{
"best_metric": 5.995427264932838,
"best_model_checkpoint": "./checkpoint-9000",
"epoch": 19.0272,
"eval_steps": 1000,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.6000000000000004e-07,
"loss": 1.0392,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.600000000000001e-07,
"loss": 0.6818,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.46e-06,
"loss": 0.345,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.9600000000000003e-06,
"loss": 0.2564,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.46e-06,
"loss": 0.2228,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.96e-06,
"loss": 0.1938,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.46e-06,
"loss": 0.1902,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.96e-06,
"loss": 0.1887,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.4600000000000005e-06,
"loss": 0.1764,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 4.960000000000001e-06,
"loss": 0.181,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 5.460000000000001e-06,
"loss": 0.1691,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.9600000000000005e-06,
"loss": 0.1618,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.460000000000001e-06,
"loss": 0.1553,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 6.96e-06,
"loss": 0.1598,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 7.4600000000000006e-06,
"loss": 0.1548,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 7.960000000000002e-06,
"loss": 0.1527,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.46e-06,
"loss": 0.1547,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.96e-06,
"loss": 0.1506,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 9.460000000000001e-06,
"loss": 0.1438,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 9.960000000000001e-06,
"loss": 0.1474,
"step": 500
},
{
"epoch": 1.0,
"learning_rate": 9.975789473684211e-06,
"loss": 0.1538,
"step": 525
},
{
"epoch": 1.0,
"learning_rate": 9.949473684210526e-06,
"loss": 0.1717,
"step": 550
},
{
"epoch": 1.01,
"learning_rate": 9.923157894736844e-06,
"loss": 0.1792,
"step": 575
},
{
"epoch": 1.01,
"learning_rate": 9.89684210526316e-06,
"loss": 0.1739,
"step": 600
},
{
"epoch": 1.01,
"learning_rate": 9.870526315789474e-06,
"loss": 0.1771,
"step": 625
},
{
"epoch": 1.01,
"learning_rate": 9.84421052631579e-06,
"loss": 0.1451,
"step": 650
},
{
"epoch": 1.02,
"learning_rate": 9.817894736842106e-06,
"loss": 0.1247,
"step": 675
},
{
"epoch": 1.02,
"learning_rate": 9.791578947368422e-06,
"loss": 0.1325,
"step": 700
},
{
"epoch": 1.02,
"learning_rate": 9.765263157894737e-06,
"loss": 0.1396,
"step": 725
},
{
"epoch": 1.02,
"learning_rate": 9.738947368421054e-06,
"loss": 0.1277,
"step": 750
},
{
"epoch": 1.03,
"learning_rate": 9.712631578947369e-06,
"loss": 0.1389,
"step": 775
},
{
"epoch": 1.03,
"learning_rate": 9.686315789473684e-06,
"loss": 0.1424,
"step": 800
},
{
"epoch": 1.03,
"learning_rate": 9.66e-06,
"loss": 0.1343,
"step": 825
},
{
"epoch": 1.03,
"learning_rate": 9.633684210526316e-06,
"loss": 0.1303,
"step": 850
},
{
"epoch": 1.04,
