      "learning_rate": 3.0660652865208646e-05,
+      "loss": 0.4925,
+      "step": 12020
+    },
+    {
+      "epoch": 1.1672811954201436,
+      "grad_norm": 2.2512724716981434,
+      "learning_rate": 3.064442858069959e-05,
+      "loss": 0.5295,
+      "step": 12030
+    },
+    {
+      "epoch": 1.168251503978265,
+      "grad_norm": 1.8002247664216207,
+      "learning_rate": 3.062820429619054e-05,
+      "loss": 0.4996,
+      "step": 12040
+    },
+    {
+      "epoch": 1.1692218125363865,
+      "grad_norm": 2.1116351677440943,
+      "learning_rate": 3.061198001168148e-05,
+      "loss": 0.5191,
+      "step": 12050
+    },
+    {
+      "epoch": 1.170192121094508,
+      "grad_norm": 1.6725748372150215,
+      "learning_rate": 3.059575572717243e-05,
+      "loss": 0.5336,
+      "step": 12060
+    },
+    {
+      "epoch": 1.1711624296526295,
+      "grad_norm": 1.9025453675497637,
+      "learning_rate": 3.057953144266338e-05,
+      "loss": 0.5188,
+      "step": 12070
+    },
+    {
+      "epoch": 1.172132738210751,
+      "grad_norm": 1.8069530215074574,
+      "learning_rate": 3.0563307158154324e-05,
+      "loss": 0.5495,
+      "step": 12080
+    },
+    {
+      "epoch": 1.1731030467688726,
+      "grad_norm": 1.4756589453476456,
+      "learning_rate": 3.0547082873645274e-05,
+      "loss": 0.5248,
+      "step": 12090
+    },
+    {
+      "epoch": 1.174073355326994,
+      "grad_norm": 1.894556403114157,
+      "learning_rate": 3.053085858913622e-05,
+      "loss": 0.5604,
+      "step": 12100
+    },
+    {
+      "epoch": 1.1750436638851154,
+      "grad_norm": 1.7674666934401635,
+      "learning_rate": 3.0514634304627167e-05,
+      "loss": 0.5355,
+      "step": 12110
+    },
+    {
+      "epoch": 1.176013972443237,
+      "grad_norm": 1.6165024582050291,
+      "learning_rate": 3.0498410020118113e-05,
+      "loss": 0.4861,
+      "step": 12120
+    },
+    {
+      "epoch": 1.1769842810013584,
+      "grad_norm": 1.7586790770235476,
+      "learning_rate": 3.048218573560906e-05,
+      "loss": 0.5009,
+      "step": 12130
+    },
+    {
+      "epoch": 1.1779545895594798,
+      "grad_norm": 1.424893034597034,
+      "learning_rate": 3.0465961451100006e-05,
+      "loss": 0.5441,
+      "step": 12140
+    },
+    {
+      "epoch": 1.1789248981176015,
+      "grad_norm": 1.79883718769592,
+      "learning_rate": 3.0449737166590952e-05,
+      "loss": 0.5313,
+      "step": 12150
+    },
+    {
+      "epoch": 1.179895206675723,
+      "grad_norm": 1.2166009796658896,
+      "learning_rate": 3.04335128820819e-05,
+      "loss": 0.5771,
+      "step": 12160
+    },
+    {
+      "epoch": 1.1808655152338443,
+      "grad_norm": 1.8319881864090852,
+      "learning_rate": 3.0417288597572845e-05,
+      "loss": 0.5341,
+      "step": 12170
+    },
+    {
+      "epoch": 1.1818358237919657,
+      "grad_norm": 2.000590117325487,
+      "learning_rate": 3.040106431306379e-05,
+      "loss": 0.5434,
+      "step": 12180
+    },
+    {
+      "epoch": 1.1828061323500874,
+      "grad_norm": 1.7597266341378504,
+      "learning_rate": 3.038484002855474e-05,
+      "loss": 0.4637,
+      "step": 12190
+    },
+    {
+      "epoch": 1.1837764409082088,
+      "grad_norm": 1.7547988130290746,
+      "learning_rate": 3.0368615744045688e-05,
+      "loss": 0.5238,
+      "step": 12200
+    },
+    {
+      "epoch": 1.1847467494663304,
+      "grad_norm": 2.072598486347679,
+      "learning_rate": 3.0352391459536634e-05,
+      "loss": 0.5198,
+      "step": 12210
+    },
+    {
+      "epoch": 1.1857170580244518,
+      "grad_norm": 1.801682666637873,
+      "learning_rate": 3.033616717502758e-05,
+      "loss": 0.5549,
+      "step": 12220
+    },
+    {
+      "epoch": 1.1866873665825732,
+      "grad_norm": 1.7660188156759438,
+      "learning_rate": 3.0319942890518534e-05,
+      "loss": 0.5415,
+      "step": 12230
+    },
+    {
+      "epoch": 1.1876576751406946,
+      "grad_norm": 1.531355321026242,
+      "learning_rate": 3.030371860600948e-05,
+      "loss": 0.5125,
+      "step": 12240
+    },
+    {
+      "epoch": 1.1886279836988163,
+      "grad_norm": 1.6289177265084471,
+      "learning_rate": 3.0287494321500426e-05,
+      "loss": 0.5131,
+      "step": 12250
+    },
+    {
+      "epoch": 1.1895982922569377,
+      "grad_norm": 2.3228651312748423,
+      "learning_rate": 3.0271270036991373e-05,
+      "loss": 0.4859,
+      "step": 12260
+    },
+    {
+      "epoch": 1.190568600815059,
+      "grad_norm": 2.076037584027454,
+      "learning_rate": 3.025504575248232e-05,
+      "loss": 0.5234,
+      "step": 12270
+    },
+    {
+      "epoch": 1.1915389093731807,
+      "grad_norm": 1.4111596404765325,
+      "learning_rate": 3.0238821467973266e-05,
+      "loss": 0.5226,
+      "step": 12280
+    },
+    {
+      "epoch": 1.1925092179313022,
+      "grad_norm": 1.976122066748625,
+      "learning_rate": 3.0222597183464212e-05,
+      "loss": 0.5753,
+      "step": 12290
+    },
+    {
+      "epoch": 1.1934795264894236,
+      "grad_norm": 1.6905440972221555,
+      "learning_rate": 3.0206372898955158e-05,
+      "loss": 0.5559,
+      "step": 12300
+    },
+    {
+      "epoch": 1.1944498350475452,
+      "grad_norm": 1.549005200001293,
+      "learning_rate": 3.0190148614446108e-05,
+      "loss": 0.5199,
+      "step": 12310
+    },
+    {
+      "epoch": 1.1954201436056666,
+      "grad_norm": 1.7886858738633538,
+      "learning_rate": 3.0173924329937054e-05,
+      "loss": 0.5524,
+      "step": 12320
+    },
+    {
+      "epoch": 1.196390452163788,
+      "grad_norm": 2.226393124620313,
+      "learning_rate": 3.0157700045428e-05,
+      "loss": 0.511,
+      "step": 12330
+    },
+    {
+      "epoch": 1.1973607607219097,
+      "grad_norm": 1.84134054205334,
+      "learning_rate": 3.0141475760918947e-05,
+      "loss": 0.5537,
+      "step": 12340
+    },
+    {
+      "epoch": 1.198331069280031,
+      "grad_norm": 1.938676760265039,
+      "learning_rate": 3.0125251476409894e-05,
+      "loss": 0.5782,
+      "step": 12350
+    },
+    {
+      "epoch": 1.1993013778381525,
+      "grad_norm": 1.6671922582631973,
+      "learning_rate": 3.010902719190084e-05,
+      "loss": 0.5146,
+      "step": 12360
+    },
+    {
+      "epoch": 1.200271686396274,
+      "grad_norm": 2.09828561481085,
+      "learning_rate": 3.0092802907391786e-05,
+      "loss": 0.5292,
+      "step": 12370
+    },
+    {
+      "epoch": 1.2012419949543955,
+      "grad_norm": 1.782738185716262,
+      "learning_rate": 3.0076578622882733e-05,
+      "loss": 0.5797,
+      "step": 12380
+    },
+    {
+      "epoch": 1.202212303512517,
+      "grad_norm": 1.7386393975624728,
+      "learning_rate": 3.006035433837368e-05,
+      "loss": 0.5459,
+      "step": 12390
+    },
+    {
+      "epoch": 1.2031826120706384,
+      "grad_norm": 1.834619533516341,
+      "learning_rate": 3.004413005386463e-05,
+      "loss": 0.5711,
+      "step": 12400
+    },
+    {
+      "epoch": 1.20415292062876,
+      "grad_norm": 1.6555949649402653,
+      "learning_rate": 3.0027905769355575e-05,
+      "loss": 0.5447,
+      "step": 12410
+    },
+    {
+      "epoch": 1.2051232291868814,
+      "grad_norm": 1.997714330931175,
+      "learning_rate": 3.001168148484652e-05,
+      "loss": 0.5001,
+      "step": 12420
+    },
+    {
+      "epoch": 1.2060935377450028,
+      "grad_norm": 1.943566532939107,
+      "learning_rate": 2.9995457200337468e-05,
+      "loss": 0.5612,
+      "step": 12430
+    },
+    {
+      "epoch": 1.2070638463031245,
+      "grad_norm": 2.295630424579193,
+      "learning_rate": 2.9979232915828414e-05,
+      "loss": 0.5443,
+      "step": 12440
+    },
+    {
+      "epoch": 1.2080341548612459,
+      "grad_norm": 1.6087261613477206,
+      "learning_rate": 2.996300863131936e-05,
+      "loss": 0.4929,
+      "step": 12450
+    },
+    {
+      "epoch": 1.2090044634193673,
+      "grad_norm": 1.7507917723489945,
+      "learning_rate": 2.9946784346810307e-05,
+      "loss": 0.4915,
+      "step": 12460
+    },
+    {
+      "epoch": 1.209974771977489,
+      "grad_norm": 1.712263283374138,
+      "learning_rate": 2.9930560062301254e-05,
+      "loss": 0.5184,
+      "step": 12470
+    },
+    {
+      "epoch": 1.2109450805356103,
+      "grad_norm": 2.1091363117437427,
+      "learning_rate": 2.9914335777792203e-05,
+      "loss": 0.5123,
+      "step": 12480
+    },
+    {
+      "epoch": 1.2119153890937318,
+      "grad_norm": 2.404178772933857,
+      "learning_rate": 2.989811149328315e-05,
+      "loss": 0.4568,
+      "step": 12490
+    },
+    {
+      "epoch": 1.2128856976518532,
+      "grad_norm": 1.6898907246159178,
+      "learning_rate": 2.9881887208774096e-05,
+      "loss": 0.5114,
+      "step": 12500
+    },
+    {
+      "epoch": 1.2128856976518532,
+      "eval_loss": 0.6506454348564148,
+      "eval_runtime": 2464.983,
+      "eval_samples_per_second": 0.727,
+      "eval_steps_per_second": 0.363,
+      "step": 12500
+    },
+    {
+      "epoch": 1.2138560062099748,
+      "grad_norm": 1.5632796322726878,
+      "learning_rate": 2.9865662924265042e-05,
+      "loss": 0.5381,
+      "step": 12510
+    },
+    {
+      "epoch": 1.2148263147680962,
+      "grad_norm": 2.2060623757238482,
+      "learning_rate": 2.984943863975599e-05,
+      "loss": 0.6286,
+      "step": 12520
+    },
+    {
+      "epoch": 1.2157966233262179,
+      "grad_norm": 1.6896138867780373,
+      "learning_rate": 2.9833214355246935e-05,
+      "loss": 0.4824,
+      "step": 12530
+    },
+    {
+      "epoch": 1.2167669318843393,
+      "grad_norm": 1.6264014630619223,
+      "learning_rate": 2.981699007073788e-05,
+      "loss": 0.5231,
+      "step": 12540
+    },
+    {
+      "epoch": 1.2177372404424607,
+      "grad_norm": 2.288555955501704,
+      "learning_rate": 2.9800765786228828e-05,
+      "loss": 0.5415,
+      "step": 12550
+    },
+    {
+      "epoch": 1.218707549000582,
+      "grad_norm": 1.6328806432164462,
+      "learning_rate": 2.9784541501719774e-05,
+      "loss": 0.5282,
+      "step": 12560
+    },
+    {
+      "epoch": 1.2196778575587037,
+      "grad_norm": 1.9940506922760688,
+      "learning_rate": 2.9768317217210724e-05,
+      "loss": 0.5283,
+      "step": 12570
+    },
+    {
+      "epoch": 1.2206481661168251,
+      "grad_norm": 1.9369189438911159,
+      "learning_rate": 2.975209293270167e-05,
+      "loss": 0.4722,
+      "step": 12580
+    },
+    {
+      "epoch": 1.2216184746749466,
+      "grad_norm": 1.150006209597975,
+      "learning_rate": 2.9735868648192617e-05,
+      "loss": 0.5184,
+      "step": 12590
+    },
+    {
+      "epoch": 1.2225887832330682,
+      "grad_norm": 1.7210167191672803,
+      "learning_rate": 2.9719644363683563e-05,
+      "loss": 0.5031,
+      "step": 12600
+    },
+    {
+      "epoch": 1.2235590917911896,
+      "grad_norm": 1.9388550988757736,
+      "learning_rate": 2.970342007917451e-05,
+      "loss": 0.5099,
+      "step": 12610
+    },
+    {
+      "epoch": 1.224529400349311,
+      "grad_norm": 1.6524197083323393,
+      "learning_rate": 2.9687195794665456e-05,
+      "loss": 0.4447,
+      "step": 12620
+    },
+    {
+      "epoch": 1.2254997089074326,
+      "grad_norm": 1.89736479001966,
+      "learning_rate": 2.9670971510156402e-05,
+      "loss": 0.4983,
+      "step": 12630
+    },
+    {
+      "epoch": 1.226470017465554,
+      "grad_norm": 1.895097527141105,
+      "learning_rate": 2.965474722564735e-05,
+      "loss": 0.5039,
+      "step": 12640
+    },
+    {
+      "epoch": 1.2274403260236755,
+      "grad_norm": 1.976305435076919,
+      "learning_rate": 2.96385229411383e-05,
+      "loss": 0.5337,
+      "step": 12650
+    },
+    {
+      "epoch": 1.2284106345817971,
+      "grad_norm": 1.6796293491451193,
+      "learning_rate": 2.9622298656629245e-05,
+      "loss": 0.506,
+      "step": 12660
+    },
+    {
+      "epoch": 1.2293809431399185,
+      "grad_norm": 1.6431597372306554,
+      "learning_rate": 2.960607437212019e-05,
+      "loss": 0.5797,
+      "step": 12670
+    },
+    {
+      "epoch": 1.23035125169804,
+      "grad_norm": 1.8020203988119472,
+      "learning_rate": 2.9589850087611138e-05,
+      "loss": 0.4957,
+      "step": 12680
+    },
+    {
+      "epoch": 1.2313215602561614,
+      "grad_norm": 2.0575623836232935,
+      "learning_rate": 2.9573625803102084e-05,
+      "loss": 0.5125,
+      "step": 12690
+    },
+    {
+      "epoch": 1.232291868814283,
+      "grad_norm": 2.0440836316269544,
+      "learning_rate": 2.955740151859303e-05,
+      "loss": 0.5186,
+      "step": 12700
+    },
+    {
+      "epoch": 1.2332621773724044,
+      "grad_norm": 1.6684531467277435,
+      "learning_rate": 2.9541177234083977e-05,
+      "loss": 0.5348,
+      "step": 12710
+    },
+    {
+      "epoch": 1.2342324859305258,
+      "grad_norm": 1.6949248831996988,
+      "learning_rate": 2.9524952949574923e-05,
+      "loss": 0.5484,
+      "step": 12720
+    },
+    {
+      "epoch": 1.2352027944886474,
+      "grad_norm": 1.773704436016878,
+      "learning_rate": 2.950872866506587e-05,
+      "loss": 0.5033,
+      "step": 12730
+    },
+    {
+      "epoch": 1.2361731030467689,
+      "grad_norm": 1.9246733654165475,
+      "learning_rate": 2.949250438055682e-05,
+      "loss": 0.5164,
+      "step": 12740
+    },
+    {
+      "epoch": 1.2371434116048903,
+      "grad_norm": 1.7869787657786207,
+      "learning_rate": 2.9476280096047766e-05,
+      "loss": 0.549,
+      "step": 12750
+    },
+    {
+      "epoch": 1.238113720163012,
+      "grad_norm": 1.5427226807712424,
+      "learning_rate": 2.9460055811538712e-05,
+      "loss": 0.5453,
+      "step": 12760
+    },
+    {
+      "epoch": 1.2390840287211333,
+      "grad_norm": 2.1243893484204706,
+      "learning_rate": 2.944383152702966e-05,
+      "loss": 0.5232,
+      "step": 12770
+    },
+    {
+      "epoch": 1.2400543372792547,
+      "grad_norm": 1.7624693076719502,
+      "learning_rate": 2.9427607242520605e-05,
+      "loss": 0.4767,
+      "step": 12780
+    },
+    {
+      "epoch": 1.2410246458373764,
+      "grad_norm": 1.8048923369800416,
+      "learning_rate": 2.941138295801155e-05,
+      "loss": 0.5786,
+      "step": 12790
+    },
+    {
+      "epoch": 1.2419949543954978,
+      "grad_norm": 1.557577350338282,
+      "learning_rate": 2.9395158673502498e-05,
+      "loss": 0.5358,
+      "step": 12800
+    },
+    {
+      "epoch": 1.2429652629536192,
+      "grad_norm": 1.7796545697030264,
+      "learning_rate": 2.9378934388993444e-05,
+      "loss": 0.5447,
+      "step": 12810
+    },
+    {
+      "epoch": 1.2439355715117406,
+      "grad_norm": 1.5233346835390529,
+      "learning_rate": 2.936271010448439e-05,
+      "loss": 0.5434,
+      "step": 12820
+    },
+    {
+      "epoch": 1.2449058800698622,
+      "grad_norm": 1.8281296123454516,
+      "learning_rate": 2.934648581997534e-05,
+      "loss": 0.5311,
+      "step": 12830
+    },
+    {
+      "epoch": 1.2458761886279837,
+      "grad_norm": 1.8761299295652716,
+      "learning_rate": 2.9330261535466287e-05,
+      "loss": 0.561,
+      "step": 12840
+    },
+    {
+      "epoch": 1.2468464971861053,
+      "grad_norm": 1.571060091128229,
+      "learning_rate": 2.9314037250957233e-05,
+      "loss": 0.5293,
+      "step": 12850
+    },
+    {
+      "epoch": 1.2478168057442267,
+      "grad_norm": 1.3554175730214915,
+      "learning_rate": 2.929781296644818e-05,
+      "loss": 0.5259,
+      "step": 12860
+    },
+    {
+      "epoch": 1.2487871143023481,
+      "grad_norm": 1.5048611859450334,
+      "learning_rate": 2.9281588681939126e-05,
+      "loss": 0.5526,
+      "step": 12870
+    },
+    {
+      "epoch": 1.2497574228604695,
+      "grad_norm": 1.944003477785508,
+      "learning_rate": 2.9265364397430072e-05,
+      "loss": 0.5319,
+      "step": 12880
+    },
+    {
+      "epoch": 1.2507277314185912,
+      "grad_norm": 1.6137183858737645,
+      "learning_rate": 2.924914011292102e-05,
+      "loss": 0.5219,
+      "step": 12890
+    },
+    {
+      "epoch": 1.2516980399767126,
+      "grad_norm": 1.8792239391042322,
+      "learning_rate": 2.9232915828411965e-05,
+      "loss": 0.4996,
+      "step": 12900
+    },
+    {
+      "epoch": 1.252668348534834,
+      "grad_norm": 1.6048948758745463,
+      "learning_rate": 2.9216691543902915e-05,
+      "loss": 0.5094,
+      "step": 12910
+    },
+    {
+      "epoch": 1.2536386570929556,
+      "grad_norm": 2.06874670039208,
+      "learning_rate": 2.920046725939386e-05,
+      "loss": 0.5412,
+      "step": 12920
+    },
+    {
+      "epoch": 1.254608965651077,
+      "grad_norm": 1.8299215479076065,
+      "learning_rate": 2.9184242974884808e-05,
+      "loss": 0.5215,
+      "step": 12930
+    },
+    {
+      "epoch": 1.2555792742091985,
+      "grad_norm": 1.768175531798484,
+      "learning_rate": 2.9168018690375754e-05,
+      "loss": 0.5466,
+      "step": 12940
+    },
+    {
+      "epoch": 1.2565495827673199,
+      "grad_norm": 2.19878862821484,
+      "learning_rate": 2.91517944058667e-05,
+      "loss": 0.5728,
+      "step": 12950
+    },
+    {
+      "epoch": 1.2575198913254415,
+      "grad_norm": 1.5410059750444967,
+      "learning_rate": 2.9135570121357647e-05,
+      "loss": 0.5248,
+      "step": 12960
+    },
+    {
+      "epoch": 1.258490199883563,
+      "grad_norm": 1.9319057652262193,
+      "learning_rate": 2.9119345836848593e-05,
+      "loss": 0.4906,
+      "step": 12970
+    },
+    {
+      "epoch": 1.2594605084416846,
+      "grad_norm": 1.4285564772766584,
+      "learning_rate": 2.910312155233954e-05,
+      "loss": 0.4976,
+      "step": 12980
+    },
+    {
+      "epoch": 1.260430816999806,
+      "grad_norm": 1.9098028563216323,
+      "learning_rate": 2.9086897267830486e-05,
+      "loss": 0.4841,
+      "step": 12990
+    },
+    {
+      "epoch": 1.2614011255579274,
+      "grad_norm": 2.155481185125366,
+      "learning_rate": 2.9070672983321436e-05,
+      "loss": 0.5486,
+      "step": 13000
+    },
+    {
+      "epoch": 1.2614011255579274,
+      "eval_loss": 0.6476565003395081,
+      "eval_runtime": 2469.4767,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 13000
+    },
+    {
+      "epoch": 1.2623714341160488,
+      "grad_norm": 1.7583302051951548,
+      "learning_rate": 2.9054448698812385e-05,
+      "loss": 0.5287,
+      "step": 13010
+    },
+    {
+      "epoch": 1.2633417426741704,
+      "grad_norm": 2.1364231232324777,
+      "learning_rate": 2.9038224414303332e-05,
+      "loss": 0.5338,
+      "step": 13020
+    },
+    {
+      "epoch": 1.2643120512322918,
+      "grad_norm": 1.858948662905276,
+      "learning_rate": 2.902200012979428e-05,
+      "loss": 0.5371,
+      "step": 13030
+    },
+    {
+      "epoch": 1.2652823597904135,
+      "grad_norm": 1.434526008901497,
+      "learning_rate": 2.9005775845285228e-05,
+      "loss": 0.5199,
+      "step": 13040
+    },
+    {
+      "epoch": 1.266252668348535,
+      "grad_norm": 1.8326954509581932,
+      "learning_rate": 2.8989551560776174e-05,
+      "loss": 0.4626,
+      "step": 13050
+    },
+    {
+      "epoch": 1.2672229769066563,
+      "grad_norm": 1.929734623290484,
+      "learning_rate": 2.897332727626712e-05,
+      "loss": 0.5095,
+      "step": 13060
+    },
+    {
+      "epoch": 1.2681932854647777,
+      "grad_norm": 1.7342388298669522,
+      "learning_rate": 2.8957102991758067e-05,
+      "loss": 0.5334,
+      "step": 13070
+    },
+    {
+      "epoch": 1.2691635940228994,
+      "grad_norm": 1.8419675684147427,
+      "learning_rate": 2.8940878707249013e-05,
+      "loss": 0.5215,
+      "step": 13080
+    },
+    {
+      "epoch": 1.2701339025810208,
+      "grad_norm": 2.0003719776284434,
+      "learning_rate": 2.892465442273996e-05,
+      "loss": 0.5291,
+      "step": 13090
+    },
+    {
+      "epoch": 1.2711042111391422,
+      "grad_norm": 1.5672070400575155,
+      "learning_rate": 2.8908430138230906e-05,
+      "loss": 0.5302,
+      "step": 13100
+    },
+    {
+      "epoch": 1.2720745196972638,
+      "grad_norm": 1.6329476161915306,
+      "learning_rate": 2.8892205853721856e-05,
+      "loss": 0.5004,
+      "step": 13110
+    },
+    {
+      "epoch": 1.2730448282553852,
+      "grad_norm": 1.3430643020518607,
+      "learning_rate": 2.8875981569212802e-05,
+      "loss": 0.5586,
+      "step": 13120
+    },
+    {
+      "epoch": 1.2740151368135066,
+      "grad_norm": 1.865116300464639,
+      "learning_rate": 2.885975728470375e-05,
+      "loss": 0.5176,
+      "step": 13130
+    },
+    {
+      "epoch": 1.274985445371628,
+      "grad_norm": 1.9537429368278099,
+      "learning_rate": 2.8843533000194695e-05,
+      "loss": 0.5167,
+      "step": 13140
+    },
+    {
+      "epoch": 1.2759557539297497,
+      "grad_norm": 2.059620522389882,
+      "learning_rate": 2.882730871568564e-05,
+      "loss": 0.4654,
+      "step": 13150
+    },
+    {
+      "epoch": 1.276926062487871,
+      "grad_norm": 2.1139159674960317,
+      "learning_rate": 2.8811084431176588e-05,
+      "loss": 0.5365,
+      "step": 13160
+    },
+    {
+      "epoch": 1.2778963710459927,
+      "grad_norm": 1.9675564766085183,
+      "learning_rate": 2.8794860146667534e-05,
+      "loss": 0.5567,
+      "step": 13170
+    },
+    {
+      "epoch": 1.2788666796041142,
+      "grad_norm": 1.4714857260607748,
+      "learning_rate": 2.877863586215848e-05,
+      "loss": 0.5543,
+      "step": 13180
+    },
+    {
+      "epoch": 1.2798369881622356,
+      "grad_norm": 1.997029571976202,
+      "learning_rate": 2.8762411577649427e-05,
+      "loss": 0.5349,
+      "step": 13190
+    },
+    {
+      "epoch": 1.280807296720357,
+      "grad_norm": 1.7739211918572755,
+      "learning_rate": 2.8746187293140377e-05,
+      "loss": 0.4893,
+      "step": 13200
+    },
+    {
+      "epoch": 1.2817776052784786,
+      "grad_norm": 1.7069871948143287,
+      "learning_rate": 2.8729963008631323e-05,
+      "loss": 0.4907,
+      "step": 13210
+    },
+    {
+      "epoch": 1.2827479138366,
+      "grad_norm": 1.9981954512910625,
+      "learning_rate": 2.871373872412227e-05,
+      "loss": 0.5672,
+      "step": 13220
+    },
+    {
+      "epoch": 1.2837182223947214,
+      "grad_norm": 1.6613620495892372,
+      "learning_rate": 2.8697514439613216e-05,
+      "loss": 0.541,
+      "step": 13230
+    },
+    {
+      "epoch": 1.284688530952843,
+      "grad_norm": 2.3272590952107675,
+      "learning_rate": 2.8681290155104162e-05,
+      "loss": 0.5411,
+      "step": 13240
+    },
+    {
+      "epoch": 1.2856588395109645,
+      "grad_norm": 2.12223916886103,
+      "learning_rate": 2.866506587059511e-05,
+      "loss": 0.4957,
+      "step": 13250
+    },
+    {
+      "epoch": 1.286629148069086,
+      "grad_norm": 1.4853001948712596,
+      "learning_rate": 2.8648841586086055e-05,
+      "loss": 0.5432,
+      "step": 13260
+    },
+    {
+      "epoch": 1.2875994566272073,
+      "grad_norm": 2.095035318310476,
+      "learning_rate": 2.8632617301577e-05,
+      "loss": 0.5466,
+      "step": 13270
+    },
+    {
+      "epoch": 1.288569765185329,
+      "grad_norm": 1.8562546228217478,
+      "learning_rate": 2.8616393017067948e-05,
+      "loss": 0.5393,
+      "step": 13280
+    },
+    {
+      "epoch": 1.2895400737434504,
+      "grad_norm": 1.960721539695441,
+      "learning_rate": 2.8600168732558898e-05,
+      "loss": 0.5737,
+      "step": 13290
+    },
+    {
+      "epoch": 1.290510382301572,
+      "grad_norm": 1.8998278241439668,
+      "learning_rate": 2.8583944448049844e-05,
+      "loss": 0.4965,
+      "step": 13300
+    },
+    {
+      "epoch": 1.2914806908596934,
+      "grad_norm": 1.648594607244266,
+      "learning_rate": 2.856772016354079e-05,
+      "loss": 0.5691,
+      "step": 13310
+    },
+    {
+      "epoch": 1.2924509994178148,
+      "grad_norm": 1.65007773003003,
+      "learning_rate": 2.8551495879031737e-05,
+      "loss": 0.5,
+      "step": 13320
+    },
+    {
+      "epoch": 1.2934213079759362,
+      "grad_norm": 2.0573651261935293,
+      "learning_rate": 2.8535271594522683e-05,
+      "loss": 0.4926,
+      "step": 13330
+    },
+    {
+      "epoch": 1.2943916165340579,
+      "grad_norm": 2.328992496419527,
+      "learning_rate": 2.851904731001363e-05,
+      "loss": 0.5221,
+      "step": 13340
+    },
+    {
+      "epoch": 1.2953619250921793,
+      "grad_norm": 1.8509833991470588,
+      "learning_rate": 2.8502823025504576e-05,
+      "loss": 0.5018,
+      "step": 13350
+    },
+    {
+      "epoch": 1.296332233650301,
+      "grad_norm": 1.895399492110669,
+      "learning_rate": 2.8486598740995522e-05,
+      "loss": 0.4724,
+      "step": 13360
+    },
+    {
+      "epoch": 1.2973025422084223,
+      "grad_norm": 1.667924410333677,
+      "learning_rate": 2.8470374456486472e-05,
+      "loss": 0.5165,
+      "step": 13370
+    },
+    {
+      "epoch": 1.2982728507665438,
+      "grad_norm": 2.1618438681975722,
+      "learning_rate": 2.845415017197742e-05,
+      "loss": 0.5208,
+      "step": 13380
+    },
+    {
+      "epoch": 1.2992431593246652,
+      "grad_norm": 2.0324515358778434,
+      "learning_rate": 2.8437925887468365e-05,
+      "loss": 0.4944,
+      "step": 13390
+    },
+    {
+      "epoch": 1.3002134678827868,
+      "grad_norm": 1.4374011148146266,
+      "learning_rate": 2.842170160295931e-05,
+      "loss": 0.556,
+      "step": 13400
+    },
+    {
+      "epoch": 1.3011837764409082,
+      "grad_norm": 2.0625833924885075,
+      "learning_rate": 2.8405477318450258e-05,
+      "loss": 0.533,
+      "step": 13410
+    },
+    {
+      "epoch": 1.3021540849990296,
+      "grad_norm": 1.7327381778160986,
+      "learning_rate": 2.8389253033941204e-05,
+      "loss": 0.5428,
+      "step": 13420
+    },
+    {
+      "epoch": 1.3031243935571513,
+      "grad_norm": 1.5980081031800506,
+      "learning_rate": 2.837302874943215e-05,
+      "loss": 0.5169,
+      "step": 13430
+    },
+    {
+      "epoch": 1.3040947021152727,
+      "grad_norm": 1.4133329538382045,
+      "learning_rate": 2.8356804464923097e-05,
+      "loss": 0.5365,
+      "step": 13440
+    },
+    {
+      "epoch": 1.305065010673394,
+      "grad_norm": 1.6052634633267397,
+      "learning_rate": 2.8340580180414043e-05,
+      "loss": 0.5062,
+      "step": 13450
+    },
+    {
+      "epoch": 1.3060353192315155,
+      "grad_norm": 1.8789186495289818,
+      "learning_rate": 2.8324355895904993e-05,
+      "loss": 0.5301,
+      "step": 13460
+    },
+    {
+      "epoch": 1.3070056277896371,
+      "grad_norm": 1.7539318856514114,
+      "learning_rate": 2.830813161139594e-05,
+      "loss": 0.5191,
+      "step": 13470
+    },
+    {
+      "epoch": 1.3079759363477585,
+      "grad_norm": 1.652081891063903,
+      "learning_rate": 2.8291907326886886e-05,
+      "loss": 0.5162,
+      "step": 13480
+    },
+    {
+      "epoch": 1.3089462449058802,
+      "grad_norm": 1.9042372320107943,
+      "learning_rate": 2.8275683042377832e-05,
+      "loss": 0.5202,
+      "step": 13490
+    },
+    {
+      "epoch": 1.3099165534640016,
+      "grad_norm": 1.9149924815733665,
+      "learning_rate": 2.825945875786878e-05,
+      "loss": 0.5785,
+      "step": 13500
+    },
+    {
+      "epoch": 1.3099165534640016,
+      "eval_loss": 0.6458503007888794,
+      "eval_runtime": 2472.3879,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 13500
+    },
+    {
+      "epoch": 1.310886862022123,
+      "grad_norm": 1.6405768771538807,
+      "learning_rate": 2.8243234473359725e-05,
+      "loss": 0.5037,
+      "step": 13510
+    },
+    {
+      "epoch": 1.3118571705802444,
+      "grad_norm": 1.8018934132744455,
+      "learning_rate": 2.822701018885067e-05,
+      "loss": 0.5465,
+      "step": 13520
+    },
+    {
+      "epoch": 1.312827479138366,
+      "grad_norm": 2.013531086384291,
+      "learning_rate": 2.8210785904341618e-05,
+      "loss": 0.5011,
+      "step": 13530
+    },
+    {
+      "epoch": 1.3137977876964875,
+      "grad_norm": 2.022552703035907,
+      "learning_rate": 2.8194561619832567e-05,
+      "loss": 0.5657,
+      "step": 13540
+    },
+    {
+      "epoch": 1.3147680962546089,
+      "grad_norm": 1.734149403661694,
+      "learning_rate": 2.8178337335323514e-05,
+      "loss": 0.5074,
+      "step": 13550
+    },
+    {
+      "epoch": 1.3157384048127305,
+      "grad_norm": 2.1508604289357476,
+      "learning_rate": 2.816211305081446e-05,
+      "loss": 0.4873,
+      "step": 13560
+    },
+    {
+      "epoch": 1.316708713370852,
+      "grad_norm": 1.9266405893568426,
+      "learning_rate": 2.8145888766305406e-05,
+      "loss": 0.5231,
+      "step": 13570
+    },
+    {
+      "epoch": 1.3176790219289733,
+      "grad_norm": 1.8623995402610134,
+      "learning_rate": 2.8129664481796353e-05,
+      "loss": 0.5748,
+      "step": 13580
+    },
+    {
+      "epoch": 1.3186493304870948,
+      "grad_norm": 1.5283427217297032,
+      "learning_rate": 2.81134401972873e-05,
+      "loss": 0.5322,
+      "step": 13590
+    },
+    {
+      "epoch": 1.3196196390452164,
+      "grad_norm": 1.7148189442620572,
+      "learning_rate": 2.8097215912778246e-05,
+      "loss": 0.4896,
+      "step": 13600
+    },
+    {
+      "epoch": 1.3205899476033378,
+      "grad_norm": 1.4473714452022712,
+      "learning_rate": 2.8080991628269192e-05,
+      "loss": 0.4974,
+      "step": 13610
+    },
+    {
+      "epoch": 1.3215602561614594,
+      "grad_norm": 1.8643931328762333,
+      "learning_rate": 2.806476734376014e-05,
+      "loss": 0.5707,
+      "step": 13620
+    },
+    {
+      "epoch": 1.3225305647195809,
+      "grad_norm": 2.1475243574365264,
+      "learning_rate": 2.8048543059251088e-05,
+      "loss": 0.5167,
+      "step": 13630
+    },
+    {
+      "epoch": 1.3235008732777023,
+      "grad_norm": 1.6293038645225775,
+      "learning_rate": 2.8032318774742035e-05,
+      "loss": 0.5268,
+      "step": 13640
+    },
+    {
+      "epoch": 1.3244711818358237,
