ydshieh's picture
ydshieh HF staff
Add model files
1d3c43e
raw
history blame
58.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.024486226497595,
"global_step": 4600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 4.9999999999999996e-06,
"loss": 155.33,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 9.999999999999999e-06,
"loss": 175.1775,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.4999999999999999e-05,
"loss": 165.6332,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 1.9999999999999998e-05,
"loss": 161.1489,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 2.4999999999999998e-05,
"loss": 176.11,
"step": 50
},
{
"epoch": 0.1,
"learning_rate": 2.9999999999999997e-05,
"loss": 153.7462,
"step": 60
},
{
"epoch": 0.12,
"learning_rate": 3.5e-05,
"loss": 170.9414,
"step": 70
},
{
"epoch": 0.14,
"learning_rate": 3.9999999999999996e-05,
"loss": 160.0549,
"step": 80
},
{
"epoch": 0.16,
"learning_rate": 4.4999999999999996e-05,
"loss": 151.4735,
"step": 90
},
{
"epoch": 0.17,
"learning_rate": 4.9999999999999996e-05,
"loss": 167.8096,
"step": 100
},
{
"epoch": 0.19,
"learning_rate": 5.499999999999999e-05,
"loss": 141.9121,
"step": 110
},
{
"epoch": 0.21,
"learning_rate": 5.9999999999999995e-05,
"loss": 153.8951,
"step": 120
},
{
"epoch": 0.23,
"learning_rate": 6.5e-05,
"loss": 131.0289,
"step": 130
},
{
"epoch": 0.24,
"learning_rate": 7e-05,
"loss": 101.1117,
"step": 140
},
{
"epoch": 0.26,
"learning_rate": 7.5e-05,
"loss": 100.5863,
"step": 150
},
{
"epoch": 0.28,
"learning_rate": 7.999999999999999e-05,
"loss": 80.6457,
"step": 160
},
{
"epoch": 0.3,
"learning_rate": 8.499999999999999e-05,
"loss": 80.4102,
"step": 170
},
{
"epoch": 0.31,
"learning_rate": 8.999999999999999e-05,
"loss": 70.8443,
"step": 180
},
{
"epoch": 0.33,
"learning_rate": 9.499999999999999e-05,
"loss": 61.107,
"step": 190
},
{
"epoch": 0.35,
"learning_rate": 9.999999999999999e-05,
"loss": 56.9596,
"step": 200
},
{
"epoch": 0.37,
"learning_rate": 0.00010499999999999999,
"loss": 40.8098,
"step": 210
},
{
"epoch": 0.38,
"learning_rate": 0.00010999999999999998,
"loss": 36.5124,
"step": 220
},
{
"epoch": 0.4,
"learning_rate": 0.000115,
"loss": 28.1064,
"step": 230
},
{
"epoch": 0.42,
"learning_rate": 0.00011999999999999999,
"loss": 20.9962,
"step": 240
},
{
"epoch": 0.44,
"learning_rate": 0.000125,
"loss": 16.4393,
"step": 250
},
{
"epoch": 0.45,
"learning_rate": 0.00013,
"loss": 10.9059,
"step": 260
},
{
"epoch": 0.47,
"learning_rate": 0.000135,
"loss": 8.8954,
"step": 270
},
{
"epoch": 0.49,
"learning_rate": 0.00014,
"loss": 7.7207,
"step": 280
},
{
"epoch": 0.51,
"learning_rate": 0.000145,
"loss": 7.0725,
"step": 290
},
{
"epoch": 0.52,
"learning_rate": 0.00015,
"loss": 6.9189,
"step": 300
},
{
"epoch": 0.54,
"learning_rate": 0.000155,
"loss": 6.9459,
"step": 310
},
{
"epoch": 0.56,
"learning_rate": 0.00015999999999999999,
"loss": 6.8881,
"step": 320
},
{
"epoch": 0.58,
"learning_rate": 0.000165,
"loss": 6.8779,
"step": 330
},
{
"epoch": 0.59,
"learning_rate": 0.00016999999999999999,
"loss": 6.897,
"step": 340
},
{
"epoch": 0.61,
"learning_rate": 0.000175,
"loss": 6.7802,
"step": 350
},
{
"epoch": 0.63,
"learning_rate": 0.00017999999999999998,
"loss": 6.8897,
"step": 360
},
{
"epoch": 0.65,
"learning_rate": 0.000185,
"loss": 6.8261,
"step": 370
},
{
"epoch": 0.66,
"learning_rate": 0.00018999999999999998,
"loss": 6.7646,
"step": 380
},
{
"epoch": 0.68,
"learning_rate": 0.000195,
"loss": 6.7816,
"step": 390
},
{
"epoch": 0.7,
"learning_rate": 0.00019999999999999998,
"loss": 6.7041,
"step": 400
},
{
"epoch": 0.72,
"learning_rate": 0.000205,
"loss": 6.7975,
"step": 410
},
{
"epoch": 0.73,
"learning_rate": 0.00020999999999999998,
"loss": 6.6509,
"step": 420
},
{
"epoch": 0.75,
"learning_rate": 0.000215,
"loss": 6.6573,
"step": 430
},
{
"epoch": 0.77,
"learning_rate": 0.00021999999999999995,
"loss": 6.6817,
"step": 440
},
{
"epoch": 0.79,
"learning_rate": 0.000225,
"loss": 6.5168,
"step": 450
},
{
"epoch": 0.8,
"learning_rate": 0.00023,
"loss": 6.7404,
"step": 460
},
{
"epoch": 0.82,
"learning_rate": 0.00023499999999999997,
"loss": 6.605,
"step": 470
},
{
"epoch": 0.84,
"learning_rate": 0.00023999999999999998,
"loss": 6.5657,
"step": 480
},
{
"epoch": 0.86,
"learning_rate": 0.000245,
"loss": 6.6316,
"step": 490
},
{
"epoch": 0.87,
"learning_rate": 0.00025,
"loss": 6.4709,
"step": 500
},
{
"epoch": 0.89,
"learning_rate": 0.00025499999999999996,
"loss": 6.6643,
"step": 510
},
{
"epoch": 0.91,
"learning_rate": 0.00026,
"loss": 6.5056,
"step": 520
},
{
"epoch": 0.93,
"learning_rate": 0.000265,
"loss": 6.5331,
"step": 530
},
{
"epoch": 0.94,
"learning_rate": 0.00027,
"loss": 6.5842,
"step": 540
},
{
"epoch": 0.96,
"learning_rate": 0.00027499999999999996,
"loss": 6.4195,
"step": 550
},
{
"epoch": 0.98,
"learning_rate": 0.00028,
"loss": 6.6377,
"step": 560
},
{
"epoch": 1.0,
"learning_rate": 0.000285,
"loss": 6.4403,
"step": 570
},
{
"epoch": 1.01,
"learning_rate": 0.00029,
"loss": 6.5407,
"step": 580
},
{
"epoch": 1.03,
"learning_rate": 0.00029499999999999996,
"loss": 6.5123,
"step": 590
},
{
"epoch": 1.05,
"learning_rate": 0.0003,
"loss": 6.4086,
"step": 600
},
{
"epoch": 1.05,
"eval_cer": 1.0,
"eval_loss": 6.509830951690674,
"eval_runtime": 1134.0889,
"eval_samples_per_second": 7.724,
"eval_wer": 1.0,
"step": 600
},
{
"epoch": 1.07,
"learning_rate": 0.00029986535008976655,
"loss": 6.5638,