"learning_rate": 9.607368421052632e-06,
"loss": 0.1586,
"step": 875
},
{
"epoch": 1.04,
"learning_rate": 9.581052631578947e-06,
"loss": 0.1224,
"step": 900
},
{
"epoch": 1.04,
"learning_rate": 9.554736842105264e-06,
"loss": 0.114,
"step": 925
},
{
"epoch": 1.04,
"learning_rate": 9.52842105263158e-06,
"loss": 0.1133,
"step": 950
},
{
"epoch": 1.05,
"learning_rate": 9.502105263157896e-06,
"loss": 0.1114,
"step": 975
},
{
"epoch": 1.05,
"learning_rate": 9.475789473684212e-06,
"loss": 0.1158,
"step": 1000
},
{
"epoch": 1.05,
"eval_loss": 0.18459251523017883,
"eval_runtime": 3880.1501,
"eval_samples_per_second": 4.221,
"eval_steps_per_second": 0.132,
"eval_wer": 8.362960845955987,
"step": 1000
},
{
"epoch": 2.0,
"learning_rate": 9.449473684210527e-06,
"loss": 0.106,
"step": 1025
},
{
"epoch": 2.0,
"learning_rate": 9.423157894736842e-06,
"loss": 0.0998,
"step": 1050
},
{
"epoch": 2.01,
"learning_rate": 9.396842105263159e-06,
"loss": 0.0933,
"step": 1075
},
{
"epoch": 2.01,
"learning_rate": 9.370526315789474e-06,
"loss": 0.1059,
"step": 1100
},
{
"epoch": 2.01,
"learning_rate": 9.34421052631579e-06,
"loss": 0.113,
"step": 1125
},
{
"epoch": 2.01,
"learning_rate": 9.317894736842105e-06,
"loss": 0.0987,
"step": 1150
},
{
"epoch": 2.02,
"learning_rate": 9.291578947368422e-06,
"loss": 0.0983,
"step": 1175
},
{
"epoch": 2.02,
"learning_rate": 9.265263157894737e-06,
"loss": 0.1105,
"step": 1200
},
{
"epoch": 2.02,
"learning_rate": 9.238947368421052e-06,
"loss": 0.1113,
"step": 1225
},
{
"epoch": 2.02,
"learning_rate": 9.21263157894737e-06,
"loss": 0.1043,
"step": 1250
},
{
"epoch": 2.03,
"learning_rate": 9.186315789473685e-06,
"loss": 0.0986,
"step": 1275
},
{
"epoch": 2.03,
"learning_rate": 9.16e-06,
"loss": 0.107,
"step": 1300
},
{
"epoch": 2.03,
"learning_rate": 9.133684210526317e-06,
"loss": 0.1048,
"step": 1325
},
{
"epoch": 2.03,
"learning_rate": 9.107368421052632e-06,
"loss": 0.0979,
"step": 1350
},
{
"epoch": 2.04,
"learning_rate": 9.081052631578949e-06,
"loss": 0.0883,
"step": 1375
},
{
"epoch": 2.04,
"learning_rate": 9.054736842105264e-06,
"loss": 0.0647,
"step": 1400
},
{
"epoch": 2.04,
"learning_rate": 9.02842105263158e-06,
"loss": 0.0602,
"step": 1425
},
{
"epoch": 2.04,
"learning_rate": 9.002105263157895e-06,
"loss": 0.0536,
"step": 1450
},
{
"epoch": 2.05,
"learning_rate": 8.97578947368421e-06,
"loss": 0.0574,
"step": 1475
},
{
"epoch": 2.05,
"learning_rate": 8.949473684210527e-06,
"loss": 0.0491,
"step": 1500
},
{
"epoch": 2.05,
"learning_rate": 8.923157894736842e-06,
"loss": 0.0555,
"step": 1525
},
{
"epoch": 3.0,
"learning_rate": 8.896842105263159e-06,
"loss": 0.0505,
"step": 1550
},
{
"epoch": 3.0,
"learning_rate": 8.870526315789474e-06,
"loss": 0.0447,
"step": 1575
},
{
"epoch": 3.01,
"learning_rate": 8.84421052631579e-06,
"loss": 0.042,
"step": 1600
},
{
"epoch": 3.01,
"learning_rate": 8.817894736842107e-06,
"loss": 0.0467,
"step": 1625
},
{
"epoch": 3.01,
"learning_rate": 8.791578947368422e-06,
"loss": 0.0451,
"step": 1650
},
{
"epoch": 3.01,
"learning_rate": 8.765263157894739e-06,
"loss": 0.0402,
"step": 1675
},