+      "grad_norm": 1.8987021880259773,
+      "learning_rate": 2.801609449023298e-05,
+      "loss": 0.5432,
+      "step": 13650
+    },
+    {
+      "epoch": 1.3254414903939453,
+      "grad_norm": 1.5753855821090716,
+      "learning_rate": 2.7999870205723927e-05,
+      "loss": 0.4827,
+      "step": 13660
+    },
+    {
+      "epoch": 1.3264117989520667,
+      "grad_norm": 1.350257936732941,
+      "learning_rate": 2.7983645921214874e-05,
+      "loss": 0.5544,
+      "step": 13670
+    },
+    {
+      "epoch": 1.3273821075101884,
+      "grad_norm": 1.9293055864878257,
+      "learning_rate": 2.796742163670582e-05,
+      "loss": 0.5441,
+      "step": 13680
+    },
+    {
+      "epoch": 1.3283524160683098,
+      "grad_norm": 2.2701614440224813,
+      "learning_rate": 2.7951197352196766e-05,
+      "loss": 0.5237,
+      "step": 13690
+    },
+    {
+      "epoch": 1.3293227246264312,
+      "grad_norm": 1.2019327117406085,
+      "learning_rate": 2.7934973067687713e-05,
+      "loss": 0.5434,
+      "step": 13700
+    },
+    {
+      "epoch": 1.3302930331845526,
+      "grad_norm": 1.406257025731888,
+      "learning_rate": 2.7918748783178663e-05,
+      "loss": 0.5063,
+      "step": 13710
+    },
+    {
+      "epoch": 1.3312633417426742,
+      "grad_norm": 1.666513597838276,
+      "learning_rate": 2.790252449866961e-05,
+      "loss": 0.5068,
+      "step": 13720
+    },
+    {
+      "epoch": 1.3322336503007957,
+      "grad_norm": 1.7668183417692156,
+      "learning_rate": 2.7886300214160555e-05,
+      "loss": 0.5141,
+      "step": 13730
+    },
+    {
+      "epoch": 1.333203958858917,
+      "grad_norm": 1.433593593379768,
+      "learning_rate": 2.7870075929651502e-05,
+      "loss": 0.4626,
+      "step": 13740
+    },
+    {
+      "epoch": 1.3341742674170387,
+      "grad_norm": 2.009965700363568,
+      "learning_rate": 2.7853851645142448e-05,
+      "loss": 0.5618,
+      "step": 13750
+    },
+    {
+      "epoch": 1.3351445759751601,
+      "grad_norm": 1.763668408586817,
+      "learning_rate": 2.7837627360633395e-05,
+      "loss": 0.5369,
+      "step": 13760
+    },
+    {
+      "epoch": 1.3361148845332815,
+      "grad_norm": 2.1411324301354053,
+      "learning_rate": 2.782140307612434e-05,
+      "loss": 0.5084,
+      "step": 13770
+    },
+    {
+      "epoch": 1.337085193091403,
+      "grad_norm": 1.436827138346562,
+      "learning_rate": 2.7805178791615287e-05,
+      "loss": 0.5032,
+      "step": 13780
+    },
+    {
+      "epoch": 1.3380555016495246,
+      "grad_norm": 1.9606877958295938,
+      "learning_rate": 2.778895450710624e-05,
+      "loss": 0.5327,
+      "step": 13790
+    },
+    {
+      "epoch": 1.339025810207646,
+      "grad_norm": 2.0522393469060396,
+      "learning_rate": 2.7772730222597187e-05,
+      "loss": 0.5398,
+      "step": 13800
+    },
+    {
+      "epoch": 1.3399961187657676,
+      "grad_norm": 1.9786107551791654,
+      "learning_rate": 2.7756505938088133e-05,
+      "loss": 0.5009,
+      "step": 13810
+    },
+    {
+      "epoch": 1.340966427323889,
+      "grad_norm": 2.090584551627846,
+      "learning_rate": 2.774028165357908e-05,
+      "loss": 0.5296,
+      "step": 13820
+    },
+    {
+      "epoch": 1.3419367358820105,
+      "grad_norm": 2.1298950160365613,
+      "learning_rate": 2.772405736907003e-05,
+      "loss": 0.4865,
+      "step": 13830
+    },
+    {
+      "epoch": 1.3429070444401319,
+      "grad_norm": 1.7629297128429193,
+      "learning_rate": 2.7707833084560976e-05,
+      "loss": 0.5324,
+      "step": 13840
+    },
+    {
+      "epoch": 1.3438773529982535,
+      "grad_norm": 1.8397197756904764,
+      "learning_rate": 2.7691608800051922e-05,
+      "loss": 0.5732,
+      "step": 13850
+    },
+    {
+      "epoch": 1.344847661556375,
+      "grad_norm": 1.8108747182833855,
+      "learning_rate": 2.767538451554287e-05,
+      "loss": 0.4947,
+      "step": 13860
+    },
+    {
+      "epoch": 1.3458179701144963,
+      "grad_norm": 1.8199614245355178,
+      "learning_rate": 2.7659160231033815e-05,
+      "loss": 0.5557,
+      "step": 13870
+    },
+    {
+      "epoch": 1.346788278672618,
+      "grad_norm": 1.3431248134057203,
+      "learning_rate": 2.764293594652476e-05,
+      "loss": 0.5404,
+      "step": 13880
+    },
+    {
+      "epoch": 1.3477585872307394,
+      "grad_norm": 1.5574090523973907,
+      "learning_rate": 2.7626711662015708e-05,
+      "loss": 0.568,
+      "step": 13890
+    },
+    {
+      "epoch": 1.3487288957888608,
+      "grad_norm": 1.4115013178507176,
+      "learning_rate": 2.7610487377506654e-05,
+      "loss": 0.4716,
+      "step": 13900
+    },
+    {
+      "epoch": 1.3496992043469822,
+      "grad_norm": 2.1040010286866444,
+      "learning_rate": 2.75942630929976e-05,
+      "loss": 0.4934,
+      "step": 13910
+    },
+    {
+      "epoch": 1.3506695129051038,
+      "grad_norm": 1.7498060957230301,
+      "learning_rate": 2.757803880848855e-05,
+      "loss": 0.5454,
+      "step": 13920
+    },
+    {
+      "epoch": 1.3516398214632253,
+      "grad_norm": 1.8910220759501428,
+      "learning_rate": 2.7561814523979497e-05,
+      "loss": 0.5641,
+      "step": 13930
+    },
+    {
+      "epoch": 1.352610130021347,
+      "grad_norm": 1.6170015266249442,
+      "learning_rate": 2.7545590239470443e-05,
+      "loss": 0.5378,
+      "step": 13940
+    },
+    {
+      "epoch": 1.3535804385794683,
+      "grad_norm": 1.7666178388963019,
+      "learning_rate": 2.752936595496139e-05,
+      "loss": 0.5448,
+      "step": 13950
+    },
+    {
+      "epoch": 1.3545507471375897,
+      "grad_norm": 2.052806059730038,
+      "learning_rate": 2.7513141670452336e-05,
+      "loss": 0.5158,
+      "step": 13960
+    },
+    {
+      "epoch": 1.3555210556957111,
+      "grad_norm": 1.9222501285118103,
+      "learning_rate": 2.7496917385943282e-05,
+      "loss": 0.4982,
+      "step": 13970
+    },
+    {
+      "epoch": 1.3564913642538328,
+      "grad_norm": 2.1370232742150903,
+      "learning_rate": 2.748069310143423e-05,
+      "loss": 0.5409,
+      "step": 13980
+    },
+    {
+      "epoch": 1.3574616728119542,
+      "grad_norm": 1.7396628585118512,
+      "learning_rate": 2.7464468816925175e-05,
+      "loss": 0.5327,
+      "step": 13990
+    },
+    {
+      "epoch": 1.3584319813700758,
+      "grad_norm": 1.990716927328341,
+      "learning_rate": 2.7448244532416125e-05,
+      "loss": 0.5291,
+      "step": 14000
+    },
+    {
+      "epoch": 1.3584319813700758,
+      "eval_loss": 0.6447646021842957,
+      "eval_runtime": 2468.8402,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 14000
+    },
+    {
+      "epoch": 1.3594022899281972,
+      "grad_norm": 1.7121198472067753,
+      "learning_rate": 2.743202024790707e-05,
+      "loss": 0.4758,
+      "step": 14010
+    },
+    {
+      "epoch": 1.3603725984863186,
+      "grad_norm": 1.6674489776942332,
+      "learning_rate": 2.7415795963398017e-05,
+      "loss": 0.5484,
+      "step": 14020
+    },
+    {
+      "epoch": 1.36134290704444,
+      "grad_norm": 1.458164078535125,
+      "learning_rate": 2.7399571678888964e-05,
+      "loss": 0.5523,
+      "step": 14030
+    },
+    {
+      "epoch": 1.3623132156025617,
+      "grad_norm": 1.8775916401381276,
+      "learning_rate": 2.738334739437991e-05,
+      "loss": 0.5254,
+      "step": 14040
+    },
+    {
+      "epoch": 1.363283524160683,
+      "grad_norm": 1.671470903392816,
+      "learning_rate": 2.7367123109870857e-05,
+      "loss": 0.5005,
+      "step": 14050
+    },
+    {
+      "epoch": 1.3642538327188045,
+      "grad_norm": 1.9806763146120538,
+      "learning_rate": 2.7350898825361803e-05,
+      "loss": 0.5134,
+      "step": 14060
+    },
+    {
+      "epoch": 1.3652241412769262,
+      "grad_norm": 1.514946401551448,
+      "learning_rate": 2.733467454085275e-05,
+      "loss": 0.5597,
+      "step": 14070
+    },
+    {
+      "epoch": 1.3661944498350476,
+      "grad_norm": 1.6635049106626887,
+      "learning_rate": 2.7318450256343696e-05,
+      "loss": 0.5377,
+      "step": 14080
+    },
+    {
+      "epoch": 1.367164758393169,
+      "grad_norm": 1.7564332843089718,
+      "learning_rate": 2.7302225971834645e-05,
+      "loss": 0.5027,
+      "step": 14090
+    },
+    {
+      "epoch": 1.3681350669512904,
+      "grad_norm": 2.0454872453162682,
+      "learning_rate": 2.7286001687325592e-05,
+      "loss": 0.5056,
+      "step": 14100
+    },
+    {
+      "epoch": 1.369105375509412,
+      "grad_norm": 2.0786210322749032,
+      "learning_rate": 2.7269777402816538e-05,
+      "loss": 0.5835,
+      "step": 14110
+    },
+    {
+      "epoch": 1.3700756840675334,
+      "grad_norm": 2.1275215610783555,
+      "learning_rate": 2.7253553118307485e-05,
+      "loss": 0.5235,
+      "step": 14120
+    },
+    {
+      "epoch": 1.371045992625655,
+      "grad_norm": 1.7013208723478948,
+      "learning_rate": 2.723732883379843e-05,
+      "loss": 0.5105,
+      "step": 14130
+    },
+    {
+      "epoch": 1.3720163011837765,
+      "grad_norm": 1.586441401545233,
+      "learning_rate": 2.7221104549289377e-05,
+      "loss": 0.5221,
+      "step": 14140
+    },
+    {
+      "epoch": 1.372986609741898,
+      "grad_norm": 1.533260185383275,
+      "learning_rate": 2.7204880264780324e-05,
+      "loss": 0.4714,
+      "step": 14150
+    },
+    {
+      "epoch": 1.3739569183000193,
+      "grad_norm": 1.99495942930363,
+      "learning_rate": 2.718865598027127e-05,
+      "loss": 0.5127,
+      "step": 14160
+    },
+    {
+      "epoch": 1.374927226858141,
+      "grad_norm": 1.5455163764317608,
+      "learning_rate": 2.717243169576222e-05,
+      "loss": 0.5331,
+      "step": 14170
+    },
+    {
+      "epoch": 1.3758975354162624,
+      "grad_norm": 2.542301246782172,
+      "learning_rate": 2.7156207411253166e-05,
+      "loss": 0.5196,
+      "step": 14180
+    },
+    {
+      "epoch": 1.3768678439743838,
+      "grad_norm": 2.367306381136639,
+      "learning_rate": 2.7139983126744113e-05,
+      "loss": 0.5173,
+      "step": 14190
+    },
+    {
+      "epoch": 1.3778381525325054,
+      "grad_norm": 1.748922567904988,
+      "learning_rate": 2.712375884223506e-05,
+      "loss": 0.5301,
+      "step": 14200
+    },
+    {
+      "epoch": 1.3788084610906268,
+      "grad_norm": 1.9185950864572987,
+      "learning_rate": 2.7107534557726005e-05,
+      "loss": 0.4242,
+      "step": 14210
+    },
+    {
+      "epoch": 1.3797787696487482,
+      "grad_norm": 1.7876319295098282,
+      "learning_rate": 2.7091310273216952e-05,
+      "loss": 0.5251,
+      "step": 14220
+    },
+    {
+      "epoch": 1.3807490782068697,
+      "grad_norm": 1.7638194028118903,
+      "learning_rate": 2.7075085988707898e-05,
+      "loss": 0.5087,
+      "step": 14230
+    },
+    {
+      "epoch": 1.3817193867649913,
+      "grad_norm": 1.781232728742025,
+      "learning_rate": 2.7058861704198845e-05,
+      "loss": 0.46,
+      "step": 14240
+    },
+    {
+      "epoch": 1.3826896953231127,
+      "grad_norm": 2.0454750065003564,
+      "learning_rate": 2.704263741968979e-05,
+      "loss": 0.542,
+      "step": 14250
+    },
+    {
+      "epoch": 1.3836600038812343,
+      "grad_norm": 2.2368401504756,
+      "learning_rate": 2.702641313518074e-05,
+      "loss": 0.501,
+      "step": 14260
+    },
+    {
+      "epoch": 1.3846303124393557,
+      "grad_norm": 1.4717190899560046,
+      "learning_rate": 2.7010188850671687e-05,
+      "loss": 0.5146,
+      "step": 14270
+    },
+    {
+      "epoch": 1.3856006209974772,
+      "grad_norm": 1.8752670210283335,
+      "learning_rate": 2.6993964566162634e-05,
+      "loss": 0.4999,
+      "step": 14280
+    },
+    {
+      "epoch": 1.3865709295555986,
+      "grad_norm": 2.065234795556922,
+      "learning_rate": 2.697774028165358e-05,
+      "loss": 0.4447,
+      "step": 14290
+    },
+    {
+      "epoch": 1.3875412381137202,
+      "grad_norm": 2.125070528749461,
+      "learning_rate": 2.6961515997144526e-05,
+      "loss": 0.4797,
+      "step": 14300
+    },
+    {
+      "epoch": 1.3885115466718416,
+      "grad_norm": 1.5859308448291956,
+      "learning_rate": 2.6945291712635473e-05,
+      "loss": 0.5336,
+      "step": 14310
+    },
+    {
+      "epoch": 1.3894818552299633,
+      "grad_norm": 2.861319646114515,
+      "learning_rate": 2.692906742812642e-05,
+      "loss": 0.4812,
+      "step": 14320
+    },
+    {
+      "epoch": 1.3904521637880847,
+      "grad_norm": 1.7597224976071428,
+      "learning_rate": 2.6912843143617365e-05,
+      "loss": 0.55,
+      "step": 14330
+    },
+    {
+      "epoch": 1.391422472346206,
+      "grad_norm": 1.681866583110354,
+      "learning_rate": 2.6896618859108312e-05,
+      "loss": 0.5046,
+      "step": 14340
+    },
+    {
+      "epoch": 1.3923927809043275,
+      "grad_norm": 1.5215374726522628,
+      "learning_rate": 2.688039457459926e-05,
+      "loss": 0.5579,
+      "step": 14350
+    },
+    {
+      "epoch": 1.3933630894624491,
+      "grad_norm": 2.079812021813028,
+      "learning_rate": 2.6864170290090208e-05,
+      "loss": 0.5495,
+      "step": 14360
+    },
+    {
+      "epoch": 1.3943333980205705,
+      "grad_norm": 1.617321032110613,
+      "learning_rate": 2.6847946005581154e-05,
+      "loss": 0.529,
+      "step": 14370
+    },
+    {
+      "epoch": 1.395303706578692,
+      "grad_norm": 1.9418293727425169,
+      "learning_rate": 2.68317217210721e-05,
+      "loss": 0.5503,
+      "step": 14380
+    },
+    {
+      "epoch": 1.3962740151368136,
+      "grad_norm": 1.6562011692834846,
+      "learning_rate": 2.6815497436563047e-05,
+      "loss": 0.5245,
+      "step": 14390
+    },
+    {
+      "epoch": 1.397244323694935,
+      "grad_norm": 1.8991877294518344,
+      "learning_rate": 2.6799273152053994e-05,
+      "loss": 0.4976,
+      "step": 14400
+    },
+    {
+      "epoch": 1.3982146322530564,
+      "grad_norm": 1.9301826202288486,
+      "learning_rate": 2.678304886754494e-05,
+      "loss": 0.4878,
+      "step": 14410
+    },
+    {
+      "epoch": 1.3991849408111778,
+      "grad_norm": 2.019905342024004,
+      "learning_rate": 2.6766824583035886e-05,
+      "loss": 0.5093,
+      "step": 14420
+    },
+    {
+      "epoch": 1.4001552493692995,
+      "grad_norm": 2.364843712084718,
+      "learning_rate": 2.6750600298526836e-05,
+      "loss": 0.5005,
+      "step": 14430
+    },
+    {
+      "epoch": 1.4011255579274209,
+      "grad_norm": 1.9751607681902115,
+      "learning_rate": 2.6734376014017782e-05,
+      "loss": 0.4958,
+      "step": 14440
+    },
+    {
+      "epoch": 1.4020958664855425,
+      "grad_norm": 1.9625121046714047,
+      "learning_rate": 2.671815172950873e-05,
+      "loss": 0.5251,
+      "step": 14450
+    },
+    {
+      "epoch": 1.403066175043664,
+      "grad_norm": 1.4828612193168078,
+      "learning_rate": 2.6701927444999675e-05,
+      "loss": 0.5496,
+      "step": 14460
+    },
+    {
+      "epoch": 1.4040364836017853,
+      "grad_norm": 1.5015720843906,
+      "learning_rate": 2.668570316049062e-05,
+      "loss": 0.535,
+      "step": 14470
+    },
+    {
+      "epoch": 1.4050067921599068,
+      "grad_norm": 1.3851757562899687,
+      "learning_rate": 2.6669478875981568e-05,
+      "loss": 0.5628,
+      "step": 14480
+    },
+    {
+      "epoch": 1.4059771007180284,
+      "grad_norm": 1.6822047277997916,
+      "learning_rate": 2.6653254591472514e-05,
+      "loss": 0.5605,
+      "step": 14490
+    },
+    {
+      "epoch": 1.4069474092761498,
+      "grad_norm": 1.9002135762249894,
+      "learning_rate": 2.663703030696346e-05,
+      "loss": 0.5348,
+      "step": 14500
+    },
+    {
+      "epoch": 1.4069474092761498,
+      "eval_loss": 0.6422961950302124,
+      "eval_runtime": 2474.9423,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.362,
+      "step": 14500
+    },
+    {
+      "epoch": 1.4079177178342712,
+      "grad_norm": 1.7272883282125664,
+      "learning_rate": 2.6620806022454407e-05,
+      "loss": 0.5518,
+      "step": 14510
+    },
+    {
+      "epoch": 1.4088880263923929,
+      "grad_norm": 1.7836182665219666,
+      "learning_rate": 2.6604581737945357e-05,
+      "loss": 0.5195,
+      "step": 14520
+    },
+    {
+      "epoch": 1.4098583349505143,
+      "grad_norm": 1.7860780128722327,
+      "learning_rate": 2.6588357453436303e-05,
+      "loss": 0.5061,
+      "step": 14530
+    },
+    {
+      "epoch": 1.4108286435086357,
+      "grad_norm": 1.8487543077145125,
+      "learning_rate": 2.657213316892725e-05,
+      "loss": 0.5261,
+      "step": 14540
+    },
+    {
+      "epoch": 1.411798952066757,
+      "grad_norm": 1.8960118987299062,
+      "learning_rate": 2.6555908884418196e-05,
+      "loss": 0.523,
+      "step": 14550
+    },
+    {
+      "epoch": 1.4127692606248787,
+      "grad_norm": 1.3767948578941858,
+      "learning_rate": 2.6539684599909142e-05,
+      "loss": 0.4654,
+      "step": 14560
+    },
+    {
+      "epoch": 1.4137395691830001,
+      "grad_norm": 1.7605168074790472,
+      "learning_rate": 2.652346031540009e-05,
+      "loss": 0.4711,
+      "step": 14570
+    },
+    {
+      "epoch": 1.4147098777411218,
+      "grad_norm": 1.424973688569186,
+      "learning_rate": 2.6507236030891042e-05,
+      "loss": 0.5639,
+      "step": 14580
+    },
+    {
+      "epoch": 1.4156801862992432,
+      "grad_norm": 1.9215824187230874,
+      "learning_rate": 2.649101174638199e-05,
+      "loss": 0.5137,
+      "step": 14590
+    },
+    {
+      "epoch": 1.4166504948573646,
+      "grad_norm": 1.439592415452285,
+      "learning_rate": 2.6474787461872935e-05,
+      "loss": 0.488,
+      "step": 14600
+    },
+    {
+      "epoch": 1.417620803415486,
+      "grad_norm": 1.6446340356434774,
+      "learning_rate": 2.645856317736388e-05,
+      "loss": 0.5324,
+      "step": 14610
+    },
+    {
+      "epoch": 1.4185911119736077,
+      "grad_norm": 1.9264081435604268,
+      "learning_rate": 2.6442338892854827e-05,
+      "loss": 0.5297,
+      "step": 14620
+    },
+    {
+      "epoch": 1.419561420531729,
+      "grad_norm": 1.5948112951701827,
+      "learning_rate": 2.6426114608345777e-05,
+      "loss": 0.5291,
+      "step": 14630
+    },
+    {
+      "epoch": 1.4205317290898507,
+      "grad_norm": 1.8654562276618851,
+      "learning_rate": 2.6409890323836724e-05,
+      "loss": 0.5588,
+      "step": 14640
+    },
+    {
+      "epoch": 1.4215020376479721,
+      "grad_norm": 1.8410472433514884,
+      "learning_rate": 2.639366603932767e-05,
+      "loss": 0.5167,
+      "step": 14650
+    },
+    {
+      "epoch": 1.4224723462060935,
+      "grad_norm": 1.9663683443045321,
+      "learning_rate": 2.6377441754818616e-05,
+      "loss": 0.5166,
+      "step": 14660
+    },
+    {
+      "epoch": 1.423442654764215,
+      "grad_norm": 1.77295818801796,
+      "learning_rate": 2.6361217470309563e-05,
+      "loss": 0.5557,
+      "step": 14670
+    },
+    {
+      "epoch": 1.4244129633223366,
+      "grad_norm": 1.542816761347417,
+      "learning_rate": 2.634499318580051e-05,
+      "loss": 0.5147,
+      "step": 14680
+    },
+    {
+      "epoch": 1.425383271880458,
+      "grad_norm": 2.13234320369828,
+      "learning_rate": 2.6328768901291456e-05,
+      "loss": 0.5421,
+      "step": 14690
+    },
+    {
+      "epoch": 1.4263535804385794,
+      "grad_norm": 1.4543036251195376,
+      "learning_rate": 2.6312544616782402e-05,
+      "loss": 0.488,
+      "step": 14700
+    },
+    {
+      "epoch": 1.427323888996701,
+      "grad_norm": 1.5267685947307574,
+      "learning_rate": 2.6296320332273348e-05,
+      "loss": 0.5352,
+      "step": 14710
+    },
+    {
+      "epoch": 1.4282941975548225,
+      "grad_norm": 2.05420183759504,
+      "learning_rate": 2.6280096047764298e-05,
+      "loss": 0.5489,
+      "step": 14720
+    },
+    {
+      "epoch": 1.4292645061129439,
+      "grad_norm": 1.5141763719248076,
+      "learning_rate": 2.6263871763255244e-05,
+      "loss": 0.5627,
+      "step": 14730
+    },
+    {
+      "epoch": 1.4302348146710653,
+      "grad_norm": 1.7887916433402153,
+      "learning_rate": 2.624764747874619e-05,
+      "loss": 0.5129,
+      "step": 14740
+    },
+    {
+      "epoch": 1.431205123229187,
+      "grad_norm": 1.9817836633125971,
+      "learning_rate": 2.6231423194237137e-05,
+      "loss": 0.5129,
+      "step": 14750
+    },
+    {
+      "epoch": 1.4321754317873083,
+      "grad_norm": 2.358539950484514,
+      "learning_rate": 2.6215198909728084e-05,
+      "loss": 0.482,
+      "step": 14760
+    },
+    {
+      "epoch": 1.43314574034543,
+      "grad_norm": 1.7444853271955691,
+      "learning_rate": 2.619897462521903e-05,
+      "loss": 0.4948,
+      "step": 14770
+    },
+    {
+      "epoch": 1.4341160489035514,
+      "grad_norm": 1.573301988778271,
+      "learning_rate": 2.6182750340709976e-05,
+      "loss": 0.4998,
+      "step": 14780
+    },
+    {
+      "epoch": 1.4350863574616728,
+      "grad_norm": 1.878570997397064,
+      "learning_rate": 2.6166526056200923e-05,
+      "loss": 0.489,
+      "step": 14790
+    },
+    {
+      "epoch": 1.4360566660197942,
+      "grad_norm": 2.0345633471458444,
+      "learning_rate": 2.615030177169187e-05,
+      "loss": 0.5395,
+      "step": 14800
+    },
+    {
+      "epoch": 1.4370269745779158,
+      "grad_norm": 1.7616258348677174,
+      "learning_rate": 2.613407748718282e-05,
+      "loss": 0.5575,
+      "step": 14810
+    },
+    {
+      "epoch": 1.4379972831360373,
+      "grad_norm": 2.202827489308336,
+      "learning_rate": 2.6117853202673765e-05,
+      "loss": 0.5378,
+      "step": 14820
+    },
+    {
+      "epoch": 1.4389675916941587,
+      "grad_norm": 2.0172801246892496,
+      "learning_rate": 2.610162891816471e-05,
+      "loss": 0.565,
+      "step": 14830
+    },
+    {
+      "epoch": 1.4399379002522803,
+      "grad_norm": 1.460733632748616,
+      "learning_rate": 2.6085404633655658e-05,
+      "loss": 0.4866,
+      "step": 14840
+    },
+    {
+      "epoch": 1.4409082088104017,
+      "grad_norm": 1.5702235924410088,
+      "learning_rate": 2.6069180349146604e-05,
+      "loss": 0.4814,
+      "step": 14850
+    },
+    {
+      "epoch": 1.4418785173685231,
+      "grad_norm": 2.3366982495878013,
+      "learning_rate": 2.605295606463755e-05,
+      "loss": 0.5204,
+      "step": 14860
+    },
+    {
+      "epoch": 1.4428488259266445,
+      "grad_norm": 2.0441313559069805,
+      "learning_rate": 2.6036731780128497e-05,
+      "loss": 0.4861,
+      "step": 14870
+    },
+    {
+      "epoch": 1.4438191344847662,
+      "grad_norm": 2.2095151787420417,
+      "learning_rate": 2.6020507495619444e-05,
+      "loss": 0.5376,
+      "step": 14880
+    },
+    {
+      "epoch": 1.4447894430428876,
+      "grad_norm": 1.9134281253559753,
+      "learning_rate": 2.6004283211110393e-05,
+      "loss": 0.5102,
+      "step": 14890
+    },
+    {
+      "epoch": 1.4457597516010092,
+      "grad_norm": 1.7832517565992747,
+      "learning_rate": 2.598805892660134e-05,
+      "loss": 0.5303,
+      "step": 14900
+    },
+    {
+      "epoch": 1.4467300601591306,
+      "grad_norm": 1.7558924047130664,
+      "learning_rate": 2.5971834642092286e-05,
+      "loss": 0.4994,
+      "step": 14910
+    },
+    {
+      "epoch": 1.447700368717252,
+      "grad_norm": 1.7883915957856114,
+      "learning_rate": 2.5955610357583233e-05,
+      "loss": 0.4677,
+      "step": 14920
+    },
+    {
+      "epoch": 1.4486706772753735,
+      "grad_norm": 2.077257189743679,
+      "learning_rate": 2.593938607307418e-05,
+      "loss": 0.515,
+      "step": 14930
+    },
+    {
+      "epoch": 1.449640985833495,
+      "grad_norm": 1.9045010104323963,
+      "learning_rate": 2.5923161788565125e-05,
+      "loss": 0.5024,
+      "step": 14940
+    },
+    {
+      "epoch": 1.4506112943916165,
+      "grad_norm": 2.106289689638874,
+      "learning_rate": 2.590693750405607e-05,
+      "loss": 0.5242,
+      "step": 14950
+    },
+    {
+      "epoch": 1.4515816029497381,
+      "grad_norm": 1.9221754156642648,
+      "learning_rate": 2.5890713219547018e-05,
+      "loss": 0.4712,
+      "step": 14960
+    },
+    {
+      "epoch": 1.4525519115078596,
+      "grad_norm": 1.8046920974227167,
+      "learning_rate": 2.5874488935037964e-05,
+      "loss": 0.5017,
+      "step": 14970
+    },
+    {
+      "epoch": 1.453522220065981,
+      "grad_norm": 1.855646189779827,
+      "learning_rate": 2.5858264650528914e-05,
+      "loss": 0.5146,
+      "step": 14980
+    },
+    {
+      "epoch": 1.4544925286241024,
+      "grad_norm": 1.8308672850602437,
+      "learning_rate": 2.584204036601986e-05,
+      "loss": 0.5293,
+      "step": 14990
+    },
+    {
+      "epoch": 1.455462837182224,
+      "grad_norm": 1.6235231403062087,
+      "learning_rate": 2.5825816081510807e-05,
+      "loss": 0.4968,
+      "step": 15000
+    },
+    {
+      "epoch": 1.455462837182224,
+      "eval_loss": 0.6395026445388794,
+      "eval_runtime": 2472.4459,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 15000
+    },
+    {
+      "epoch": 1.4564331457403454,
+      "grad_norm": 1.9674866651938567,
+      "learning_rate": 2.5809591797001753e-05,
+      "loss": 0.5402,
+      "step": 15010
+    },
+    {
+      "epoch": 1.4574034542984668,
+      "grad_norm": 1.7667679072452773,
+      "learning_rate": 2.57933675124927e-05,
+      "loss": 0.5598,
+      "step": 15020
+    },
+    {
+      "epoch": 1.4583737628565885,
+      "grad_norm": 1.6236668551346622,
+      "learning_rate": 2.5777143227983646e-05,
+      "loss": 0.4819,
+      "step": 15030
+    },
+    {
+      "epoch": 1.45934407141471,
+      "grad_norm": 1.5162521396018838,
+      "learning_rate": 2.5760918943474592e-05,
+      "loss": 0.5609,
+      "step": 15040
+    },
+    {
+      "epoch": 1.4603143799728313,
+      "grad_norm": 2.248488105603888,
+      "learning_rate": 2.574469465896554e-05,
+      "loss": 0.4987,
+      "step": 15050
+    },
+    {
+      "epoch": 1.4612846885309527,
+      "grad_norm": 1.5876501247710053,
+      "learning_rate": 2.572847037445649e-05,
+      "loss": 0.4794,
+      "step": 15060
+    },
+    {
+      "epoch": 1.4622549970890744,
+      "grad_norm": 1.9484672186526921,
+      "learning_rate": 2.5712246089947435e-05,
+      "loss": 0.5115,
+      "step": 15070
+    },
+    {
+      "epoch": 1.4632253056471958,
+      "grad_norm": 1.9931906967691606,
+      "learning_rate": 2.569602180543838e-05,
+      "loss": 0.4848,
+      "step": 15080
+    },
+    {
+      "epoch": 1.4641956142053174,
+      "grad_norm": 1.9798564517490436,
+      "learning_rate": 2.5679797520929328e-05,
+      "loss": 0.5136,
+      "step": 15090
+    },
+    {
+      "epoch": 1.4651659227634388,
+      "grad_norm": 1.7543081259876927,
+      "learning_rate": 2.5663573236420274e-05,
+      "loss": 0.5179,
+      "step": 15100
+    },
+    {
+      "epoch": 1.4661362313215602,
+      "grad_norm": 1.797610941181892,
+      "learning_rate": 2.564734895191122e-05,
+      "loss": 0.5134,
+      "step": 15110
+    },
+    {
+      "epoch": 1.4671065398796816,
+      "grad_norm": 1.6254067495518276,
+      "learning_rate": 2.5631124667402167e-05,
+      "loss": 0.497,
+      "step": 15120
+    },
+    {
+      "epoch": 1.4680768484378033,
+      "grad_norm": 2.0916697354749743,
+      "learning_rate": 2.5614900382893113e-05,
+      "loss": 0.5035,
+      "step": 15130
+    },
+    {
+      "epoch": 1.4690471569959247,
+      "grad_norm": 1.7408478320862355,
+      "learning_rate": 2.559867609838406e-05,
+      "loss": 0.5165,
+      "step": 15140
+    },
+    {
+      "epoch": 1.470017465554046,
+      "grad_norm": 1.6971971097300078,
+      "learning_rate": 2.558245181387501e-05,
+      "loss": 0.5416,
+      "step": 15150
+    },
+    {
+      "epoch": 1.4709877741121677,
+      "grad_norm": 1.7615798102754638,
+      "learning_rate": 2.5566227529365956e-05,
+      "loss": 0.5028,
+      "step": 15160
+    },
+    {
+      "epoch": 1.4719580826702892,
+      "grad_norm": 2.126034120697344,
+      "learning_rate": 2.5550003244856902e-05,
+      "loss": 0.4983,
+      "step": 15170
+    },
+    {
+      "epoch": 1.4729283912284106,
+      "grad_norm": 1.7930301180931063,
+      "learning_rate": 2.553377896034785e-05,
+      "loss": 0.4836,
+      "step": 15180
+    },
+    {
+      "epoch": 1.473898699786532,
+      "grad_norm": 1.945479896374108,
+      "learning_rate": 2.5517554675838795e-05,
+      "loss": 0.4874,
+      "step": 15190
+    },
+    {
+      "epoch": 1.4748690083446536,
+      "grad_norm": 1.659537976782885,
+      "learning_rate": 2.550133039132974e-05,
+      "loss": 0.485,
+      "step": 15200
+    },
+    {
+      "epoch": 1.475839316902775,
+      "grad_norm": 1.7852824594767274,
+      "learning_rate": 2.5485106106820688e-05,
+      "loss": 0.5314,
+      "step": 15210
+    },
+    {
+      "epoch": 1.4768096254608967,
+      "grad_norm": 1.7757412059616349,
+      "learning_rate": 2.5468881822311634e-05,
+      "loss": 0.4906,
+      "step": 15220
+    },
+    {
+      "epoch": 1.477779934019018,
+      "grad_norm": 1.7940563485432668,
+      "learning_rate": 2.5452657537802584e-05,
+      "loss": 0.4973,
+      "step": 15230
+    },
+    {
+      "epoch": 1.4787502425771395,
+      "grad_norm": 1.8223131394278327,
+      "learning_rate": 2.543643325329353e-05,
+      "loss": 0.5122,
+      "step": 15240
+    },
+    {
+      "epoch": 1.479720551135261,
+      "grad_norm": 1.9217304666232693,
+      "learning_rate": 2.5420208968784477e-05,
+      "loss": 0.5497,
+      "step": 15250
+    },
+    {
+      "epoch": 1.4806908596933825,
+      "grad_norm": 1.6084194486566938,
+      "learning_rate": 2.5403984684275423e-05,
+      "loss": 0.4971,
+      "step": 15260
+    },
+    {
+      "epoch": 1.481661168251504,
+      "grad_norm": 2.1056710345080827,
+      "learning_rate": 2.538776039976637e-05,
+      "loss": 0.5379,
+      "step": 15270
+    },
+    {