"step": 610
},
{
"epoch": 1.08,
"learning_rate": 0.0002997307001795332,
"loss": 6.4348,
"step": 620
},
{
"epoch": 1.1,
"learning_rate": 0.0002995960502692998,
"loss": 6.5179,
"step": 630
},
{
"epoch": 1.12,
"learning_rate": 0.0002994614003590664,
"loss": 6.4513,
"step": 640
},
{
"epoch": 1.14,
"learning_rate": 0.000299326750448833,
"loss": 6.4153,
"step": 650
},
{
"epoch": 1.15,
"learning_rate": 0.0002991921005385996,
"loss": 6.5478,
"step": 660
},
{
"epoch": 1.17,
"learning_rate": 0.0002990574506283662,
"loss": 6.3301,
"step": 670
},
{
"epoch": 1.19,
"learning_rate": 0.0002989228007181328,
"loss": 6.5422,
"step": 680
},
{
"epoch": 1.21,
"learning_rate": 0.00029878815080789945,
"loss": 6.4761,
"step": 690
},
{
"epoch": 1.22,
"learning_rate": 0.0002986535008976661,
"loss": 6.3739,
"step": 700
},
{
"epoch": 1.24,
"learning_rate": 0.00029851885098743265,
"loss": 6.5702,
"step": 710
},
{
"epoch": 1.26,
"learning_rate": 0.0002983842010771993,
"loss": 6.3541,
"step": 720
},
{
"epoch": 1.28,
"learning_rate": 0.00029824955116696585,
"loss": 6.5389,
"step": 730
},
{
"epoch": 1.29,
"learning_rate": 0.0002981149012567325,
"loss": 6.4416,
"step": 740
},
{
"epoch": 1.31,
"learning_rate": 0.00029798025134649905,
"loss": 6.3941,
"step": 750
},
{
"epoch": 1.33,
"learning_rate": 0.0002978456014362657,
"loss": 6.784,
"step": 760
},
{
"epoch": 1.35,
"learning_rate": 0.0002977109515260323,
"loss": 6.3669,
"step": 770
},
{
"epoch": 1.36,
"learning_rate": 0.0002975763016157989,
"loss": 6.5267,
"step": 780
},
{
"epoch": 1.38,
"learning_rate": 0.0002974416517055655,
"loss": 6.4389,
"step": 790
},
{
"epoch": 1.4,
"learning_rate": 0.0002973070017953321,
"loss": 6.3461,
"step": 800
},
{
"epoch": 1.42,
"learning_rate": 0.0002971723518850987,
"loss": 6.5165,
"step": 810
},
{
"epoch": 1.43,
"learning_rate": 0.00029703770197486533,
"loss": 6.3583,
"step": 820
},
{
"epoch": 1.45,
"learning_rate": 0.00029690305206463196,
"loss": 6.5063,
"step": 830
},
{
"epoch": 1.47,
"learning_rate": 0.00029676840215439853,
"loss": 6.4102,
"step": 840
},
{
"epoch": 1.49,
"learning_rate": 0.00029663375224416516,
"loss": 6.3466,
"step": 850
},
{
"epoch": 1.5,
"learning_rate": 0.00029649910233393173,
"loss": 6.5043,
"step": 860
},
{
"epoch": 1.52,
"learning_rate": 0.00029636445242369836,
"loss": 6.3248,
"step": 870
},
{
"epoch": 1.54,
"learning_rate": 0.000296229802513465,
"loss": 6.4586,
"step": 880
},
{
"epoch": 1.56,
"learning_rate": 0.00029609515260323155,
"loss": 6.4547,
"step": 890
},
{
"epoch": 1.57,
"learning_rate": 0.0002959605026929982,
"loss": 6.3167,
"step": 900
},
{
"epoch": 1.59,
"learning_rate": 0.00029582585278276475,
"loss": 6.4654,
"step": 910
},
{
"epoch": 1.61,
"learning_rate": 0.0002956912028725314,
"loss": 6.2638,
"step": 920
},
{
"epoch": 1.63,
"learning_rate": 0.000295556552962298,
"loss": 6.4081,
"step": 930
},
{
"epoch": 1.64,
"learning_rate": 0.00029542190305206463,
"loss": 6.367,
"step": 940
},
{
"epoch": 1.66,
"learning_rate": 0.0002952872531418312,
"loss": 6.3051,
"step": 950
},
{
"epoch": 1.68,
"learning_rate": 0.00029515260323159783,
"loss": 6.481,
"step": 960
},
{
"epoch": 1.7,
"learning_rate": 0.0002950179533213644,
"loss": 6.2663,
"step": 970
},
{
"epoch": 1.71,
"learning_rate": 0.00029488330341113103,
"loss": 6.4301,
"step": 980
},
{
"epoch": 1.73,
"learning_rate": 0.00029474865350089766,
"loss": 6.2978,
"step": 990
},
{
"epoch": 1.75,
"learning_rate": 0.0002946140035906643,
"loss": 6.2719,
"step": 1000
},
{
"epoch": 1.77,
"learning_rate": 0.00029447935368043086,
"loss": 6.41,
"step": 1010
},
{
"epoch": 1.78,
"learning_rate": 0.0002943447037701975,
"loss": 6.2289,
"step": 1020
},
{
"epoch": 1.8,
"learning_rate": 0.00029421005385996406,
"loss": 6.3948,
"step": 1030
},
{
"epoch": 1.82,
"learning_rate": 0.0002940754039497307,
"loss": 6.3203,
"step": 1040
},
{
"epoch": 1.84,
"learning_rate": 0.00029394075403949726,
"loss": 6.1822,
"step": 1050
},
{
"epoch": 1.85,
"learning_rate": 0.0002938061041292639,
"loss": 6.3675,
"step": 1060
},
{
"epoch": 1.87,
"learning_rate": 0.0002936714542190305,
"loss": 6.2043,
"step": 1070
},
{
"epoch": 1.89,
"learning_rate": 0.0002935368043087971,
"loss": 6.3436,
"step": 1080
},
{
"epoch": 1.91,
"learning_rate": 0.0002934021543985637,
"loss": 6.2709,
"step": 1090
},
{
"epoch": 1.92,
"learning_rate": 0.0002932675044883303,
"loss": 6.171,
"step": 1100
},
{
"epoch": 1.94,
"learning_rate": 0.0002931328545780969,
"loss": 6.296,
"step": 1110
},
{
"epoch": 1.96,
"learning_rate": 0.00029299820466786354,
"loss": 6.1652,
"step": 1120
},
{
"epoch": 1.98,
"learning_rate": 0.00029286355475763016,
"loss": 6.2756,
"step": 1130
},
{
"epoch": 1.99,
"learning_rate": 0.00029272890484739674,
"loss": 6.1802,
"step": 1140
},
{
"epoch": 2.01,
"learning_rate": 0.00029259425493716336,
"loss": 6.1664,
"step": 1150
},
{
"epoch": 2.03,
"learning_rate": 0.00029245960502692994,
"loss": 6.1878,
"step": 1160
},
{
"epoch": 2.05,
"learning_rate": 0.00029232495511669656,
"loss": 5.9369,
"step": 1170
},
{
"epoch": 2.06,
"learning_rate": 0.0002921903052064632,
"loss": 6.1776,
"step": 1180
},
{
"epoch": 2.08,
"learning_rate": 0.00029205565529622976,
"loss": 6.0142,
"step": 1190
},
{
"epoch": 2.1,
"learning_rate": 0.0002919210053859964,
"loss": 6.1265,
"step": 1200
},
{
"epoch": 2.1,
"eval_cer": 0.9626404073685787,
"eval_loss": 5.982735633850098,
"eval_runtime": 1169.314,
"eval_samples_per_second": 7.492,