{
"epoch": 3.02,
"learning_rate": 8.738947368421053e-06,
"loss": 0.0448,
"step": 1700
},
{
"epoch": 3.02,
"learning_rate": 8.712631578947368e-06,
"loss": 0.0484,
"step": 1725
},
{
"epoch": 3.02,
"learning_rate": 8.686315789473685e-06,
"loss": 0.0451,
"step": 1750
},
{
"epoch": 3.02,
"learning_rate": 8.66e-06,
"loss": 0.0437,
"step": 1775
},
{
"epoch": 3.03,
"learning_rate": 8.633684210526317e-06,
"loss": 0.043,
"step": 1800
},
{
"epoch": 3.03,
"learning_rate": 8.607368421052632e-06,
"loss": 0.0474,
"step": 1825
},
{
"epoch": 3.03,
"learning_rate": 8.581052631578948e-06,
"loss": 0.0402,
"step": 1850
},
{
"epoch": 3.03,
"learning_rate": 8.554736842105263e-06,
"loss": 0.0413,
"step": 1875
},
{
"epoch": 3.04,
"learning_rate": 8.528421052631578e-06,
"loss": 0.0279,
"step": 1900
},
{
"epoch": 3.04,
"learning_rate": 8.502105263157897e-06,
"loss": 0.0214,
"step": 1925
},
{
"epoch": 3.04,
"learning_rate": 8.475789473684212e-06,
"loss": 0.0216,
"step": 1950
},
{
"epoch": 3.04,
"learning_rate": 8.449473684210527e-06,
"loss": 0.021,
"step": 1975
},
{
"epoch": 3.05,
"learning_rate": 8.423157894736843e-06,
"loss": 0.0184,
"step": 2000
},
{
"epoch": 3.05,
"eval_loss": 0.2016635239124298,
"eval_runtime": 3865.7622,
"eval_samples_per_second": 4.237,
"eval_steps_per_second": 0.132,
"eval_wer": 8.062875107173479,
"step": 2000
},
{
"epoch": 3.05,
"learning_rate": 8.396842105263158e-06,
"loss": 0.0186,
"step": 2025
},
{
"epoch": 4.0,
"learning_rate": 8.370526315789475e-06,
"loss": 0.026,
"step": 2050
},
{
"epoch": 4.0,
"learning_rate": 8.34421052631579e-06,
"loss": 0.075,
"step": 2075
},
{
"epoch": 4.01,
"learning_rate": 8.317894736842107e-06,
"loss": 0.095,
"step": 2100
},
{
"epoch": 4.01,
"learning_rate": 8.291578947368422e-06,
"loss": 0.0799,
"step": 2125
},
{
"epoch": 4.01,
"learning_rate": 8.265263157894737e-06,
"loss": 0.066,
"step": 2150
},
{
"epoch": 4.01,
"learning_rate": 8.238947368421053e-06,
"loss": 0.0641,
"step": 2175
},
{
"epoch": 4.02,
"learning_rate": 8.212631578947368e-06,
"loss": 0.0569,
"step": 2200
},
{
"epoch": 4.02,
"learning_rate": 8.186315789473685e-06,
"loss": 0.0416,
"step": 2225
},
{
"epoch": 4.02,
"learning_rate": 8.16e-06,
"loss": 0.0692,
"step": 2250
},
{
"epoch": 4.02,
"learning_rate": 8.133684210526316e-06,
"loss": 0.0776,
"step": 2275
},
{
"epoch": 4.03,
"learning_rate": 8.107368421052633e-06,
"loss": 0.0889,
"step": 2300
},
{
"epoch": 4.03,
"learning_rate": 8.081052631578948e-06,
"loss": 0.0608,
"step": 2325
},
{
"epoch": 4.03,
"learning_rate": 8.054736842105265e-06,
"loss": 0.0635,
"step": 2350
},
{
"epoch": 4.03,
"learning_rate": 8.02842105263158e-06,
"loss": 0.07,
"step": 2375
},
{
"epoch": 4.04,
"learning_rate": 8.002105263157895e-06,
"loss": 0.0728,
"step": 2400
},
{
"epoch": 4.04,
"learning_rate": 7.975789473684211e-06,
"loss": 0.0753,
"step": 2425
},
{
"epoch": 4.04,
"learning_rate": 7.949473684210526e-06,
"loss": 0.0722,
"step": 2450
},
{
"epoch": 4.04,
"learning_rate": 7.923157894736843e-06,
"loss": 0.0716,
"step": 2475
},
{
"epoch": 4.05,
"learning_rate": 7.896842105263158e-06,
"loss": 0.0623,