+      "epoch": 1.4826314768096256,
+      "grad_norm": 2.2545586744739015,
+      "learning_rate": 2.5371536115257316e-05,
+      "loss": 0.4239,
+      "step": 15280
+    },
+    {
+      "epoch": 1.483601785367747,
+      "grad_norm": 2.0015642103213063,
+      "learning_rate": 2.5355311830748262e-05,
+      "loss": 0.5283,
+      "step": 15290
+    },
+    {
+      "epoch": 1.4845720939258684,
+      "grad_norm": 2.0206628423435387,
+      "learning_rate": 2.533908754623921e-05,
+      "loss": 0.4533,
+      "step": 15300
+    },
+    {
+      "epoch": 1.4855424024839898,
+      "grad_norm": 2.0392064010865263,
+      "learning_rate": 2.5322863261730155e-05,
+      "loss": 0.516,
+      "step": 15310
+    },
+    {
+      "epoch": 1.4865127110421115,
+      "grad_norm": 1.6397938987534753,
+      "learning_rate": 2.5306638977221105e-05,
+      "loss": 0.4915,
+      "step": 15320
+    },
+    {
+      "epoch": 1.4874830196002329,
+      "grad_norm": 2.0519457829568615,
+      "learning_rate": 2.529041469271205e-05,
+      "loss": 0.4938,
+      "step": 15330
+    },
+    {
+      "epoch": 1.4884533281583543,
+      "grad_norm": 1.883771979267065,
+      "learning_rate": 2.5274190408202998e-05,
+      "loss": 0.5454,
+      "step": 15340
+    },
+    {
+      "epoch": 1.489423636716476,
+      "grad_norm": 1.8963017429804823,
+      "learning_rate": 2.5257966123693944e-05,
+      "loss": 0.4733,
+      "step": 15350
+    },
+    {
+      "epoch": 1.4903939452745973,
+      "grad_norm": 2.005144587549119,
+      "learning_rate": 2.5241741839184897e-05,
+      "loss": 0.4967,
+      "step": 15360
+    },
+    {
+      "epoch": 1.4913642538327188,
+      "grad_norm": 2.151826141060965,
+      "learning_rate": 2.5225517554675843e-05,
+      "loss": 0.5466,
+      "step": 15370
+    },
+    {
+      "epoch": 1.4923345623908402,
+      "grad_norm": 1.6751197267270117,
+      "learning_rate": 2.520929327016679e-05,
+      "loss": 0.5456,
+      "step": 15380
+    },
+    {
+      "epoch": 1.4933048709489618,
+      "grad_norm": 1.9009493553059222,
+      "learning_rate": 2.5193068985657736e-05,
+      "loss": 0.483,
+      "step": 15390
+    },
+    {
+      "epoch": 1.4942751795070832,
+      "grad_norm": 2.1457921969425757,
+      "learning_rate": 2.5176844701148683e-05,
+      "loss": 0.5458,
+      "step": 15400
+    },
+    {
+      "epoch": 1.4952454880652049,
+      "grad_norm": 1.9369152546010822,
+      "learning_rate": 2.516062041663963e-05,
+      "loss": 0.5477,
+      "step": 15410
+    },
+    {
+      "epoch": 1.4962157966233263,
+      "grad_norm": 1.9226350314538543,
+      "learning_rate": 2.5144396132130575e-05,
+      "loss": 0.5469,
+      "step": 15420
+    },
+    {
+      "epoch": 1.4971861051814477,
+      "grad_norm": 1.768970891771466,
+      "learning_rate": 2.5128171847621522e-05,
+      "loss": 0.4789,
+      "step": 15430
+    },
+    {
+      "epoch": 1.498156413739569,
+      "grad_norm": 1.6324753131013463,
+      "learning_rate": 2.511194756311247e-05,
+      "loss": 0.4984,
+      "step": 15440
+    },
+    {
+      "epoch": 1.4991267222976907,
+      "grad_norm": 1.5500314116241656,
+      "learning_rate": 2.5095723278603418e-05,
+      "loss": 0.5313,
+      "step": 15450
+    },
+    {
+      "epoch": 1.5000970308558121,
+      "grad_norm": 1.8222773979858036,
+      "learning_rate": 2.5079498994094364e-05,
+      "loss": 0.4935,
+      "step": 15460
+    },
+    {
+      "epoch": 1.5010673394139338,
+      "grad_norm": 1.6457715681286798,
+      "learning_rate": 2.506327470958531e-05,
+      "loss": 0.4739,
+      "step": 15470
+    },
+    {
+      "epoch": 1.5020376479720552,
+      "grad_norm": 1.6810719922194195,
+      "learning_rate": 2.5047050425076257e-05,
+      "loss": 0.478,
+      "step": 15480
+    },
+    {
+      "epoch": 1.5030079565301766,
+      "grad_norm": 1.9891968136548206,
+      "learning_rate": 2.5030826140567203e-05,
+      "loss": 0.5365,
+      "step": 15490
+    },
+    {
+      "epoch": 1.503978265088298,
+      "grad_norm": 1.836679974736217,
+      "learning_rate": 2.501460185605815e-05,
+      "loss": 0.5497,
+      "step": 15500
+    },
+    {
+      "epoch": 1.503978265088298,
+      "eval_loss": 0.6403182148933411,
+      "eval_runtime": 2470.3613,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 15500
+    },
+    {
+      "epoch": 1.5049485736464194,
+      "grad_norm": 1.9808212956686209,
+      "learning_rate": 2.4998377571549096e-05,
+      "loss": 0.5195,
+      "step": 15510
+    },
+    {
+      "epoch": 1.505918882204541,
+      "grad_norm": 1.696799638403568,
+      "learning_rate": 2.4982153287040043e-05,
+      "loss": 0.481,
+      "step": 15520
+    },
+    {
+      "epoch": 1.5068891907626625,
+      "grad_norm": 1.559487018818546,
+      "learning_rate": 2.496592900253099e-05,
+      "loss": 0.4849,
+      "step": 15530
+    },
+    {
+      "epoch": 1.5078594993207841,
+      "grad_norm": 2.0106819741019635,
+      "learning_rate": 2.4949704718021935e-05,
+      "loss": 0.5068,
+      "step": 15540
+    },
+    {
+      "epoch": 1.5088298078789055,
+      "grad_norm": 1.554558550656832,
+      "learning_rate": 2.4933480433512885e-05,
+      "loss": 0.5542,
+      "step": 15550
+    },
+    {
+      "epoch": 1.509800116437027,
+      "grad_norm": 2.1570334538409597,
+      "learning_rate": 2.491725614900383e-05,
+      "loss": 0.5537,
+      "step": 15560
+    },
+    {
+      "epoch": 1.5107704249951484,
+      "grad_norm": 2.082418119992958,
+      "learning_rate": 2.4901031864494778e-05,
+      "loss": 0.4495,
+      "step": 15570
+    },
+    {
+      "epoch": 1.51174073355327,
+      "grad_norm": 1.7903747855746062,
+      "learning_rate": 2.4884807579985724e-05,
+      "loss": 0.4755,
+      "step": 15580
+    },
+    {
+      "epoch": 1.5127110421113914,
+      "grad_norm": 1.8907675365997436,
+      "learning_rate": 2.486858329547667e-05,
+      "loss": 0.5377,
+      "step": 15590
+    },
+    {
+      "epoch": 1.513681350669513,
+      "grad_norm": 1.6823616833089232,
+      "learning_rate": 2.4852359010967617e-05,
+      "loss": 0.4747,
+      "step": 15600
+    },
+    {
+      "epoch": 1.5146516592276345,
+      "grad_norm": 1.8263050453987495,
+      "learning_rate": 2.4836134726458567e-05,
+      "loss": 0.4962,
+      "step": 15610
+    },
+    {
+      "epoch": 1.5156219677857559,
+      "grad_norm": 1.855857638521949,
+      "learning_rate": 2.4819910441949513e-05,
+      "loss": 0.5306,
+      "step": 15620
+    },
+    {
+      "epoch": 1.5165922763438773,
+      "grad_norm": 1.7196056977172254,
+      "learning_rate": 2.480368615744046e-05,
+      "loss": 0.507,
+      "step": 15630
+    },
+    {
+      "epoch": 1.5175625849019987,
+      "grad_norm": 2.2274627334622354,
+      "learning_rate": 2.4787461872931406e-05,
+      "loss": 0.4741,
+      "step": 15640
+    },
+    {
+      "epoch": 1.5185328934601203,
+      "grad_norm": 1.7506693514893716,
+      "learning_rate": 2.4771237588422352e-05,
+      "loss": 0.4512,
+      "step": 15650
+    },
+    {
+      "epoch": 1.519503202018242,
+      "grad_norm": 1.5102515980183955,
+      "learning_rate": 2.47550133039133e-05,
+      "loss": 0.5656,
+      "step": 15660
+    },
+    {
+      "epoch": 1.5204735105763634,
+      "grad_norm": 1.7638756809733325,
+      "learning_rate": 2.4738789019404245e-05,
+      "loss": 0.5209,
+      "step": 15670
+    },
+    {
+      "epoch": 1.5214438191344848,
+      "grad_norm": 1.67804925991046,
+      "learning_rate": 2.472256473489519e-05,
+      "loss": 0.5442,
+      "step": 15680
+    },
+    {
+      "epoch": 1.5224141276926062,
+      "grad_norm": 1.6503421070503612,
+      "learning_rate": 2.470634045038614e-05,
+      "loss": 0.5516,
+      "step": 15690
+    },
+    {
+      "epoch": 1.5233844362507276,
+      "grad_norm": 1.7388522657721959,
+      "learning_rate": 2.4690116165877088e-05,
+      "loss": 0.4894,
+      "step": 15700
+    },
+    {
+      "epoch": 1.5243547448088492,
+      "grad_norm": 1.955414988998346,
+      "learning_rate": 2.4673891881368034e-05,
+      "loss": 0.5493,
+      "step": 15710
+    },
+    {
+      "epoch": 1.5253250533669707,
+      "grad_norm": 2.1089020897683324,
+      "learning_rate": 2.465766759685898e-05,
+      "loss": 0.5173,
+      "step": 15720
+    },
+    {
+      "epoch": 1.5262953619250923,
+      "grad_norm": 2.1627868553616416,
+      "learning_rate": 2.4641443312349927e-05,
+      "loss": 0.5165,
+      "step": 15730
+    },
+    {
+      "epoch": 1.5272656704832137,
+      "grad_norm": 1.9351405181934673,
+      "learning_rate": 2.4625219027840873e-05,
+      "loss": 0.4873,
+      "step": 15740
+    },
+    {
+      "epoch": 1.5282359790413351,
+      "grad_norm": 1.8488349390401109,
+      "learning_rate": 2.460899474333182e-05,
+      "loss": 0.5121,
+      "step": 15750
+    },
+    {
+      "epoch": 1.5292062875994565,
+      "grad_norm": 1.9814673588646732,
+      "learning_rate": 2.4592770458822766e-05,
+      "loss": 0.4942,
+      "step": 15760
+    },
+    {
+      "epoch": 1.530176596157578,
+      "grad_norm": 1.8985936905786858,
+      "learning_rate": 2.4576546174313712e-05,
+      "loss": 0.5227,
+      "step": 15770
+    },
+    {
+      "epoch": 1.5311469047156996,
+      "grad_norm": 2.144939705295627,
+      "learning_rate": 2.4560321889804662e-05,
+      "loss": 0.5584,
+      "step": 15780
+    },
+    {
+      "epoch": 1.5321172132738212,
+      "grad_norm": 2.196125539672085,
+      "learning_rate": 2.454409760529561e-05,
+      "loss": 0.4592,
+      "step": 15790
+    },
+    {
+      "epoch": 1.5330875218319426,
+      "grad_norm": 1.5896290702039733,
+      "learning_rate": 2.4527873320786555e-05,
+      "loss": 0.5183,
+      "step": 15800
+    },
+    {
+      "epoch": 1.534057830390064,
+      "grad_norm": 1.3926787086096604,
+      "learning_rate": 2.45116490362775e-05,
+      "loss": 0.4929,
+      "step": 15810
+    },
+    {
+      "epoch": 1.5350281389481855,
+      "grad_norm": 1.8947619224673768,
+      "learning_rate": 2.4495424751768448e-05,
+      "loss": 0.487,
+      "step": 15820
+    },
+    {
+      "epoch": 1.5359984475063069,
+      "grad_norm": 1.9366715293465946,
+      "learning_rate": 2.4479200467259394e-05,
+      "loss": 0.5126,
+      "step": 15830
+    },
+    {
+      "epoch": 1.5369687560644285,
+      "grad_norm": 1.8854258468503662,
+      "learning_rate": 2.446297618275034e-05,
+      "loss": 0.4732,
+      "step": 15840
+    },
+    {
+      "epoch": 1.53793906462255,
+      "grad_norm": 1.7897154496692322,
+      "learning_rate": 2.4446751898241287e-05,
+      "loss": 0.4869,
+      "step": 15850
+    },
+    {
+      "epoch": 1.5389093731806716,
+      "grad_norm": 1.9691769621461568,
+      "learning_rate": 2.4430527613732233e-05,
+      "loss": 0.5204,
+      "step": 15860
+    },
+    {
+      "epoch": 1.539879681738793,
+      "grad_norm": 1.5159729039901195,
+      "learning_rate": 2.4414303329223183e-05,
+      "loss": 0.5067,
+      "step": 15870
+    },
+    {
+      "epoch": 1.5408499902969144,
+      "grad_norm": 2.0085743327171364,
+      "learning_rate": 2.439807904471413e-05,
+      "loss": 0.5855,
+      "step": 15880
+    },
+    {
+      "epoch": 1.5418202988550358,
+      "grad_norm": 1.808826518921202,
+      "learning_rate": 2.4381854760205076e-05,
+      "loss": 0.511,
+      "step": 15890
+    },
+    {
+      "epoch": 1.5427906074131574,
+      "grad_norm": 2.144838816182226,
+      "learning_rate": 2.4365630475696022e-05,
+      "loss": 0.5151,
+      "step": 15900
+    },
+    {
+      "epoch": 1.5437609159712788,
+      "grad_norm": 1.9134282869959454,
+      "learning_rate": 2.434940619118697e-05,
+      "loss": 0.547,
+      "step": 15910
+    },
+    {
+      "epoch": 1.5447312245294005,
+      "grad_norm": 1.4718585956659067,
+      "learning_rate": 2.4333181906677915e-05,
+      "loss": 0.5035,
+      "step": 15920
+    },
+    {
+      "epoch": 1.545701533087522,
+      "grad_norm": 1.9846910792449015,
+      "learning_rate": 2.431695762216886e-05,
+      "loss": 0.4603,
+      "step": 15930
+    },
+    {
+      "epoch": 1.5466718416456433,
+      "grad_norm": 2.237689406521008,
+      "learning_rate": 2.430073333765981e-05,
+      "loss": 0.4835,
+      "step": 15940
+    },
+    {
+      "epoch": 1.5476421502037647,
+      "grad_norm": 1.8521877806585876,
+      "learning_rate": 2.4284509053150757e-05,
+      "loss": 0.486,
+      "step": 15950
+    },
+    {
+      "epoch": 1.5486124587618861,
+      "grad_norm": 1.6410601205984716,
+      "learning_rate": 2.4268284768641704e-05,
+      "loss": 0.5362,
+      "step": 15960
+    },
+    {
+      "epoch": 1.5495827673200078,
+      "grad_norm": 2.0223928767474524,
+      "learning_rate": 2.4252060484132653e-05,
+      "loss": 0.5954,
+      "step": 15970
+    },
+    {
+      "epoch": 1.5505530758781294,
+      "grad_norm": 2.531432809831189,
+      "learning_rate": 2.42358361996236e-05,
+      "loss": 0.5618,
+      "step": 15980
+    },
+    {
+      "epoch": 1.5515233844362508,
+      "grad_norm": 2.0740258540735788,
+      "learning_rate": 2.4219611915114546e-05,
+      "loss": 0.4733,
+      "step": 15990
+    },
+    {
+      "epoch": 1.5524936929943722,
+      "grad_norm": 1.7587497399634056,
+      "learning_rate": 2.4203387630605493e-05,
+      "loss": 0.5099,
+      "step": 16000
+    },
+    {
+      "epoch": 1.5524936929943722,
+      "eval_loss": 0.635991096496582,
+      "eval_runtime": 2468.0382,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 16000
+    },
+    {
+      "epoch": 1.5534640015524936,
+      "grad_norm": 1.670336846718715,
+      "learning_rate": 2.418716334609644e-05,
+      "loss": 0.4553,
+      "step": 16010
+    },
+    {
+      "epoch": 1.554434310110615,
+      "grad_norm": 2.0177796015243237,
+      "learning_rate": 2.4170939061587385e-05,
+      "loss": 0.5115,
+      "step": 16020
+    },
+    {
+      "epoch": 1.5554046186687367,
+      "grad_norm": 1.8858883624862002,
+      "learning_rate": 2.4154714777078332e-05,
+      "loss": 0.5172,
+      "step": 16030
+    },
+    {
+      "epoch": 1.556374927226858,
+      "grad_norm": 2.0359679981766647,
+      "learning_rate": 2.4138490492569278e-05,
+      "loss": 0.4589,
+      "step": 16040
+    },
+    {
+      "epoch": 1.5573452357849797,
+      "grad_norm": 1.2444792958823963,
+      "learning_rate": 2.4122266208060228e-05,
+      "loss": 0.5134,
+      "step": 16050
+    },
+    {
+      "epoch": 1.5583155443431012,
+      "grad_norm": 1.5062153000531946,
+      "learning_rate": 2.4106041923551174e-05,
+      "loss": 0.4869,
+      "step": 16060
+    },
+    {
+      "epoch": 1.5592858529012226,
+      "grad_norm": 2.2139374575219364,
+      "learning_rate": 2.408981763904212e-05,
+      "loss": 0.4811,
+      "step": 16070
+    },
+    {
+      "epoch": 1.560256161459344,
+      "grad_norm": 1.7542756535220294,
+      "learning_rate": 2.4073593354533067e-05,
+      "loss": 0.4895,
+      "step": 16080
+    },
+    {
+      "epoch": 1.5612264700174654,
+      "grad_norm": 1.9663056975292839,
+      "learning_rate": 2.4057369070024013e-05,
+      "loss": 0.5046,
+      "step": 16090
+    },
+    {
+      "epoch": 1.562196778575587,
+      "grad_norm": 1.760446721510282,
+      "learning_rate": 2.404114478551496e-05,
+      "loss": 0.495,
+      "step": 16100
+    },
+    {
+      "epoch": 1.5631670871337087,
+      "grad_norm": 1.6132518828061082,
+      "learning_rate": 2.4024920501005906e-05,
+      "loss": 0.4985,
+      "step": 16110
+    },
+    {
+      "epoch": 1.56413739569183,
+      "grad_norm": 1.7640461059330637,
+      "learning_rate": 2.4008696216496853e-05,
+      "loss": 0.5186,
+      "step": 16120
+    },
+    {
+      "epoch": 1.5651077042499515,
+      "grad_norm": 1.769814216957158,
+      "learning_rate": 2.39924719319878e-05,
+      "loss": 0.5413,
+      "step": 16130
+    },
+    {
+      "epoch": 1.566078012808073,
+      "grad_norm": 2.293241964432802,
+      "learning_rate": 2.397624764747875e-05,
+      "loss": 0.4809,
+      "step": 16140
+    },
+    {
+      "epoch": 1.5670483213661943,
+      "grad_norm": 1.8759559371438301,
+      "learning_rate": 2.3960023362969695e-05,
+      "loss": 0.5262,
+      "step": 16150
+    },
+    {
+      "epoch": 1.568018629924316,
+      "grad_norm": 2.3087090615314114,
+      "learning_rate": 2.394379907846064e-05,
+      "loss": 0.4841,
+      "step": 16160
+    },
+    {
+      "epoch": 1.5689889384824374,
+      "grad_norm": 1.6108810314362396,
+      "learning_rate": 2.3927574793951588e-05,
+      "loss": 0.5188,
+      "step": 16170
+    },
+    {
+      "epoch": 1.569959247040559,
+      "grad_norm": 2.241869048530712,
+      "learning_rate": 2.3911350509442534e-05,
+      "loss": 0.5005,
+      "step": 16180
+    },
+    {
+      "epoch": 1.5709295555986804,
+      "grad_norm": 2.1127428890150215,
+      "learning_rate": 2.389512622493348e-05,
+      "loss": 0.5063,
+      "step": 16190
+    },
+    {
+      "epoch": 1.5718998641568018,
+      "grad_norm": 1.7046555426171013,
+      "learning_rate": 2.3878901940424427e-05,
+      "loss": 0.5154,
+      "step": 16200
+    },
+    {
+      "epoch": 1.5728701727149232,
+      "grad_norm": 1.6974188480153705,
+      "learning_rate": 2.3862677655915373e-05,
+      "loss": 0.4775,
+      "step": 16210
+    },
+    {
+      "epoch": 1.5738404812730449,
+      "grad_norm": 1.9773772891509165,
+      "learning_rate": 2.384645337140632e-05,
+      "loss": 0.489,
+      "step": 16220
+    },
+    {
+      "epoch": 1.5748107898311663,
+      "grad_norm": 2.3323506244488645,
+      "learning_rate": 2.383022908689727e-05,
+      "loss": 0.5252,
+      "step": 16230
+    },
+    {
+      "epoch": 1.575781098389288,
+      "grad_norm": 1.4580240462073564,
+      "learning_rate": 2.3814004802388216e-05,
+      "loss": 0.4809,
+      "step": 16240
+    },
+    {
+      "epoch": 1.5767514069474093,
+      "grad_norm": 1.9799244985970428,
+      "learning_rate": 2.3797780517879162e-05,
+      "loss": 0.4938,
+      "step": 16250
+    },
+    {
+      "epoch": 1.5777217155055308,
+      "grad_norm": 2.2967773637028315,
+      "learning_rate": 2.378155623337011e-05,
+      "loss": 0.48,
+      "step": 16260
+    },
+    {
+      "epoch": 1.5786920240636522,
+      "grad_norm": 1.6715177397086511,
+      "learning_rate": 2.3765331948861055e-05,
+      "loss": 0.5112,
+      "step": 16270
+    },
+    {
+      "epoch": 1.5796623326217736,
+      "grad_norm": 1.7543224867437088,
+      "learning_rate": 2.3749107664352e-05,
+      "loss": 0.5355,
+      "step": 16280
+    },
+    {
+      "epoch": 1.5806326411798952,
+      "grad_norm": 1.7059823329477148,
+      "learning_rate": 2.3732883379842948e-05,
+      "loss": 0.4885,
+      "step": 16290
+    },
+    {
+      "epoch": 1.5816029497380169,
+      "grad_norm": 1.808906067511985,
+      "learning_rate": 2.3716659095333894e-05,
+      "loss": 0.4883,
+      "step": 16300
+    },
+    {
+      "epoch": 1.5825732582961383,
+      "grad_norm": 1.456280506138187,
+      "learning_rate": 2.3700434810824844e-05,
+      "loss": 0.4986,
+      "step": 16310
+    },
+    {
+      "epoch": 1.5835435668542597,
+      "grad_norm": 1.7844142401171645,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.5726,
+      "step": 16320
+    },
+    {
+      "epoch": 1.584513875412381,
+      "grad_norm": 1.8511812344459693,
+      "learning_rate": 2.3667986241806737e-05,
+      "loss": 0.5418,
+      "step": 16330
+    },
+    {
+      "epoch": 1.5854841839705025,
+      "grad_norm": 1.6886781249735945,
+      "learning_rate": 2.3651761957297687e-05,
+      "loss": 0.4871,
+      "step": 16340
+    },
+    {
+      "epoch": 1.5864544925286241,
+      "grad_norm": 1.6386080299672316,
+      "learning_rate": 2.3635537672788633e-05,
+      "loss": 0.5502,
+      "step": 16350
+    },
+    {
+      "epoch": 1.5874248010867456,
+      "grad_norm": 1.6152930914041828,
+      "learning_rate": 2.361931338827958e-05,
+      "loss": 0.4602,
+      "step": 16360
+    },
+    {
+      "epoch": 1.5883951096448672,
+      "grad_norm": 1.9896290423806842,
+      "learning_rate": 2.3603089103770526e-05,
+      "loss": 0.5517,
+      "step": 16370
+    },
+    {
+      "epoch": 1.5893654182029886,
+      "grad_norm": 1.785358677833909,
+      "learning_rate": 2.3586864819261472e-05,
+      "loss": 0.5108,
+      "step": 16380
+    },
+    {
+      "epoch": 1.59033572676111,
+      "grad_norm": 1.8919452155428813,
+      "learning_rate": 2.357064053475242e-05,
+      "loss": 0.5586,
+      "step": 16390
+    },
+    {
+      "epoch": 1.5913060353192314,
+      "grad_norm": 2.1765937296152376,
+      "learning_rate": 2.3554416250243365e-05,
+      "loss": 0.4978,
+      "step": 16400
+    },
+    {
+      "epoch": 1.5922763438773528,
+      "grad_norm": 1.9463613536174769,
+      "learning_rate": 2.3538191965734315e-05,
+      "loss": 0.534,
+      "step": 16410
+    },
+    {
+      "epoch": 1.5932466524354745,
+      "grad_norm": 1.483428953982187,
+      "learning_rate": 2.352196768122526e-05,
+      "loss": 0.4481,
+      "step": 16420
+    },
+    {
+      "epoch": 1.594216960993596,
+      "grad_norm": 1.8946130151442728,
+      "learning_rate": 2.3505743396716207e-05,
+      "loss": 0.4499,
+      "step": 16430
+    },
+    {
+      "epoch": 1.5951872695517175,
+      "grad_norm": 1.496172632655443,
+      "learning_rate": 2.3489519112207154e-05,
+      "loss": 0.4719,
+      "step": 16440
+    },
+    {
+      "epoch": 1.596157578109839,
+      "grad_norm": 1.5582678654830617,
+      "learning_rate": 2.34732948276981e-05,
+      "loss": 0.5023,
+      "step": 16450
+    },
+    {
+      "epoch": 1.5971278866679604,
+      "grad_norm": 1.9101490466966036,
+      "learning_rate": 2.3457070543189047e-05,
+      "loss": 0.4442,
+      "step": 16460
+    },
+    {
+      "epoch": 1.5980981952260818,
+      "grad_norm": 2.5712850986928757,
+      "learning_rate": 2.3440846258679993e-05,
+      "loss": 0.5008,
+      "step": 16470
+    },
+    {
+      "epoch": 1.5990685037842034,
+      "grad_norm": 2.1177587573552805,
+      "learning_rate": 2.342462197417094e-05,
+      "loss": 0.5014,
+      "step": 16480
+    },
+    {
+      "epoch": 1.6000388123423248,
+      "grad_norm": 2.4990786716780455,
+      "learning_rate": 2.3408397689661886e-05,
+      "loss": 0.4583,
+      "step": 16490
+    },
+    {
+      "epoch": 1.6010091209004464,
+      "grad_norm": 1.91427095412827,
+      "learning_rate": 2.3392173405152835e-05,
+      "loss": 0.5037,
+      "step": 16500
+    },
+    {
+      "epoch": 1.6010091209004464,
+      "eval_loss": 0.6359612345695496,
+      "eval_runtime": 2467.4852,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 16500
+    },
+    {
+      "epoch": 1.6019794294585679,
+      "grad_norm": 1.8370649227416649,
+      "learning_rate": 2.3375949120643782e-05,
+      "loss": 0.5399,
+      "step": 16510
+    },
+    {
+      "epoch": 1.6029497380166893,
+      "grad_norm": 1.6932084838539212,
+      "learning_rate": 2.3359724836134728e-05,
+      "loss": 0.5212,
+      "step": 16520
+    },
+    {
+      "epoch": 1.6039200465748107,
+      "grad_norm": 1.68790954641985,
+      "learning_rate": 2.3343500551625675e-05,
+      "loss": 0.4837,
+      "step": 16530
+    },
+    {
+      "epoch": 1.6048903551329323,
+      "grad_norm": 1.5773573029174093,
+      "learning_rate": 2.332727626711662e-05,
+      "loss": 0.5125,
+      "step": 16540
+    },
+    {
+      "epoch": 1.6058606636910537,
+      "grad_norm": 1.7640635185794997,
+      "learning_rate": 2.3311051982607567e-05,
+      "loss": 0.5013,
+      "step": 16550
+    },
+    {
+      "epoch": 1.6068309722491754,
+      "grad_norm": 1.7477056146726457,
+      "learning_rate": 2.3294827698098514e-05,
+      "loss": 0.5095,
+      "step": 16560
+    },
+    {
+      "epoch": 1.6078012808072968,
+      "grad_norm": 1.0733428522302542,
+      "learning_rate": 2.327860341358946e-05,
+      "loss": 0.5,
+      "step": 16570
+    },
+    {
+      "epoch": 1.6087715893654182,
+      "grad_norm": 1.9564292094193194,
+      "learning_rate": 2.326237912908041e-05,
+      "loss": 0.5058,
+      "step": 16580
+    },
+    {
+      "epoch": 1.6097418979235396,
+      "grad_norm": 2.1045720941333466,
+      "learning_rate": 2.3246154844571356e-05,
+      "loss": 0.5101,
+      "step": 16590
+    },
+    {
+      "epoch": 1.610712206481661,
+      "grad_norm": 2.171187966701804,
+      "learning_rate": 2.3229930560062303e-05,
+      "loss": 0.5069,
+      "step": 16600
+    },
+    {
+      "epoch": 1.6116825150397827,
+      "grad_norm": 1.9610313890988438,
+      "learning_rate": 2.321370627555325e-05,
+      "loss": 0.4904,
+      "step": 16610
+    },
+    {
+      "epoch": 1.6126528235979043,
+      "grad_norm": 1.941252888287077,
+      "learning_rate": 2.3197481991044195e-05,
+      "loss": 0.4865,
+      "step": 16620
+    },
+    {
+      "epoch": 1.6136231321560257,
+      "grad_norm": 1.897218836962928,
+      "learning_rate": 2.3181257706535142e-05,
+      "loss": 0.4966,
+      "step": 16630
+    },
+    {
+      "epoch": 1.6145934407141471,
+      "grad_norm": 2.201820668610594,
+      "learning_rate": 2.3165033422026088e-05,
+      "loss": 0.4895,
+      "step": 16640
+    },
+    {
+      "epoch": 1.6155637492722685,
+      "grad_norm": 1.5720192215380684,
+      "learning_rate": 2.3148809137517035e-05,
+      "loss": 0.4918,
+      "step": 16650
+    },
+    {
+      "epoch": 1.61653405783039,
+      "grad_norm": 2.227704713223705,
+      "learning_rate": 2.313258485300798e-05,
+      "loss": 0.5251,
+      "step": 16660
+    },
+    {
+      "epoch": 1.6175043663885116,
+      "grad_norm": 2.0170271312798707,
+      "learning_rate": 2.311636056849893e-05,
+      "loss": 0.4965,
+      "step": 16670
+    },
+    {
+      "epoch": 1.618474674946633,
+      "grad_norm": 1.9688156858912216,
+      "learning_rate": 2.3100136283989877e-05,
+      "loss": 0.4685,
+      "step": 16680
+    },
+    {
+      "epoch": 1.6194449835047546,
+      "grad_norm": 2.0958524562959675,
+      "learning_rate": 2.3083911999480824e-05,
+      "loss": 0.4884,
+      "step": 16690
+    },
+    {
+      "epoch": 1.620415292062876,
+      "grad_norm": 2.0973339157843447,
+      "learning_rate": 2.306768771497177e-05,
+      "loss": 0.4394,
+      "step": 16700
+    },
+    {
+      "epoch": 1.6213856006209975,
+      "grad_norm": 1.5852203386480506,
+      "learning_rate": 2.3051463430462716e-05,
+      "loss": 0.5068,
+      "step": 16710
+    },
+    {
+      "epoch": 1.6223559091791189,
+      "grad_norm": 1.907182215766834,
+      "learning_rate": 2.3035239145953663e-05,
+      "loss": 0.461,
+      "step": 16720
+    },
+    {
+      "epoch": 1.6233262177372403,
+      "grad_norm": 1.9903511844679715,
+      "learning_rate": 2.3019014861444612e-05,
+      "loss": 0.4861,
+      "step": 16730
+    },
+    {
+      "epoch": 1.624296526295362,
+      "grad_norm": 1.8734845445192632,
+      "learning_rate": 2.300279057693556e-05,
+      "loss": 0.5203,
+      "step": 16740
+    },
+    {
+      "epoch": 1.6252668348534836,
+      "grad_norm": 1.4830887784438085,
+      "learning_rate": 2.2986566292426505e-05,
+      "loss": 0.5504,
+      "step": 16750
+    },
+    {
+      "epoch": 1.626237143411605,
+      "grad_norm": 1.9852817046238187,
+      "learning_rate": 2.297034200791745e-05,
+      "loss": 0.5727,
+      "step": 16760
+    },
+    {
+      "epoch": 1.6272074519697264,
+      "grad_norm": 1.9574779506866022,
+      "learning_rate": 2.29541177234084e-05,
+      "loss": 0.4932,
+      "step": 16770
+    },
+    {
+      "epoch": 1.6281777605278478,
+      "grad_norm": 1.8045639100087034,
+      "learning_rate": 2.2937893438899348e-05,
+      "loss": 0.4874,
+      "step": 16780
+    },
+    {
+      "epoch": 1.6291480690859692,
+      "grad_norm": 1.9467881794111201,
+      "learning_rate": 2.2921669154390294e-05,
+      "loss": 0.4231,
+      "step": 16790
+    },
+    {
+      "epoch": 1.6301183776440908,
+      "grad_norm": 2.115958262042766,
+      "learning_rate": 2.290544486988124e-05,
+      "loss": 0.5085,
+      "step": 16800
+    },
+    {
+      "epoch": 1.6310886862022123,
+      "grad_norm": 1.9242055352600025,
+      "learning_rate": 2.2889220585372187e-05,
+      "loss": 0.4961,
+      "step": 16810
+    },
+    {
+      "epoch": 1.632058994760334,
+      "grad_norm": 1.7473969264432643,
+      "learning_rate": 2.2872996300863133e-05,
+      "loss": 0.4896,
+      "step": 16820
+    },
+    {
+      "epoch": 1.6330293033184553,
+      "grad_norm": 1.9086620383723776,
+      "learning_rate": 2.285677201635408e-05,
+      "loss": 0.5793,
+      "step": 16830
+    },
+    {
+      "epoch": 1.6339996118765767,
+      "grad_norm": 1.8179786231819581,
+      "learning_rate": 2.2840547731845026e-05,
+      "loss": 0.47,
+      "step": 16840
+    },
+    {
+      "epoch": 1.6349699204346981,
+      "grad_norm": 2.3742959935176993,
+      "learning_rate": 2.2824323447335972e-05,
+      "loss": 0.4601,
+      "step": 16850
+    },
+    {
+      "epoch": 1.6359402289928198,
+      "grad_norm": 1.5831090068828182,
+      "learning_rate": 2.2808099162826922e-05,
+      "loss": 0.527,
+      "step": 16860
+    },
+    {
+      "epoch": 1.6369105375509412,
+      "grad_norm": 2.429466466556868,
+      "learning_rate": 2.279187487831787e-05,
+      "loss": 0.5041,
+      "step": 16870
+    },
+    {
+      "epoch": 1.6378808461090628,
+      "grad_norm": 2.0666827078950676,
+      "learning_rate": 2.2775650593808815e-05,
+      "loss": 0.5093,
+      "step": 16880
+    },
+    {
+      "epoch": 1.6388511546671842,
+      "grad_norm": 1.7376049636619024,
+      "learning_rate": 2.275942630929976e-05,
+      "loss": 0.4899,
+      "step": 16890
+    },
+    {
+      "epoch": 1.6398214632253056,
+      "grad_norm": 1.9743952968077716,
+      "learning_rate": 2.2743202024790708e-05,
+      "loss": 0.4483,
+      "step": 16900
+    },
+    {