"eval_wer": 1.0,
"step": 1200
},
{
"epoch": 2.12,
"learning_rate": 0.00029178635547576296,
"loss": 5.9301,
"step": 1210
},
{
"epoch": 2.13,
"learning_rate": 0.0002916517055655296,
"loss": 5.9003,
"step": 1220
},
{
"epoch": 2.15,
"learning_rate": 0.0002915170556552962,
"loss": 5.7485,
"step": 1230
},
{
"epoch": 2.17,
"learning_rate": 0.00029138240574506284,
"loss": 5.6099,
"step": 1240
},
{
"epoch": 2.19,
"learning_rate": 0.0002912477558348294,
"loss": 5.5064,
"step": 1250
},
{
"epoch": 2.2,
"learning_rate": 0.00029111310592459604,
"loss": 5.3511,
"step": 1260
},
{
"epoch": 2.22,
"learning_rate": 0.0002909784560143626,
"loss": 5.1143,
"step": 1270
},
{
"epoch": 2.24,
"learning_rate": 0.00029084380610412924,
"loss": 4.9744,
"step": 1280
},
{
"epoch": 2.26,
"learning_rate": 0.00029070915619389587,
"loss": 4.6601,
"step": 1290
},
{
"epoch": 2.27,
"learning_rate": 0.00029057450628366244,
"loss": 4.3969,
"step": 1300
},
{
"epoch": 2.29,
"learning_rate": 0.00029043985637342907,
"loss": 4.2051,
"step": 1310
},
{
"epoch": 2.31,
"learning_rate": 0.00029030520646319564,
"loss": 4.1824,
"step": 1320
},
{
"epoch": 2.33,
"learning_rate": 0.00029017055655296227,
"loss": 4.0095,
"step": 1330
},
{
"epoch": 2.34,
"learning_rate": 0.00029003590664272884,
"loss": 3.8963,
"step": 1340
},
{
"epoch": 2.36,
"learning_rate": 0.00028990125673249547,
"loss": 3.7867,
"step": 1350
},
{
"epoch": 2.38,
"learning_rate": 0.0002897666068222621,
"loss": 3.7578,
"step": 1360
},
{
"epoch": 2.4,
"learning_rate": 0.0002896319569120287,
"loss": 3.7065,
"step": 1370
},
{
"epoch": 2.41,
"learning_rate": 0.0002894973070017953,
"loss": 3.5965,
"step": 1380
},
{
"epoch": 2.43,
"learning_rate": 0.0002893626570915619,
"loss": 3.5597,
"step": 1390
},
{
"epoch": 2.45,
"learning_rate": 0.0002892280071813285,
"loss": 3.5877,
"step": 1400
},
{
"epoch": 2.47,
"learning_rate": 0.0002890933572710951,
"loss": 3.5316,
"step": 1410
},
{
"epoch": 2.48,
"learning_rate": 0.00028895870736086175,
"loss": 3.5905,
"step": 1420
},
{
"epoch": 2.5,
"learning_rate": 0.00028882405745062837,
"loss": 3.5515,
"step": 1430
},
{
"epoch": 2.52,
"learning_rate": 0.00028868940754039495,
"loss": 3.4082,
"step": 1440
},
{
"epoch": 2.53,
"learning_rate": 0.00028855475763016157,
"loss": 3.5074,
"step": 1450
},
{
"epoch": 2.55,
"learning_rate": 0.00028842010771992814,
"loss": 3.5773,
"step": 1460
},
{
"epoch": 2.57,
"learning_rate": 0.00028828545780969477,
"loss": 3.426,
"step": 1470
},
{
"epoch": 2.59,
"learning_rate": 0.0002881508078994614,
"loss": 3.3261,
"step": 1480
},
{
"epoch": 2.6,
"learning_rate": 0.00028801615798922797,
"loss": 3.3198,
"step": 1490
},
{
"epoch": 2.62,
"learning_rate": 0.0002878815080789946,
"loss": 3.3809,
"step": 1500
},
{
"epoch": 2.64,
"learning_rate": 0.00028774685816876117,
"loss": 3.3378,
"step": 1510
},
{
"epoch": 2.66,
"learning_rate": 0.0002876122082585278,
"loss": 3.2242,
"step": 1520
},
{
"epoch": 2.67,
"learning_rate": 0.0002874775583482944,
"loss": 3.2431,
"step": 1530
},
{
"epoch": 2.69,
"learning_rate": 0.00028734290843806105,
"loss": 3.2502,
"step": 1540
},
{
"epoch": 2.71,
"learning_rate": 0.0002872082585278276,
"loss": 3.2542,
"step": 1550
},
{
"epoch": 2.73,
"learning_rate": 0.00028707360861759425,
"loss": 3.1088,
"step": 1560
},
{
"epoch": 2.74,
"learning_rate": 0.0002869389587073608,
"loss": 3.0092,
"step": 1570
},
{
"epoch": 2.76,
"learning_rate": 0.00028680430879712745,
"loss": 3.0566,
"step": 1580
},
{
"epoch": 2.78,
"learning_rate": 0.000286669658886894,
"loss": 3.0401,
"step": 1590
},
{
"epoch": 2.8,
"learning_rate": 0.00028653500897666065,
"loss": 2.9808,
"step": 1600
},
{
"epoch": 2.81,
"learning_rate": 0.0002864003590664273,
"loss": 3.0996,
"step": 1610
},
{
"epoch": 2.83,
"learning_rate": 0.00028626570915619385,
"loss": 3.0149,
"step": 1620
},
{
"epoch": 2.85,
"learning_rate": 0.0002861310592459605,
"loss": 3.0153,
"step": 1630
},
{
"epoch": 2.87,
"learning_rate": 0.00028599640933572705,
"loss": 3.0267,
"step": 1640
},
{
"epoch": 2.88,
"learning_rate": 0.0002858617594254937,
"loss": 2.8632,
"step": 1650
},
{
"epoch": 2.9,
"learning_rate": 0.0002857271095152603,
"loss": 2.8897,
"step": 1660
},
{
"epoch": 2.92,
"learning_rate": 0.00028559245960502693,
"loss": 2.8773,
"step": 1670
},
{
"epoch": 2.94,
"learning_rate": 0.0002854578096947935,
"loss": 2.8656,
"step": 1680
},
{
"epoch": 2.95,
"learning_rate": 0.00028532315978456013,
"loss": 2.8248,
"step": 1690
},
{
"epoch": 2.97,
"learning_rate": 0.0002851885098743267,
"loss": 2.9358,
"step": 1700
},
{
"epoch": 2.99,
"learning_rate": 0.0002850538599640933,
"loss": 2.9389,
"step": 1710
},
{
"epoch": 3.01,
"learning_rate": 0.00028491921005385995,
"loss": 2.7758,
"step": 1720
},
{
"epoch": 3.02,
"learning_rate": 0.0002847845601436266,
"loss": 2.8451,
"step": 1730
},
{
"epoch": 3.04,
"learning_rate": 0.00028464991023339315,
"loss": 2.7879,
"step": 1740
},
{
"epoch": 3.06,
"learning_rate": 0.0002845152603231597,
"loss": 2.7708,
"step": 1750
},
{
"epoch": 3.08,
"learning_rate": 0.00028438061041292635,
"loss": 2.7697,
"step": 1760
},
{
"epoch": 3.09,
"learning_rate": 0.000284245960502693,
"loss": 2.6994,
"step": 1770
},
{
"epoch": 3.11,
"learning_rate": 0.0002841113105924596,
"loss": 2.646,
"step": 1780
},
{
"epoch": 3.13,
"learning_rate": 0.0002839766606822262,
"loss": 2.7175,
"step": 1790
},
{
"epoch": 3.15,
"learning_rate": 0.0002838420107719928,
"loss": 2.5818,