"step": 2500
},
{
"epoch": 4.05,
"learning_rate": 7.870526315789475e-06,
"loss": 0.1082,
"step": 2525
},
{
"epoch": 4.05,
"learning_rate": 7.84421052631579e-06,
"loss": 0.1029,
"step": 2550
},
{
"epoch": 5.0,
"learning_rate": 7.817894736842105e-06,
"loss": 0.0832,
"step": 2575
},
{
"epoch": 5.0,
"learning_rate": 7.791578947368423e-06,
"loss": 0.0951,
"step": 2600
},
{
"epoch": 5.01,
"learning_rate": 7.765263157894738e-06,
"loss": 0.0834,
"step": 2625
},
{
"epoch": 5.01,
"learning_rate": 7.738947368421053e-06,
"loss": 0.0849,
"step": 2650
},
{
"epoch": 5.01,
"learning_rate": 7.71263157894737e-06,
"loss": 0.0726,
"step": 2675
},
{
"epoch": 5.01,
"learning_rate": 7.686315789473685e-06,
"loss": 0.072,
"step": 2700
},
{
"epoch": 5.02,
"learning_rate": 7.660000000000001e-06,
"loss": 0.0778,
"step": 2725
},
{
"epoch": 5.02,
"learning_rate": 7.633684210526316e-06,
"loss": 0.0741,
"step": 2750
},
{
"epoch": 5.02,
"learning_rate": 7.607368421052632e-06,
"loss": 0.0629,
"step": 2775
},
{
"epoch": 5.02,
"learning_rate": 7.581052631578948e-06,
"loss": 0.0336,
"step": 2800
},
{
"epoch": 5.03,
"learning_rate": 7.554736842105264e-06,
"loss": 0.0293,
"step": 2825
},
{
"epoch": 5.03,
"learning_rate": 7.5284210526315794e-06,
"loss": 0.0305,
"step": 2850
},
{
"epoch": 5.03,
"learning_rate": 7.502105263157895e-06,
"loss": 0.0355,
"step": 2875
},
{
"epoch": 5.03,
"learning_rate": 7.475789473684211e-06,
"loss": 0.0327,
"step": 2900
},
{
"epoch": 5.04,
"learning_rate": 7.449473684210526e-06,
"loss": 0.0261,
"step": 2925
},
{
"epoch": 5.04,
"learning_rate": 7.4231578947368436e-06,
"loss": 0.0327,
"step": 2950
},
{
"epoch": 5.04,
"learning_rate": 7.3968421052631585e-06,
"loss": 0.0446,
"step": 2975
},
{
"epoch": 5.04,
"learning_rate": 7.370526315789474e-06,
"loss": 0.0522,
"step": 3000
},
{
"epoch": 5.04,
"eval_loss": 0.19403910636901855,
"eval_runtime": 3809.2059,
"eval_samples_per_second": 4.3,
"eval_steps_per_second": 0.134,
"eval_wer": 8.117747927979423,
"step": 3000
},
{
"epoch": 5.05,
"learning_rate": 7.34421052631579e-06,
"loss": 0.0625,
"step": 3025
},
{
"epoch": 5.05,
"learning_rate": 7.317894736842106e-06,
"loss": 0.0489,
"step": 3050
},
{
"epoch": 6.0,
"learning_rate": 7.291578947368422e-06,
"loss": 0.0567,
"step": 3075
},
{
"epoch": 6.0,
"learning_rate": 7.265263157894738e-06,
"loss": 0.0779,
"step": 3100
},
{
"epoch": 6.01,
"learning_rate": 7.2389473684210534e-06,
"loss": 0.0796,
"step": 3125
},
{
"epoch": 6.01,
"learning_rate": 7.212631578947369e-06,
"loss": 0.0788,
"step": 3150
},
{
"epoch": 6.01,
"learning_rate": 7.186315789473684e-06,
"loss": 0.0913,
"step": 3175
},
{
"epoch": 6.01,
"learning_rate": 7.16e-06,
"loss": 0.0812,
"step": 3200
},
{
"epoch": 6.02,
"learning_rate": 7.133684210526316e-06,
"loss": 0.0739,
"step": 3225
},
{
"epoch": 6.02,
"learning_rate": 7.107368421052632e-06,
"loss": 0.0875,
"step": 3250
},
{
"epoch": 6.02,
"learning_rate": 7.0810526315789475e-06,
"loss": 0.0726,
"step": 3275
},
{
"epoch": 6.02,
"learning_rate": 7.054736842105264e-06,
"loss": 0.0785,
"step": 3300
},
{
"epoch": 6.03,
"learning_rate": 7.02842105263158e-06,