+      "epoch": 1.640791771783427,
+      "grad_norm": 1.7020982614990035,
+      "learning_rate": 2.2726977740281654e-05,
+      "loss": 0.4746,
+      "step": 16910
+    },
+    {
+      "epoch": 1.6417620803415485,
+      "grad_norm": 1.5288620918579647,
+      "learning_rate": 2.27107534557726e-05,
+      "loss": 0.5386,
+      "step": 16920
+    },
+    {
+      "epoch": 1.64273238889967,
+      "grad_norm": 2.2418459177194725,
+      "learning_rate": 2.2694529171263547e-05,
+      "loss": 0.5221,
+      "step": 16930
+    },
+    {
+      "epoch": 1.6437026974577917,
+      "grad_norm": 1.7801442444217204,
+      "learning_rate": 2.2678304886754497e-05,
+      "loss": 0.502,
+      "step": 16940
+    },
+    {
+      "epoch": 1.6446730060159132,
+      "grad_norm": 2.750428317407271,
+      "learning_rate": 2.2662080602245443e-05,
+      "loss": 0.4947,
+      "step": 16950
+    },
+    {
+      "epoch": 1.6456433145740346,
+      "grad_norm": 1.5310655899660852,
+      "learning_rate": 2.264585631773639e-05,
+      "loss": 0.5228,
+      "step": 16960
+    },
+    {
+      "epoch": 1.646613623132156,
+      "grad_norm": 1.7325660800102027,
+      "learning_rate": 2.2629632033227336e-05,
+      "loss": 0.4724,
+      "step": 16970
+    },
+    {
+      "epoch": 1.6475839316902774,
+      "grad_norm": 3.028000821794418,
+      "learning_rate": 2.2613407748718282e-05,
+      "loss": 0.5444,
+      "step": 16980
+    },
+    {
+      "epoch": 1.648554240248399,
+      "grad_norm": 1.8847091529095503,
+      "learning_rate": 2.259718346420923e-05,
+      "loss": 0.5095,
+      "step": 16990
+    },
+    {
+      "epoch": 1.6495245488065204,
+      "grad_norm": 1.584479445600283,
+      "learning_rate": 2.2580959179700175e-05,
+      "loss": 0.5471,
+      "step": 17000
+    },
+    {
+      "epoch": 1.6495245488065204,
+      "eval_loss": 0.6354050636291504,
+      "eval_runtime": 2468.5352,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.363,
+      "step": 17000
+    },
+    {
+      "epoch": 1.650494857364642,
+      "grad_norm": 1.668904824652418,
+      "learning_rate": 2.256473489519112e-05,
+      "loss": 0.4838,
+      "step": 17010
+    },
+    {
+      "epoch": 1.6514651659227635,
+      "grad_norm": 1.7782439469811153,
+      "learning_rate": 2.2548510610682068e-05,
+      "loss": 0.5225,
+      "step": 17020
+    },
+    {
+      "epoch": 1.652435474480885,
+      "grad_norm": 2.1175538832822576,
+      "learning_rate": 2.2532286326173017e-05,
+      "loss": 0.5438,
+      "step": 17030
+    },
+    {
+      "epoch": 1.6534057830390063,
+      "grad_norm": 1.9031650594744414,
+      "learning_rate": 2.2516062041663964e-05,
+      "loss": 0.4456,
+      "step": 17040
+    },
+    {
+      "epoch": 1.6543760915971277,
+      "grad_norm": 1.8418596497226953,
+      "learning_rate": 2.249983775715491e-05,
+      "loss": 0.5148,
+      "step": 17050
+    },
+    {
+      "epoch": 1.6553464001552494,
+      "grad_norm": 2.001058155829245,
+      "learning_rate": 2.2483613472645857e-05,
+      "loss": 0.5381,
+      "step": 17060
+    },
+    {
+      "epoch": 1.656316708713371,
+      "grad_norm": 1.6070553409553883,
+      "learning_rate": 2.2467389188136803e-05,
+      "loss": 0.5199,
+      "step": 17070
+    },
+    {
+      "epoch": 1.6572870172714924,
+      "grad_norm": 1.8977480591408646,
+      "learning_rate": 2.245116490362775e-05,
+      "loss": 0.5132,
+      "step": 17080
+    },
+    {
+      "epoch": 1.6582573258296138,
+      "grad_norm": 1.8011894993916608,
+      "learning_rate": 2.2434940619118696e-05,
+      "loss": 0.5448,
+      "step": 17090
+    },
+    {
+      "epoch": 1.6592276343877352,
+      "grad_norm": 2.0284255015217565,
+      "learning_rate": 2.2418716334609642e-05,
+      "loss": 0.4717,
+      "step": 17100
+    },
+    {
+      "epoch": 1.6601979429458567,
+      "grad_norm": 1.885999888526123,
+      "learning_rate": 2.2402492050100592e-05,
+      "loss": 0.5183,
+      "step": 17110
+    },
+    {
+      "epoch": 1.6611682515039783,
+      "grad_norm": 1.9056970812089797,
+      "learning_rate": 2.2386267765591538e-05,
+      "loss": 0.4393,
+      "step": 17120
+    },
+    {
+      "epoch": 1.6621385600620997,
+      "grad_norm": 2.267841170403769,
+      "learning_rate": 2.2370043481082488e-05,
+      "loss": 0.4882,
+      "step": 17130
+    },
+    {
+      "epoch": 1.6631088686202213,
+      "grad_norm": 1.5488822989395437,
+      "learning_rate": 2.2353819196573434e-05,
+      "loss": 0.5294,
+      "step": 17140
+    },
+    {
+      "epoch": 1.6640791771783427,
+      "grad_norm": 1.992034372217963,
+      "learning_rate": 2.233759491206438e-05,
+      "loss": 0.4925,
+      "step": 17150
+    },
+    {
+      "epoch": 1.6650494857364642,
+      "grad_norm": 1.7050665334433257,
+      "learning_rate": 2.2321370627555327e-05,
+      "loss": 0.5237,
+      "step": 17160
+    },
+    {
+      "epoch": 1.6660197942945856,
+      "grad_norm": 1.8164813364072325,
+      "learning_rate": 2.2305146343046274e-05,
+      "loss": 0.5128,
+      "step": 17170
+    },
+    {
+      "epoch": 1.6669901028527072,
+      "grad_norm": 1.7407964429381908,
+      "learning_rate": 2.228892205853722e-05,
+      "loss": 0.4812,
+      "step": 17180
+    },
+    {
+      "epoch": 1.6679604114108286,
+      "grad_norm": 2.220803029174025,
+      "learning_rate": 2.2272697774028166e-05,
+      "loss": 0.4936,
+      "step": 17190
+    },
+    {
+      "epoch": 1.6689307199689503,
+      "grad_norm": 1.7697338691239146,
+      "learning_rate": 2.2256473489519113e-05,
+      "loss": 0.5208,
+      "step": 17200
+    },
+    {
+      "epoch": 1.6699010285270717,
+      "grad_norm": 2.2487670732225085,
+      "learning_rate": 2.224024920501006e-05,
+      "loss": 0.4833,
+      "step": 17210
+    },
+    {
+      "epoch": 1.670871337085193,
+      "grad_norm": 1.888912866564453,
+      "learning_rate": 2.222402492050101e-05,
+      "loss": 0.5287,
+      "step": 17220
+    },
+    {
+      "epoch": 1.6718416456433145,
+      "grad_norm": 1.7902985235611324,
+      "learning_rate": 2.2207800635991955e-05,
+      "loss": 0.5271,
+      "step": 17230
+    },
+    {
+      "epoch": 1.672811954201436,
+      "grad_norm": 2.2626922106136607,
+      "learning_rate": 2.21915763514829e-05,
+      "loss": 0.5062,
+      "step": 17240
+    },
+    {
+      "epoch": 1.6737822627595575,
+      "grad_norm": 2.0509489741448923,
+      "learning_rate": 2.2175352066973848e-05,
+      "loss": 0.4586,
+      "step": 17250
+    },
+    {
+      "epoch": 1.6747525713176792,
+      "grad_norm": 1.7028309180471788,
+      "learning_rate": 2.2159127782464794e-05,
+      "loss": 0.5591,
+      "step": 17260
+    },
+    {
+      "epoch": 1.6757228798758006,
+      "grad_norm": 1.8148060135409945,
+      "learning_rate": 2.214290349795574e-05,
+      "loss": 0.4708,
+      "step": 17270
+    },
+    {
+      "epoch": 1.676693188433922,
+      "grad_norm": 1.7415357367009103,
+      "learning_rate": 2.2126679213446687e-05,
+      "loss": 0.4877,
+      "step": 17280
+    },
+    {
+      "epoch": 1.6776634969920434,
+      "grad_norm": 1.6879714610876217,
+      "learning_rate": 2.2110454928937634e-05,
+      "loss": 0.4838,
+      "step": 17290
+    },
+    {
+      "epoch": 1.6786338055501648,
+      "grad_norm": 1.3869597913004987,
+      "learning_rate": 2.2094230644428583e-05,
+      "loss": 0.496,
+      "step": 17300
+    },
+    {
+      "epoch": 1.6796041141082865,
+      "grad_norm": 1.4730335934961858,
+      "learning_rate": 2.207800635991953e-05,
+      "loss": 0.4926,
+      "step": 17310
+    },
+    {
+      "epoch": 1.6805744226664079,
+      "grad_norm": 1.771999456693497,
+      "learning_rate": 2.2061782075410476e-05,
+      "loss": 0.4993,
+      "step": 17320
+    },
+    {
+      "epoch": 1.6815447312245295,
+      "grad_norm": 1.6689193723022826,
+      "learning_rate": 2.2045557790901423e-05,
+      "loss": 0.4815,
+      "step": 17330
+    },
+    {
+      "epoch": 1.682515039782651,
+      "grad_norm": 2.1543699729043717,
+      "learning_rate": 2.202933350639237e-05,
+      "loss": 0.4772,
+      "step": 17340
+    },
+    {
+      "epoch": 1.6834853483407723,
+      "grad_norm": 1.7964447035669882,
+      "learning_rate": 2.2013109221883315e-05,
+      "loss": 0.4872,
+      "step": 17350
+    },
+    {
+      "epoch": 1.6844556568988938,
+      "grad_norm": 1.5499702179216124,
+      "learning_rate": 2.199688493737426e-05,
+      "loss": 0.5115,
+      "step": 17360
+    },
+    {
+      "epoch": 1.6854259654570152,
+      "grad_norm": 1.736034409254174,
+      "learning_rate": 2.1980660652865208e-05,
+      "loss": 0.5004,
+      "step": 17370
+    },
+    {
+      "epoch": 1.6863962740151368,
+      "grad_norm": 1.6670883800573928,
+      "learning_rate": 2.1964436368356154e-05,
+      "loss": 0.5172,
+      "step": 17380
+    },
+    {
+      "epoch": 1.6873665825732584,
+      "grad_norm": 1.8976878178656118,
+      "learning_rate": 2.1948212083847104e-05,
+      "loss": 0.5056,
+      "step": 17390
+    },
+    {
+      "epoch": 1.6883368911313799,
+      "grad_norm": 1.7825783485749558,
+      "learning_rate": 2.193198779933805e-05,
+      "loss": 0.5079,
+      "step": 17400
+    },
+    {
+      "epoch": 1.6893071996895013,
+      "grad_norm": 1.963783922685446,
+      "learning_rate": 2.1915763514828997e-05,
+      "loss": 0.5595,
+      "step": 17410
+    },
+    {
+      "epoch": 1.6902775082476227,
+      "grad_norm": 1.577100524669314,
+      "learning_rate": 2.1899539230319943e-05,
+      "loss": 0.5122,
+      "step": 17420
+    },
+    {
+      "epoch": 1.691247816805744,
+      "grad_norm": 1.9224227001223828,
+      "learning_rate": 2.188331494581089e-05,
+      "loss": 0.5179,
+      "step": 17430
+    },
+    {
+      "epoch": 1.6922181253638657,
+      "grad_norm": 1.968807335226898,
+      "learning_rate": 2.1867090661301836e-05,
+      "loss": 0.4811,
+      "step": 17440
+    },
+    {
+      "epoch": 1.6931884339219871,
+      "grad_norm": 1.8656524316794496,
+      "learning_rate": 2.1850866376792782e-05,
+      "loss": 0.5562,
+      "step": 17450
+    },
+    {
+      "epoch": 1.6941587424801088,
+      "grad_norm": 1.936995738098722,
+      "learning_rate": 2.183464209228373e-05,
+      "loss": 0.5128,
+      "step": 17460
+    },
+    {
+      "epoch": 1.6951290510382302,
+      "grad_norm": 1.716335988732862,
+      "learning_rate": 2.181841780777468e-05,
+      "loss": 0.4878,
+      "step": 17470
+    },
+    {
+      "epoch": 1.6960993595963516,
+      "grad_norm": 1.8671035370037838,
+      "learning_rate": 2.1802193523265625e-05,
+      "loss": 0.5631,
+      "step": 17480
+    },
+    {
+      "epoch": 1.697069668154473,
+      "grad_norm": 2.10031146751523,
+      "learning_rate": 2.178596923875657e-05,
+      "loss": 0.4911,
+      "step": 17490
+    },
+    {
+      "epoch": 1.6980399767125947,
+      "grad_norm": 1.4651955164387034,
+      "learning_rate": 2.1769744954247518e-05,
+      "loss": 0.4841,
+      "step": 17500
+    },
+    {
+      "epoch": 1.6980399767125947,
+      "eval_loss": 0.6339951157569885,
+      "eval_runtime": 2473.3585,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 17500
+    },
+    {
+      "epoch": 1.699010285270716,
+      "grad_norm": 1.9647331919080824,
+      "learning_rate": 2.1753520669738468e-05,
+      "loss": 0.4788,
+      "step": 17510
+    },
+    {
+      "epoch": 1.6999805938288377,
+      "grad_norm": 1.9100412693293238,
+      "learning_rate": 2.1737296385229414e-05,
+      "loss": 0.475,
+      "step": 17520
+    },
+    {
+      "epoch": 1.7009509023869591,
+      "grad_norm": 1.7020406495492606,
+      "learning_rate": 2.172107210072036e-05,
+      "loss": 0.4908,
+      "step": 17530
+    },
+    {
+      "epoch": 1.7019212109450805,
+      "grad_norm": 1.4911580170253027,
+      "learning_rate": 2.1704847816211307e-05,
+      "loss": 0.4832,
+      "step": 17540
+    },
+    {
+      "epoch": 1.702891519503202,
+      "grad_norm": 2.086292759031188,
+      "learning_rate": 2.1688623531702253e-05,
+      "loss": 0.5129,
+      "step": 17550
+    },
+    {
+      "epoch": 1.7038618280613234,
+      "grad_norm": 1.5359738880009668,
+      "learning_rate": 2.16723992471932e-05,
+      "loss": 0.4512,
+      "step": 17560
+    },
+    {
+      "epoch": 1.704832136619445,
+      "grad_norm": 1.850242190602371,
+      "learning_rate": 2.165617496268415e-05,
+      "loss": 0.5224,
+      "step": 17570
+    },
+    {
+      "epoch": 1.7058024451775666,
+      "grad_norm": 1.2439323678887082,
+      "learning_rate": 2.1639950678175096e-05,
+      "loss": 0.4625,
+      "step": 17580
+    },
+    {
+      "epoch": 1.706772753735688,
+      "grad_norm": 1.5186947478659967,
+      "learning_rate": 2.1623726393666042e-05,
+      "loss": 0.5527,
+      "step": 17590
+    },
+    {
+      "epoch": 1.7077430622938095,
+      "grad_norm": 2.1869553971014035,
+      "learning_rate": 2.160750210915699e-05,
+      "loss": 0.499,
+      "step": 17600
+    },
+    {
+      "epoch": 1.7087133708519309,
+      "grad_norm": 2.1896368464295968,
+      "learning_rate": 2.1591277824647935e-05,
+      "loss": 0.5182,
+      "step": 17610
+    },
+    {
+      "epoch": 1.7096836794100523,
+      "grad_norm": 2.5508627899244853,
+      "learning_rate": 2.157505354013888e-05,
+      "loss": 0.529,
+      "step": 17620
+    },
+    {
+      "epoch": 1.710653987968174,
+      "grad_norm": 2.0451418684052527,
+      "learning_rate": 2.1558829255629828e-05,
+      "loss": 0.4789,
+      "step": 17630
+    },
+    {
+      "epoch": 1.7116242965262953,
+      "grad_norm": 2.358852269881603,
+      "learning_rate": 2.1542604971120774e-05,
+      "loss": 0.4701,
+      "step": 17640
+    },
+    {
+      "epoch": 1.712594605084417,
+      "grad_norm": 1.65876380363299,
+      "learning_rate": 2.152638068661172e-05,
+      "loss": 0.5078,
+      "step": 17650
+    },
+    {
+      "epoch": 1.7135649136425384,
+      "grad_norm": 1.819260137292149,
+      "learning_rate": 2.151015640210267e-05,
+      "loss": 0.5115,
+      "step": 17660
+    },
+    {
+      "epoch": 1.7145352222006598,
+      "grad_norm": 1.9278786735506124,
+      "learning_rate": 2.1493932117593616e-05,
+      "loss": 0.526,
+      "step": 17670
+    },
+    {
+      "epoch": 1.7155055307587812,
+      "grad_norm": 1.647908501988862,
+      "learning_rate": 2.1477707833084563e-05,
+      "loss": 0.4455,
+      "step": 17680
+    },
+    {
+      "epoch": 1.7164758393169026,
+      "grad_norm": 2.211325829098572,
+      "learning_rate": 2.146148354857551e-05,
+      "loss": 0.5251,
+      "step": 17690
+    },
+    {
+      "epoch": 1.7174461478750243,
+      "grad_norm": 2.068602886260178,
+      "learning_rate": 2.1445259264066456e-05,
+      "loss": 0.5173,
+      "step": 17700
+    },
+    {
+      "epoch": 1.718416456433146,
+      "grad_norm": 1.7976369940318933,
+      "learning_rate": 2.1429034979557402e-05,
+      "loss": 0.4997,
+      "step": 17710
+    },
+    {
+      "epoch": 1.7193867649912673,
+      "grad_norm": 1.8518665259593716,
+      "learning_rate": 2.141281069504835e-05,
+      "loss": 0.5221,
+      "step": 17720
+    },
+    {
+      "epoch": 1.7203570735493887,
+      "grad_norm": 2.162010082710502,
+      "learning_rate": 2.1396586410539295e-05,
+      "loss": 0.5096,
+      "step": 17730
+    },
+    {
+      "epoch": 1.7213273821075101,
+      "grad_norm": 2.3175061118466704,
+      "learning_rate": 2.138036212603024e-05,
+      "loss": 0.5203,
+      "step": 17740
+    },
+    {
+      "epoch": 1.7222976906656315,
+      "grad_norm": 2.0036851029306364,
+      "learning_rate": 2.136413784152119e-05,
+      "loss": 0.5312,
+      "step": 17750
+    },
+    {
+      "epoch": 1.7232679992237532,
+      "grad_norm": 2.232799745735125,
+      "learning_rate": 2.1347913557012137e-05,
+      "loss": 0.477,
+      "step": 17760
+    },
+    {
+      "epoch": 1.7242383077818746,
+      "grad_norm": 2.0579250461700225,
+      "learning_rate": 2.1331689272503084e-05,
+      "loss": 0.5104,
+      "step": 17770
+    },
+    {
+      "epoch": 1.7252086163399962,
+      "grad_norm": 1.9067412132990194,
+      "learning_rate": 2.131546498799403e-05,
+      "loss": 0.4663,
+      "step": 17780
+    },
+    {
+      "epoch": 1.7261789248981176,
+      "grad_norm": 1.6166777364493794,
+      "learning_rate": 2.1299240703484976e-05,
+      "loss": 0.5091,
+      "step": 17790
+    },
+    {
+      "epoch": 1.727149233456239,
+      "grad_norm": 1.8400065423176315,
+      "learning_rate": 2.1283016418975923e-05,
+      "loss": 0.4577,
+      "step": 17800
+    },
+    {
+      "epoch": 1.7281195420143605,
+      "grad_norm": 1.8343636819724598,
+      "learning_rate": 2.126679213446687e-05,
+      "loss": 0.5171,
+      "step": 17810
+    },
+    {
+      "epoch": 1.729089850572482,
+      "grad_norm": 2.035066789255769,
+      "learning_rate": 2.1250567849957816e-05,
+      "loss": 0.4809,
+      "step": 17820
+    },
+    {
+      "epoch": 1.7300601591306035,
+      "grad_norm": 1.9750470965118032,
+      "learning_rate": 2.1234343565448765e-05,
+      "loss": 0.4681,
+      "step": 17830
+    },
+    {
+      "epoch": 1.7310304676887251,
+      "grad_norm": 1.707449728088738,
+      "learning_rate": 2.1218119280939712e-05,
+      "loss": 0.4768,
+      "step": 17840
+    },
+    {
+      "epoch": 1.7320007762468466,
+      "grad_norm": 2.091683187172791,
+      "learning_rate": 2.1201894996430658e-05,
+      "loss": 0.4967,
+      "step": 17850
+    },
+    {
+      "epoch": 1.732971084804968,
+      "grad_norm": 1.8317230891625513,
+      "learning_rate": 2.1185670711921604e-05,
+      "loss": 0.4617,
+      "step": 17860
+    },
+    {
+      "epoch": 1.7339413933630894,
+      "grad_norm": 1.8581590049450867,
+      "learning_rate": 2.116944642741255e-05,
+      "loss": 0.453,
+      "step": 17870
+    },
+    {
+      "epoch": 1.7349117019212108,
+      "grad_norm": 2.394138871900483,
+      "learning_rate": 2.1153222142903497e-05,
+      "loss": 0.5231,
+      "step": 17880
+    },
+    {
+      "epoch": 1.7358820104793324,
+      "grad_norm": 1.8303455092887013,
+      "learning_rate": 2.1136997858394444e-05,
+      "loss": 0.5045,
+      "step": 17890
+    },
+    {
+      "epoch": 1.736852319037454,
+      "grad_norm": 1.7040025873966649,
+      "learning_rate": 2.1120773573885393e-05,
+      "loss": 0.5185,
+      "step": 17900
+    },
+    {
+      "epoch": 1.7378226275955755,
+      "grad_norm": 1.721795865054807,
+      "learning_rate": 2.110454928937634e-05,
+      "loss": 0.5153,
+      "step": 17910
+    },
+    {
+      "epoch": 1.738792936153697,
+      "grad_norm": 2.195542603436347,
+      "learning_rate": 2.1088325004867286e-05,
+      "loss": 0.515,
+      "step": 17920
+    },
+    {
+      "epoch": 1.7397632447118183,
+      "grad_norm": 1.9475497865392444,
+      "learning_rate": 2.1072100720358236e-05,
+      "loss": 0.6134,
+      "step": 17930
+    },
+    {
+      "epoch": 1.7407335532699397,
+      "grad_norm": 2.007827581925297,
+      "learning_rate": 2.1055876435849182e-05,
+      "loss": 0.4483,
+      "step": 17940
+    },
+    {
+      "epoch": 1.7417038618280614,
+      "grad_norm": 2.427289180607889,
+      "learning_rate": 2.103965215134013e-05,
+      "loss": 0.4791,
+      "step": 17950
+    },
+    {
+      "epoch": 1.7426741703861828,
+      "grad_norm": 2.090548009394418,
+      "learning_rate": 2.1023427866831075e-05,
+      "loss": 0.4485,
+      "step": 17960
+    },
+    {
+      "epoch": 1.7436444789443044,
+      "grad_norm": 1.4231148878110762,
+      "learning_rate": 2.100720358232202e-05,
+      "loss": 0.4819,
+      "step": 17970
+    },
+    {
+      "epoch": 1.7446147875024258,
+      "grad_norm": 1.952695177483555,
+      "learning_rate": 2.0990979297812968e-05,
+      "loss": 0.4615,
+      "step": 17980
+    },
+    {
+      "epoch": 1.7455850960605472,
+      "grad_norm": 1.5076062020543768,
+      "learning_rate": 2.0974755013303914e-05,
+      "loss": 0.5145,
+      "step": 17990
+    },
+    {
+      "epoch": 1.7465554046186686,
+      "grad_norm": 2.385317508868708,
+      "learning_rate": 2.095853072879486e-05,
+      "loss": 0.4667,
+      "step": 18000
+    },
+    {
+      "epoch": 1.7465554046186686,
+      "eval_loss": 0.6351094841957092,
+      "eval_runtime": 2471.0101,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 18000
+    },
+    {
+      "epoch": 1.74752571317679,
+      "grad_norm": 1.5204879350818516,
+      "learning_rate": 2.0942306444285807e-05,
+      "loss": 0.4808,
+      "step": 18010
+    },
+    {
+      "epoch": 1.7484960217349117,
+      "grad_norm": 2.363404313519829,
+      "learning_rate": 2.0926082159776757e-05,
+      "loss": 0.4829,
+      "step": 18020
+    },
+    {
+      "epoch": 1.7494663302930333,
+      "grad_norm": 1.771660432749116,
+      "learning_rate": 2.0909857875267703e-05,
+      "loss": 0.4838,
+      "step": 18030
+    },
+    {
+      "epoch": 1.7504366388511547,
+      "grad_norm": 1.754309450094576,
+      "learning_rate": 2.089363359075865e-05,
+      "loss": 0.5289,
+      "step": 18040
+    },
+    {
+      "epoch": 1.7514069474092762,
+      "grad_norm": 1.7394710721825013,
+      "learning_rate": 2.0877409306249596e-05,
+      "loss": 0.4681,
+      "step": 18050
+    },
+    {
+      "epoch": 1.7523772559673976,
+      "grad_norm": 1.903002291309086,
+      "learning_rate": 2.0861185021740542e-05,
+      "loss": 0.4352,
+      "step": 18060
+    },
+    {
+      "epoch": 1.753347564525519,
+      "grad_norm": 1.8599137833426156,
+      "learning_rate": 2.084496073723149e-05,
+      "loss": 0.5158,
+      "step": 18070
+    },
+    {
+      "epoch": 1.7543178730836406,
+      "grad_norm": 2.416977234001108,
+      "learning_rate": 2.0828736452722435e-05,
+      "loss": 0.458,
+      "step": 18080
+    },
+    {
+      "epoch": 1.755288181641762,
+      "grad_norm": 2.0471286021548876,
+      "learning_rate": 2.081251216821338e-05,
+      "loss": 0.5093,
+      "step": 18090
+    },
+    {
+      "epoch": 1.7562584901998837,
+      "grad_norm": 1.676028789821819,
+      "learning_rate": 2.079628788370433e-05,
+      "loss": 0.4682,
+      "step": 18100
+    },
+    {
+      "epoch": 1.757228798758005,
+      "grad_norm": 2.27723774043769,
+      "learning_rate": 2.0780063599195278e-05,
+      "loss": 0.4844,
+      "step": 18110
+    },
+    {
+      "epoch": 1.7581991073161265,
+      "grad_norm": 1.9065062110447546,
+      "learning_rate": 2.0763839314686224e-05,
+      "loss": 0.4913,
+      "step": 18120
+    },
+    {
+      "epoch": 1.759169415874248,
+      "grad_norm": 1.499218120899005,
+      "learning_rate": 2.074761503017717e-05,
+      "loss": 0.4764,
+      "step": 18130
+    },
+    {
+      "epoch": 1.7601397244323695,
+      "grad_norm": 1.7111231535151605,
+      "learning_rate": 2.0731390745668117e-05,
+      "loss": 0.4376,
+      "step": 18140
+    },
+    {
+      "epoch": 1.761110032990491,
+      "grad_norm": 2.1731054208801406,
+      "learning_rate": 2.0715166461159063e-05,
+      "loss": 0.5115,
+      "step": 18150
+    },
+    {
+      "epoch": 1.7620803415486126,
+      "grad_norm": 2.537816428144228,
+      "learning_rate": 2.069894217665001e-05,
+      "loss": 0.4872,
+      "step": 18160
+    },
+    {
+      "epoch": 1.763050650106734,
+      "grad_norm": 1.8860270703852153,
+      "learning_rate": 2.0682717892140956e-05,
+      "loss": 0.544,
+      "step": 18170
+    },
+    {
+      "epoch": 1.7640209586648554,
+      "grad_norm": 2.021054656113362,
+      "learning_rate": 2.0666493607631902e-05,
+      "loss": 0.4595,
+      "step": 18180
+    },
+    {
+      "epoch": 1.7649912672229768,
+      "grad_norm": 1.7916097460233942,
+      "learning_rate": 2.0650269323122852e-05,
+      "loss": 0.5098,
+      "step": 18190
+    },
+    {
+      "epoch": 1.7659615757810982,
+      "grad_norm": 2.3100480138537,
+      "learning_rate": 2.06340450386138e-05,
+      "loss": 0.4801,
+      "step": 18200
+    },
+    {
+      "epoch": 1.7669318843392199,
+      "grad_norm": 2.4377132473167413,
+      "learning_rate": 2.0617820754104745e-05,
+      "loss": 0.4507,
+      "step": 18210
+    },
+    {
+      "epoch": 1.7679021928973415,
+      "grad_norm": 2.303644526731782,
+      "learning_rate": 2.060159646959569e-05,
+      "loss": 0.4557,
+      "step": 18220
+    },
+    {
+      "epoch": 1.768872501455463,
+      "grad_norm": 2.104570073246246,
+      "learning_rate": 2.0585372185086638e-05,
+      "loss": 0.4657,
+      "step": 18230
+    },
+    {
+      "epoch": 1.7698428100135843,
+      "grad_norm": 1.902037691736978,
+      "learning_rate": 2.0569147900577584e-05,
+      "loss": 0.4885,
+      "step": 18240
+    },
+    {
+      "epoch": 1.7708131185717058,
+      "grad_norm": 1.8321649978057852,
+      "learning_rate": 2.055292361606853e-05,
+      "loss": 0.4953,
+      "step": 18250
+    },
+    {
+      "epoch": 1.7717834271298272,
+      "grad_norm": 1.7318686323178791,
+      "learning_rate": 2.0536699331559477e-05,
+      "loss": 0.4993,
+      "step": 18260
+    },
+    {
+      "epoch": 1.7727537356879488,
+      "grad_norm": 1.6339592524860076,
+      "learning_rate": 2.0520475047050423e-05,
+      "loss": 0.5311,
+      "step": 18270
+    },
+    {
+      "epoch": 1.7737240442460702,
+      "grad_norm": 2.0377735402561448,
+      "learning_rate": 2.0504250762541373e-05,
+      "loss": 0.4785,
+      "step": 18280
+    },
+    {
+      "epoch": 1.7746943528041919,
+      "grad_norm": 1.4282663698707199,
+      "learning_rate": 2.0488026478032323e-05,
+      "loss": 0.5077,
+      "step": 18290
+    },
+    {
+      "epoch": 1.7756646613623133,
+      "grad_norm": 1.9149293437143238,
+      "learning_rate": 2.047180219352327e-05,
+      "loss": 0.5319,
+      "step": 18300
+    },
+    {
+      "epoch": 1.7766349699204347,
+      "grad_norm": 2.1578236484875077,
+      "learning_rate": 2.0455577909014215e-05,
+      "loss": 0.4928,
+      "step": 18310
+    },
+    {
+      "epoch": 1.777605278478556,
+      "grad_norm": 2.107005500307039,
+      "learning_rate": 2.0439353624505162e-05,
+      "loss": 0.4969,
+      "step": 18320
+    },
+    {
+      "epoch": 1.7785755870366775,
+      "grad_norm": 1.9793180118959242,
+      "learning_rate": 2.0423129339996108e-05,
+      "loss": 0.4585,
+      "step": 18330
+    },
+    {
+      "epoch": 1.7795458955947991,
+      "grad_norm": 1.8333976168598185,
+      "learning_rate": 2.0406905055487055e-05,
+      "loss": 0.4982,
+      "step": 18340
+    },
+    {
+      "epoch": 1.7805162041529208,
+      "grad_norm": 1.9049818949759387,
+      "learning_rate": 2.0390680770978e-05,
+      "loss": 0.494,
+      "step": 18350
+    },
+    {
+      "epoch": 1.7814865127110422,
+      "grad_norm": 1.9993776566406598,
+      "learning_rate": 2.0374456486468947e-05,
+      "loss": 0.4918,
+      "step": 18360
+    },
+    {
+      "epoch": 1.7824568212691636,
+      "grad_norm": 1.901331422597336,
+      "learning_rate": 2.0358232201959894e-05,
+      "loss": 0.5138,
+      "step": 18370
+    },
+    {
+      "epoch": 1.783427129827285,
+      "grad_norm": 1.6972755513639741,
+      "learning_rate": 2.0342007917450843e-05,
+      "loss": 0.4715,
+      "step": 18380
+    },
+    {
+      "epoch": 1.7843974383854064,
+      "grad_norm": 1.7683176039213586,
+      "learning_rate": 2.032578363294179e-05,
+      "loss": 0.4917,
+      "step": 18390
+    },
+    {
+      "epoch": 1.785367746943528,
+      "grad_norm": 2.473174371442726,
+      "learning_rate": 2.0309559348432736e-05,
+      "loss": 0.5202,
+      "step": 18400
+    },
+    {
+      "epoch": 1.7863380555016495,
+      "grad_norm": 1.5926803956510256,
+      "learning_rate": 2.0293335063923683e-05,
+      "loss": 0.4988,
+      "step": 18410
+    },
+    {
+      "epoch": 1.7873083640597711,
+      "grad_norm": 2.291647967076523,
+      "learning_rate": 2.027711077941463e-05,
+      "loss": 0.4824,
+      "step": 18420
+    },
+    {
+      "epoch": 1.7882786726178925,
+      "grad_norm": 1.8562307329324135,
+      "learning_rate": 2.0260886494905575e-05,
+      "loss": 0.4563,
+      "step": 18430
+    },
+    {
+      "epoch": 1.789248981176014,
+      "grad_norm": 2.0376982504169225,
+      "learning_rate": 2.0244662210396522e-05,
+      "loss": 0.5306,
+      "step": 18440
+    },
+    {
+      "epoch": 1.7902192897341354,
+      "grad_norm": 2.2229061869152287,
+      "learning_rate": 2.0228437925887468e-05,
+      "loss": 0.4992,
+      "step": 18450
+    },
+    {
+      "epoch": 1.791189598292257,
+      "grad_norm": 1.9381177105445806,
+      "learning_rate": 2.0212213641378418e-05,
+      "loss": 0.5163,
+      "step": 18460
+    },
+    {
+      "epoch": 1.7921599068503784,
+      "grad_norm": 2.2038889339934276,
+      "learning_rate": 2.0195989356869364e-05,
+      "loss": 0.4872,
+      "step": 18470
+    },
+    {
+      "epoch": 1.7931302154085,
+      "grad_norm": 1.7467002610372642,
+      "learning_rate": 2.017976507236031e-05,
+      "loss": 0.4876,
+      "step": 18480
+    },
+    {
+      "epoch": 1.7941005239666215,
+      "grad_norm": 1.80516559021005,
+      "learning_rate": 2.0163540787851257e-05,
+      "loss": 0.5155,
+      "step": 18490
+    },
+    {
+      "epoch": 1.7950708325247429,
+      "grad_norm": 1.6841944257951464,
+      "learning_rate": 2.0147316503342203e-05,
+      "loss": 0.485,
+      "step": 18500
+    },
+    {
+      "epoch": 1.7950708325247429,
+      "eval_loss": 0.634519636631012,
+      "eval_runtime": 2476.7893,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.362,
+      "step": 18500
+    },
+    {
+      "epoch": 1.7960411410828643,
+      "grad_norm": 1.9439021855874303,
+      "learning_rate": 2.013109221883315e-05,
+      "loss": 0.452,
+      "step": 18510
+    },
+    {
+      "epoch": 1.7970114496409857,
+      "grad_norm": 2.2513317080100226,
+      "learning_rate": 2.0114867934324096e-05,
+      "loss": 0.5325,
+      "step": 18520