"step": 1800
},
{
"epoch": 3.15,
"eval_cer1": 0.5706086897808987,
"eval_cer2": 0.5588515176374077,
"eval_cer3": 0.5401176086246324,
"eval_cer4": 0.5407772304324029,
"eval_cer5": 0.5651804670912951,
"eval_loss": 2.5096516609191895,
"eval_runtime": 1141.0448,
"eval_samples_per_second": 7.677,
"step": 1800
},
{
"epoch": 3.16,
"learning_rate": 0.0002837073608617594,
"loss": 2.7269,
"step": 1810
},
{
"epoch": 3.18,
"learning_rate": 0.000283572710951526,
"loss": 2.8614,
"step": 1820
},
{
"epoch": 3.2,
"learning_rate": 0.00028343806104129263,
"loss": 2.6653,
"step": 1830
},
{
"epoch": 3.22,
"learning_rate": 0.00028330341113105926,
"loss": 2.689,
"step": 1840
},
{
"epoch": 3.23,
"learning_rate": 0.00028316876122082583,
"loss": 2.5852,
"step": 1850
},
{
"epoch": 3.25,
"learning_rate": 0.00028303411131059246,
"loss": 2.6783,
"step": 1860
},
{
"epoch": 3.27,
"learning_rate": 0.00028289946140035903,
"loss": 2.5618,
"step": 1870
},
{
"epoch": 3.29,
"learning_rate": 0.00028276481149012566,
"loss": 2.6193,
"step": 1880
},
{
"epoch": 3.3,
"learning_rate": 0.00028263016157989223,
"loss": 2.5681,
"step": 1890
},
{
"epoch": 3.32,
"learning_rate": 0.00028249551166965886,
"loss": 2.5623,
"step": 1900
},
{
"epoch": 3.32,
"eval_cer1": 0.5584551500624557,
"eval_cer2": 0.5466119770303528,
"eval_cer3": 0.522999019928128,
"eval_cer4": 0.524195885250652,
"eval_cer5": 0.5435244161358811,
"eval_loss": 2.379859447479248,
"eval_runtime": 1137.4958,
"eval_samples_per_second": 7.701,
"step": 1900
},
{
"epoch": 3.34,
"learning_rate": 0.0002823608617594255,
"loss": 2.5353,
"step": 1910
},
{
"epoch": 3.36,
"learning_rate": 0.00028222621184919206,
"loss": 2.5668,
"step": 1920
},
{
"epoch": 3.37,
"learning_rate": 0.0002820915619389587,
"loss": 2.4749,
"step": 1930
},
{
"epoch": 3.39,
"learning_rate": 0.00028195691202872526,
"loss": 2.4638,
"step": 1940
},
{
"epoch": 3.41,
"learning_rate": 0.0002818222621184919,
"loss": 2.53,
"step": 1950
},
{
"epoch": 3.43,
"learning_rate": 0.0002816876122082585,
"loss": 2.3977,
"step": 1960
},
{
"epoch": 3.44,
"learning_rate": 0.00028155296229802514,
"loss": 2.4476,
"step": 1970
},
{
"epoch": 3.46,
"learning_rate": 0.0002814183123877917,
"loss": 2.3519,
"step": 1980
},
{
"epoch": 3.48,
"learning_rate": 0.00028128366247755834,
"loss": 2.3922,
"step": 1990
},
{
"epoch": 3.5,
"learning_rate": 0.0002811490125673249,
"loss": 2.3761,
"step": 2000
},
{
"epoch": 3.51,
"learning_rate": 0.00028101436265709154,
"loss": 2.4301,
"step": 2010
},
{
"epoch": 3.53,
"learning_rate": 0.00028087971274685816,
"loss": 2.5032,
"step": 2020
},
{
"epoch": 3.55,
"learning_rate": 0.00028074506283662473,
"loss": 2.4313,
"step": 2030
},
{
"epoch": 3.57,
"learning_rate": 0.00028061041292639136,
"loss": 2.3814,
"step": 2040
},
{
"epoch": 3.58,
"learning_rate": 0.00028047576301615793,
"loss": 2.4016,
"step": 2050
},
{
"epoch": 3.6,
"learning_rate": 0.00028034111310592456,
"loss": 2.2827,
"step": 2060
},
{
"epoch": 3.62,
"learning_rate": 0.0002802064631956912,
"loss": 2.3379,
"step": 2070
},
{
"epoch": 3.64,
"learning_rate": 0.0002800718132854578,
"loss": 2.3315,
"step": 2080
},
{
"epoch": 3.65,
"learning_rate": 0.0002799371633752244,
"loss": 2.2943,
"step": 2090
},
{
"epoch": 3.67,
"learning_rate": 0.000279802513464991,
"loss": 2.4029,
"step": 2100
},
{
"epoch": 3.69,
"learning_rate": 0.0002796678635547576,
"loss": 2.3145,
"step": 2110
},
{
"epoch": 3.71,
"learning_rate": 0.0002795332136445242,
"loss": 2.2538,
"step": 2120
},
{
"epoch": 3.72,
"learning_rate": 0.00027939856373429084,
"loss": 2.2717,
"step": 2130
},
{
"epoch": 3.74,
"learning_rate": 0.0002792639138240574,
"loss": 2.1891,
"step": 2140
},
{
"epoch": 3.76,
"learning_rate": 0.00027912926391382404,
"loss": 2.2323,
"step": 2150
},
{
"epoch": 3.78,
"learning_rate": 0.00027899461400359067,
"loss": 2.2378,
"step": 2160
},
{
"epoch": 3.79,
"learning_rate": 0.00027885996409335724,
"loss": 2.2133,
"step": 2170
},
{
"epoch": 3.81,
"learning_rate": 0.0002787253141831238,
"loss": 2.1628,
"step": 2180
},
{
"epoch": 3.83,
"learning_rate": 0.00027859066427289044,
"loss": 2.1824,
"step": 2190
},
{
"epoch": 3.85,
"learning_rate": 0.00027845601436265706,
"loss": 2.1867,
"step": 2200
},
{
"epoch": 1.93,
"learning_rate": 0.00028929521276595743,
"loss": 2.1458,
"step": 2210
},
{
"epoch": 1.94,
"learning_rate": 0.00028922872340425527,
"loss": 2.2512,
"step": 2220
},
{
"epoch": 1.95,
"learning_rate": 0.00028916223404255317,
"loss": 2.2017,
"step": 2230
},
{
"epoch": 1.96,
"learning_rate": 0.000289095744680851,
"loss": 2.2122,
"step": 2240
},
{
"epoch": 1.97,
"learning_rate": 0.0002890292553191489,
"loss": 2.2943,
"step": 2250
},
{
"epoch": 1.98,
"learning_rate": 0.00028896276595744675,
"loss": 2.1862,
"step": 2260
},
{
"epoch": 1.99,
"learning_rate": 0.00028889627659574465,
"loss": 2.1958,
"step": 2270
},
{
"epoch": 1.99,
"learning_rate": 0.00028882978723404254,
"loss": 2.1499,
"step": 2280
},
{
"epoch": 2.0,
"learning_rate": 0.0002887632978723404,
"loss": 2.4003,
"step": 2290
},
{
"epoch": 2.01,
"learning_rate": 0.0002886968085106383,
"loss": 2.205,
"step": 2300
},
{
"epoch": 2.02,
"learning_rate": 0.0002886303191489361,
"loss": 2.2114,
"step": 2310
},
{
"epoch": 2.03,
"learning_rate": 0.000288563829787234,
"loss": 2.2381,
"step": 2320
},
{
"epoch": 2.04,
"learning_rate": 0.00028849734042553187,
"loss": 2.0985,
"step": 2330
},
{
"epoch": 2.05,