"loss": 0.0673,
"step": 3325
},
{
"epoch": 6.03,
"learning_rate": 7.002105263157896e-06,
"loss": 0.0854,
"step": 3350
},
{
"epoch": 6.03,
"learning_rate": 6.975789473684212e-06,
"loss": 0.0731,
"step": 3375
},
{
"epoch": 6.03,
"learning_rate": 6.9494736842105275e-06,
"loss": 0.0661,
"step": 3400
},
{
"epoch": 6.04,
"learning_rate": 6.9231578947368424e-06,
"loss": 0.0791,
"step": 3425
},
{
"epoch": 6.04,
"learning_rate": 6.896842105263158e-06,
"loss": 0.0677,
"step": 3450
},
{
"epoch": 6.04,
"learning_rate": 6.870526315789474e-06,
"loss": 0.0758,
"step": 3475
},
{
"epoch": 6.04,
"learning_rate": 6.84421052631579e-06,
"loss": 0.0765,
"step": 3500
},
{
"epoch": 6.05,
"learning_rate": 6.817894736842106e-06,
"loss": 0.0767,
"step": 3525
},
{
"epoch": 6.05,
"learning_rate": 6.7915789473684215e-06,
"loss": 0.077,
"step": 3550
},
{
"epoch": 6.05,
"learning_rate": 6.765263157894737e-06,
"loss": 0.0881,
"step": 3575
},
{
"epoch": 7.0,
"learning_rate": 6.738947368421052e-06,
"loss": 0.0449,
"step": 3600
},
{
"epoch": 7.0,
"learning_rate": 6.71263157894737e-06,
"loss": 0.0307,
"step": 3625
},
{
"epoch": 7.01,
"learning_rate": 6.686315789473685e-06,
"loss": 0.0221,
"step": 3650
},
{
"epoch": 7.01,
"learning_rate": 6.660000000000001e-06,
"loss": 0.04,
"step": 3675
},
{
"epoch": 7.01,
"learning_rate": 6.6336842105263164e-06,
"loss": 0.0498,
"step": 3700
},
{
"epoch": 7.01,
"learning_rate": 6.607368421052632e-06,
"loss": 0.0461,
"step": 3725
},
{
"epoch": 7.02,
"learning_rate": 6.581052631578948e-06,
"loss": 0.0457,
"step": 3750
},
{
"epoch": 7.02,
"learning_rate": 6.554736842105264e-06,
"loss": 0.0471,
"step": 3775
},
{
"epoch": 7.02,
"learning_rate": 6.52842105263158e-06,
"loss": 0.0455,
"step": 3800
},
{
"epoch": 7.02,
"learning_rate": 6.5021052631578955e-06,
"loss": 0.0437,
"step": 3825
},
{
"epoch": 7.03,
"learning_rate": 6.4757894736842105e-06,
"loss": 0.0467,
"step": 3850
},
{
"epoch": 7.03,
"learning_rate": 6.449473684210526e-06,
"loss": 0.0541,
"step": 3875
},
{
"epoch": 7.03,
"learning_rate": 6.423157894736842e-06,
"loss": 0.0459,
"step": 3900
},
{
"epoch": 7.03,
"learning_rate": 6.396842105263158e-06,
"loss": 0.0519,
"step": 3925
},
{
"epoch": 7.04,
"learning_rate": 6.370526315789474e-06,
"loss": 0.0608,
"step": 3950
},
{
"epoch": 7.04,
"learning_rate": 6.3442105263157904e-06,
"loss": 0.0527,
"step": 3975
},
{
"epoch": 7.04,
"learning_rate": 6.317894736842106e-06,
"loss": 0.0595,
"step": 4000
},
{
"epoch": 7.04,
"eval_loss": 0.17417794466018677,
"eval_runtime": 3815.6056,
"eval_samples_per_second": 4.293,
"eval_steps_per_second": 0.134,
"eval_wer": 7.469562732209202,
"step": 4000
},
{
"epoch": 7.04,
"learning_rate": 6.291578947368422e-06,
"loss": 0.071,
"step": 4025
},
{
"epoch": 7.05,
"learning_rate": 6.265263157894738e-06,
"loss": 0.0852,
"step": 4050
},
{
"epoch": 7.05,
"learning_rate": 6.238947368421054e-06,
"loss": 0.0862,
"step": 4075
},
{
"epoch": 8.0,
"learning_rate": 6.212631578947369e-06,
"loss": 0.0674,
"step": 4100
},
{
"epoch": 8.0,
"learning_rate": 6.1863157894736845e-06,
"loss": 0.0502,
"step": 4125
},
{