+    },
+    {
+      "epoch": 1.7979817581991073,
+      "grad_norm": 2.08111292912013,
+      "learning_rate": 2.0098643649815043e-05,
+      "loss": 0.5473,
+      "step": 18530
+    },
+    {
+      "epoch": 1.798952066757229,
+      "grad_norm": 1.5733834399311872,
+      "learning_rate": 2.008241936530599e-05,
+      "loss": 0.477,
+      "step": 18540
+    },
+    {
+      "epoch": 1.7999223753153504,
+      "grad_norm": 2.8502682562631954,
+      "learning_rate": 2.006619508079694e-05,
+      "loss": 0.5206,
+      "step": 18550
+    },
+    {
+      "epoch": 1.8008926838734718,
+      "grad_norm": 1.753275957528235,
+      "learning_rate": 2.0049970796287885e-05,
+      "loss": 0.4547,
+      "step": 18560
+    },
+    {
+      "epoch": 1.8018629924315932,
+      "grad_norm": 2.1314672834115473,
+      "learning_rate": 2.003374651177883e-05,
+      "loss": 0.5419,
+      "step": 18570
+    },
+    {
+      "epoch": 1.8028333009897146,
+      "grad_norm": 1.9602167097784182,
+      "learning_rate": 2.0017522227269778e-05,
+      "loss": 0.5155,
+      "step": 18580
+    },
+    {
+      "epoch": 1.8038036095478363,
+      "grad_norm": 1.938753313860514,
+      "learning_rate": 2.0001297942760724e-05,
+      "loss": 0.5068,
+      "step": 18590
+    },
+    {
+      "epoch": 1.8047739181059577,
+      "grad_norm": 2.5245908920622924,
+      "learning_rate": 1.998507365825167e-05,
+      "loss": 0.4823,
+      "step": 18600
+    },
+    {
+      "epoch": 1.8057442266640793,
+      "grad_norm": 1.8472353172834934,
+      "learning_rate": 1.9968849373742617e-05,
+      "loss": 0.5307,
+      "step": 18610
+    },
+    {
+      "epoch": 1.8067145352222007,
+      "grad_norm": 2.119846040052731,
+      "learning_rate": 1.9952625089233563e-05,
+      "loss": 0.5497,
+      "step": 18620
+    },
+    {
+      "epoch": 1.8076848437803221,
+      "grad_norm": 1.8254352104689588,
+      "learning_rate": 1.9936400804724513e-05,
+      "loss": 0.5343,
+      "step": 18630
+    },
+    {
+      "epoch": 1.8086551523384435,
+      "grad_norm": 2.286926975318207,
+      "learning_rate": 1.992017652021546e-05,
+      "loss": 0.5079,
+      "step": 18640
+    },
+    {
+      "epoch": 1.809625460896565,
+      "grad_norm": 2.145456846814811,
+      "learning_rate": 1.9903952235706406e-05,
+      "loss": 0.4792,
+      "step": 18650
+    },
+    {
+      "epoch": 1.8105957694546866,
+      "grad_norm": 2.1237990360529615,
+      "learning_rate": 1.9887727951197352e-05,
+      "loss": 0.5221,
+      "step": 18660
+    },
+    {
+      "epoch": 1.8115660780128082,
+      "grad_norm": 1.9365349043963929,
+      "learning_rate": 1.98715036666883e-05,
+      "loss": 0.4994,
+      "step": 18670
+    },
+    {
+      "epoch": 1.8125363865709296,
+      "grad_norm": 2.1477510328366543,
+      "learning_rate": 1.985527938217925e-05,
+      "loss": 0.5051,
+      "step": 18680
+    },
+    {
+      "epoch": 1.813506695129051,
+      "grad_norm": 2.24009021245378,
+      "learning_rate": 1.9839055097670195e-05,
+      "loss": 0.4411,
+      "step": 18690
+    },
+    {
+      "epoch": 1.8144770036871725,
+      "grad_norm": 2.161147622755333,
+      "learning_rate": 1.982283081316114e-05,
+      "loss": 0.4947,
+      "step": 18700
+    },
+    {
+      "epoch": 1.8154473122452939,
+      "grad_norm": 2.5777752480302296,
+      "learning_rate": 1.9806606528652088e-05,
+      "loss": 0.5027,
+      "step": 18710
+    },
+    {
+      "epoch": 1.8164176208034155,
+      "grad_norm": 1.5289437881489423,
+      "learning_rate": 1.9790382244143034e-05,
+      "loss": 0.5276,
+      "step": 18720
+    },
+    {
+      "epoch": 1.817387929361537,
+      "grad_norm": 1.808692912848697,
+      "learning_rate": 1.977415795963398e-05,
+      "loss": 0.5155,
+      "step": 18730
+    },
+    {
+      "epoch": 1.8183582379196586,
+      "grad_norm": 1.7568825751279589,
+      "learning_rate": 1.975793367512493e-05,
+      "loss": 0.4658,
+      "step": 18740
+    },
+    {
+      "epoch": 1.81932854647778,
+      "grad_norm": 1.5226619183552335,
+      "learning_rate": 1.9741709390615877e-05,
+      "loss": 0.4764,
+      "step": 18750
+    },
+    {
+      "epoch": 1.8202988550359014,
+      "grad_norm": 1.5664856097734037,
+      "learning_rate": 1.9725485106106823e-05,
+      "loss": 0.4503,
+      "step": 18760
+    },
+    {
+      "epoch": 1.8212691635940228,
+      "grad_norm": 1.7127846075920363,
+      "learning_rate": 1.970926082159777e-05,
+      "loss": 0.4558,
+      "step": 18770
+    },
+    {
+      "epoch": 1.8222394721521444,
+      "grad_norm": 2.1030557687163007,
+      "learning_rate": 1.9693036537088716e-05,
+      "loss": 0.5578,
+      "step": 18780
+    },
+    {
+      "epoch": 1.8232097807102658,
+      "grad_norm": 2.0907960296944474,
+      "learning_rate": 1.9676812252579662e-05,
+      "loss": 0.4974,
+      "step": 18790
+    },
+    {
+      "epoch": 1.8241800892683875,
+      "grad_norm": 1.4612012638589904,
+      "learning_rate": 1.966058796807061e-05,
+      "loss": 0.4931,
+      "step": 18800
+    },
+    {
+      "epoch": 1.825150397826509,
+      "grad_norm": 1.6970216305431254,
+      "learning_rate": 1.9644363683561555e-05,
+      "loss": 0.4691,
+      "step": 18810
+    },
+    {
+      "epoch": 1.8261207063846303,
+      "grad_norm": 2.4415425125364125,
+      "learning_rate": 1.9628139399052505e-05,
+      "loss": 0.5249,
+      "step": 18820
+    },
+    {
+      "epoch": 1.8270910149427517,
+      "grad_norm": 1.7771990898290009,
+      "learning_rate": 1.961191511454345e-05,
+      "loss": 0.5019,
+      "step": 18830
+    },
+    {
+      "epoch": 1.8280613235008731,
+      "grad_norm": 1.7839717190667712,
+      "learning_rate": 1.9595690830034397e-05,
+      "loss": 0.5052,
+      "step": 18840
+    },
+    {
+      "epoch": 1.8290316320589948,
+      "grad_norm": 2.209198913639542,
+      "learning_rate": 1.9579466545525344e-05,
+      "loss": 0.5235,
+      "step": 18850
+    },
+    {
+      "epoch": 1.8300019406171164,
+      "grad_norm": 2.200681320097298,
+      "learning_rate": 1.956324226101629e-05,
+      "loss": 0.5302,
+      "step": 18860
+    },
+    {
+      "epoch": 1.8309722491752378,
+      "grad_norm": 2.000921134970765,
+      "learning_rate": 1.9547017976507237e-05,
+      "loss": 0.4648,
+      "step": 18870
+    },
+    {
+      "epoch": 1.8319425577333592,
+      "grad_norm": 2.623565257277865,
+      "learning_rate": 1.9530793691998183e-05,
+      "loss": 0.4855,
+      "step": 18880
+    },
+    {
+      "epoch": 1.8329128662914806,
+      "grad_norm": 1.8550961862060569,
+      "learning_rate": 1.951456940748913e-05,
+      "loss": 0.5135,
+      "step": 18890
+    },
+    {
+      "epoch": 1.833883174849602,
+      "grad_norm": 2.0777083761268718,
+      "learning_rate": 1.9498345122980076e-05,
+      "loss": 0.5236,
+      "step": 18900
+    },
+    {
+      "epoch": 1.8348534834077237,
+      "grad_norm": 2.2339320492583337,
+      "learning_rate": 1.9482120838471025e-05,
+      "loss": 0.5336,
+      "step": 18910
+    },
+    {
+      "epoch": 1.835823791965845,
+      "grad_norm": 2.2130179741978173,
+      "learning_rate": 1.9465896553961972e-05,
+      "loss": 0.4468,
+      "step": 18920
+    },
+    {
+      "epoch": 1.8367941005239667,
+      "grad_norm": 1.9136609274036684,
+      "learning_rate": 1.9449672269452918e-05,
+      "loss": 0.516,
+      "step": 18930
+    },
+    {
+      "epoch": 1.8377644090820882,
+      "grad_norm": 1.9319944175941055,
+      "learning_rate": 1.9433447984943865e-05,
+      "loss": 0.4837,
+      "step": 18940
+    },
+    {
+      "epoch": 1.8387347176402096,
+      "grad_norm": 1.8144243996167448,
+      "learning_rate": 1.941722370043481e-05,
+      "loss": 0.5388,
+      "step": 18950
+    },
+    {
+      "epoch": 1.839705026198331,
+      "grad_norm": 1.9871922674209272,
+      "learning_rate": 1.9400999415925757e-05,
+      "loss": 0.5543,
+      "step": 18960
+    },
+    {
+      "epoch": 1.8406753347564524,
+      "grad_norm": 2.1436846647503707,
+      "learning_rate": 1.9384775131416704e-05,
+      "loss": 0.4787,
+      "step": 18970
+    },
+    {
+      "epoch": 1.841645643314574,
+      "grad_norm": 1.926685455696982,
+      "learning_rate": 1.936855084690765e-05,
+      "loss": 0.5096,
+      "step": 18980
+    },
+    {
+      "epoch": 1.8426159518726957,
+      "grad_norm": 2.342596116742962,
+      "learning_rate": 1.93523265623986e-05,
+      "loss": 0.522,
+      "step": 18990
+    },
+    {
+      "epoch": 1.843586260430817,
+      "grad_norm": 2.255501885540124,
+      "learning_rate": 1.9336102277889546e-05,
+      "loss": 0.5377,
+      "step": 19000
+    },
+    {
+      "epoch": 1.843586260430817,
+      "eval_loss": 0.6318312883377075,
+      "eval_runtime": 2473.7549,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.362,
+      "step": 19000
+    },
+    {
+      "epoch": 1.8445565689889385,
+      "grad_norm": 2.1596583204757884,
+      "learning_rate": 1.9319877993380493e-05,
+      "loss": 0.5629,
+      "step": 19010
+    },
+    {
+      "epoch": 1.84552687754706,
+      "grad_norm": 1.8190379524057834,
+      "learning_rate": 1.930365370887144e-05,
+      "loss": 0.4796,
+      "step": 19020
+    },
+    {
+      "epoch": 1.8464971861051813,
+      "grad_norm": 1.019855687196459,
+      "learning_rate": 1.9287429424362385e-05,
+      "loss": 0.471,
+      "step": 19030
+    },
+    {
+      "epoch": 1.847467494663303,
+      "grad_norm": 1.9569552509020551,
+      "learning_rate": 1.9271205139853332e-05,
+      "loss": 0.5126,
+      "step": 19040
+    },
+    {
+      "epoch": 1.8484378032214244,
+      "grad_norm": 2.1870238918367524,
+      "learning_rate": 1.9254980855344278e-05,
+      "loss": 0.4865,
+      "step": 19050
+    },
+    {
+      "epoch": 1.849408111779546,
+      "grad_norm": 1.594729244733264,
+      "learning_rate": 1.9238756570835225e-05,
+      "loss": 0.4802,
+      "step": 19060
+    },
+    {
+      "epoch": 1.8503784203376674,
+      "grad_norm": 2.1426009139912843,
+      "learning_rate": 1.9222532286326174e-05,
+      "loss": 0.5303,
+      "step": 19070
+    },
+    {
+      "epoch": 1.8513487288957888,
+      "grad_norm": 1.7760248540972265,
+      "learning_rate": 1.920630800181712e-05,
+      "loss": 0.4948,
+      "step": 19080
+    },
+    {
+      "epoch": 1.8523190374539102,
+      "grad_norm": 1.6725671761260958,
+      "learning_rate": 1.919008371730807e-05,
+      "loss": 0.4922,
+      "step": 19090
+    },
+    {
+      "epoch": 1.8532893460120319,
+      "grad_norm": 1.9565774055440965,
+      "learning_rate": 1.9173859432799017e-05,
+      "loss": 0.5658,
+      "step": 19100
+    },
+    {
+      "epoch": 1.8542596545701533,
+      "grad_norm": 2.231350831039019,
+      "learning_rate": 1.9157635148289963e-05,
+      "loss": 0.4906,
+      "step": 19110
+    },
+    {
+      "epoch": 1.855229963128275,
+      "grad_norm": 1.5993699848453529,
+      "learning_rate": 1.914141086378091e-05,
+      "loss": 0.4796,
+      "step": 19120
+    },
+    {
+      "epoch": 1.8562002716863963,
+      "grad_norm": 2.0780226359466387,
+      "learning_rate": 1.9125186579271856e-05,
+      "loss": 0.4773,
+      "step": 19130
+    },
+    {
+      "epoch": 1.8571705802445178,
+      "grad_norm": 2.01763881138038,
+      "learning_rate": 1.9108962294762802e-05,
+      "loss": 0.4984,
+      "step": 19140
+    },
+    {
+      "epoch": 1.8581408888026392,
+      "grad_norm": 2.240723917017317,
+      "learning_rate": 1.909273801025375e-05,
+      "loss": 0.4559,
+      "step": 19150
+    },
+    {
+      "epoch": 1.8591111973607606,
+      "grad_norm": 1.849249605812625,
+      "learning_rate": 1.9076513725744695e-05,
+      "loss": 0.5445,
+      "step": 19160
+    },
+    {
+      "epoch": 1.8600815059188822,
+      "grad_norm": 1.7656331962779384,
+      "learning_rate": 1.906028944123564e-05,
+      "loss": 0.466,
+      "step": 19170
+    },
+    {
+      "epoch": 1.8610518144770039,
+      "grad_norm": 1.9723478771428924,
+      "learning_rate": 1.904406515672659e-05,
+      "loss": 0.4951,
+      "step": 19180
+    },
+    {
+      "epoch": 1.8620221230351253,
+      "grad_norm": 2.133741060165629,
+      "learning_rate": 1.9027840872217538e-05,
+      "loss": 0.5064,
+      "step": 19190
+    },
+    {
+      "epoch": 1.8629924315932467,
+      "grad_norm": 2.267317304504692,
+      "learning_rate": 1.9011616587708484e-05,
+      "loss": 0.4774,
+      "step": 19200
+    },
+    {
+      "epoch": 1.863962740151368,
+      "grad_norm": 1.6590470147649787,
+      "learning_rate": 1.899539230319943e-05,
+      "loss": 0.4684,
+      "step": 19210
+    },
+    {
+      "epoch": 1.8649330487094895,
+      "grad_norm": 1.7428570168415043,
+      "learning_rate": 1.8979168018690377e-05,
+      "loss": 0.494,
+      "step": 19220
+    },
+    {
+      "epoch": 1.8659033572676111,
+      "grad_norm": 1.6212375134091286,
+      "learning_rate": 1.8962943734181323e-05,
+      "loss": 0.481,
+      "step": 19230
+    },
+    {
+      "epoch": 1.8668736658257326,
+      "grad_norm": 2.0261745446913717,
+      "learning_rate": 1.894671944967227e-05,
+      "loss": 0.5107,
+      "step": 19240
+    },
+    {
+      "epoch": 1.8678439743838542,
+      "grad_norm": 1.7751731877262686,
+      "learning_rate": 1.8930495165163216e-05,
+      "loss": 0.4615,
+      "step": 19250
+    },
+    {
+      "epoch": 1.8688142829419756,
+      "grad_norm": 1.5566178962662132,
+      "learning_rate": 1.8914270880654162e-05,
+      "loss": 0.5061,
+      "step": 19260
+    },
+    {
+      "epoch": 1.869784591500097,
+      "grad_norm": 2.0648415585072137,
+      "learning_rate": 1.8898046596145112e-05,
+      "loss": 0.4721,
+      "step": 19270
+    },
+    {
+      "epoch": 1.8707549000582184,
+      "grad_norm": 1.8255664078089486,
+      "learning_rate": 1.888182231163606e-05,
+      "loss": 0.4652,
+      "step": 19280
+    },
+    {
+      "epoch": 1.8717252086163398,
+      "grad_norm": 2.144758151637292,
+      "learning_rate": 1.8865598027127005e-05,
+      "loss": 0.4391,
+      "step": 19290
+    },
+    {
+      "epoch": 1.8726955171744615,
+      "grad_norm": 1.803874603824675,
+      "learning_rate": 1.884937374261795e-05,
+      "loss": 0.492,
+      "step": 19300
+    },
+    {
+      "epoch": 1.8736658257325831,
+      "grad_norm": 2.1603319792398685,
+      "learning_rate": 1.8833149458108898e-05,
+      "loss": 0.4602,
+      "step": 19310
+    },
+    {
+      "epoch": 1.8746361342907045,
+      "grad_norm": 1.931469165672542,
+      "learning_rate": 1.8816925173599844e-05,
+      "loss": 0.479,
+      "step": 19320
+    },
+    {
+      "epoch": 1.875606442848826,
+      "grad_norm": 2.289215850770512,
+      "learning_rate": 1.880070088909079e-05,
+      "loss": 0.475,
+      "step": 19330
+    },
+    {
+      "epoch": 1.8765767514069474,
+      "grad_norm": 2.0548701793141384,
+      "learning_rate": 1.8784476604581737e-05,
+      "loss": 0.4523,
+      "step": 19340
+    },
+    {
+      "epoch": 1.8775470599650688,
+      "grad_norm": 1.6304292402973444,
+      "learning_rate": 1.8768252320072687e-05,
+      "loss": 0.4704,
+      "step": 19350
+    },
+    {
+      "epoch": 1.8785173685231904,
+      "grad_norm": 1.3760981254681826,
+      "learning_rate": 1.8752028035563633e-05,
+      "loss": 0.509,
+      "step": 19360
+    },
+    {
+      "epoch": 1.8794876770813118,
+      "grad_norm": 2.1389308936606364,
+      "learning_rate": 1.873580375105458e-05,
+      "loss": 0.4791,
+      "step": 19370
+    },
+    {
+      "epoch": 1.8804579856394334,
+      "grad_norm": 2.0673076161036663,
+      "learning_rate": 1.8719579466545526e-05,
+      "loss": 0.4902,
+      "step": 19380
+    },
+    {
+      "epoch": 1.8814282941975549,
+      "grad_norm": 1.861431210462014,
+      "learning_rate": 1.8703355182036472e-05,
+      "loss": 0.4719,
+      "step": 19390
+    },
+    {
+      "epoch": 1.8823986027556763,
+      "grad_norm": 1.9788908151587594,
+      "learning_rate": 1.868713089752742e-05,
+      "loss": 0.5127,
+      "step": 19400
+    },
+    {
+      "epoch": 1.8833689113137977,
+      "grad_norm": 1.9487695948736958,
+      "learning_rate": 1.8670906613018365e-05,
+      "loss": 0.4469,
+      "step": 19410
+    },
+    {
+      "epoch": 1.8843392198719193,
+      "grad_norm": 1.8048413444236069,
+      "learning_rate": 1.865468232850931e-05,
+      "loss": 0.5345,
+      "step": 19420
+    },
+    {
+      "epoch": 1.8853095284300407,
+      "grad_norm": 1.8168395167881652,
+      "learning_rate": 1.8638458044000258e-05,
+      "loss": 0.4957,
+      "step": 19430
+    },
+    {
+      "epoch": 1.8862798369881624,
+      "grad_norm": 1.9900462235084369,
+      "learning_rate": 1.8622233759491207e-05,
+      "loss": 0.4928,
+      "step": 19440
+    },
+    {
+      "epoch": 1.8872501455462838,
+      "grad_norm": 1.9978639659195843,
+      "learning_rate": 1.8606009474982154e-05,
+      "loss": 0.5085,
+      "step": 19450
+    },
+    {
+      "epoch": 1.8882204541044052,
+      "grad_norm": 1.9870398133754268,
+      "learning_rate": 1.8589785190473104e-05,
+      "loss": 0.4857,
+      "step": 19460
+    },
+    {
+      "epoch": 1.8891907626625266,
+      "grad_norm": 1.9723654086733247,
+      "learning_rate": 1.857356090596405e-05,
+      "loss": 0.4658,
+      "step": 19470
+    },
+    {
+      "epoch": 1.890161071220648,
+      "grad_norm": 1.8141064421371964,
+      "learning_rate": 1.8557336621454996e-05,
+      "loss": 0.4881,
+      "step": 19480
+    },
+    {
+      "epoch": 1.8911313797787697,
+      "grad_norm": 1.6779273277493256,
+      "learning_rate": 1.8541112336945943e-05,
+      "loss": 0.4798,
+      "step": 19490
+    },
+    {
+      "epoch": 1.8921016883368913,
+      "grad_norm": 1.901625964641795,
+      "learning_rate": 1.852488805243689e-05,
+      "loss": 0.5199,
+      "step": 19500
+    },
+    {
+      "epoch": 1.8921016883368913,
+      "eval_loss": 0.631912350654602,
+      "eval_runtime": 2473.0661,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.362,
+      "step": 19500
+    },
+    {
+      "epoch": 1.8930719968950127,
+      "grad_norm": 1.7019367377384935,
+      "learning_rate": 1.8508663767927836e-05,
+      "loss": 0.4693,
+      "step": 19510
+    },
+    {
+      "epoch": 1.8940423054531341,
+      "grad_norm": 1.9777898503801572,
+      "learning_rate": 1.8492439483418782e-05,
+      "loss": 0.4907,
+      "step": 19520
+    },
+    {
+      "epoch": 1.8950126140112555,
+      "grad_norm": 2.613311730337197,
+      "learning_rate": 1.8476215198909728e-05,
+      "loss": 0.5179,
+      "step": 19530
+    },
+    {
+      "epoch": 1.895982922569377,
+      "grad_norm": 1.3976847929675515,
+      "learning_rate": 1.8459990914400678e-05,
+      "loss": 0.4316,
+      "step": 19540
+    },
+    {
+      "epoch": 1.8969532311274986,
+      "grad_norm": 2.1428783623843715,
+      "learning_rate": 1.8443766629891624e-05,
+      "loss": 0.5275,
+      "step": 19550
+    },
+    {
+      "epoch": 1.89792353968562,
+      "grad_norm": 1.6878966830886573,
+      "learning_rate": 1.842754234538257e-05,
+      "loss": 0.5061,
+      "step": 19560
+    },
+    {
+      "epoch": 1.8988938482437416,
+      "grad_norm": 1.6660923395832288,
+      "learning_rate": 1.8411318060873517e-05,
+      "loss": 0.4846,
+      "step": 19570
+    },
+    {
+      "epoch": 1.899864156801863,
+      "grad_norm": 1.88350077160719,
+      "learning_rate": 1.8395093776364464e-05,
+      "loss": 0.4558,
+      "step": 19580
+    },
+    {
+      "epoch": 1.9008344653599845,
+      "grad_norm": 1.8180585161160603,
+      "learning_rate": 1.837886949185541e-05,
+      "loss": 0.4462,
+      "step": 19590
+    },
+    {
+      "epoch": 1.9018047739181059,
+      "grad_norm": 2.17218483201875,
+      "learning_rate": 1.8362645207346356e-05,
+      "loss": 0.4732,
+      "step": 19600
+    },
+    {
+      "epoch": 1.9027750824762273,
+      "grad_norm": 2.150072249394045,
+      "learning_rate": 1.8346420922837303e-05,
+      "loss": 0.4807,
+      "step": 19610
+    },
+    {
+      "epoch": 1.903745391034349,
+      "grad_norm": 2.186590190383392,
+      "learning_rate": 1.8330196638328253e-05,
+      "loss": 0.5064,
+      "step": 19620
+    },
+    {
+      "epoch": 1.9047156995924706,
+      "grad_norm": 1.9773135303028604,
+      "learning_rate": 1.83139723538192e-05,
+      "loss": 0.4987,
+      "step": 19630
+    },
+    {
+      "epoch": 1.905686008150592,
+      "grad_norm": 1.7722699958849875,
+      "learning_rate": 1.8297748069310145e-05,
+      "loss": 0.493,
+      "step": 19640
+    },
+    {
+      "epoch": 1.9066563167087134,
+      "grad_norm": 2.398748653045921,
+      "learning_rate": 1.828152378480109e-05,
+      "loss": 0.4632,
+      "step": 19650
+    },
+    {
+      "epoch": 1.9076266252668348,
+      "grad_norm": 1.9381638418073313,
+      "learning_rate": 1.8265299500292038e-05,
+      "loss": 0.4879,
+      "step": 19660
+    },
+    {
+      "epoch": 1.9085969338249562,
+      "grad_norm": 1.806142449327403,
+      "learning_rate": 1.8249075215782984e-05,
+      "loss": 0.4426,
+      "step": 19670
+    },
+    {
+      "epoch": 1.9095672423830778,
+      "grad_norm": 1.8375555769454368,
+      "learning_rate": 1.823285093127393e-05,
+      "loss": 0.5133,
+      "step": 19680
+    },
+    {
+      "epoch": 1.9105375509411993,
+      "grad_norm": 2.1900711143018703,
+      "learning_rate": 1.8216626646764877e-05,
+      "loss": 0.4783,
+      "step": 19690
+    },
+    {
+      "epoch": 1.911507859499321,
+      "grad_norm": 1.7860565394856096,
+      "learning_rate": 1.8200402362255824e-05,
+      "loss": 0.5432,
+      "step": 19700
+    },
+    {
+      "epoch": 1.9124781680574423,
+      "grad_norm": 1.5287991043658957,
+      "learning_rate": 1.8184178077746773e-05,
+      "loss": 0.4818,
+      "step": 19710
+    },
+    {
+      "epoch": 1.9134484766155637,
+      "grad_norm": 1.9162344859036078,
+      "learning_rate": 1.816795379323772e-05,
+      "loss": 0.5134,
+      "step": 19720
+    },
+    {
+      "epoch": 1.9144187851736851,
+      "grad_norm": 1.9361948714776511,
+      "learning_rate": 1.8151729508728666e-05,
+      "loss": 0.4705,
+      "step": 19730
+    },
+    {
+      "epoch": 1.9153890937318068,
+      "grad_norm": 1.7948070446854587,
+      "learning_rate": 1.8135505224219613e-05,
+      "loss": 0.4713,
+      "step": 19740
+    },
+    {
+      "epoch": 1.9163594022899282,
+      "grad_norm": 1.94331048872498,
+      "learning_rate": 1.811928093971056e-05,
+      "loss": 0.4636,
+      "step": 19750
+    },
+    {
+      "epoch": 1.9173297108480498,
+      "grad_norm": 1.9083792067669565,
+      "learning_rate": 1.8103056655201505e-05,
+      "loss": 0.4985,
+      "step": 19760
+    },
+    {
+      "epoch": 1.9183000194061712,
+      "grad_norm": 1.7889544788537133,
+      "learning_rate": 1.808683237069245e-05,
+      "loss": 0.465,
+      "step": 19770
+    },
+    {
+      "epoch": 1.9192703279642926,
+      "grad_norm": 2.1133403229015606,
+      "learning_rate": 1.8070608086183398e-05,
+      "loss": 0.4962,
+      "step": 19780
+    },
+    {
+      "epoch": 1.920240636522414,
+      "grad_norm": 2.119511574698095,
+      "learning_rate": 1.8054383801674344e-05,
+      "loss": 0.5265,
+      "step": 19790
+    },
+    {
+      "epoch": 1.9212109450805355,
+      "grad_norm": 1.8530048672687303,
+      "learning_rate": 1.8038159517165294e-05,
+      "loss": 0.4674,
+      "step": 19800
+    },
+    {
+      "epoch": 1.922181253638657,
+      "grad_norm": 2.060163549018839,
+      "learning_rate": 1.802193523265624e-05,
+      "loss": 0.5232,
+      "step": 19810
+    },
+    {
+      "epoch": 1.9231515621967787,
+      "grad_norm": 1.582662573057051,
+      "learning_rate": 1.8005710948147187e-05,
+      "loss": 0.5146,
+      "step": 19820
+    },
+    {
+      "epoch": 1.9241218707549002,
+      "grad_norm": 1.9973478575975316,
+      "learning_rate": 1.7989486663638133e-05,
+      "loss": 0.5294,
+      "step": 19830
+    },
+    {
+      "epoch": 1.9250921793130216,
+      "grad_norm": 2.2324903918022088,
+      "learning_rate": 1.797326237912908e-05,
+      "loss": 0.5039,
+      "step": 19840
+    },
+    {
+      "epoch": 1.926062487871143,
+      "grad_norm": 2.24550841489829,
+      "learning_rate": 1.7957038094620026e-05,
+      "loss": 0.4763,
+      "step": 19850
+    },
+    {
+      "epoch": 1.9270327964292644,
+      "grad_norm": 1.9055133702740896,
+      "learning_rate": 1.7940813810110976e-05,
+      "loss": 0.4978,
+      "step": 19860
+    },
+    {
+      "epoch": 1.928003104987386,
+      "grad_norm": 1.6202922943730298,
+      "learning_rate": 1.7924589525601922e-05,
+      "loss": 0.4467,
+      "step": 19870
+    },
+    {
+      "epoch": 1.9289734135455074,
+      "grad_norm": 1.7946217889300042,
+      "learning_rate": 1.790836524109287e-05,
+      "loss": 0.4975,
+      "step": 19880
+    },
+    {
+      "epoch": 1.929943722103629,
+      "grad_norm": 2.0837037474907776,
+      "learning_rate": 1.7892140956583815e-05,
+      "loss": 0.4488,
+      "step": 19890
+    },
+    {
+      "epoch": 1.9309140306617505,
+      "grad_norm": 2.284301097020319,
+      "learning_rate": 1.7875916672074765e-05,
+      "loss": 0.5315,
+      "step": 19900
+    },
+    {
+      "epoch": 1.931884339219872,
+      "grad_norm": 1.6410720035809963,
+      "learning_rate": 1.785969238756571e-05,
+      "loss": 0.5064,
+      "step": 19910
+    },
+    {
+      "epoch": 1.9328546477779933,
+      "grad_norm": 1.913707413180516,
+      "learning_rate": 1.7843468103056658e-05,
+      "loss": 0.5226,
+      "step": 19920
+    },
+    {
+      "epoch": 1.9338249563361147,
+      "grad_norm": 1.972749697853565,
+      "learning_rate": 1.7827243818547604e-05,
+      "loss": 0.4794,
+      "step": 19930
+    },
+    {
+      "epoch": 1.9347952648942364,
+      "grad_norm": 2.4245035874963663,
+      "learning_rate": 1.781101953403855e-05,
+      "loss": 0.472,
+      "step": 19940
+    },
+    {
+      "epoch": 1.935765573452358,
+      "grad_norm": 1.6532981118899497,
+      "learning_rate": 1.7794795249529497e-05,
+      "loss": 0.4926,
+      "step": 19950
+    },
+    {
+      "epoch": 1.9367358820104794,
+      "grad_norm": 2.301368908039538,
+      "learning_rate": 1.7778570965020443e-05,
+      "loss": 0.4847,
+      "step": 19960
+    },
+    {
+      "epoch": 1.9377061905686008,
+      "grad_norm": 1.9798901797296686,
+      "learning_rate": 1.776234668051139e-05,
+      "loss": 0.5041,
+      "step": 19970
+    },
+    {
+      "epoch": 1.9386764991267222,
+      "grad_norm": 2.0914880764103825,
+      "learning_rate": 1.774612239600234e-05,
+      "loss": 0.4543,
+      "step": 19980
+    },
+    {
+      "epoch": 1.9396468076848437,
+      "grad_norm": 1.1710098992705869,
+      "learning_rate": 1.7729898111493286e-05,
+      "loss": 0.4568,
+      "step": 19990
+    },
+    {
+      "epoch": 1.9406171162429653,
+      "grad_norm": 2.475821372597333,
+      "learning_rate": 1.7713673826984232e-05,
+      "loss": 0.503,
+      "step": 20000
+    },
+    {
+      "epoch": 1.9406171162429653,
+      "eval_loss": 0.6306876540184021,
+      "eval_runtime": 2470.8815,
+      "eval_samples_per_second": 0.725,
+      "eval_steps_per_second": 0.363,
+      "step": 20000
+    },
+    {
+      "epoch": 1.9415874248010867,
+      "grad_norm": 1.8135605973580828,
+      "learning_rate": 1.769744954247518e-05,
+      "loss": 0.5225,
+      "step": 20010
+    },
+    {
+      "epoch": 1.9425577333592083,
+      "grad_norm": 1.8818540999249187,
+      "learning_rate": 1.7681225257966125e-05,
+      "loss": 0.4411,
+      "step": 20020
+    },
+    {
+      "epoch": 1.9435280419173298,
+      "grad_norm": 2.3455403519185767,
+      "learning_rate": 1.766500097345707e-05,
+      "loss": 0.4618,
+      "step": 20030
+    },
+    {
+      "epoch": 1.9444983504754512,
+      "grad_norm": 1.8781392474090728,
+      "learning_rate": 1.7648776688948018e-05,
+      "loss": 0.4757,
+      "step": 20040
+    },
+    {
+      "epoch": 1.9454686590335726,
+      "grad_norm": 2.4837262255821004,
+      "learning_rate": 1.7632552404438964e-05,
+      "loss": 0.4983,
+      "step": 20050
+    },
+    {
+      "epoch": 1.9464389675916942,
+      "grad_norm": 2.0371146667611963,
+      "learning_rate": 1.761632811992991e-05,
+      "loss": 0.4684,
+      "step": 20060
+    },
+    {
+      "epoch": 1.9474092761498156,
+      "grad_norm": 1.9070841836100068,
+      "learning_rate": 1.760010383542086e-05,
+      "loss": 0.5148,
+      "step": 20070
+    },
+    {
+      "epoch": 1.9483795847079373,
+      "grad_norm": 1.6764022613616618,
+      "learning_rate": 1.7583879550911806e-05,
+      "loss": 0.5129,
+      "step": 20080
+    },
+    {
+      "epoch": 1.9493498932660587,
+      "grad_norm": 1.801575932413747,
+      "learning_rate": 1.7567655266402753e-05,
+      "loss": 0.4924,
+      "step": 20090
+    },
+    {
+      "epoch": 1.95032020182418,
+      "grad_norm": 1.417352324977605,
+      "learning_rate": 1.75514309818937e-05,
+      "loss": 0.4871,
+      "step": 20100
+    },
+    {
+      "epoch": 1.9512905103823015,
+      "grad_norm": 1.7430495978746263,
+      "learning_rate": 1.7535206697384646e-05,
+      "loss": 0.4906,
+      "step": 20110
+    },
+    {
+      "epoch": 1.952260818940423,
+      "grad_norm": 1.9378606596579797,
+      "learning_rate": 1.7518982412875592e-05,
+      "loss": 0.4441,
+      "step": 20120
+    },
+    {
+      "epoch": 1.9532311274985446,
+      "grad_norm": 2.024032521983931,
+      "learning_rate": 1.750275812836654e-05,
+      "loss": 0.4574,
+      "step": 20130
+    },
+    {
+      "epoch": 1.9542014360566662,
+      "grad_norm": 1.9748616447898866,
+      "learning_rate": 1.7486533843857485e-05,
+      "loss": 0.4955,
+      "step": 20140
+    },
+    {
+      "epoch": 1.9551717446147876,
+      "grad_norm": 2.0789037317300876,