"learning_rate": 0.00028843085106382976,
"loss": 2.3753,
"step": 2340
},
{
"epoch": 2.06,
"learning_rate": 0.00028836436170212766,
"loss": 2.1873,
"step": 2350
},
{
"epoch": 2.06,
"learning_rate": 0.0002882978723404255,
"loss": 2.2122,
"step": 2360
},
{
"epoch": 2.07,
"learning_rate": 0.0002882313829787234,
"loss": 2.0525,
"step": 2370
},
{
"epoch": 2.08,
"learning_rate": 0.00028816489361702124,
"loss": 2.1569,
"step": 2380
},
{
"epoch": 2.09,
"learning_rate": 0.00028809840425531914,
"loss": 2.2276,
"step": 2390
},
{
"epoch": 2.1,
"learning_rate": 0.000288031914893617,
"loss": 2.0076,
"step": 2400
},
{
"epoch": 2.1,
"eval_cer1": 0.48924749333243306,
"eval_cer2": 0.47238720262510253,
"eval_cer3": 0.4526951976478275,
"eval_cer4": 0.45085160500982,
"eval_cer5": 0.46989384288747343,
"eval_loss": 1.9238340854644775,
"eval_runtime": 1120.0867,
"eval_samples_per_second": 7.821,
"step": 2400
},
{
"epoch": 2.11,
"learning_rate": 0.0002879654255319149,
"loss": 1.9779,
"step": 2410
},
{
"epoch": 2.12,
"learning_rate": 0.0002878989361702127,
"loss": 2.1389,
"step": 2420
},
{
"epoch": 2.13,
"learning_rate": 0.0002878324468085106,
"loss": 2.2141,
"step": 2430
},
{
"epoch": 2.13,
"learning_rate": 0.0002877659574468085,
"loss": 2.251,
"step": 2440
},
{
"epoch": 2.14,
"learning_rate": 0.00028769946808510636,
"loss": 2.2231,
"step": 2450
},
{
"epoch": 2.15,
"learning_rate": 0.00028763297872340426,
"loss": 2.155,
"step": 2460
},
{
"epoch": 2.16,
"learning_rate": 0.0002875664893617021,
"loss": 2.1766,
"step": 2470
},
{
"epoch": 2.17,
"learning_rate": 0.0002875,
"loss": 2.079,
"step": 2480
},
{
"epoch": 2.18,
"learning_rate": 0.00028743351063829784,
"loss": 2.1769,
"step": 2490
},
{
"epoch": 2.19,
"learning_rate": 0.00028736702127659573,
"loss": 2.131,
"step": 2500
},
{
"epoch": 2.2,
"learning_rate": 0.00028730053191489363,
"loss": 2.1497,
"step": 2510
},
{
"epoch": 2.2,
"learning_rate": 0.0002872340425531915,
"loss": 2.1463,
"step": 2520
},
{
"epoch": 2.21,
"learning_rate": 0.00028716755319148937,
"loss": 2.1748,
"step": 2530
},
{
"epoch": 2.22,
"learning_rate": 0.0002871010638297872,
"loss": 2.2521,
"step": 2540
},
{
"epoch": 2.23,
"learning_rate": 0.00028703457446808506,
"loss": 2.0849,
"step": 2550
},
{
"epoch": 2.24,
"learning_rate": 0.00028696808510638295,
"loss": 2.0427,
"step": 2560
},
{
"epoch": 2.25,
"learning_rate": 0.0002869015957446808,
"loss": 2.0855,
"step": 2570
},
{
"epoch": 2.26,
"learning_rate": 0.0002868351063829787,
"loss": 2.0552,
"step": 2580
},
{
"epoch": 2.27,
"learning_rate": 0.00028676861702127654,
"loss": 2.1035,
"step": 2590
},
{
"epoch": 2.27,
"learning_rate": 0.00028670212765957443,
"loss": 2.1035,
"step": 2600
},
{
"epoch": 2.28,
"learning_rate": 0.00028663563829787233,
"loss": 2.1345,
"step": 2610
},
{
"epoch": 2.29,
"learning_rate": 0.0002865691489361702,
"loss": 2.1135,
"step": 2620
},
{
"epoch": 2.3,
"learning_rate": 0.00028650265957446807,
"loss": 2.1364,
"step": 2630
},
{
"epoch": 2.31,
"learning_rate": 0.0002864361702127659,
"loss": 2.3248,
"step": 2640
},
{
"epoch": 2.32,
"learning_rate": 0.0002863696808510638,
"loss": 2.3336,
"step": 2650
},
{
"epoch": 2.33,
"learning_rate": 0.00028630319148936165,
"loss": 2.1158,
"step": 2660
},
{
"epoch": 2.34,
"learning_rate": 0.00028623670212765955,
"loss": 2.1155,
"step": 2670
},
{
"epoch": 2.34,
"learning_rate": 0.0002861702127659574,
"loss": 2.1989,
"step": 2680
},
{
"epoch": 2.35,
"learning_rate": 0.0002861037234042553,
"loss": 2.2268,
"step": 2690
},
{
"epoch": 2.36,
"learning_rate": 0.0002860372340425532,
"loss": 2.0541,
"step": 2700
},
{
"epoch": 2.37,
"learning_rate": 0.00028597074468085103,
"loss": 1.9792,
"step": 2710
},
{
"epoch": 2.38,
"learning_rate": 0.0002859042553191489,
"loss": 2.2246,
"step": 2720
},
{
"epoch": 2.39,
"learning_rate": 0.00028583776595744677,
"loss": 1.9866,
"step": 2730
},
{
"epoch": 2.4,
"learning_rate": 0.00028577127659574467,
"loss": 2.1589,
"step": 2740
},
{
"epoch": 2.41,
"learning_rate": 0.0002857047872340425,
"loss": 2.0232,
"step": 2750
},
{
"epoch": 2.41,
"learning_rate": 0.0002856382978723404,
"loss": 2.0295,
"step": 2760
},
{
"epoch": 2.42,
"learning_rate": 0.0002855718085106383,
"loss": 2.0344,
"step": 2770
},
{
"epoch": 2.43,
"learning_rate": 0.00028550531914893615,
"loss": 2.0738,
"step": 2780
},
{
"epoch": 2.44,
"learning_rate": 0.00028543882978723404,
"loss": 2.2399,
"step": 2790
},
{
"epoch": 2.45,
"learning_rate": 0.0002853723404255319,
"loss": 2.1269,
"step": 2800
},
{
"epoch": 2.46,
"learning_rate": 0.0002853058510638298,
"loss": 2.1242,
"step": 2810
},
{
"epoch": 2.47,
"learning_rate": 0.0002852393617021276,
"loss": 2.1683,
"step": 2820
},
{
"epoch": 2.48,
"learning_rate": 0.0002851728723404255,
"loss": 1.9675,
"step": 2830
},
{
"epoch": 2.48,
"learning_rate": 0.00028510638297872336,
"loss": 2.0978,
"step": 2840
},
{
"epoch": 2.49,
"learning_rate": 0.00028503989361702126,
"loss": 2.1067,
"step": 2850
},
{
"epoch": 2.5,
"learning_rate": 0.00028497340425531916,
"loss": 1.9776,
"step": 2860
},
{
"epoch": 2.51,
"learning_rate": 0.000284906914893617,
"loss": 1.9753,
"step": 2870
},
{
"epoch": 2.52,
"learning_rate": 0.0002848404255319149,
"loss": 2.0909,
"step": 2880
},
{
"epoch": 2.53,
"learning_rate": 0.00028477393617021274,
"loss": 1.9942,
"step": 2890
},
{
"epoch": 2.54,
"learning_rate": 0.0002847074468085106,
"loss": 1.9181,
"step": 2900
},
{
"epoch": 2.55,
"learning_rate": 0.0002846409574468085,