"epoch": 8.01,
"learning_rate": 6.16e-06,
"loss": 0.0521,
"step": 4150
},
{
"epoch": 8.01,
"learning_rate": 6.133684210526316e-06,
"loss": 0.053,
"step": 4175
},
{
"epoch": 8.01,
"learning_rate": 6.107368421052632e-06,
"loss": 0.0488,
"step": 4200
},
{
"epoch": 8.01,
"learning_rate": 6.081052631578948e-06,
"loss": 0.0407,
"step": 4225
},
{
"epoch": 8.02,
"learning_rate": 6.054736842105264e-06,
"loss": 0.0444,
"step": 4250
},
{
"epoch": 8.02,
"learning_rate": 6.0284210526315786e-06,
"loss": 0.0424,
"step": 4275
},
{
"epoch": 8.02,
"learning_rate": 6.002105263157896e-06,
"loss": 0.0419,
"step": 4300
},
{
"epoch": 8.02,
"learning_rate": 5.975789473684212e-06,
"loss": 0.0413,
"step": 4325
},
{
"epoch": 8.03,
"learning_rate": 5.949473684210527e-06,
"loss": 0.0437,
"step": 4350
},
{
"epoch": 8.03,
"learning_rate": 5.923157894736843e-06,
"loss": 0.0403,
"step": 4375
},
{
"epoch": 8.03,
"learning_rate": 5.8968421052631585e-06,
"loss": 0.0395,
"step": 4400
},
{
"epoch": 8.03,
"learning_rate": 5.870526315789474e-06,
"loss": 0.0417,
"step": 4425
},
{
"epoch": 8.04,
"learning_rate": 5.84421052631579e-06,
"loss": 0.0411,
"step": 4450
},
{
"epoch": 8.04,
"learning_rate": 5.817894736842106e-06,
"loss": 0.0405,
"step": 4475
},
{
"epoch": 8.04,
"learning_rate": 5.791578947368422e-06,
"loss": 0.0401,
"step": 4500
},
{
"epoch": 8.04,
"learning_rate": 5.765263157894737e-06,
"loss": 0.0399,
"step": 4525
},
{
"epoch": 8.05,
"learning_rate": 5.7389473684210526e-06,
"loss": 0.0453,
"step": 4550
},
{
"epoch": 8.05,
"learning_rate": 5.712631578947368e-06,
"loss": 0.0763,
"step": 4575
},
{
"epoch": 8.05,
"learning_rate": 5.686315789473684e-06,
"loss": 0.0666,
"step": 4600
},
{
"epoch": 9.0,
"learning_rate": 5.66e-06,
"loss": 0.0202,
"step": 4625
},
{
"epoch": 9.0,
"learning_rate": 5.633684210526317e-06,
"loss": 0.0095,
"step": 4650
},
{
"epoch": 9.01,
"learning_rate": 5.6073684210526325e-06,
"loss": 0.0089,
"step": 4675
},
{
"epoch": 9.01,
"learning_rate": 5.581052631578948e-06,
"loss": 0.0166,
"step": 4700
},
{
"epoch": 9.01,
"learning_rate": 5.554736842105264e-06,
"loss": 0.017,
"step": 4725
},
{
"epoch": 9.01,
"learning_rate": 5.52842105263158e-06,
"loss": 0.0175,
"step": 4750
},
{
"epoch": 9.02,
"learning_rate": 5.502105263157895e-06,
"loss": 0.0173,
"step": 4775
},
{
"epoch": 9.02,
"learning_rate": 5.475789473684211e-06,
"loss": 0.0167,
"step": 4800
},
{
"epoch": 9.02,
"learning_rate": 5.4494736842105266e-06,
"loss": 0.017,
"step": 4825
},
{
"epoch": 9.02,
"learning_rate": 5.423157894736842e-06,
"loss": 0.0165,
"step": 4850
},
{
"epoch": 9.03,
"learning_rate": 5.396842105263158e-06,
"loss": 0.0163,
"step": 4875
},
{
"epoch": 9.03,
"learning_rate": 5.370526315789474e-06,
"loss": 0.0196,
"step": 4900
},
{
"epoch": 9.03,
"learning_rate": 5.34421052631579e-06,
"loss": 0.0175,
"step": 4925
},
{
"epoch": 9.03,
"learning_rate": 5.317894736842105e-06,
"loss": 0.0191,
"step": 4950
},
{
"epoch": 9.04,
"learning_rate": 5.291578947368422e-06,
"loss": 0.0208,
"step": 4975
},
{
"epoch": 9.04,
"learning_rate": 5.265263157894738e-06,
"loss": 0.0179,
"step": 5000