+      "learning_rate": 1.7470309559348435e-05,
+      "loss": 0.5176,
+      "step": 20150
+    },
+    {
+      "epoch": 1.956142053172909,
+      "grad_norm": 1.7669158731638315,
+      "learning_rate": 1.745408527483938e-05,
+      "loss": 0.4242,
+      "step": 20160
+    },
+    {
+      "epoch": 1.9571123617310304,
+      "grad_norm": 2.15917406316717,
+      "learning_rate": 1.7437860990330327e-05,
+      "loss": 0.457,
+      "step": 20170
+    },
+    {
+      "epoch": 1.9580826702891518,
+      "grad_norm": 1.7798687601178151,
+      "learning_rate": 1.7421636705821274e-05,
+      "loss": 0.467,
+      "step": 20180
+    },
+    {
+      "epoch": 1.9590529788472735,
+      "grad_norm": 2.2050151044492154,
+      "learning_rate": 1.740541242131222e-05,
+      "loss": 0.4562,
+      "step": 20190
+    },
+    {
+      "epoch": 1.9600232874053949,
+      "grad_norm": 1.7571600556165854,
+      "learning_rate": 1.7389188136803166e-05,
+      "loss": 0.4951,
+      "step": 20200
+    },
+    {
+      "epoch": 1.9609935959635165,
+      "grad_norm": 2.370912740726444,
+      "learning_rate": 1.7372963852294113e-05,
+      "loss": 0.5177,
+      "step": 20210
+    },
+    {
+      "epoch": 1.961963904521638,
+      "grad_norm": 1.9598530660426143,
+      "learning_rate": 1.735673956778506e-05,
+      "loss": 0.4898,
+      "step": 20220
+    },
+    {
+      "epoch": 1.9629342130797593,
+      "grad_norm": 1.8458335418058476,
+      "learning_rate": 1.7340515283276006e-05,
+      "loss": 0.4685,
+      "step": 20230
+    },
+    {
+      "epoch": 1.9639045216378808,
+      "grad_norm": 2.086603579490736,
+      "learning_rate": 1.7324290998766955e-05,
+      "loss": 0.4655,
+      "step": 20240
+    },
+    {
+      "epoch": 1.9648748301960022,
+      "grad_norm": 1.8861003799697185,
+      "learning_rate": 1.7308066714257902e-05,
+      "loss": 0.5012,
+      "step": 20250
+    },
+    {
+      "epoch": 1.9658451387541238,
+      "grad_norm": 2.0410888085121655,
+      "learning_rate": 1.729184242974885e-05,
+      "loss": 0.4686,
+      "step": 20260
+    },
+    {
+      "epoch": 1.9668154473122454,
+      "grad_norm": 2.0486563420463293,
+      "learning_rate": 1.7275618145239798e-05,
+      "loss": 0.4971,
+      "step": 20270
+    },
+    {
+      "epoch": 1.9677857558703669,
+      "grad_norm": 2.053946424813733,
+      "learning_rate": 1.7259393860730744e-05,
+      "loss": 0.5002,
+      "step": 20280
+    },
+    {
+      "epoch": 1.9687560644284883,
+      "grad_norm": 1.933061743356704,
+      "learning_rate": 1.724316957622169e-05,
+      "loss": 0.5072,
+      "step": 20290
+    },
+    {
+      "epoch": 1.9697263729866097,
+      "grad_norm": 2.066749720982141,
+      "learning_rate": 1.7226945291712637e-05,
+      "loss": 0.4928,
+      "step": 20300
+    },
+    {
+      "epoch": 1.970696681544731,
+      "grad_norm": 1.55984280015245,
+      "learning_rate": 1.7210721007203583e-05,
+      "loss": 0.4681,
+      "step": 20310
+    },
+    {
+      "epoch": 1.9716669901028527,
+      "grad_norm": 2.0658531156309254,
+      "learning_rate": 1.719449672269453e-05,
+      "loss": 0.4451,
+      "step": 20320
+    },
+    {
+      "epoch": 1.9726372986609741,
+      "grad_norm": 1.56362708496437,
+      "learning_rate": 1.7178272438185476e-05,
+      "loss": 0.5221,
+      "step": 20330
+    },
+    {
+      "epoch": 1.9736076072190958,
+      "grad_norm": 2.161774232597424,
+      "learning_rate": 1.7162048153676426e-05,
+      "loss": 0.5527,
+      "step": 20340
+    },
+    {
+      "epoch": 1.9745779157772172,
+      "grad_norm": 2.411930244708304,
+      "learning_rate": 1.7145823869167372e-05,
+      "loss": 0.4742,
+      "step": 20350
+    },
+    {
+      "epoch": 1.9755482243353386,
+      "grad_norm": 2.0274695839961803,
+      "learning_rate": 1.712959958465832e-05,
+      "loss": 0.5055,
+      "step": 20360
+    },
+    {
+      "epoch": 1.97651853289346,
+      "grad_norm": 1.9046814027988292,
+      "learning_rate": 1.7113375300149265e-05,
+      "loss": 0.5018,
+      "step": 20370
+    },
+    {
+      "epoch": 1.9774888414515817,
+      "grad_norm": 2.6745574351193437,
+      "learning_rate": 1.709715101564021e-05,
+      "loss": 0.5013,
+      "step": 20380
+    },
+    {
+      "epoch": 1.978459150009703,
+      "grad_norm": 1.983809422276857,
+      "learning_rate": 1.7080926731131158e-05,
+      "loss": 0.506,
+      "step": 20390
+    },
+    {
+      "epoch": 1.9794294585678247,
+      "grad_norm": 1.8313417501684919,
+      "learning_rate": 1.7064702446622104e-05,
+      "loss": 0.4913,
+      "step": 20400
+    },
+    {
+      "epoch": 1.9803997671259461,
+      "grad_norm": 1.9488632706180251,
+      "learning_rate": 1.704847816211305e-05,
+      "loss": 0.4753,
+      "step": 20410
+    },
+    {
+      "epoch": 1.9813700756840675,
+      "grad_norm": 2.1235067779704386,
+      "learning_rate": 1.7032253877603997e-05,
+      "loss": 0.5076,
+      "step": 20420
+    },
+    {
+      "epoch": 1.982340384242189,
+      "grad_norm": 1.9628606937438253,
+      "learning_rate": 1.7016029593094947e-05,
+      "loss": 0.4488,
+      "step": 20430
+    },
+    {
+      "epoch": 1.9833106928003104,
+      "grad_norm": 2.053405187246396,
+      "learning_rate": 1.6999805308585893e-05,
+      "loss": 0.4492,
+      "step": 20440
+    },
+    {
+      "epoch": 1.984281001358432,
+      "grad_norm": 2.027877392955927,
+      "learning_rate": 1.698358102407684e-05,
+      "loss": 0.4481,
+      "step": 20450
+    },
+    {
+      "epoch": 1.9852513099165536,
+      "grad_norm": 2.6424653212507585,
+      "learning_rate": 1.6967356739567786e-05,
+      "loss": 0.4866,
+      "step": 20460
+    },
+    {
+      "epoch": 1.986221618474675,
+      "grad_norm": 1.798153398364296,
+      "learning_rate": 1.6951132455058732e-05,
+      "loss": 0.4901,
+      "step": 20470
+    },
+    {
+      "epoch": 1.9871919270327965,
+      "grad_norm": 1.9375715199504966,
+      "learning_rate": 1.693490817054968e-05,
+      "loss": 0.5145,
+      "step": 20480
+    },
+    {
+      "epoch": 1.9881622355909179,
+      "grad_norm": 2.227384783436997,
+      "learning_rate": 1.6918683886040625e-05,
+      "loss": 0.5234,
+      "step": 20490
+    },
+    {
+      "epoch": 1.9891325441490393,
+      "grad_norm": 1.6406328006727198,
+      "learning_rate": 1.690245960153157e-05,
+      "loss": 0.4938,
+      "step": 20500
+    },
+    {
+      "epoch": 1.9891325441490393,
+      "eval_loss": 0.6277603507041931,
+      "eval_runtime": 3080.184,
+      "eval_samples_per_second": 0.582,
+      "eval_steps_per_second": 0.291,
+      "step": 20500
+    },
+    {
+      "epoch": 1.990102852707161,
+      "grad_norm": 2.189175771584087,
+      "learning_rate": 1.688623531702252e-05,
+      "loss": 0.4815,
+      "step": 20510
+    },
+    {
+      "epoch": 1.9910731612652823,
+      "grad_norm": 1.9037268932948403,
+      "learning_rate": 1.6870011032513468e-05,
+      "loss": 0.4675,
+      "step": 20520
+    },
+    {
+      "epoch": 1.992043469823404,
+      "grad_norm": 2.1370328259985096,
+      "learning_rate": 1.6853786748004414e-05,
+      "loss": 0.4696,
+      "step": 20530
+    },
+    {
+      "epoch": 1.9930137783815254,
+      "grad_norm": 2.407631343670518,
+      "learning_rate": 1.683756246349536e-05,
+      "loss": 0.4904,
+      "step": 20540
+    },
+    {
+      "epoch": 1.9939840869396468,
+      "grad_norm": 2.113566676023369,
+      "learning_rate": 1.6821338178986307e-05,
+      "loss": 0.4713,
+      "step": 20550
+    },
+    {
+      "epoch": 1.9949543954977682,
+      "grad_norm": 1.7998412815701594,
+      "learning_rate": 1.6805113894477253e-05,
+      "loss": 0.4638,
+      "step": 20560
+    },
+    {
+      "epoch": 1.9959247040558896,
+      "grad_norm": 2.129222373307167,
+      "learning_rate": 1.67888896099682e-05,
+      "loss": 0.4567,
+      "step": 20570
+    },
+    {
+      "epoch": 1.9968950126140113,
+      "grad_norm": 2.0439893320145197,
+      "learning_rate": 1.6772665325459146e-05,
+      "loss": 0.499,
+      "step": 20580
+    },
+    {
+      "epoch": 1.997865321172133,
+      "grad_norm": 1.996718548398611,
+      "learning_rate": 1.6756441040950092e-05,
+      "loss": 0.4657,
+      "step": 20590
+    },
+    {
+      "epoch": 1.9988356297302543,
+      "grad_norm": 1.8047941252901814,
+      "learning_rate": 1.6740216756441042e-05,
+      "loss": 0.4723,
+      "step": 20600
+    },
+    {
+      "epoch": 1.9998059382883757,
+      "grad_norm": 1.8452676210065997,
+      "learning_rate": 1.672399247193199e-05,
+      "loss": 0.4802,
+      "step": 20610
+    },
+    {
+      "epoch": 2.000776246846497,
+      "grad_norm": 2.2446601261564254,
+      "learning_rate": 1.6707768187422935e-05,
+      "loss": 0.501,
+      "step": 20620
+    },
+    {
+      "epoch": 2.0017465554046185,
+      "grad_norm": 1.6471605415361736,
+      "learning_rate": 1.669154390291388e-05,
+      "loss": 0.4735,
+      "step": 20630
+    },
+    {
+      "epoch": 2.00271686396274,
+      "grad_norm": 2.038536155314213,
+      "learning_rate": 1.667531961840483e-05,
+      "loss": 0.4575,
+      "step": 20640
+    },
+    {
+      "epoch": 2.003687172520862,
+      "grad_norm": 2.1416733149321834,
+      "learning_rate": 1.6659095333895777e-05,
+      "loss": 0.5152,
+      "step": 20650
+    },
+    {
+      "epoch": 2.0046574810789832,
+      "grad_norm": 1.9639254031456526,
+      "learning_rate": 1.6642871049386724e-05,
+      "loss": 0.4424,
+      "step": 20660
+    },
+    {
+      "epoch": 2.0056277896371046,
+      "grad_norm": 2.1201945824283777,
+      "learning_rate": 1.662664676487767e-05,
+      "loss": 0.5194,
+      "step": 20670
+    },
+    {
+      "epoch": 2.006598098195226,
+      "grad_norm": 2.0259029186156616,
+      "learning_rate": 1.6610422480368617e-05,
+      "loss": 0.5085,
+      "step": 20680
+    },
+    {
+      "epoch": 2.0075684067533475,
+      "grad_norm": 1.5035201214590004,
+      "learning_rate": 1.6594198195859563e-05,
+      "loss": 0.445,
+      "step": 20690
+    },
+    {
+      "epoch": 2.008538715311469,
+      "grad_norm": 1.655200868630425,
+      "learning_rate": 1.6577973911350513e-05,
+      "loss": 0.4247,
+      "step": 20700
+    },
+    {
+      "epoch": 2.0095090238695907,
+      "grad_norm": 2.1707069905375764,
+      "learning_rate": 1.656174962684146e-05,
+      "loss": 0.4783,
+      "step": 20710
+    },
+    {
+      "epoch": 2.010479332427712,
+      "grad_norm": 1.8212536925620644,
+      "learning_rate": 1.6545525342332405e-05,
+      "loss": 0.4972,
+      "step": 20720
+    },
+    {
+      "epoch": 2.0114496409858336,
+      "grad_norm": 2.126405409204561,
+      "learning_rate": 1.6529301057823352e-05,
+      "loss": 0.4878,
+      "step": 20730
+    },
+    {
+      "epoch": 2.012419949543955,
+      "grad_norm": 2.077712603387953,
+      "learning_rate": 1.6513076773314298e-05,
+      "loss": 0.4571,
+      "step": 20740
+    },
+    {
+      "epoch": 2.0133902581020764,
+      "grad_norm": 2.1605056660181345,
+      "learning_rate": 1.6496852488805245e-05,
+      "loss": 0.5012,
+      "step": 20750
+    },
+    {
+      "epoch": 2.014360566660198,
+      "grad_norm": 1.8021397618450556,
+      "learning_rate": 1.648062820429619e-05,
+      "loss": 0.5049,
+      "step": 20760
+    },
+    {
+      "epoch": 2.015330875218319,
+      "grad_norm": 1.8857630619437384,
+      "learning_rate": 1.6464403919787137e-05,
+      "loss": 0.4726,
+      "step": 20770
+    },
+    {
+      "epoch": 2.016301183776441,
+      "grad_norm": 1.7074209121094917,
+      "learning_rate": 1.6448179635278084e-05,
+      "loss": 0.4914,
+      "step": 20780
+    },
+    {
+      "epoch": 2.0172714923345625,
+      "grad_norm": 2.6019622648688614,
+      "learning_rate": 1.6431955350769033e-05,
+      "loss": 0.4239,
+      "step": 20790
+    },
+    {
+      "epoch": 2.018241800892684,
+      "grad_norm": 1.6879446237580993,
+      "learning_rate": 1.641573106625998e-05,
+      "loss": 0.4657,
+      "step": 20800
+    },
+    {
+      "epoch": 2.0192121094508053,
+      "grad_norm": 1.4603799114693456,
+      "learning_rate": 1.6399506781750926e-05,
+      "loss": 0.4503,
+      "step": 20810
+    },
+    {
+      "epoch": 2.0201824180089267,
+      "grad_norm": 1.948639651180534,
+      "learning_rate": 1.6383282497241873e-05,
+      "loss": 0.4544,
+      "step": 20820
+    },
+    {
+      "epoch": 2.021152726567048,
+      "grad_norm": 2.1458744835535617,
+      "learning_rate": 1.636705821273282e-05,
+      "loss": 0.4632,
+      "step": 20830
+    },
+    {
+      "epoch": 2.02212303512517,
+      "grad_norm": 2.3393805927580997,
+      "learning_rate": 1.6350833928223765e-05,
+      "loss": 0.4686,
+      "step": 20840
+    },
+    {
+      "epoch": 2.0230933436832914,
+      "grad_norm": 1.8274491944101148,
+      "learning_rate": 1.6334609643714712e-05,
+      "loss": 0.4509,
+      "step": 20850
+    },
+    {
+      "epoch": 2.024063652241413,
+      "grad_norm": 2.5507345485151767,
+      "learning_rate": 1.6318385359205658e-05,
+      "loss": 0.5145,
+      "step": 20860
+    },
+    {
+      "epoch": 2.0250339607995342,
+      "grad_norm": 1.878444628312939,
+      "learning_rate": 1.6302161074696608e-05,
+      "loss": 0.449,
+      "step": 20870
+    },
+    {
+      "epoch": 2.0260042693576557,
+      "grad_norm": 2.1096395212213235,
+      "learning_rate": 1.6285936790187554e-05,
+      "loss": 0.409,
+      "step": 20880
+    },
+    {
+      "epoch": 2.026974577915777,
+      "grad_norm": 1.9597465473748303,
+      "learning_rate": 1.62697125056785e-05,
+      "loss": 0.4909,
+      "step": 20890
+    },
+    {
+      "epoch": 2.0279448864738985,
+      "grad_norm": 1.4262456717146317,
+      "learning_rate": 1.6253488221169447e-05,
+      "loss": 0.4933,
+      "step": 20900
+    },
+    {
+      "epoch": 2.0289151950320203,
+      "grad_norm": 1.7254635997166883,
+      "learning_rate": 1.6237263936660393e-05,
+      "loss": 0.4581,
+      "step": 20910
+    },
+    {
+      "epoch": 2.0298855035901417,
+      "grad_norm": 1.8255091897283775,
+      "learning_rate": 1.622103965215134e-05,
+      "loss": 0.462,
+      "step": 20920
+    },
+    {
+      "epoch": 2.030855812148263,
+      "grad_norm": 1.592461465195459,
+      "learning_rate": 1.6204815367642286e-05,
+      "loss": 0.5343,
+      "step": 20930
+    },
+    {
+      "epoch": 2.0318261207063846,
+      "grad_norm": 2.4124278045088445,
+      "learning_rate": 1.6188591083133233e-05,
+      "loss": 0.4738,
+      "step": 20940
+    },
+    {
+      "epoch": 2.032796429264506,
+      "grad_norm": 1.4707790381425363,
+      "learning_rate": 1.617236679862418e-05,
+      "loss": 0.4183,
+      "step": 20950
+    },
+    {
+      "epoch": 2.0337667378226274,
+      "grad_norm": 1.7383346761795593,
+      "learning_rate": 1.615614251411513e-05,
+      "loss": 0.4351,
+      "step": 20960
+    },
+    {
+      "epoch": 2.0347370463807493,
+      "grad_norm": 2.292481354882268,
+      "learning_rate": 1.6139918229606075e-05,
+      "loss": 0.4294,
+      "step": 20970
+    },
+    {
+      "epoch": 2.0357073549388707,
+      "grad_norm": 2.158168642341278,
+      "learning_rate": 1.612369394509702e-05,
+      "loss": 0.4703,
+      "step": 20980
+    },
+    {
+      "epoch": 2.036677663496992,
+      "grad_norm": 2.283161751864384,
+      "learning_rate": 1.6107469660587968e-05,
+      "loss": 0.5128,
+      "step": 20990
+    },
+    {
+      "epoch": 2.0376479720551135,
+      "grad_norm": 2.125582882132147,
+      "learning_rate": 1.6091245376078914e-05,
+      "loss": 0.4652,
+      "step": 21000
+    },
+    {
+      "epoch": 2.0376479720551135,
+      "eval_loss": 0.6309866905212402,
+      "eval_runtime": 3136.9547,
+      "eval_samples_per_second": 0.571,
+      "eval_steps_per_second": 0.286,
+      "step": 21000
+    },
+    {
+      "epoch": 2.038618280613235,
+      "grad_norm": 1.9414092633891158,
+      "learning_rate": 1.607502109156986e-05,
+      "loss": 0.4451,
+      "step": 21010
+    },
+    {
+      "epoch": 2.0395885891713563,
+      "grad_norm": 2.0603594802374863,
+      "learning_rate": 1.6058796807060807e-05,
+      "loss": 0.4569,
+      "step": 21020
+    },
+    {
+      "epoch": 2.040558897729478,
+      "grad_norm": 1.6323777472844119,
+      "learning_rate": 1.6042572522551757e-05,
+      "loss": 0.4764,
+      "step": 21030
+    },
+    {
+      "epoch": 2.0415292062875996,
+      "grad_norm": 1.8941882942793062,
+      "learning_rate": 1.6026348238042703e-05,
+      "loss": 0.4514,
+      "step": 21040
+    },
+    {
+      "epoch": 2.042499514845721,
+      "grad_norm": 1.905361739266791,
+      "learning_rate": 1.601012395353365e-05,
+      "loss": 0.4681,
+      "step": 21050
+    },
+    {
+      "epoch": 2.0434698234038424,
+      "grad_norm": 2.357513508114439,
+      "learning_rate": 1.59938996690246e-05,
+      "loss": 0.5029,
+      "step": 21060
+    },
+    {
+      "epoch": 2.044440131961964,
+      "grad_norm": 2.045526641253892,
+      "learning_rate": 1.5977675384515546e-05,
+      "loss": 0.5031,
+      "step": 21070
+    },
+    {
+      "epoch": 2.0454104405200852,
+      "grad_norm": 1.8018214238085957,
+      "learning_rate": 1.5961451100006492e-05,
+      "loss": 0.475,
+      "step": 21080
+    },
+    {
+      "epoch": 2.0463807490782067,
+      "grad_norm": 2.157846458503379,
+      "learning_rate": 1.594522681549744e-05,
+      "loss": 0.4922,
+      "step": 21090
+    },
+    {
+      "epoch": 2.0473510576363285,
+      "grad_norm": 2.2821214882561773,
+      "learning_rate": 1.5929002530988385e-05,
+      "loss": 0.4361,
+      "step": 21100
+    },
+    {
+      "epoch": 2.04832136619445,
+      "grad_norm": 2.4410635323771395,
+      "learning_rate": 1.591277824647933e-05,
+      "loss": 0.4426,
+      "step": 21110
+    },
+    {
+      "epoch": 2.0492916747525713,
+      "grad_norm": 1.847883527949176,
+      "learning_rate": 1.5896553961970278e-05,
+      "loss": 0.4998,
+      "step": 21120
+    },
+    {
+      "epoch": 2.0502619833106928,
+      "grad_norm": 2.086929164640455,
+      "learning_rate": 1.5880329677461224e-05,
+      "loss": 0.5218,
+      "step": 21130
+    },
+    {
+      "epoch": 2.051232291868814,
+      "grad_norm": 1.8955619202400775,
+      "learning_rate": 1.5864105392952174e-05,
+      "loss": 0.519,
+      "step": 21140
+    },
+    {
+      "epoch": 2.0522026004269356,
+      "grad_norm": 2.43252621811095,
+      "learning_rate": 1.584788110844312e-05,
+      "loss": 0.4439,
+      "step": 21150
+    },
+    {
+      "epoch": 2.0531729089850574,
+      "grad_norm": 2.0827457082492464,
+      "learning_rate": 1.5831656823934067e-05,
+      "loss": 0.5047,
+      "step": 21160
+    },
+    {
+      "epoch": 2.054143217543179,
+      "grad_norm": 1.9212860192837349,
+      "learning_rate": 1.5815432539425013e-05,
+      "loss": 0.4655,
+      "step": 21170
+    },
+    {
+      "epoch": 2.0551135261013003,
+      "grad_norm": 2.090169452444841,
+      "learning_rate": 1.579920825491596e-05,
+      "loss": 0.4421,
+      "step": 21180
+    },
+    {
+      "epoch": 2.0560838346594217,
+      "grad_norm": 1.9503749019739738,
+      "learning_rate": 1.5782983970406906e-05,
+      "loss": 0.442,
+      "step": 21190
+    },
+    {
+      "epoch": 2.057054143217543,
+      "grad_norm": 1.724201210329928,
+      "learning_rate": 1.5766759685897852e-05,
+      "loss": 0.3953,
+      "step": 21200
+    },
+    {
+      "epoch": 2.0580244517756645,
+      "grad_norm": 2.081302027118628,
+      "learning_rate": 1.57505354013888e-05,
+      "loss": 0.4846,
+      "step": 21210
+    },
+    {
+      "epoch": 2.0589947603337864,
+      "grad_norm": 2.5592097491782546,
+      "learning_rate": 1.5734311116879745e-05,
+      "loss": 0.4428,
+      "step": 21220
+    },
+    {
+      "epoch": 2.059965068891908,
+      "grad_norm": 1.9807395474177856,
+      "learning_rate": 1.5718086832370695e-05,
+      "loss": 0.4843,
+      "step": 21230
+    },
+    {
+      "epoch": 2.060935377450029,
+      "grad_norm": 1.946678515294379,
+      "learning_rate": 1.570186254786164e-05,
+      "loss": 0.5123,
+      "step": 21240
+    },
+    {
+      "epoch": 2.0619056860081506,
+      "grad_norm": 1.704583779463954,
+      "learning_rate": 1.5685638263352587e-05,
+      "loss": 0.5018,
+      "step": 21250
+    },
+    {
+      "epoch": 2.062875994566272,
+      "grad_norm": 1.9509206749604142,
+      "learning_rate": 1.5669413978843534e-05,
+      "loss": 0.4156,
+      "step": 21260
+    },
+    {
+      "epoch": 2.0638463031243934,
+      "grad_norm": 2.0838442571499365,
+      "learning_rate": 1.565318969433448e-05,
+      "loss": 0.4632,
+      "step": 21270
+    },
+    {
+      "epoch": 2.064816611682515,
+      "grad_norm": 1.9349794015053616,
+      "learning_rate": 1.5636965409825427e-05,
+      "loss": 0.4493,
+      "step": 21280
+    },
+    {
+      "epoch": 2.0657869202406367,
+      "grad_norm": 2.188378390390559,
+      "learning_rate": 1.5620741125316373e-05,
+      "loss": 0.5279,
+      "step": 21290
+    },
+    {
+      "epoch": 2.066757228798758,
+      "grad_norm": 1.8328420572385127,
+      "learning_rate": 1.560451684080732e-05,
+      "loss": 0.4611,
+      "step": 21300
+    },
+    {
+      "epoch": 2.0677275373568795,
+      "grad_norm": 1.563376875838364,
+      "learning_rate": 1.5588292556298266e-05,
+      "loss": 0.4449,
+      "step": 21310
+    },
+    {
+      "epoch": 2.068697845915001,
+      "grad_norm": 2.231697202375159,
+      "learning_rate": 1.5572068271789215e-05,
+      "loss": 0.4763,
+      "step": 21320
+    },
+    {
+      "epoch": 2.0696681544731224,
+      "grad_norm": 2.116198927861082,
+      "learning_rate": 1.5555843987280162e-05,
+      "loss": 0.4313,
+      "step": 21330
+    },
+    {
+      "epoch": 2.0706384630312438,
+      "grad_norm": 2.3801041085422763,
+      "learning_rate": 1.5539619702771108e-05,
+      "loss": 0.4643,
+      "step": 21340
+    },
+    {
+      "epoch": 2.0716087715893656,
+      "grad_norm": 1.8571501672365118,
+      "learning_rate": 1.5523395418262055e-05,
+      "loss": 0.4423,
+      "step": 21350
+    },
+    {
+      "epoch": 2.072579080147487,
+      "grad_norm": 1.6559131513123913,
+      "learning_rate": 1.5507171133753e-05,
+      "loss": 0.4437,
+      "step": 21360
+    },
+    {
+      "epoch": 2.0735493887056085,
+      "grad_norm": 2.137893390544345,
+      "learning_rate": 1.5490946849243947e-05,
+      "loss": 0.4732,
+      "step": 21370
+    },
+    {
+      "epoch": 2.07451969726373,
+      "grad_norm": 1.4742723623250382,
+      "learning_rate": 1.5474722564734894e-05,
+      "loss": 0.412,
+      "step": 21380
+    },
+    {
+      "epoch": 2.0754900058218513,
+      "grad_norm": 2.1318829799319543,
+      "learning_rate": 1.545849828022584e-05,
+      "loss": 0.4916,
+      "step": 21390
+    },
+    {
+      "epoch": 2.0764603143799727,
+      "grad_norm": 2.307110089522299,
+      "learning_rate": 1.544227399571679e-05,
+      "loss": 0.4124,
+      "step": 21400
+    },
+    {
+      "epoch": 2.077430622938094,
+      "grad_norm": 2.3097275184207637,
+      "learning_rate": 1.5426049711207736e-05,
+      "loss": 0.4799,
+      "step": 21410
+    },
+    {
+      "epoch": 2.078400931496216,
+      "grad_norm": 1.5854013780070428,
+      "learning_rate": 1.5409825426698686e-05,
+      "loss": 0.4759,
+      "step": 21420
+    },
+    {
+      "epoch": 2.0793712400543374,
+      "grad_norm": 2.008961619138051,
+      "learning_rate": 1.5393601142189632e-05,
+      "loss": 0.4844,
+      "step": 21430
+    },
+    {
+      "epoch": 2.080341548612459,
+      "grad_norm": 1.848635541993877,
+      "learning_rate": 1.537737685768058e-05,
+      "loss": 0.4558,
+      "step": 21440
+    },
+    {
+      "epoch": 2.08131185717058,
+      "grad_norm": 2.3862164923195217,
+      "learning_rate": 1.5361152573171525e-05,
+      "loss": 0.4514,
+      "step": 21450
+    },
+    {
+      "epoch": 2.0822821657287016,
+      "grad_norm": 2.322912088388782,
+      "learning_rate": 1.534492828866247e-05,
+      "loss": 0.4507,
+      "step": 21460
+    },
+    {
+      "epoch": 2.083252474286823,
+      "grad_norm": 2.1765100761424074,
+      "learning_rate": 1.5328704004153418e-05,
+      "loss": 0.4114,
+      "step": 21470
+    },
+    {
+      "epoch": 2.084222782844945,
+      "grad_norm": 1.915448768034906,
+      "learning_rate": 1.5312479719644364e-05,
+      "loss": 0.4794,
+      "step": 21480
+    },
+    {
+      "epoch": 2.0851930914030663,
+      "grad_norm": 1.955422403883371,
+      "learning_rate": 1.529625543513531e-05,
+      "loss": 0.485,
+      "step": 21490
+    },
+    {
+      "epoch": 2.0861633999611877,
+      "grad_norm": 1.5866416015413762,
+      "learning_rate": 1.528003115062626e-05,
+      "loss": 0.4961,
+      "step": 21500
+    },
+    {
+      "epoch": 2.0861633999611877,
+      "eval_loss": 0.6304420232772827,
+      "eval_runtime": 3075.3205,
+      "eval_samples_per_second": 0.583,
+      "eval_steps_per_second": 0.291,
+      "step": 21500
+    },
+    {
+      "epoch": 2.087133708519309,
+      "grad_norm": 1.8961317659006691,
+      "learning_rate": 1.5263806866117207e-05,
+      "loss": 0.4803,
+      "step": 21510
+    },
+    {
+      "epoch": 2.0881040170774305,
+      "grad_norm": 2.1574061097476633,
+      "learning_rate": 1.5247582581608153e-05,
+      "loss": 0.4897,
+      "step": 21520
+    },
+    {
+      "epoch": 2.089074325635552,
+      "grad_norm": 2.080433640436673,
+      "learning_rate": 1.52313582970991e-05,
+      "loss": 0.4623,
+      "step": 21530
+    },
+    {
+      "epoch": 2.0900446341936734,
+      "grad_norm": 2.274765795002327,
+      "learning_rate": 1.5215134012590046e-05,
+      "loss": 0.5149,
+      "step": 21540
+    },
+    {
+      "epoch": 2.0910149427517952,
+      "grad_norm": 2.514598169769852,
+      "learning_rate": 1.5198909728080992e-05,
+      "loss": 0.4028,
+      "step": 21550
+    },
+    {
+      "epoch": 2.0919852513099166,
+      "grad_norm": 2.1324330512224057,
+      "learning_rate": 1.5182685443571939e-05,
+      "loss": 0.5213,
+      "step": 21560
+    },
+    {
+      "epoch": 2.092955559868038,
+      "grad_norm": 2.0829108139421106,
+      "learning_rate": 1.5166461159062887e-05,
+      "loss": 0.4338,
+      "step": 21570
+    },
+    {
+      "epoch": 2.0939258684261595,
+      "grad_norm": 1.9606921161193192,
+      "learning_rate": 1.5150236874553833e-05,
+      "loss": 0.5347,
+      "step": 21580
+    },
+    {
+      "epoch": 2.094896176984281,
+      "grad_norm": 1.9007868794459526,
+      "learning_rate": 1.513401259004478e-05,
+      "loss": 0.4698,
+      "step": 21590
+    },
+    {
+      "epoch": 2.0958664855424023,
+      "grad_norm": 2.2070560808992385,
+      "learning_rate": 1.5117788305535726e-05,
+      "loss": 0.4992,
+      "step": 21600
+    },
+    {
+      "epoch": 2.096836794100524,
+      "grad_norm": 1.7052502823407865,
+      "learning_rate": 1.5101564021026674e-05,
+      "loss": 0.4448,
+      "step": 21610
+    },
+    {
+      "epoch": 2.0978071026586456,
+      "grad_norm": 1.9838818965293705,
+      "learning_rate": 1.508533973651762e-05,
+      "loss": 0.452,
+      "step": 21620
+    },
+    {
+      "epoch": 2.098777411216767,
+      "grad_norm": 2.361870038528201,
+      "learning_rate": 1.5069115452008567e-05,
+      "loss": 0.4772,
+      "step": 21630
+    },
+    {
+      "epoch": 2.0997477197748884,
+      "grad_norm": 2.3190708510226767,
+      "learning_rate": 1.5052891167499513e-05,
+      "loss": 0.4732,
+      "step": 21640
+    },
+    {
+      "epoch": 2.10071802833301,
+      "grad_norm": 2.0341040427547967,
+      "learning_rate": 1.5036666882990461e-05,
+      "loss": 0.5013,
+      "step": 21650
+    },
+    {
+      "epoch": 2.101688336891131,
+      "grad_norm": 2.2854031285851337,
+      "learning_rate": 1.5020442598481408e-05,
+      "loss": 0.4587,
+      "step": 21660
+    },
+    {
+      "epoch": 2.102658645449253,
+      "grad_norm": 2.171534197336386,
+      "learning_rate": 1.5004218313972354e-05,
+      "loss": 0.4685,
+      "step": 21670
+    },
+    {
+      "epoch": 2.1036289540073745,
+      "grad_norm": 2.237731272885117,
+      "learning_rate": 1.49879940294633e-05,
+      "loss": 0.4714,
+      "step": 21680
+    },
+    {
+      "epoch": 2.104599262565496,
+      "grad_norm": 1.83625835750722,
+      "learning_rate": 1.4971769744954247e-05,
+      "loss": 0.4858,
+      "step": 21690
+    },
+    {
+      "epoch": 2.1055695711236173,
+      "grad_norm": 2.1226910702010935,
+      "learning_rate": 1.4955545460445195e-05,
+      "loss": 0.4735,
+      "step": 21700
+    },
+    {
+      "epoch": 2.1065398796817387,
+      "grad_norm": 2.4412848671448764,
+      "learning_rate": 1.4939321175936141e-05,
+      "loss": 0.4955,
+      "step": 21710
+    },
+    {
+      "epoch": 2.10751018823986,
+      "grad_norm": 2.171686191103959,
+      "learning_rate": 1.4923096891427088e-05,
+      "loss": 0.4882,
+      "step": 21720
+    },
+    {
+      "epoch": 2.1084804967979816,
+      "grad_norm": 1.9028000906788562,
+      "learning_rate": 1.4906872606918034e-05,
+      "loss": 0.5008,
+      "step": 21730
+    },
+    {
+      "epoch": 2.1094508053561034,
+      "grad_norm": 2.032206148469331,
+      "learning_rate": 1.4890648322408982e-05,
+      "loss": 0.4688,
+      "step": 21740
+    },
+    {
+      "epoch": 2.110421113914225,
+      "grad_norm": 2.2254255720592457,
+      "learning_rate": 1.4874424037899929e-05,
+      "loss": 0.434,
+      "step": 21750
+    },
+    {
+      "epoch": 2.1113914224723462,
+      "grad_norm": 2.013615294714695,
+      "learning_rate": 1.4858199753390875e-05,
+      "loss": 0.4429,
+      "step": 21760
+    },
+    {
+      "epoch": 2.1123617310304676,
+      "grad_norm": 1.625353905024779,
+      "learning_rate": 1.4841975468881821e-05,
+      "loss": 0.4197,
+      "step": 21770
+    },