"loss": 2.1184,
"step": 2910
},
{
"epoch": 2.55,
"learning_rate": 0.0002845744680851063,
"loss": 2.0608,
"step": 2920
},
{
"epoch": 2.56,
"learning_rate": 0.0002845079787234042,
"loss": 2.1055,
"step": 2930
},
{
"epoch": 2.57,
"learning_rate": 0.0002844414893617021,
"loss": 2.1383,
"step": 2940
},
{
"epoch": 2.58,
"learning_rate": 0.00028437499999999996,
"loss": 1.97,
"step": 2950
},
{
"epoch": 2.59,
"learning_rate": 0.00028430851063829786,
"loss": 2.0785,
"step": 2960
},
{
"epoch": 2.6,
"learning_rate": 0.0002842420212765957,
"loss": 2.1369,
"step": 2970
},
{
"epoch": 2.61,
"learning_rate": 0.0002841755319148936,
"loss": 2.0299,
"step": 2980
},
{
"epoch": 2.62,
"learning_rate": 0.00028410904255319144,
"loss": 2.0835,
"step": 2990
},
{
"epoch": 2.62,
"learning_rate": 0.00028404255319148934,
"loss": 1.9336,
"step": 3000
},
{
"epoch": 2.62,
"eval_cer1": 0.47790418959521963,
"eval_cer2": 0.46027891714520097,
"eval_cer3": 0.4367200261352499,
"eval_cer4": 0.43565472165877844,
"eval_cer5": 0.4569851380042463,
"eval_loss": 1.8061408996582031,
"eval_runtime": 1152.9636,
"eval_samples_per_second": 7.598,
"step": 3000
},
{
"epoch": 2.63,
"learning_rate": 0.0002839760638297872,
"loss": 1.9459,
"step": 3010
},
{
"epoch": 2.64,
"learning_rate": 0.0002839095744680851,
"loss": 1.8683,
"step": 3020
},
{
"epoch": 2.65,
"learning_rate": 0.000283843085106383,
"loss": 1.8963,
"step": 3030
},
{
"epoch": 2.66,
"learning_rate": 0.0002837765957446808,
"loss": 2.0346,
"step": 3040
},
{
"epoch": 2.67,
"learning_rate": 0.0002837101063829787,
"loss": 2.0598,
"step": 3050
},
{
"epoch": 2.68,
"learning_rate": 0.00028364361702127656,
"loss": 2.1834,
"step": 3060
},
{
"epoch": 2.69,
"learning_rate": 0.00028357712765957445,
"loss": 2.0357,
"step": 3070
},
{
"epoch": 2.69,
"learning_rate": 0.0002835106382978723,
"loss": 2.0925,
"step": 3080
},
{
"epoch": 2.7,
"learning_rate": 0.0002834441489361702,
"loss": 2.2516,
"step": 3090
},
{
"epoch": 2.71,
"learning_rate": 0.0002833776595744681,
"loss": 2.2491,
"step": 3100
},
{
"epoch": 2.72,
"learning_rate": 0.00028331117021276593,
"loss": 2.2018,
"step": 3110
},
{
"epoch": 2.73,
"learning_rate": 0.00028324468085106383,
"loss": 2.1119,
"step": 3120
},
{
"epoch": 2.74,
"learning_rate": 0.00028317819148936167,
"loss": 2.1143,
"step": 3130
},
{
"epoch": 2.75,
"learning_rate": 0.00028311170212765957,
"loss": 2.3422,
"step": 3140
},
{
"epoch": 2.76,
"learning_rate": 0.0002830452127659574,
"loss": 2.1345,
"step": 3150
},
{
"epoch": 2.76,
"learning_rate": 0.0002829787234042553,
"loss": 2.2023,
"step": 3160
},
{
"epoch": 2.77,
"learning_rate": 0.00028291223404255315,
"loss": 2.0969,
"step": 3170
},
{
"epoch": 2.78,
"learning_rate": 0.00028284574468085105,
"loss": 2.1492,
"step": 3180
},
{
"epoch": 2.79,
"learning_rate": 0.00028277925531914894,
"loss": 2.1605,
"step": 3190
},
{
"epoch": 2.8,
"learning_rate": 0.0002827127659574468,
"loss": 2.0944,
"step": 3200
},
{
"epoch": 2.81,
"learning_rate": 0.0002826462765957447,
"loss": 1.9815,
"step": 3210
},
{
"epoch": 2.82,
"learning_rate": 0.00028257978723404253,
"loss": 2.0554,
"step": 3220
},
{
"epoch": 2.83,
"learning_rate": 0.0002825132978723404,
"loss": 2.007,
"step": 3230
},
{
"epoch": 2.83,
"learning_rate": 0.00028244680851063827,
"loss": 2.0277,
"step": 3240
},
{
"epoch": 2.84,
"learning_rate": 0.0002823803191489361,
"loss": 1.9326,
"step": 3250
},
{
"epoch": 2.85,
"learning_rate": 0.000282313829787234,
"loss": 1.8932,
"step": 3260
},
{
"epoch": 2.86,
"learning_rate": 0.00028224734042553185,
"loss": 2.0032,
"step": 3270
},
{
"epoch": 2.87,
"learning_rate": 0.00028218085106382975,
"loss": 2.1045,
"step": 3280
},
{
"epoch": 2.88,
"learning_rate": 0.00028211436170212764,
"loss": 1.976,
"step": 3290
},
{
"epoch": 2.89,
"learning_rate": 0.0002820478723404255,
"loss": 1.8952,
"step": 3300
},
{
"epoch": 2.9,
"learning_rate": 0.0002819813829787234,
"loss": 1.988,
"step": 3310
},
{
"epoch": 2.9,
"learning_rate": 0.0002819148936170212,
"loss": 2.1143,
"step": 3320
},
{
"epoch": 2.91,
"learning_rate": 0.0002818484042553191,
"loss": 1.9619,
"step": 3330
},
{
"epoch": 2.92,
"learning_rate": 0.00028178191489361697,
"loss": 2.0878,
"step": 3340
},
{
"epoch": 2.93,
"learning_rate": 0.00028171542553191486,
"loss": 1.9068,
"step": 3350
},
{
"epoch": 2.94,
"learning_rate": 0.00028164893617021276,
"loss": 2.0037,
"step": 3360
},
{
"epoch": 2.95,
"learning_rate": 0.0002815824468085106,
"loss": 1.9381,
"step": 3370
},
{
"epoch": 2.96,
"learning_rate": 0.0002815159574468085,
"loss": 1.9552,
"step": 3380
},
{
"epoch": 2.97,
"learning_rate": 0.00028144946808510634,
"loss": 1.9375,
"step": 3390
},
{
"epoch": 2.97,
"learning_rate": 0.00028138297872340424,
"loss": 2.0038,
"step": 3400
},
{
"epoch": 2.98,
"learning_rate": 0.0002813164893617021,
"loss": 1.9518,
"step": 3410
},
{
"epoch": 2.99,
"learning_rate": 0.00028125,
"loss": 2.2452,
"step": 3420
},
{
"epoch": 3.0,
"learning_rate": 0.0002811835106382979,
"loss": 2.4543,
"step": 3430
},
{
"epoch": 3.01,
"learning_rate": 0.0002811170212765957,
"loss": 2.0192,
"step": 3440
},
{
"epoch": 3.02,
"learning_rate": 0.0002810505319148936,
"loss": 2.1015,
"step": 3450
},
{
"epoch": 3.03,
"learning_rate": 0.00028098404255319146,
"loss": 1.971,
"step": 3460
},
{
"epoch": 3.04,
"learning_rate": 0.00028091755319148936,
"loss": 1.9138,
"step": 3470
},
{
"epoch": 3.04,
"learning_rate": 0.0002808510638297872,