+    {
+      "epoch": 2.113332039588589,
+      "grad_norm": 2.6956426206670074,
+      "learning_rate": 1.482575118437277e-05,
+      "loss": 0.4412,
+      "step": 21780
+    },
+    {
+      "epoch": 2.1143023481467105,
+      "grad_norm": 1.642670039289611,
+      "learning_rate": 1.4809526899863716e-05,
+      "loss": 0.4365,
+      "step": 21790
+    },
+    {
+      "epoch": 2.1152726567048323,
+      "grad_norm": 2.047556877474702,
+      "learning_rate": 1.4793302615354662e-05,
+      "loss": 0.4882,
+      "step": 21800
+    },
+    {
+      "epoch": 2.1162429652629537,
+      "grad_norm": 2.666095406948268,
+      "learning_rate": 1.4777078330845612e-05,
+      "loss": 0.4885,
+      "step": 21810
+    },
+    {
+      "epoch": 2.117213273821075,
+      "grad_norm": 2.0148741202390736,
+      "learning_rate": 1.4760854046336558e-05,
+      "loss": 0.5092,
+      "step": 21820
+    },
+    {
+      "epoch": 2.1181835823791966,
+      "grad_norm": 1.5220212421711388,
+      "learning_rate": 1.4744629761827505e-05,
+      "loss": 0.5007,
+      "step": 21830
+    },
+    {
+      "epoch": 2.119153890937318,
+      "grad_norm": 2.2566622563684446,
+      "learning_rate": 1.4728405477318453e-05,
+      "loss": 0.4386,
+      "step": 21840
+    },
+    {
+      "epoch": 2.1201241994954394,
+      "grad_norm": 2.13286935915719,
+      "learning_rate": 1.47121811928094e-05,
+      "loss": 0.4559,
+      "step": 21850
+    },
+    {
+      "epoch": 2.1210945080535613,
+      "grad_norm": 1.8220197546440986,
+      "learning_rate": 1.4695956908300346e-05,
+      "loss": 0.451,
+      "step": 21860
+    },
+    {
+      "epoch": 2.1220648166116827,
+      "grad_norm": 2.2468599066846653,
+      "learning_rate": 1.4679732623791292e-05,
+      "loss": 0.446,
+      "step": 21870
+    },
+    {
+      "epoch": 2.123035125169804,
+      "grad_norm": 1.8918874944552218,
+      "learning_rate": 1.466350833928224e-05,
+      "loss": 0.459,
+      "step": 21880
+    },
+    {
+      "epoch": 2.1240054337279255,
+      "grad_norm": 2.247255412788798,
+      "learning_rate": 1.4647284054773186e-05,
+      "loss": 0.5151,
+      "step": 21890
+    },
+    {
+      "epoch": 2.124975742286047,
+      "grad_norm": 1.8291853119382993,
+      "learning_rate": 1.4631059770264133e-05,
+      "loss": 0.4557,
+      "step": 21900
+    },
+    {
+      "epoch": 2.1259460508441683,
+      "grad_norm": 2.2000080164154654,
+      "learning_rate": 1.461483548575508e-05,
+      "loss": 0.5106,
+      "step": 21910
+    },
+    {
+      "epoch": 2.1269163594022897,
+      "grad_norm": 1.608928034740487,
+      "learning_rate": 1.4598611201246026e-05,
+      "loss": 0.4696,
+      "step": 21920
+    },
+    {
+      "epoch": 2.1278866679604116,
+      "grad_norm": 1.7796242187219558,
+      "learning_rate": 1.4582386916736974e-05,
+      "loss": 0.4372,
+      "step": 21930
+    },
+    {
+      "epoch": 2.128856976518533,
+      "grad_norm": 1.9355986409124974,
+      "learning_rate": 1.456616263222792e-05,
+      "loss": 0.4601,
+      "step": 21940
+    },
+    {
+      "epoch": 2.1298272850766544,
+      "grad_norm": 2.5044478412060776,
+      "learning_rate": 1.4549938347718866e-05,
+      "loss": 0.5153,
+      "step": 21950
+    },
+    {
+      "epoch": 2.130797593634776,
+      "grad_norm": 2.1644310902185264,
+      "learning_rate": 1.4533714063209813e-05,
+      "loss": 0.4755,
+      "step": 21960
+    },
+    {
+      "epoch": 2.1317679021928972,
+      "grad_norm": 1.9759401921373174,
+      "learning_rate": 1.451748977870076e-05,
+      "loss": 0.5306,
+      "step": 21970
+    },
+    {
+      "epoch": 2.1327382107510187,
+      "grad_norm": 2.5662041065947827,
+      "learning_rate": 1.4501265494191707e-05,
+      "loss": 0.4936,
+      "step": 21980
+    },
+    {
+      "epoch": 2.13370851930914,
+      "grad_norm": 1.834252286116916,
+      "learning_rate": 1.4485041209682654e-05,
+      "loss": 0.4248,
+      "step": 21990
+    },
+    {
+      "epoch": 2.134678827867262,
+      "grad_norm": 2.048099727707521,
+      "learning_rate": 1.44688169251736e-05,
+      "loss": 0.4573,
+      "step": 22000
+    },
+    {
+      "epoch": 2.134678827867262,
+      "eval_loss": 0.6303107142448425,
+      "eval_runtime": 3417.8488,
+      "eval_samples_per_second": 0.524,
+      "eval_steps_per_second": 0.262,
+      "step": 22000
+    },
+    {
+      "epoch": 2.1356491364253833,
+      "grad_norm": 2.032462630747061,
+      "learning_rate": 1.4452592640664548e-05,
+      "loss": 0.4496,
+      "step": 22010
+    },
+    {
+      "epoch": 2.1366194449835048,
+      "grad_norm": 2.0538977423462548,
+      "learning_rate": 1.4436368356155494e-05,
+      "loss": 0.4512,
+      "step": 22020
+    },
+    {
+      "epoch": 2.137589753541626,
+      "grad_norm": 2.0559328662146403,
+      "learning_rate": 1.442014407164644e-05,
+      "loss": 0.4469,
+      "step": 22030
+    },
+    {
+      "epoch": 2.1385600620997476,
+      "grad_norm": 1.8412343311132753,
+      "learning_rate": 1.4403919787137387e-05,
+      "loss": 0.449,
+      "step": 22040
+    },
+    {
+      "epoch": 2.139530370657869,
+      "grad_norm": 2.0513694288656765,
+      "learning_rate": 1.4387695502628334e-05,
+      "loss": 0.4709,
+      "step": 22050
+    },
+    {
+      "epoch": 2.140500679215991,
+      "grad_norm": 1.7890079195029862,
+      "learning_rate": 1.4371471218119282e-05,
+      "loss": 0.4419,
+      "step": 22060
+    },
+    {
+      "epoch": 2.1414709877741123,
+      "grad_norm": 2.535357630446327,
+      "learning_rate": 1.4355246933610228e-05,
+      "loss": 0.4305,
+      "step": 22070
+    },
+    {
+      "epoch": 2.1424412963322337,
+      "grad_norm": 2.2977476166849957,
+      "learning_rate": 1.4339022649101174e-05,
+      "loss": 0.4664,
+      "step": 22080
+    },
+    {
+      "epoch": 2.143411604890355,
+      "grad_norm": 1.8778176122292005,
+      "learning_rate": 1.432279836459212e-05,
+      "loss": 0.4321,
+      "step": 22090
+    },
+    {
+      "epoch": 2.1443819134484765,
+      "grad_norm": 2.2740974899219526,
+      "learning_rate": 1.4306574080083069e-05,
+      "loss": 0.4582,
+      "step": 22100
+    },
+    {
+      "epoch": 2.145352222006598,
+      "grad_norm": 1.7411166330088808,
+      "learning_rate": 1.4290349795574015e-05,
+      "loss": 0.4703,
+      "step": 22110
+    },
+    {
+      "epoch": 2.1463225305647198,
+      "grad_norm": 2.270025807621109,
+      "learning_rate": 1.4274125511064962e-05,
+      "loss": 0.462,
+      "step": 22120
+    },
+    {
+      "epoch": 2.147292839122841,
+      "grad_norm": 2.047404098376008,
+      "learning_rate": 1.4257901226555908e-05,
+      "loss": 0.4572,
+      "step": 22130
+    },
+    {
+      "epoch": 2.1482631476809626,
+      "grad_norm": 1.9025461949345281,
+      "learning_rate": 1.4241676942046856e-05,
+      "loss": 0.4523,
+      "step": 22140
+    },
+    {
+      "epoch": 2.149233456239084,
+      "grad_norm": 2.41657182633304,
+      "learning_rate": 1.4225452657537803e-05,
+      "loss": 0.4485,
+      "step": 22150
+    },
+    {
+      "epoch": 2.1502037647972054,
+      "grad_norm": 2.1984178815028246,
+      "learning_rate": 1.4209228373028749e-05,
+      "loss": 0.4744,
+      "step": 22160
+    },
+    {
+      "epoch": 2.151174073355327,
+      "grad_norm": 2.1671082483900044,
+      "learning_rate": 1.4193004088519695e-05,
+      "loss": 0.4277,
+      "step": 22170
+    },
+    {
+      "epoch": 2.1521443819134483,
+      "grad_norm": 2.2526638629347118,
+      "learning_rate": 1.4176779804010643e-05,
+      "loss": 0.4078,
+      "step": 22180
+    },
+    {
+      "epoch": 2.15311469047157,
+      "grad_norm": 2.4484821608494665,
+      "learning_rate": 1.416055551950159e-05,
+      "loss": 0.4726,
+      "step": 22190
+    },
+    {
+      "epoch": 2.1540849990296915,
+      "grad_norm": 2.145545596752741,
+      "learning_rate": 1.414433123499254e-05,
+      "loss": 0.5181,
+      "step": 22200
+    },
+    {
+      "epoch": 2.155055307587813,
+      "grad_norm": 2.2593836123374,
+      "learning_rate": 1.4128106950483486e-05,
+      "loss": 0.4986,
+      "step": 22210
+    },
+    {
+      "epoch": 2.1560256161459344,
+      "grad_norm": 1.8007881267775863,
+      "learning_rate": 1.4111882665974432e-05,
+      "loss": 0.4483,
+      "step": 22220
+    },
+    {
+      "epoch": 2.1569959247040558,
+      "grad_norm": 1.348362904445229,
+      "learning_rate": 1.4095658381465379e-05,
+      "loss": 0.4637,
+      "step": 22230
+    },
+    {
+      "epoch": 2.157966233262177,
+      "grad_norm": 2.1469770203677445,
+      "learning_rate": 1.4079434096956327e-05,
+      "loss": 0.494,
+      "step": 22240
+    },
+    {
+      "epoch": 2.158936541820299,
+      "grad_norm": 2.337402150331418,
+      "learning_rate": 1.4063209812447273e-05,
+      "loss": 0.4761,
+      "step": 22250
+    },
+    {
+      "epoch": 2.1599068503784205,
+      "grad_norm": 1.7961706912599362,
+      "learning_rate": 1.404698552793822e-05,
+      "loss": 0.4683,
+      "step": 22260
+    },
+    {
+      "epoch": 2.160877158936542,
+      "grad_norm": 1.6086385155901042,
+      "learning_rate": 1.4030761243429166e-05,
+      "loss": 0.4305,
+      "step": 22270
+    },
+    {
+      "epoch": 2.1618474674946633,
+      "grad_norm": 1.853121432266295,
+      "learning_rate": 1.4014536958920112e-05,
+      "loss": 0.4808,
+      "step": 22280
+    },
+    {
+      "epoch": 2.1628177760527847,
+      "grad_norm": 2.1594671768519094,
+      "learning_rate": 1.399831267441106e-05,
+      "loss": 0.4279,
+      "step": 22290
+    },
+    {
+      "epoch": 2.163788084610906,
+      "grad_norm": 1.8445913886566945,
+      "learning_rate": 1.3982088389902007e-05,
+      "loss": 0.4905,
+      "step": 22300
+    },
+    {
+      "epoch": 2.164758393169028,
+      "grad_norm": 1.8975990263034965,
+      "learning_rate": 1.3965864105392953e-05,
+      "loss": 0.4888,
+      "step": 22310
+    },
+    {
+      "epoch": 2.1657287017271494,
+      "grad_norm": 2.015470924856379,
+      "learning_rate": 1.39496398208839e-05,
+      "loss": 0.5232,
+      "step": 22320
+    },
+    {
+      "epoch": 2.166699010285271,
+      "grad_norm": 2.4943071641031698,
+      "learning_rate": 1.3933415536374848e-05,
+      "loss": 0.4827,
+      "step": 22330
+    },
+    {
+      "epoch": 2.167669318843392,
+      "grad_norm": 1.5900299707664105,
+      "learning_rate": 1.3917191251865794e-05,
+      "loss": 0.4963,
+      "step": 22340
+    },
+    {
+      "epoch": 2.1686396274015136,
+      "grad_norm": 1.8705765102756569,
+      "learning_rate": 1.390096696735674e-05,
+      "loss": 0.4389,
+      "step": 22350
+    },
+    {
+      "epoch": 2.169609935959635,
+      "grad_norm": 1.9703821664953534,
+      "learning_rate": 1.3884742682847687e-05,
+      "loss": 0.3945,
+      "step": 22360
+    },
+    {
+      "epoch": 2.1705802445177564,
+      "grad_norm": 1.6647445519639283,
+      "learning_rate": 1.3868518398338635e-05,
+      "loss": 0.4163,
+      "step": 22370
+    },
+    {
+      "epoch": 2.1715505530758783,
+      "grad_norm": 1.617333797152806,
+      "learning_rate": 1.3852294113829581e-05,
+      "loss": 0.4943,
+      "step": 22380
+    },
+    {
+      "epoch": 2.1725208616339997,
+      "grad_norm": 2.1150619610025796,
+      "learning_rate": 1.3836069829320528e-05,
+      "loss": 0.4941,
+      "step": 22390
+    },
+    {
+      "epoch": 2.173491170192121,
+      "grad_norm": 2.70908196165092,
+      "learning_rate": 1.3819845544811474e-05,
+      "loss": 0.448,
+      "step": 22400
+    },
+    {
+      "epoch": 2.1744614787502425,
+      "grad_norm": 2.195631441394506,
+      "learning_rate": 1.3803621260302422e-05,
+      "loss": 0.4611,
+      "step": 22410
+    },
+    {
+      "epoch": 2.175431787308364,
+      "grad_norm": 1.602301760768748,
+      "learning_rate": 1.3787396975793368e-05,
+      "loss": 0.4837,
+      "step": 22420
+    },
+    {
+      "epoch": 2.1764020958664854,
+      "grad_norm": 1.7676765995955546,
+      "learning_rate": 1.3771172691284315e-05,
+      "loss": 0.4798,
+      "step": 22430
+    },
+    {
+      "epoch": 2.177372404424607,
+      "grad_norm": 2.211793623579749,
+      "learning_rate": 1.3754948406775261e-05,
+      "loss": 0.4607,
+      "step": 22440
+    },
+    {
+      "epoch": 2.1783427129827286,
+      "grad_norm": 1.9526273269630037,
+      "learning_rate": 1.3738724122266208e-05,
+      "loss": 0.4638,
+      "step": 22450
+    },
+    {
+      "epoch": 2.17931302154085,
+      "grad_norm": 2.063947829078263,
+      "learning_rate": 1.3722499837757156e-05,
+      "loss": 0.4548,
+      "step": 22460
+    },
+    {
+      "epoch": 2.1802833300989715,
+      "grad_norm": 2.127208435257803,
+      "learning_rate": 1.3706275553248102e-05,
+      "loss": 0.4332,
+      "step": 22470
+    },
+    {
+      "epoch": 2.181253638657093,
+      "grad_norm": 2.098766619473584,
+      "learning_rate": 1.3690051268739048e-05,
+      "loss": 0.4822,
+      "step": 22480
+    },
+    {
+      "epoch": 2.1822239472152143,
+      "grad_norm": 1.5960382905615234,
+      "learning_rate": 1.3673826984229995e-05,
+      "loss": 0.4394,
+      "step": 22490
+    },
+    {
+      "epoch": 2.183194255773336,
+      "grad_norm": 2.5521137154858087,
+      "learning_rate": 1.3657602699720943e-05,
+      "loss": 0.4568,
+      "step": 22500
+    },
+    {
+      "epoch": 2.183194255773336,
+      "eval_loss": 0.6301902532577515,
+      "eval_runtime": 3074.8498,
+      "eval_samples_per_second": 0.583,
+      "eval_steps_per_second": 0.291,
+      "step": 22500
+    },
+    {
+      "epoch": 2.1841645643314576,
+      "grad_norm": 1.8964546731319356,
+      "learning_rate": 1.364137841521189e-05,
+      "loss": 0.4569,
+      "step": 22510
+    },
+    {
+      "epoch": 2.185134872889579,
+      "grad_norm": 2.1836501057996918,
+      "learning_rate": 1.3625154130702836e-05,
+      "loss": 0.4497,
+      "step": 22520
+    },
+    {
+      "epoch": 2.1861051814477004,
+      "grad_norm": 1.6786624803632986,
+      "learning_rate": 1.3608929846193782e-05,
+      "loss": 0.4229,
+      "step": 22530
+    },
+    {
+      "epoch": 2.187075490005822,
+      "grad_norm": 2.1475126662304675,
+      "learning_rate": 1.359270556168473e-05,
+      "loss": 0.4412,
+      "step": 22540
+    },
+    {
+      "epoch": 2.188045798563943,
+      "grad_norm": 2.322023881303393,
+      "learning_rate": 1.3576481277175676e-05,
+      "loss": 0.4235,
+      "step": 22550
+    },
+    {
+      "epoch": 2.1890161071220646,
+      "grad_norm": 2.208490633488266,
+      "learning_rate": 1.3560256992666623e-05,
+      "loss": 0.4997,
+      "step": 22560
+    },
+    {
+      "epoch": 2.1899864156801865,
+      "grad_norm": 1.869675214539207,
+      "learning_rate": 1.354403270815757e-05,
+      "loss": 0.4635,
+      "step": 22570
+    },
+    {
+      "epoch": 2.190956724238308,
+      "grad_norm": 1.7398189038479703,
+      "learning_rate": 1.3527808423648516e-05,
+      "loss": 0.4128,
+      "step": 22580
+    },
+    {
+      "epoch": 2.1919270327964293,
+      "grad_norm": 2.266171424803214,
+      "learning_rate": 1.3511584139139465e-05,
+      "loss": 0.4651,
+      "step": 22590
+    },
+    {
+      "epoch": 2.1928973413545507,
+      "grad_norm": 2.3614429446369063,
+      "learning_rate": 1.3495359854630413e-05,
+      "loss": 0.4123,
+      "step": 22600
+    },
+    {
+      "epoch": 2.193867649912672,
+      "grad_norm": 1.724551648002317,
+      "learning_rate": 1.347913557012136e-05,
+      "loss": 0.4481,
+      "step": 22610
+    },
+    {
+      "epoch": 2.1948379584707935,
+      "grad_norm": 2.0558570962807945,
+      "learning_rate": 1.3462911285612306e-05,
+      "loss": 0.4431,
+      "step": 22620
+    },
+    {
+      "epoch": 2.195808267028915,
+      "grad_norm": 1.943776934136987,
+      "learning_rate": 1.3446687001103253e-05,
+      "loss": 0.5128,
+      "step": 22630
+    },
+    {
+      "epoch": 2.196778575587037,
+      "grad_norm": 1.958173689925549,
+      "learning_rate": 1.34304627165942e-05,
+      "loss": 0.4628,
+      "step": 22640
+    },
+    {
+      "epoch": 2.1977488841451582,
+      "grad_norm": 1.990366090746634,
+      "learning_rate": 1.3414238432085147e-05,
+      "loss": 0.459,
+      "step": 22650
+    },
+    {
+      "epoch": 2.1987191927032796,
+      "grad_norm": 1.8373680505548435,
+      "learning_rate": 1.3398014147576093e-05,
+      "loss": 0.4397,
+      "step": 22660
+    },
+    {
+      "epoch": 2.199689501261401,
+      "grad_norm": 1.8078209399386682,
+      "learning_rate": 1.338178986306704e-05,
+      "loss": 0.5215,
+      "step": 22670
+    },
+    {
+      "epoch": 2.2006598098195225,
+      "grad_norm": 2.400832940310904,
+      "learning_rate": 1.3365565578557986e-05,
+      "loss": 0.4926,
+      "step": 22680
+    },
+    {
+      "epoch": 2.201630118377644,
+      "grad_norm": 2.055045167286068,
+      "learning_rate": 1.3349341294048934e-05,
+      "loss": 0.4461,
+      "step": 22690
+    },
+    {
+      "epoch": 2.2026004269357657,
+      "grad_norm": 1.8144420639570567,
+      "learning_rate": 1.333311700953988e-05,
+      "loss": 0.4112,
+      "step": 22700
+    },
+    {
+      "epoch": 2.203570735493887,
+      "grad_norm": 2.391519552042341,
+      "learning_rate": 1.3316892725030827e-05,
+      "loss": 0.439,
+      "step": 22710
+    },
+    {
+      "epoch": 2.2045410440520086,
+      "grad_norm": 1.563384599944884,
+      "learning_rate": 1.3300668440521773e-05,
+      "loss": 0.4694,
+      "step": 22720
+    },
+    {
+      "epoch": 2.20551135261013,
+      "grad_norm": 2.3546727224779715,
+      "learning_rate": 1.3284444156012721e-05,
+      "loss": 0.4355,
+      "step": 22730
+    },
+    {
+      "epoch": 2.2064816611682514,
+      "grad_norm": 1.832975462434467,
+      "learning_rate": 1.3268219871503668e-05,
+      "loss": 0.4984,
+      "step": 22740
+    },
+    {
+      "epoch": 2.207451969726373,
+      "grad_norm": 1.7975073035604554,
+      "learning_rate": 1.3251995586994614e-05,
+      "loss": 0.4584,
+      "step": 22750
+    },
+    {
+      "epoch": 2.2084222782844947,
+      "grad_norm": 2.243306278714619,
+      "learning_rate": 1.323577130248556e-05,
+      "loss": 0.4365,
+      "step": 22760
+    },
+    {
+      "epoch": 2.209392586842616,
+      "grad_norm": 2.0297904158946043,
+      "learning_rate": 1.3219547017976509e-05,
+      "loss": 0.5148,
+      "step": 22770
+    },
+    {
+      "epoch": 2.2103628954007375,
+      "grad_norm": 1.6322562865594754,
+      "learning_rate": 1.3203322733467455e-05,
+      "loss": 0.4782,
+      "step": 22780
+    },
+    {
+      "epoch": 2.211333203958859,
+      "grad_norm": 2.277603717413092,
+      "learning_rate": 1.3187098448958401e-05,
+      "loss": 0.4899,
+      "step": 22790
+    },
+    {
+      "epoch": 2.2123035125169803,
+      "grad_norm": 1.2452738117030233,
+      "learning_rate": 1.3170874164449348e-05,
+      "loss": 0.4459,
+      "step": 22800
+    },
+    {
+      "epoch": 2.2132738210751017,
+      "grad_norm": 2.3624119652920177,
+      "learning_rate": 1.3154649879940294e-05,
+      "loss": 0.5033,
+      "step": 22810
+    },
+    {
+      "epoch": 2.214244129633223,
+      "grad_norm": 1.6861367814447958,
+      "learning_rate": 1.3138425595431242e-05,
+      "loss": 0.4434,
+      "step": 22820
+    },
+    {
+      "epoch": 2.215214438191345,
+      "grad_norm": 2.5489904753238166,
+      "learning_rate": 1.3122201310922189e-05,
+      "loss": 0.4465,
+      "step": 22830
+    },
+    {
+      "epoch": 2.2161847467494664,
+      "grad_norm": 1.5652883952842678,
+      "learning_rate": 1.3105977026413135e-05,
+      "loss": 0.4648,
+      "step": 22840
+    },
+    {
+      "epoch": 2.217155055307588,
+      "grad_norm": 2.106609892171019,
+      "learning_rate": 1.3089752741904081e-05,
+      "loss": 0.4809,
+      "step": 22850
+    },
+    {
+      "epoch": 2.2181253638657092,
+      "grad_norm": 2.7417498454164653,
+      "learning_rate": 1.307352845739503e-05,
+      "loss": 0.4652,
+      "step": 22860
+    },
+    {
+      "epoch": 2.2190956724238307,
+      "grad_norm": 2.0375355388485645,
+      "learning_rate": 1.3057304172885976e-05,
+      "loss": 0.499,
+      "step": 22870
+    },
+    {
+      "epoch": 2.220065980981952,
+      "grad_norm": 1.6341471214683112,
+      "learning_rate": 1.3041079888376922e-05,
+      "loss": 0.4935,
+      "step": 22880
+    },
+    {
+      "epoch": 2.221036289540074,
+      "grad_norm": 2.3570884682492346,
+      "learning_rate": 1.3024855603867869e-05,
+      "loss": 0.4884,
+      "step": 22890
+    },
+    {
+      "epoch": 2.2220065980981953,
+      "grad_norm": 1.9493074304368556,
+      "learning_rate": 1.3008631319358817e-05,
+      "loss": 0.4838,
+      "step": 22900
+    },
+    {
+      "epoch": 2.2229769066563168,
+      "grad_norm": 2.1315651431097224,
+      "learning_rate": 1.2992407034849763e-05,
+      "loss": 0.4978,
+      "step": 22910
+    },
+    {
+      "epoch": 2.223947215214438,
+      "grad_norm": 1.8843986392844572,
+      "learning_rate": 1.297618275034071e-05,
+      "loss": 0.4922,
+      "step": 22920
+    },
+    {
+      "epoch": 2.2249175237725596,
+      "grad_norm": 2.0055582988037313,
+      "learning_rate": 1.2959958465831656e-05,
+      "loss": 0.4702,
+      "step": 22930
+    },
+    {
+      "epoch": 2.225887832330681,
+      "grad_norm": 1.8513569200366902,
+      "learning_rate": 1.2943734181322604e-05,
+      "loss": 0.467,
+      "step": 22940
+    },
+    {
+      "epoch": 2.226858140888803,
+      "grad_norm": 2.3018937212184682,
+      "learning_rate": 1.292750989681355e-05,
+      "loss": 0.4632,
+      "step": 22950
+    },
+    {
+      "epoch": 2.2278284494469243,
+      "grad_norm": 2.1028486436310905,
+      "learning_rate": 1.2911285612304497e-05,
+      "loss": 0.4329,
+      "step": 22960
+    },
+    {
+      "epoch": 2.2287987580050457,
+      "grad_norm": 2.3891744137342994,
+      "learning_rate": 1.2895061327795443e-05,
+      "loss": 0.4549,
+      "step": 22970
+    },
+    {
+      "epoch": 2.229769066563167,
+      "grad_norm": 1.8338746725123478,
+      "learning_rate": 1.2878837043286393e-05,
+      "loss": 0.4518,
+      "step": 22980
+    },
+    {
+      "epoch": 2.2307393751212885,
+      "grad_norm": 2.5499388804291483,
+      "learning_rate": 1.286261275877734e-05,
+      "loss": 0.4852,
+      "step": 22990
+    },
+    {
+      "epoch": 2.23170968367941,
+      "grad_norm": 1.6648341489250056,
+      "learning_rate": 1.2846388474268287e-05,
+      "loss": 0.4266,
+      "step": 23000
+    },
+    {
+      "epoch": 2.23170968367941,
+      "eval_loss": 0.6283465623855591,
+      "eval_runtime": 3074.8923,
+      "eval_samples_per_second": 0.583,
+      "eval_steps_per_second": 0.291,
+      "step": 23000
+    },
+    {
+      "epoch": 2.2326799922375313,
+      "grad_norm": 2.0022361963538584,
+      "learning_rate": 1.2830164189759234e-05,
+      "loss": 0.5047,
+      "step": 23010
+    },
+    {
+      "epoch": 2.233650300795653,
+      "grad_norm": 1.5655335338455796,
+      "learning_rate": 1.281393990525018e-05,
+      "loss": 0.4699,
+      "step": 23020
+    },
+    {
+      "epoch": 2.2346206093537746,
+      "grad_norm": 2.1600874630795452,
+      "learning_rate": 1.2797715620741127e-05,
+      "loss": 0.4684,
+      "step": 23030
+    },
+    {
+      "epoch": 2.235590917911896,
+      "grad_norm": 1.6958961004239297,
+      "learning_rate": 1.2781491336232073e-05,
+      "loss": 0.4769,
+      "step": 23040
+    },
+    {
+      "epoch": 2.2365612264700174,
+      "grad_norm": 2.290659253729356,
+      "learning_rate": 1.2765267051723021e-05,
+      "loss": 0.5084,
+      "step": 23050
+    },
+    {
+      "epoch": 2.237531535028139,
+      "grad_norm": 2.250294105691862,
+      "learning_rate": 1.2749042767213967e-05,
+      "loss": 0.4562,
+      "step": 23060
+    },
+    {
+      "epoch": 2.2385018435862603,
+      "grad_norm": 1.93879453393384,
+      "learning_rate": 1.2732818482704914e-05,
+      "loss": 0.4687,
+      "step": 23070
+    },
+    {
+      "epoch": 2.239472152144382,
+      "grad_norm": 2.3403124810757805,
+      "learning_rate": 1.271659419819586e-05,
+      "loss": 0.493,
+      "step": 23080
+    },
+    {
+      "epoch": 2.2404424607025035,
+      "grad_norm": 1.800829852558906,
+      "learning_rate": 1.2700369913686808e-05,
+      "loss": 0.4786,
+      "step": 23090
+    },
+    {
+      "epoch": 2.241412769260625,
+      "grad_norm": 2.0849042725619165,
+      "learning_rate": 1.2684145629177755e-05,
+      "loss": 0.4682,
+      "step": 23100
+    },
+    {
+      "epoch": 2.2423830778187464,
+      "grad_norm": 2.209642518744148,
+      "learning_rate": 1.2667921344668701e-05,
+      "loss": 0.4705,
+      "step": 23110
+    },
+    {
+      "epoch": 2.2433533863768678,
+      "grad_norm": 2.04809833807039,
+      "learning_rate": 1.2651697060159647e-05,
+      "loss": 0.4652,
+      "step": 23120
+    },
+    {
+      "epoch": 2.244323694934989,
+      "grad_norm": 2.134292380445789,
+      "learning_rate": 1.2635472775650595e-05,
+      "loss": 0.4375,
+      "step": 23130
+    },
+    {
+      "epoch": 2.245294003493111,
+      "grad_norm": 2.0308679578505466,
+      "learning_rate": 1.2619248491141542e-05,
+      "loss": 0.4256,
+      "step": 23140
+    },
+    {
+      "epoch": 2.2462643120512324,
+      "grad_norm": 1.7740882276229983,
+      "learning_rate": 1.2603024206632488e-05,
+      "loss": 0.4754,
+      "step": 23150
+    },
+    {
+      "epoch": 2.247234620609354,
+      "grad_norm": 2.845899626015525,
+      "learning_rate": 1.2586799922123435e-05,
+      "loss": 0.4465,
+      "step": 23160
+    },
+    {
+      "epoch": 2.2482049291674753,
+      "grad_norm": 1.6656047995660936,
+      "learning_rate": 1.2570575637614383e-05,
+      "loss": 0.4645,
+      "step": 23170
+    },
+    {
+      "epoch": 2.2491752377255967,
+      "grad_norm": 2.1317439051006675,
+      "learning_rate": 1.2554351353105329e-05,
+      "loss": 0.4215,
+      "step": 23180
+    },
+    {
+      "epoch": 2.250145546283718,
+      "grad_norm": 1.8180573164715403,
+      "learning_rate": 1.2538127068596275e-05,
+      "loss": 0.448,
+      "step": 23190
+    },
+    {
+      "epoch": 2.2511158548418395,
+      "grad_norm": 1.7997832378510108,
+      "learning_rate": 1.2521902784087222e-05,
+      "loss": 0.5032,
+      "step": 23200
+    },
+    {
+      "epoch": 2.2520861633999614,
+      "grad_norm": 1.97843541131739,
+      "learning_rate": 1.2505678499578168e-05,
+      "loss": 0.4577,
+      "step": 23210
+    },
+    {
+      "epoch": 2.253056471958083,
+      "grad_norm": 2.1980311567337045,
+      "learning_rate": 1.2489454215069116e-05,
+      "loss": 0.4492,
+      "step": 23220
+    },
+    {
+      "epoch": 2.254026780516204,
+      "grad_norm": 1.888365535194607,
+      "learning_rate": 1.2473229930560063e-05,
+      "loss": 0.4416,
+      "step": 23230
+    },
+    {
+      "epoch": 2.2549970890743256,
+      "grad_norm": 1.7544850776435907,
+      "learning_rate": 1.2457005646051009e-05,
+      "loss": 0.4803,
+      "step": 23240
+    },
+    {
+      "epoch": 2.255967397632447,
+      "grad_norm": 1.8829778599554805,
+      "learning_rate": 1.2440781361541955e-05,
+      "loss": 0.4227,
+      "step": 23250
+    },
+    {
+      "epoch": 2.2569377061905684,
+      "grad_norm": 2.1628337950810956,
+      "learning_rate": 1.2424557077032903e-05,
+      "loss": 0.4467,
+      "step": 23260
+    },
+    {
+      "epoch": 2.25790801474869,
+      "grad_norm": 2.1336139136695254,
+      "learning_rate": 1.240833279252385e-05,
+      "loss": 0.4798,
+      "step": 23270
+    },
+    {
+      "epoch": 2.2588783233068117,
+      "grad_norm": 2.3462272747330277,
+      "learning_rate": 1.2392108508014798e-05,
+      "loss": 0.4894,
+      "step": 23280
+    },
+    {
+      "epoch": 2.259848631864933,
+      "grad_norm": 1.9473113253648615,
+      "learning_rate": 1.2375884223505744e-05,
+      "loss": 0.4353,
+      "step": 23290
+    },
+    {
+      "epoch": 2.2608189404230545,
+      "grad_norm": 2.357093728494574,
+      "learning_rate": 1.235965993899669e-05,
+      "loss": 0.4508,
+      "step": 23300
+    },
+    {
+      "epoch": 2.261789248981176,
+      "grad_norm": 2.0651529569693468,
+      "learning_rate": 1.2343435654487639e-05,
+      "loss": 0.5106,
+      "step": 23310
+    },
+    {
+      "epoch": 2.2627595575392974,
+      "grad_norm": 1.8536850282087043,
+      "learning_rate": 1.2327211369978585e-05,
+      "loss": 0.446,
+      "step": 23320
+    },
+    {
+      "epoch": 2.263729866097419,
+      "grad_norm": 2.129566666742622,
+      "learning_rate": 1.2310987085469532e-05,
+      "loss": 0.4467,
+      "step": 23330
+    },
+    {
+      "epoch": 2.2647001746555406,
+      "grad_norm": 1.73183433586886,
+      "learning_rate": 1.2294762800960478e-05,
+      "loss": 0.4712,
+      "step": 23340
+    },
+    {
+      "epoch": 2.265670483213662,
+      "grad_norm": 1.5251850887891814,
+      "learning_rate": 1.2278538516451426e-05,
+      "loss": 0.5013,
+      "step": 23350
+    },
+    {
+      "epoch": 2.2666407917717835,
+      "grad_norm": 2.184835227449666,
+      "learning_rate": 1.2262314231942372e-05,
+      "loss": 0.4714,
+      "step": 23360
+    },
+    {
+      "epoch": 2.267611100329905,
+      "grad_norm": 2.278133559610058,
+      "learning_rate": 1.2246089947433319e-05,
+      "loss": 0.4707,
+      "step": 23370
+    },
+    {
+      "epoch": 2.2685814088880263,
+      "grad_norm": 2.043400099097976,
+      "learning_rate": 1.2229865662924265e-05,
+      "loss": 0.4398,
+      "step": 23380
+    },
+    {
+      "epoch": 2.2695517174461477,
+      "grad_norm": 2.1473483931604145,
+      "learning_rate": 1.2213641378415212e-05,
+      "loss": 0.4229,
+      "step": 23390
+    },
+    {
+      "epoch": 2.2705220260042696,
+      "grad_norm": 1.9230359161534956,
+      "learning_rate": 1.219741709390616e-05,
+      "loss": 0.425,
+      "step": 23400
+    },
+    {
+      "epoch": 2.271492334562391,
+      "grad_norm": 2.0484743523955773,
+      "learning_rate": 1.2181192809397106e-05,
+      "loss": 0.4034,
+      "step": 23410
+    },
+    {
+      "epoch": 2.2724626431205124,
+      "grad_norm": 2.018457750509176,
+      "learning_rate": 1.2164968524888052e-05,
+      "loss": 0.5061,
+      "step": 23420
+    },
+    {
+      "epoch": 2.273432951678634,
+      "grad_norm": 2.0919643594364494,
+      "learning_rate": 1.2148744240378999e-05,
+      "loss": 0.432,
+      "step": 23430
+    },
+    {
+      "epoch": 2.274403260236755,
+      "grad_norm": 2.4420554524762395,
+      "learning_rate": 1.2132519955869947e-05,
+      "loss": 0.4734,
+      "step": 23440
+    },
+    {
+      "epoch": 2.2753735687948766,
+      "grad_norm": 2.2048878285052034,
+      "learning_rate": 1.2116295671360893e-05,
+      "loss": 0.5,
+      "step": 23450
+    },
+    {
+      "epoch": 2.276343877352998,
+      "grad_norm": 2.385315906124704,
+      "learning_rate": 1.210007138685184e-05,
+      "loss": 0.3946,
+      "step": 23460
+    },
+    {
+      "epoch": 2.27731418591112,
+      "grad_norm": 2.1039427863509936,
+      "learning_rate": 1.2083847102342788e-05,
+      "loss": 0.4509,
+      "step": 23470
+    },
+    {
+      "epoch": 2.2782844944692413,
+      "grad_norm": 2.1731393161499977,
+      "learning_rate": 1.2067622817833734e-05,
+      "loss": 0.5216,
+      "step": 23480
+    },
+    {
+      "epoch": 2.2792548030273627,
+      "grad_norm": 2.133407830915671,
+      "learning_rate": 1.2051398533324682e-05,
+      "loss": 0.4652,
+      "step": 23490
+    },
+    {
+      "epoch": 2.280225111585484,
+      "grad_norm": 1.8143906080604477,
+      "learning_rate": 1.2035174248815629e-05,
+      "loss": 0.4608,
+      "step": 23500
+    },
+    {
+      "epoch": 2.280225111585484,
+      "eval_loss": 0.6306902766227722,
+      "eval_runtime": 3119.399,
+      "eval_samples_per_second": 0.574,
+      "eval_steps_per_second": 0.287,
+      "step": 23500
+    },
+    {
+      "epoch": 2.2811954201436055,
+      "grad_norm": 2.2848754365231545,
+      "learning_rate": 1.2018949964306575e-05,
+      "loss": 0.5139,
+      "step": 23510
+    },
+    {
+      "epoch": 2.282165728701727,
+      "grad_norm": 2.354718632997715,
+      "learning_rate": 1.2002725679797521e-05,
+      "loss": 0.4255,
+      "step": 23520
+    },
+    {
+      "epoch": 2.283136037259849,
+      "grad_norm": 2.256014847978423,
+      "learning_rate": 1.198650139528847e-05,
+      "loss": 0.4393,
+      "step": 23530
+    },
+    {
+      "epoch": 2.2841063458179702,
+      "grad_norm": 2.016243274514756,
+      "learning_rate": 1.1970277110779416e-05,
+      "loss": 0.4194,
+      "step": 23540
+    },
+    {
+      "epoch": 2.2850766543760916,
+      "grad_norm": 2.4759209758692426,
+      "learning_rate": 1.1954052826270362e-05,
+      "loss": 0.4757,
+      "step": 23550
+    },
+    {
+      "epoch": 2.286046962934213,
+      "grad_norm": 2.2547867685912486,
+      "learning_rate": 1.1937828541761309e-05,
+      "loss": 0.4511,
+      "step": 23560
+    },
+    {
+      "epoch": 2.2870172714923345,
+      "grad_norm": 1.9593024304910214,
+      "learning_rate": 1.1921604257252255e-05,
+      "loss": 0.5043,
+      "step": 23570
+    },
+    {
+      "epoch": 2.287987580050456,
+      "grad_norm": 2.2107379578337567,
+      "learning_rate": 1.1905379972743203e-05,
+      "loss": 0.4349,
+      "step": 23580
+    },
+    {
+      "epoch": 2.2889578886085777,
+      "grad_norm": 1.8779808167419767,
+      "learning_rate": 1.188915568823415e-05,
+      "loss": 0.4063,
+      "step": 23590
+    },
+    {
+      "epoch": 2.289928197166699,
+      "grad_norm": 2.1111765494708052,
+      "learning_rate": 1.1872931403725096e-05,
+      "loss": 0.4327,
+      "step": 23600
+    },
+    {
+      "epoch": 2.2908985057248206,
+      "grad_norm": 1.906738344577065,
+      "learning_rate": 1.1856707119216042e-05,
+      "loss": 0.4676,
+      "step": 23610
+    },
+    {
+      "epoch": 2.291868814282942,
+      "grad_norm": 2.6990569903001167,
+      "learning_rate": 1.184048283470699e-05,
+      "loss": 0.4759,
+      "step": 23620
+    },
+    {
+      "epoch": 2.2928391228410634,
+      "grad_norm": 2.39173938568666,
+      "learning_rate": 1.1824258550197937e-05,
+      "loss": 0.4565,
+      "step": 23630
+    },
+    {
+      "epoch": 2.293809431399185,
+      "grad_norm": 2.264327763939328,
+      "learning_rate": 1.1808034265688883e-05,
+      "loss": 0.5,
+      "step": 23640
+    },
+    {
+      "epoch": 2.294779739957306,
+      "grad_norm": 2.5489345132458494,
+      "learning_rate": 1.179180998117983e-05,
+      "loss": 0.4545,
+      "step": 23650
+    },
+    {
+      "epoch": 2.295750048515428,
+      "grad_norm": 1.5435526436096745,
+      "learning_rate": 1.1775585696670777e-05,
+      "loss": 0.444,
+      "step": 23660
+    },
+    {
+      "epoch": 2.2967203570735495,
+      "grad_norm": 2.096262775761873,
+      "learning_rate": 1.1759361412161725e-05,
+      "loss": 0.4651,
+      "step": 23670
+    },
+    {
+      "epoch": 2.297690665631671,
+      "grad_norm": 2.190613074347611,
+      "learning_rate": 1.1743137127652672e-05,
+      "loss": 0.5474,
+      "step": 23680
+    },
+    {
+      "epoch": 2.2986609741897923,
+      "grad_norm": 2.077801881915876,
+      "learning_rate": 1.1726912843143618e-05,
+      "loss": 0.4762,
+      "step": 23690
+    },
+    {
+      "epoch": 2.2996312827479137,
+      "grad_norm": 2.01599660916335,
+      "learning_rate": 1.1710688558634565e-05,
+      "loss": 0.4819,
+      "step": 23700
+    },
+    {
+      "epoch": 2.300601591306035,
+      "grad_norm": 1.7711529107558897,
+      "learning_rate": 1.1694464274125513e-05,
+      "loss": 0.5082,
+      "step": 23710
+    },
+    {
+      "epoch": 2.3015718998641566,
+      "grad_norm": 2.015443958364855,
+      "learning_rate": 1.1678239989616459e-05,
+      "loss": 0.4578,
+      "step": 23720
+    },
+    {
+      "epoch": 2.3025422084222784,
+      "grad_norm": 1.6478971592533067,
+      "learning_rate": 1.1662015705107405e-05,
+      "loss": 0.4285,
+      "step": 23730
+    },
+    {
+      "epoch": 2.3035125169804,
+      "grad_norm": 1.722295133645004,
+      "learning_rate": 1.1645791420598352e-05,
+      "loss": 0.4678,
+      "step": 23740
+    },
+    {
+      "epoch": 2.3044828255385212,
+      "grad_norm": 2.004493677850927,
+      "learning_rate": 1.16295671360893e-05,
+      "loss": 0.412,
+      "step": 23750
+    },
+    {
+      "epoch": 2.3054531340966427,
+      "grad_norm": 2.45778069486736,
+      "learning_rate": 1.1613342851580246e-05,
+      "loss": 0.5124,
+      "step": 23760
+    },
+    {
+      "epoch": 2.306423442654764,
+      "grad_norm": 2.040288471406675,
+      "learning_rate": 1.1597118567071193e-05,
+      "loss": 0.448,
+      "step": 23770
+    },
+    {
+      "epoch": 2.307393751212886,
+      "grad_norm": 1.995303973637335,
+      "learning_rate": 1.1580894282562139e-05,
+      "loss": 0.5022,
+      "step": 23780
+    },
+    {
+      "epoch": 2.3083640597710073,
+      "grad_norm": 2.080262776846013,
+      "learning_rate": 1.1564669998053085e-05,
+      "loss": 0.4962,
+      "step": 23790
+    },
+    {
+      "epoch": 2.3093343683291288,
+      "grad_norm": 2.2695683008172756,
+      "learning_rate": 1.1548445713544034e-05,
+      "loss": 0.5149,
+      "step": 23800
+    },
+    {
+      "epoch": 2.31030467688725,
+      "grad_norm": 2.2271193293614724,
+      "learning_rate": 1.153222142903498e-05,
+      "loss": 0.4527,
+      "step": 23810
+    },
+    {
+      "epoch": 2.3112749854453716,
+      "grad_norm": 1.8037542108398639,
+      "learning_rate": 1.1515997144525926e-05,
+      "loss": 0.4569,
+      "step": 23820
+    },
+    {
+      "epoch": 2.312245294003493,
+      "grad_norm": 1.8828460178890125,
+      "learning_rate": 1.1499772860016873e-05,
+      "loss": 0.446,
+      "step": 23830
+    },
+    {
+      "epoch": 2.3132156025616144,
+      "grad_norm": 2.148750511684889,
+      "learning_rate": 1.148354857550782e-05,
+      "loss": 0.4258,
+      "step": 23840
+    },
+    {
+      "epoch": 2.3141859111197363,
+      "grad_norm": 1.9299282888568785,
+      "learning_rate": 1.1467324290998767e-05,
+      "loss": 0.443,
+      "step": 23850
+    },
+    {
+      "epoch": 2.3151562196778577,
+      "grad_norm": 2.306570936387397,
+      "learning_rate": 1.1451100006489715e-05,
+      "loss": 0.4742,
+      "step": 23860
+    },
+    {
+      "epoch": 2.316126528235979,
+      "grad_norm": 1.652399835965133,
+      "learning_rate": 1.1434875721980662e-05,
+      "loss": 0.4434,
+      "step": 23870
+    },
+    {
+      "epoch": 2.3170968367941005,
+      "grad_norm": 1.8666656145624976,
+      "learning_rate": 1.1418651437471608e-05,
+      "loss": 0.4165,
+      "step": 23880
+    },
+    {
+      "epoch": 2.318067145352222,
+      "grad_norm": 1.799707347697279,
+      "learning_rate": 1.1402427152962556e-05,
+      "loss": 0.4525,
+      "step": 23890
+    },
+    {
+      "epoch": 2.3190374539103433,
+      "grad_norm": 1.8900931059214818,
+      "learning_rate": 1.1386202868453502e-05,
+      "loss": 0.4675,
+      "step": 23900
+    },
+    {
+      "epoch": 2.3200077624684647,
+      "grad_norm": 2.2479167053649896,
+      "learning_rate": 1.1369978583944449e-05,
+      "loss": 0.4624,
+      "step": 23910
+    },
+    {
+      "epoch": 2.3209780710265866,
+      "grad_norm": 2.2784052431273616,
+      "learning_rate": 1.1353754299435395e-05,
+      "loss": 0.4219,
+      "step": 23920
+    },
+    {
+      "epoch": 2.321948379584708,
+      "grad_norm": 2.1485033156725417,
+      "learning_rate": 1.1337530014926343e-05,
+      "loss": 0.4118,
+      "step": 23930
+    },
+    {
+      "epoch": 2.3229186881428294,
+      "grad_norm": 2.0610231375794075,
+      "learning_rate": 1.132130573041729e-05,
+      "loss": 0.4672,
+      "step": 23940
+    },
+    {
+      "epoch": 2.323888996700951,
+      "grad_norm": 1.8373253686424216,
+      "learning_rate": 1.1305081445908236e-05,
+      "loss": 0.4892,
+      "step": 23950
+    },
+    {
+      "epoch": 2.3248593052590723,
+      "grad_norm": 2.0784968854146375,
+      "learning_rate": 1.1288857161399182e-05,
+      "loss": 0.426,
+      "step": 23960
+    },
+    {
+      "epoch": 2.325829613817194,
+      "grad_norm": 1.8246399051495223,
+      "learning_rate": 1.1272632876890129e-05,
+      "loss": 0.4707,
+      "step": 23970
+    },
+    {
+      "epoch": 2.3267999223753155,
+      "grad_norm": 1.8607594933847929,
+      "learning_rate": 1.1256408592381077e-05,
+      "loss": 0.4787,
+      "step": 23980
+    },
+    {
+      "epoch": 2.327770230933437,
+      "grad_norm": 2.1338034995885673,
+      "learning_rate": 1.1240184307872023e-05,
+      "loss": 0.4188,
+      "step": 23990
+    },
+    {
+      "epoch": 2.3287405394915583,
+      "grad_norm": 1.691316437747956,
+      "learning_rate": 1.122396002336297e-05,
+      "loss": 0.4683,
+      "step": 24000
+    },
+    {
+      "epoch": 2.3287405394915583,
+      "eval_loss": 0.6269034147262573,
+      "eval_runtime": 3077.756,
+      "eval_samples_per_second": 0.582,
+      "eval_steps_per_second": 0.291,
+      "step": 24000
+    },
+    {
+      "epoch": 2.3297108480496798,
+      "grad_norm": 2.040492940843433,
+      "learning_rate": 1.1207735738853916e-05,
+      "loss": 0.4569,
+      "step": 24010
+    },
+    {
+      "epoch": 2.330681156607801,
+      "grad_norm": 1.7971740640523082,
+      "learning_rate": 1.1191511454344864e-05,
+      "loss": 0.4495,
+      "step": 24020
+    },
+    {
+      "epoch": 2.3316514651659226,
+      "grad_norm": 1.939848570121438,
+      "learning_rate": 1.117528716983581e-05,
+      "loss": 0.4794,
+      "step": 24030
+    },
+    {
+      "epoch": 2.3326217737240444,
+      "grad_norm": 2.032618108671819,
+      "learning_rate": 1.1159062885326757e-05,
+      "loss": 0.4446,
+      "step": 24040
+    },
+    {
+      "epoch": 2.333592082282166,
+      "grad_norm": 1.9156260447966438,
+      "learning_rate": 1.1142838600817703e-05,
+      "loss": 0.4687,
+      "step": 24050
+    },
+    {
+      "epoch": 2.3345623908402873,
+      "grad_norm": 1.9351404363288518,
+      "learning_rate": 1.1126614316308651e-05,
+      "loss": 0.5079,
+      "step": 24060
+    },
+    {
+      "epoch": 2.3355326993984087,
+      "grad_norm": 1.9150652539150566,
+      "learning_rate": 1.11103900317996e-05,
+      "loss": 0.4673,
+      "step": 24070
+    },
+    {
+      "epoch": 2.33650300795653,
+      "grad_norm": 1.8679864315268337,
+      "learning_rate": 1.1094165747290546e-05,
+      "loss": 0.4291,
+      "step": 24080
+    },
+    {
+      "epoch": 2.3374733165146515,
+      "grad_norm": 2.414373392352731,
+      "learning_rate": 1.1077941462781492e-05,
+      "loss": 0.4738,
+      "step": 24090
+    },
+    {
+      "epoch": 2.338443625072773,
+      "grad_norm": 2.5480843998856466,
+      "learning_rate": 1.1061717178272439e-05,
+      "loss": 0.4281,
+      "step": 24100
+    },
+    {
+      "epoch": 2.339413933630895,
+      "grad_norm": 1.7367372410706,
+      "learning_rate": 1.1045492893763387e-05,
+      "loss": 0.483,
+      "step": 24110
+    },
+    {
+      "epoch": 2.340384242189016,
+      "grad_norm": 1.8020643362909123,
+      "learning_rate": 1.1029268609254333e-05,
+      "loss": 0.4534,
+      "step": 24120
+    },
+    {
+      "epoch": 2.3413545507471376,
+      "grad_norm": 2.0532213004634117,
+      "learning_rate": 1.101304432474528e-05,
+      "loss": 0.4691,
+      "step": 24130
+    },
+    {
+      "epoch": 2.342324859305259,
+      "grad_norm": 1.825257767760241,
+      "learning_rate": 1.0996820040236226e-05,
+      "loss": 0.488,
+      "step": 24140
+    },
+    {
+      "epoch": 2.3432951678633804,
+      "grad_norm": 2.0243831093236513,
+      "learning_rate": 1.0980595755727172e-05,
+      "loss": 0.4846,
+      "step": 24150
+    },
+    {
+      "epoch": 2.344265476421502,
+      "grad_norm": 2.249331730728062,
+      "learning_rate": 1.096437147121812e-05,
+      "loss": 0.4578,
+      "step": 24160
+    },
+    {
+      "epoch": 2.3452357849796237,
+      "grad_norm": 2.147870816517021,
+      "learning_rate": 1.0948147186709067e-05,
+      "loss": 0.4453,
+      "step": 24170
+    },
+    {
+      "epoch": 2.346206093537745,
+      "grad_norm": 1.8861077026920583,
+      "learning_rate": 1.0931922902200013e-05,
+      "loss": 0.4214,
+      "step": 24180
+    },
+    {
+      "epoch": 2.3471764020958665,
+      "grad_norm": 1.9335635244208784,
+      "learning_rate": 1.091569861769096e-05,
+      "loss": 0.4053,
+      "step": 24190
+    },
+    {
+      "epoch": 2.348146710653988,
+      "grad_norm": 2.1792571618590117,
+      "learning_rate": 1.0899474333181907e-05,
+      "loss": 0.4502,
+      "step": 24200
+    },
+    {
+      "epoch": 2.3491170192121094,
+      "grad_norm": 1.8035101955175616,
+      "learning_rate": 1.0883250048672854e-05,
+      "loss": 0.4777,
+      "step": 24210
+    },
+    {
+      "epoch": 2.3500873277702308,
+      "grad_norm": 2.3748193703673506,
+      "learning_rate": 1.08670257641638e-05,
+      "loss": 0.4348,
+      "step": 24220
+    },
+    {
+      "epoch": 2.3510576363283526,
+      "grad_norm": 2.093876681748244,
+      "learning_rate": 1.0850801479654747e-05,
+      "loss": 0.4801,
+      "step": 24230
+    },
+    {
+      "epoch": 2.352027944886474,
+      "grad_norm": 2.0917507142883025,
+      "learning_rate": 1.0834577195145695e-05,
+      "loss": 0.4622,
+      "step": 24240
+    },
+    {
+      "epoch": 2.3529982534445955,
+      "grad_norm": 2.4449607859132816,
+      "learning_rate": 1.0818352910636643e-05,
+      "loss": 0.462,
+      "step": 24250
+    },
+    {
+      "epoch": 2.353968562002717,
+      "grad_norm": 2.1889912797955877,
+      "learning_rate": 1.080212862612759e-05,
+      "loss": 0.4614,
+      "step": 24260
+    },
+    {
+      "epoch": 2.3549388705608383,
+      "grad_norm": 1.7339445133001758,
+      "learning_rate": 1.0785904341618536e-05,
+      "loss": 0.4477,
+      "step": 24270
+    },
+    {
+      "epoch": 2.3559091791189597,
+      "grad_norm": 2.2974387900444064,
+      "learning_rate": 1.0769680057109482e-05,
+      "loss": 0.488,
+      "step": 24280
+    },
+    {
+      "epoch": 2.356879487677081,
+      "grad_norm": 1.9216704451577764,
+      "learning_rate": 1.075345577260043e-05,
+      "loss": 0.4602,
+      "step": 24290
+    },
+    {
+      "epoch": 2.357849796235203,
+      "grad_norm": 1.595623484078005,
+      "learning_rate": 1.0737231488091376e-05,
+      "loss": 0.4618,
+      "step": 24300
+    },
+    {
+      "epoch": 2.3588201047933244,
+      "grad_norm": 2.1389381123821973,
+      "learning_rate": 1.0721007203582323e-05,
+      "loss": 0.4109,
+      "step": 24310
+    },
+    {
+      "epoch": 2.359790413351446,
+      "grad_norm": 2.1703895980886294,
+      "learning_rate": 1.070478291907327e-05,
+      "loss": 0.4755,
+      "step": 24320
+    },
+    {
+      "epoch": 2.360760721909567,
+      "grad_norm": 2.101496536635865,
+      "learning_rate": 1.0688558634564216e-05,
+      "loss": 0.4371,
+      "step": 24330
+    },
+    {
+      "epoch": 2.3617310304676886,
+      "grad_norm": 2.201243439271482,
+      "learning_rate": 1.0672334350055164e-05,
+      "loss": 0.4252,
+      "step": 24340
+    },
+    {
+      "epoch": 2.36270133902581,
+      "grad_norm": 2.067190247810472,
+      "learning_rate": 1.065611006554611e-05,
+      "loss": 0.4684,
+      "step": 24350
+    },
+    {
+      "epoch": 2.3636716475839314,
+      "grad_norm": 1.593808072912525,
+      "learning_rate": 1.0639885781037056e-05,
+      "loss": 0.4474,
+      "step": 24360
+    },
+    {
+      "epoch": 2.3646419561420533,
+      "grad_norm": 2.3672169952188344,
+      "learning_rate": 1.0623661496528003e-05,
+      "loss": 0.4543,
+      "step": 24370
+    },
+    {
+      "epoch": 2.3656122647001747,
+      "grad_norm": 2.173568364373994,
+      "learning_rate": 1.060743721201895e-05,
+      "loss": 0.4172,
+      "step": 24380
+    },
+    {
+      "epoch": 2.366582573258296,
+      "grad_norm": 2.0059231401496005,
+      "learning_rate": 1.0591212927509897e-05,
+      "loss": 0.4735,
+      "step": 24390
+    },
+    {
+      "epoch": 2.3675528818164175,
+      "grad_norm": 2.0851213756560187,
+      "learning_rate": 1.0574988643000844e-05,
+      "loss": 0.4363,
+      "step": 24400
+    },
+    {
+      "epoch": 2.368523190374539,
+      "grad_norm": 2.05634160028851,
+      "learning_rate": 1.055876435849179e-05,
+      "loss": 0.5026,
+      "step": 24410
+    },
+    {
+      "epoch": 2.369493498932661,
+      "grad_norm": 2.2318992281649583,
+      "learning_rate": 1.0542540073982738e-05,
+      "loss": 0.4272,
+      "step": 24420
+    },
+    {
+      "epoch": 2.3704638074907822,
+      "grad_norm": 2.3837912348792862,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.4873,
+      "step": 24430
+    },
+    {
+      "epoch": 2.3714341160489036,
+      "grad_norm": 1.9839906718631033,
+      "learning_rate": 1.051009150496463e-05,
+      "loss": 0.4689,
+      "step": 24440
+    },
+    {
+      "epoch": 2.372404424607025,
+      "grad_norm": 2.9092572102641374,
+      "learning_rate": 1.0493867220455579e-05,
+      "loss": 0.455,
+      "step": 24450
+    },
+    {
+      "epoch": 2.3733747331651465,
+      "grad_norm": 1.783405495522693,
+      "learning_rate": 1.0477642935946525e-05,
+      "loss": 0.47,
+      "step": 24460
+    },
+    {
+      "epoch": 2.374345041723268,
+      "grad_norm": 2.4188709581674,
+      "learning_rate": 1.0461418651437473e-05,
+      "loss": 0.4488,
+      "step": 24470
+    },
+    {
+      "epoch": 2.3753153502813893,
+      "grad_norm": 2.014782666224647,
+      "learning_rate": 1.044519436692842e-05,
+      "loss": 0.4533,
+      "step": 24480
+    },
+    {
+      "epoch": 2.376285658839511,
+      "grad_norm": 2.174178898015034,
+      "learning_rate": 1.0428970082419366e-05,
+      "loss": 0.444,
+      "step": 24490
+    },
+    {
+      "epoch": 2.3772559673976326,
+      "grad_norm": 2.485611502054494,
+      "learning_rate": 1.0412745797910313e-05,
+      "loss": 0.5027,
+      "step": 24500
+    },
+    {
+      "epoch": 2.3772559673976326,
+      "eval_loss": 0.6276779174804688,
+      "eval_runtime": 3079.5235,
+      "eval_samples_per_second": 0.582,
+      "eval_steps_per_second": 0.291,
+      "step": 24500
+    },
+    {
+      "epoch": 2.378226275955754,
+      "grad_norm": 2.5871995720796184,
+      "learning_rate": 1.0396521513401259e-05,
+      "loss": 0.4382,
+      "step": 24510
+    },
+    {
+      "epoch": 2.3791965845138754,
+      "grad_norm": 1.536190288984018,
+      "learning_rate": 1.0380297228892207e-05,
+      "loss": 0.4473,
+      "step": 24520
+    },
+    {
+      "epoch": 2.380166893071997,
+      "grad_norm": 1.6459160347807742,
+      "learning_rate": 1.0364072944383153e-05,
+      "loss": 0.4078,
+      "step": 24530
+    },
+    {
+      "epoch": 2.381137201630118,
+      "grad_norm": 2.152035121079775,
+      "learning_rate": 1.03478486598741e-05,
+      "loss": 0.4447,
+      "step": 24540
+    },
+    {
+      "epoch": 2.3821075101882396,
+      "grad_norm": 2.369193332564519,
+      "learning_rate": 1.0331624375365046e-05,
+      "loss": 0.3994,
+      "step": 24550
+    },
+    {
+      "epoch": 2.3830778187463615,
+      "grad_norm": 2.1514780345815256,
+      "learning_rate": 1.0315400090855994e-05,
+      "loss": 0.496,
+      "step": 24560
+    },
+    {
+      "epoch": 2.384048127304483,
+      "grad_norm": 1.7911554347140786,
+      "learning_rate": 1.029917580634694e-05,
+      "loss": 0.4701,
+      "step": 24570
+    },
+    {
+      "epoch": 2.3850184358626043,
+      "grad_norm": 2.0153664778745854,
+      "learning_rate": 1.0282951521837887e-05,
+      "loss": 0.4219,
+      "step": 24580
+    },
+    {
+      "epoch": 2.3859887444207257,
+      "grad_norm": 1.80318954794296,
+      "learning_rate": 1.0266727237328833e-05,
+      "loss": 0.4683,
+      "step": 24590
+    },
+    {
+      "epoch": 2.386959052978847,
+      "grad_norm": 2.588414199711988,
+      "learning_rate": 1.0250502952819781e-05,
+      "loss": 0.5015,
+      "step": 24600
+    },
+    {
+      "epoch": 2.387929361536969,
+      "grad_norm": 2.4217752650788613,
+      "learning_rate": 1.0234278668310728e-05,
+      "loss": 0.4975,
+      "step": 24610
+    },
+    {
+      "epoch": 2.3888996700950904,
+      "grad_norm": 1.794705542002023,
+      "learning_rate": 1.0218054383801674e-05,
+      "loss": 0.4836,
+      "step": 24620
+    },
+    {
+      "epoch": 2.389869978653212,
+      "grad_norm": 1.7631883574466254,
+      "learning_rate": 1.020183009929262e-05,
+      "loss": 0.5004,
+      "step": 24630
+    },
+    {
+      "epoch": 2.3908402872113332,
+      "grad_norm": 2.510901529741251,
+      "learning_rate": 1.0185605814783569e-05,
+      "loss": 0.4626,
+      "step": 24640
+    },
+    {
+      "epoch": 2.3918105957694547,
+      "grad_norm": 2.0262233819009157,
+      "learning_rate": 1.0169381530274517e-05,
+      "loss": 0.4289,
+      "step": 24650
+    },
+    {
+      "epoch": 2.392780904327576,
+      "grad_norm": 2.039691163120118,
+      "learning_rate": 1.0153157245765463e-05,
+      "loss": 0.4712,
+      "step": 24660
+    },
+    {
+      "epoch": 2.3937512128856975,
+      "grad_norm": 2.3530360716419665,
+      "learning_rate": 1.013693296125641e-05,
+      "loss": 0.4661,
+      "step": 24670
+    },
+    {
+      "epoch": 2.3947215214438193,
+      "grad_norm": 1.8731826227321207,
+      "learning_rate": 1.0120708676747356e-05,
+      "loss": 0.4602,
+      "step": 24680
+    },
+    {
+      "epoch": 2.3956918300019407,
+      "grad_norm": 2.1068678736668067,
+      "learning_rate": 1.0104484392238304e-05,
+      "loss": 0.3962,
+      "step": 24690
+    },
+    {
+      "epoch": 2.396662138560062,
+      "grad_norm": 2.0442268653871443,
+      "learning_rate": 1.008826010772925e-05,
+      "loss": 0.4241,
+      "step": 24700
+    },
+    {
+      "epoch": 2.3976324471181836,
+      "grad_norm": 1.823987854926844,
+      "learning_rate": 1.0072035823220197e-05,
+      "loss": 0.4332,
+      "step": 24710
+    },
+    {
+      "epoch": 2.398602755676305,
+      "grad_norm": 2.0466476195429433,
+      "learning_rate": 1.0055811538711143e-05,
+      "loss": 0.4973,
+      "step": 24720
+    },
+    {
+      "epoch": 2.3995730642344264,
+      "grad_norm": 2.2091928863349706,
+      "learning_rate": 1.003958725420209e-05,
+      "loss": 0.4561,
+      "step": 24730
+    },
+    {
+      "epoch": 2.400543372792548,
+      "grad_norm": 2.0728809538988417,
+      "learning_rate": 1.0023362969693038e-05,
+      "loss": 0.4311,
+      "step": 24740
+    },
+    {
+      "epoch": 2.4015136813506697,
+      "grad_norm": 2.5482352204418612,
+      "learning_rate": 1.0007138685183984e-05,
+      "loss": 0.5071,
+      "step": 24750
+    },
+    {
+      "epoch": 2.402483989908791,
+      "grad_norm": 1.9610940178733647,
+      "learning_rate": 9.99091440067493e-06,
+      "loss": 0.4401,
+      "step": 24760
+    },
+    {
+      "epoch": 2.4034542984669125,
+      "grad_norm": 2.3445676431912448,
+      "learning_rate": 9.974690116165877e-06,
+      "loss": 0.4482,
+      "step": 24770
+    },
+    {
+      "epoch": 2.404424607025034,
+      "grad_norm": 1.6828772581664517,
+      "learning_rate": 9.958465831656825e-06,
+      "loss": 0.4489,
+      "step": 24780
+    },
+    {
+      "epoch": 2.4053949155831553,
+      "grad_norm": 2.126865830634638,
+      "learning_rate": 9.942241547147771e-06,
+      "loss": 0.4136,
+      "step": 24790
+    },
+    {
+      "epoch": 2.4063652241412767,
+      "grad_norm": 1.6333253702974164,
+      "learning_rate": 9.926017262638718e-06,
+      "loss": 0.4479,
+      "step": 24800
+    },
+    {
+      "epoch": 2.4073355326993986,
+      "grad_norm": 2.05443116143666,
+      "learning_rate": 9.909792978129664e-06,
+      "loss": 0.4806,
+      "step": 24810
+    },
+    {
+      "epoch": 2.40830584125752,
+      "grad_norm": 2.1173326880866234,
+      "learning_rate": 9.893568693620612e-06,
+      "loss": 0.496,
+      "step": 24820
+    },
+    {
+      "epoch": 2.4092761498156414,
+      "grad_norm": 2.6271672766867185,
+      "learning_rate": 9.877344409111558e-06,
+      "loss": 0.4953,
+      "step": 24830
+    },
+    {
+      "epoch": 2.410246458373763,
+      "grad_norm": 1.7740927010946907,
+      "learning_rate": 9.861120124602506e-06,
+      "loss": 0.445,
+      "step": 24840
+    },
+    {
+      "epoch": 2.4112167669318842,
+      "grad_norm": 1.7938164516772293,
+      "learning_rate": 9.844895840093453e-06,
+      "loss": 0.4651,
+      "step": 24850
+    },
+    {
+      "epoch": 2.4121870754900057,
+      "grad_norm": 1.876181751989744,
+      "learning_rate": 9.8286715555844e-06,
+      "loss": 0.4742,
+      "step": 24860
+    },
+    {
+      "epoch": 2.4131573840481275,
+      "grad_norm": 2.268711698650067,
+      "learning_rate": 9.812447271075347e-06,
+      "loss": 0.4858,
+      "step": 24870
+    },
+    {
+      "epoch": 2.414127692606249,
+      "grad_norm": 2.002023556609326,
+      "learning_rate": 9.796222986566294e-06,
+      "loss": 0.437,
+      "step": 24880
+    },
+    {
+      "epoch": 2.4150980011643703,
+      "grad_norm": 2.2377584218251787,
+      "learning_rate": 9.77999870205724e-06,
+      "loss": 0.4631,
+      "step": 24890
+    },
+    {
+      "epoch": 2.4160683097224918,
+      "grad_norm": 2.048732665820942,
+      "learning_rate": 9.763774417548186e-06,
+      "loss": 0.4828,
+      "step": 24900
+    },
+    {
+      "epoch": 2.417038618280613,
+      "grad_norm": 1.6448348280754015,
+      "learning_rate": 9.747550133039133e-06,
+      "loss": 0.5178,
+      "step": 24910
+    },
+    {
+      "epoch": 2.4180089268387346,
+      "grad_norm": 1.5590200283867917,
+      "learning_rate": 9.731325848530081e-06,
+      "loss": 0.4582,
+      "step": 24920
+    },
+    {
+      "epoch": 2.418979235396856,
+      "grad_norm": 1.9618039375978873,
+      "learning_rate": 9.715101564021027e-06,
+      "loss": 0.4207,
+      "step": 24930
+    },
+    {
+      "epoch": 2.419949543954978,
+      "grad_norm": 2.067363699561904,
+      "learning_rate": 9.698877279511974e-06,
+      "loss": 0.4825,
+      "step": 24940
+    },
+    {
+      "epoch": 2.4209198525130993,
+      "grad_norm": 2.1824993479688635,
+      "learning_rate": 9.68265299500292e-06,
+      "loss": 0.4507,
+      "step": 24950
+    },
+    {
+      "epoch": 2.4218901610712207,
+      "grad_norm": 1.9138947437636256,
+      "learning_rate": 9.666428710493868e-06,
+      "loss": 0.4576,
+      "step": 24960
+    },
+    {
+      "epoch": 2.422860469629342,
+      "grad_norm": 1.6115918485232108,
+      "learning_rate": 9.650204425984815e-06,
+      "loss": 0.4705,
+      "step": 24970
+    },
+    {
+      "epoch": 2.4238307781874635,
+      "grad_norm": 2.250793958728259,
+      "learning_rate": 9.633980141475761e-06,
+      "loss": 0.4521,
+      "step": 24980
+    },
+    {
+      "epoch": 2.424801086745585,
+      "grad_norm": 1.8523875454672862,
+      "learning_rate": 9.617755856966707e-06,
+      "loss": 0.4397,
+      "step": 24990
+    },
+    {
+      "epoch": 2.4257713953037063,
+      "grad_norm": 1.904730107613101,
+      "learning_rate": 9.601531572457655e-06,
+      "loss": 0.4564,
+      "step": 25000
+    },
+    {
+      "epoch": 2.4257713953037063,
+      "eval_loss": 0.6268747448921204,
+      "eval_runtime": 3077.7529,
+      "eval_samples_per_second": 0.582,
+      "eval_steps_per_second": 0.291,
+      "step": 25000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 30918,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0052615872446464e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}