"loss": 1.938,
"step": 3480
},
{
"epoch": 3.05,
"learning_rate": 0.0002807845744680851,
"loss": 1.9073,
"step": 3490
},
{
"epoch": 3.06,
"learning_rate": 0.00028071808510638294,
"loss": 1.8454,
"step": 3500
},
{
"epoch": 3.07,
"learning_rate": 0.00028065159574468083,
"loss": 1.9872,
"step": 3510
},
{
"epoch": 3.08,
"learning_rate": 0.00028058510638297873,
"loss": 1.8779,
"step": 3520
},
{
"epoch": 3.09,
"learning_rate": 0.0002805186170212766,
"loss": 1.9469,
"step": 3530
},
{
"epoch": 3.1,
"learning_rate": 0.00028045212765957447,
"loss": 2.0581,
"step": 3540
},
{
"epoch": 3.11,
"learning_rate": 0.0002803856382978723,
"loss": 1.8872,
"step": 3550
},
{
"epoch": 3.11,
"learning_rate": 0.0002803191489361702,
"loss": 2.0156,
"step": 3560
},
{
"epoch": 3.12,
"learning_rate": 0.00028025265957446805,
"loss": 2.1001,
"step": 3570
},
{
"epoch": 3.13,
"learning_rate": 0.00028018617021276595,
"loss": 2.1155,
"step": 3580
},
{
"epoch": 3.14,
"learning_rate": 0.00028011968085106385,
"loss": 2.121,
"step": 3590
},
{
"epoch": 3.15,
"learning_rate": 0.00028005319148936164,
"loss": 2.0027,
"step": 3600
},
{
"epoch": 3.15,
"eval_cer1": 0.47597987913979944,
"eval_cer2": 0.45689909762100084,
"eval_cer3": 0.4369160405096374,
"eval_cer4": 0.43597668952638524,
"eval_cer5": 0.4563906581740977,
"eval_loss": 1.800094723701477,
"eval_runtime": 1141.7387,
"eval_samples_per_second": 7.673,
"step": 3600
},
{
"epoch": 3.16,
"learning_rate": 0.00027998670212765953,
"loss": 1.9812,
"step": 3610
},
{
"epoch": 3.17,
"learning_rate": 0.00027992021276595743,
"loss": 2.0124,
"step": 3620
},
{
"epoch": 3.18,
"learning_rate": 0.0002798537234042553,
"loss": 1.9499,
"step": 3630
},
{
"epoch": 3.18,
"learning_rate": 0.00027978723404255317,
"loss": 1.9615,
"step": 3640
},
{
"epoch": 3.19,
"learning_rate": 0.000279720744680851,
"loss": 1.9805,
"step": 3650
},
{
"epoch": 3.2,
"learning_rate": 0.0002796542553191489,
"loss": 1.9327,
"step": 3660
},
{
"epoch": 3.21,
"learning_rate": 0.00027958776595744675,
"loss": 2.004,
"step": 3670
},
{
"epoch": 3.22,
"learning_rate": 0.00027952127659574465,
"loss": 2.0279,
"step": 3680
},
{
"epoch": 3.23,
"learning_rate": 0.00027945478723404255,
"loss": 1.9121,
"step": 3690
},
{
"epoch": 3.24,
"learning_rate": 0.0002793882978723404,
"loss": 1.9119,
"step": 3700
},
{
"epoch": 3.25,
"learning_rate": 0.0002793218085106383,
"loss": 2.0259,
"step": 3710
},
{
"epoch": 3.25,
"learning_rate": 0.00027925531914893613,
"loss": 1.9752,
"step": 3720
},
{
"epoch": 3.26,
"learning_rate": 0.000279188829787234,
"loss": 1.9925,
"step": 3730
},
{
"epoch": 3.27,
"learning_rate": 0.00027912234042553187,
"loss": 1.9936,
"step": 3740
},
{
"epoch": 3.28,
"learning_rate": 0.00027905585106382977,
"loss": 1.8995,
"step": 3750
},
{
"epoch": 3.29,
"learning_rate": 0.0002789893617021276,
"loss": 1.9581,
"step": 3760
},
{
"epoch": 3.3,
"learning_rate": 0.0002789228723404255,
"loss": 1.8888,
"step": 3770
},
{
"epoch": 3.31,
"learning_rate": 0.0002788563829787234,
"loss": 1.9329,
"step": 3780
},
{
"epoch": 3.32,
"learning_rate": 0.00027878989361702124,
"loss": 2.0032,
"step": 3790
},
{
"epoch": 3.32,
"learning_rate": 0.00027872340425531914,
"loss": 1.9863,
"step": 3800
},
{
"epoch": 3.33,
"learning_rate": 0.000278656914893617,
"loss": 1.9392,
"step": 3810
},
{
"epoch": 3.34,
"learning_rate": 0.0002785904255319149,
"loss": 1.8759,
"step": 3820
},
{
"epoch": 3.35,
"learning_rate": 0.0002785239361702127,
"loss": 1.9832,
"step": 3830
},
{
"epoch": 3.36,
"learning_rate": 0.0002784574468085106,
"loss": 1.9111,
"step": 3840
},
{
"epoch": 3.37,
"learning_rate": 0.0002783909574468085,
"loss": 1.9362,
"step": 3850
},
{
"epoch": 3.38,
"learning_rate": 0.00027832446808510636,
"loss": 2.1686,
"step": 3860
},
{
"epoch": 3.39,
"learning_rate": 0.00027825797872340426,
"loss": 1.8924,
"step": 3870
},
{
"epoch": 3.39,
"learning_rate": 0.0002781914893617021,
"loss": 2.0049,
"step": 3880
},
{
"epoch": 3.4,
"learning_rate": 0.000278125,
"loss": 1.871,
"step": 3890
},
{
"epoch": 3.41,
"learning_rate": 0.00027805851063829784,
"loss": 1.7214,
"step": 3900
},
{
"epoch": 3.42,
"learning_rate": 0.00027799202127659574,
"loss": 1.8852,
"step": 3910
},
{
"epoch": 3.43,
"learning_rate": 0.0002779255319148936,
"loss": 1.8057,
"step": 3920
},
{
"epoch": 3.44,
"learning_rate": 0.0002778590425531915,
"loss": 2.0098,
"step": 3930
},
{
"epoch": 3.45,
"learning_rate": 0.0002777925531914894,
"loss": 1.8596,
"step": 3940
},
{
"epoch": 3.46,
"learning_rate": 0.0002777260638297872,
"loss": 1.881,
"step": 3950
},
{
"epoch": 3.46,
"learning_rate": 0.00027765957446808506,
"loss": 2.0048,
"step": 3960
},
{
"epoch": 3.47,
"learning_rate": 0.00027759308510638296,
"loss": 1.9547,
"step": 3970
},
{
"epoch": 3.48,
"learning_rate": 0.0002775265957446808,
"loss": 2.0153,
"step": 3980
},
{
"epoch": 3.49,
"learning_rate": 0.0002774601063829787,
"loss": 1.8253,
"step": 3990
},
{
"epoch": 3.5,
"learning_rate": 0.00027739361702127654,
"loss": 1.9285,
"step": 4000
},
{
"epoch": 3.51,
"learning_rate": 0.00027732712765957444,
"loss": 1.9543,
"step": 4010
},
{
"epoch": 3.52,
"learning_rate": 0.00027726063829787233,
"loss": 1.8083,
"step": 4020
},
{
"epoch": 3.53,
"learning_rate": 0.0002771941489361702,
"loss": 1.8892,
"step": 4030
},
{
"epoch": 3.53,
"learning_rate": 0.00027712765957446807,
"loss": 1.8685,
"step": 4040
},
{
"epoch": 3.54,
"learning_rate": 0.0002770611702127659,
"loss": 1.7633,
"step": 4050
},
{
"epoch": 3.55,
"learning_rate": 0.0002769946808510638,
"loss": 1.801,
"step": 4060
},
{
"epoch": 3.56,
"learning_rate": 0.00027692819148936166,
"loss": 1.8555,
"step": 4070
},
{
"epoch": 3.57,
"learning_rate": 0.00027686170212765955,
"loss": 1.8881,
"step": 4080
},
{
"epoch": 3.58,
"learning_rate": 0.0002767952127659574,
"loss": 1.8753,
"step": 4090
},
{
"epoch": 3.59,
"learning_rate": 0.0002767287234042553,
"loss": 1.8913,
"step": 4100
},
{
"epoch": 3.6,
"learning_rate": 0.0002766622340425532,
"loss": 2.0024,
"step": 4110
},
{
"epoch": 3.6,
"learning_rate": 0.00027659574468085103,
"loss": 2.0692,
"step": 4120
},
{
"epoch": 3.61,
"learning_rate": 0.00027652925531914893,
"loss": 1.9269,
"step": 4130
},
{
"epoch": 3.62,
"learning_rate": 0.00027646276595744677,
"loss": 1.9242,
"step": 4140
},
{
"epoch": 3.63,
"learning_rate": 0.00027639627659574467,
"loss": 1.8783,
"step": 4150
},
{
"epoch": 3.64,
"learning_rate": 0.0002763297872340425,
"loss": 1.7259,
"step": 4160
},
{
"epoch": 3.65,
"learning_rate": 0.0002762632978723404,
"loss": 1.7952,
"step": 4170
},
{
"epoch": 3.66,
"learning_rate": 0.0002761968085106383,
"loss": 1.7268,
"step": 4180
},
{
"epoch": 3.67,
"learning_rate": 0.00027613031914893615,
"loss": 1.804,
"step": 4190
},
{
"epoch": 3.67,
"learning_rate": 0.00027606382978723404,
"loss": 1.7527,
"step": 4200
},
{
"epoch": 3.67,
"eval_cer1": 0.45643293609263696,
"eval_cer2": 0.4330434782608696,
"eval_cer3": 0.41613851682456715,
"eval_cer4": 0.41276280627193407,
"eval_cer5": 0.4343099787685775,
"eval_loss": 1.6673153638839722,
"eval_runtime": 1162.9114,
"eval_samples_per_second": 7.533,
"step": 4200
},
{
"epoch": 3.68,
"learning_rate": 0.0002759973404255319,
"loss": 1.8746,
"step": 4210
},
{
"epoch": 3.69,
"learning_rate": 0.0002759308510638298,
"loss": 1.8192,
"step": 4220
},
{
"epoch": 3.7,
"learning_rate": 0.0002758643617021276,
"loss": 1.9139,
"step": 4230
},
{
"epoch": 3.71,
"learning_rate": 0.0002757978723404255,
"loss": 1.8454,
"step": 4240
},
{
"epoch": 3.72,
"learning_rate": 0.00027573138297872337,
"loss": 1.8008,
"step": 4250
},
{
"epoch": 3.73,
"learning_rate": 0.00027566489361702126,
"loss": 1.8388,
"step": 4260
},
{
"epoch": 3.74,
"learning_rate": 0.00027559840425531916,
"loss": 1.6761,
"step": 4270
},
{
"epoch": 3.74,
"learning_rate": 0.000275531914893617,
"loss": 1.7665,
"step": 4280
},
{
"epoch": 3.75,
"learning_rate": 0.0002754654255319149,
"loss": 1.7748,
"step": 4290
},
{
"epoch": 3.76,
"learning_rate": 0.00027539893617021274,
"loss": 1.7312,
"step": 4300
},
{
"epoch": 3.77,
"learning_rate": 0.0002753324468085106,
"loss": 1.849,
"step": 4310
},
{
"epoch": 3.78,
"learning_rate": 0.0002752659574468085,
"loss": 1.8165,
"step": 4320
},
{
"epoch": 3.79,
"learning_rate": 0.0002751994680851063,
"loss": 1.8257,
"step": 4330
},
{
"epoch": 3.8,
"learning_rate": 0.0002751329787234042,
"loss": 1.7717,
"step": 4340
},
{
"epoch": 3.81,
"learning_rate": 0.00027506648936170207,
"loss": 1.8083,
"step": 4350
},
{
"epoch": 3.81,
"learning_rate": 0.00027499999999999996,
"loss": 1.8256,
"step": 4360
},
{
"epoch": 3.82,
"learning_rate": 0.00027493351063829786,
"loss": 1.716,
"step": 4370
},
{
"epoch": 3.83,
"learning_rate": 0.0002748670212765957,
"loss": 1.8734,
"step": 4380
},
{
"epoch": 3.84,
"learning_rate": 0.0002748005319148936,
"loss": 1.7568,
"step": 4390
},
{
"epoch": 3.85,
"learning_rate": 0.00027473404255319144,
"loss": 1.8443,
"step": 4400
},
{
"epoch": 3.86,
"learning_rate": 0.00027466755319148934,
"loss": 1.8433,
"step": 4410
},
{
"epoch": 3.87,
"learning_rate": 0.0002746010638297872,
"loss": 1.7142,
"step": 4420
},
{
"epoch": 3.88,
"learning_rate": 0.0002745345744680851,
"loss": 1.8259,
"step": 4430
},
{
"epoch": 3.88,
"learning_rate": 0.000274468085106383,
"loss": 1.7237,
"step": 4440
},
{
"epoch": 3.89,
"learning_rate": 0.0002744015957446808,
"loss": 1.7902,
"step": 4450
},
{
"epoch": 3.9,
"learning_rate": 0.0002743351063829787,
"loss": 1.7954,
"step": 4460
},
{
"epoch": 3.91,
"learning_rate": 0.00027426861702127656,
"loss": 1.7452,
"step": 4470
},
{
"epoch": 3.92,
"learning_rate": 0.00027420212765957445,
"loss": 1.7052,
"step": 4480
},
{
"epoch": 3.93,
"learning_rate": 0.0002741356382978723,
"loss": 1.8716,
"step": 4490
},
{
"epoch": 3.94,
"learning_rate": 0.0002740691489361702,
"loss": 1.9407,
"step": 4500
},
{
"epoch": 3.95,
"learning_rate": 0.00027400265957446804,
"loss": 1.7709,
"step": 4510
},
{
"epoch": 3.95,
"learning_rate": 0.00027393617021276593,
"loss": 1.8658,
"step": 4520
},
{
"epoch": 3.96,
"learning_rate": 0.00027386968085106383,
"loss": 1.9818,
"step": 4530
},
{
"epoch": 3.97,
"learning_rate": 0.0002738031914893617,
"loss": 1.8163,
"step": 4540
},
{
"epoch": 3.98,
"learning_rate": 0.00027373670212765957,
"loss": 1.7863,
"step": 4550
},
{
"epoch": 3.99,
"learning_rate": 0.0002736702127659574,
"loss": 1.8471,
"step": 4560
},
{
"epoch": 4.0,
"learning_rate": 0.0002736037234042553,
"loss": 1.8029,
"step": 4570
},
{
"epoch": 4.01,
"learning_rate": 0.00027353723404255315,
"loss": 1.8216,
"step": 4580
},
{
"epoch": 4.02,
"learning_rate": 0.00027347074468085105,
"loss": 1.6946,
"step": 4590
},
{
"epoch": 4.02,
"learning_rate": 0.00027340425531914895,
"loss": 1.8107,
"step": 4600
}
],
"max_steps": 45720,
"num_train_epochs": 40,
"total_flos": 4.155029507216204e+19,
"trial_name": null,
"trial_params": null
}