},
{
"epoch": 9.04,
"eval_loss": 0.18989641964435577,
"eval_runtime": 3828.2662,
"eval_samples_per_second": 4.279,
"eval_steps_per_second": 0.134,
"eval_wer": 7.30951700485853,
"step": 5000
},
{
"epoch": 9.04,
"learning_rate": 5.238947368421053e-06,
"loss": 0.0236,
"step": 5025
},
{
"epoch": 9.04,
"learning_rate": 5.212631578947369e-06,
"loss": 0.026,
"step": 5050
},
{
"epoch": 9.05,
"learning_rate": 5.186315789473685e-06,
"loss": 0.0397,
"step": 5075
},
{
"epoch": 9.05,
"learning_rate": 5.1600000000000006e-06,
"loss": 0.0346,
"step": 5100
},
{
"epoch": 10.0,
"learning_rate": 5.133684210526316e-06,
"loss": 0.0298,
"step": 5125
},
{
"epoch": 10.0,
"learning_rate": 5.107368421052632e-06,
"loss": 0.0471,
"step": 5150
},
{
"epoch": 10.01,
"learning_rate": 5.081052631578948e-06,
"loss": 0.042,
"step": 5175
},
{
"epoch": 10.01,
"learning_rate": 5.054736842105263e-06,
"loss": 0.0448,
"step": 5200
},
{
"epoch": 10.01,
"learning_rate": 5.028421052631579e-06,
"loss": 0.0414,
"step": 5225
},
{
"epoch": 10.01,
"learning_rate": 5.002105263157895e-06,
"loss": 0.0417,
"step": 5250
},
{
"epoch": 10.02,
"learning_rate": 4.975789473684211e-06,
"loss": 0.0429,
"step": 5275
},
{
"epoch": 10.02,
"learning_rate": 4.949473684210527e-06,
"loss": 0.0384,
"step": 5300
},
{
"epoch": 10.02,
"learning_rate": 4.923157894736842e-06,
"loss": 0.037,
"step": 5325
},
{
"epoch": 10.02,
"learning_rate": 4.896842105263158e-06,
"loss": 0.0385,
"step": 5350
},
{
"epoch": 10.03,
"learning_rate": 4.870526315789474e-06,
"loss": 0.0377,
"step": 5375
},
{
"epoch": 10.03,
"learning_rate": 4.84421052631579e-06,
"loss": 0.0447,
"step": 5400
},
{
"epoch": 10.03,
"learning_rate": 4.817894736842106e-06,
"loss": 0.042,
"step": 5425
},
{
"epoch": 10.03,
"learning_rate": 4.791578947368421e-06,
"loss": 0.0379,
"step": 5450
},
{
"epoch": 10.04,
"learning_rate": 4.765263157894737e-06,
"loss": 0.0473,
"step": 5475
},
{
"epoch": 10.04,
"learning_rate": 4.738947368421053e-06,
"loss": 0.048,
"step": 5500
},
{
"epoch": 10.04,
"learning_rate": 4.712631578947369e-06,
"loss": 0.0422,
"step": 5525
},
{
"epoch": 10.04,
"learning_rate": 4.6863157894736845e-06,
"loss": 0.0438,
"step": 5550
},
{
"epoch": 10.05,
"learning_rate": 4.66e-06,
"loss": 0.0454,
"step": 5575
},
{
"epoch": 10.05,
"learning_rate": 4.633684210526316e-06,
"loss": 0.0592,
"step": 5600
},
{
"epoch": 10.05,
"learning_rate": 4.607368421052632e-06,
"loss": 0.0732,
"step": 5625
},
{
"epoch": 11.0,
"learning_rate": 4.581052631578948e-06,
"loss": 0.0635,
"step": 5650
},
{
"epoch": 11.0,
"learning_rate": 4.5547368421052636e-06,
"loss": 0.0748,
"step": 5675
},
{
"epoch": 11.01,
"learning_rate": 4.528421052631579e-06,
"loss": 0.0636,
"step": 5700
},
{
"epoch": 11.01,
"learning_rate": 4.502105263157895e-06,
"loss": 0.0374,
"step": 5725
},
{
"epoch": 11.01,
"learning_rate": 4.475789473684211e-06,
"loss": 0.0678,
"step": 5750
},
{
"epoch": 11.01,
"learning_rate": 4.449473684210527e-06,
"loss": 0.0655,
"step": 5775
},
{
"epoch": 11.02,
"learning_rate": 4.423157894736843e-06,
"loss": 0.0494,
"step": 5800
},
{
"epoch": 11.02,
"learning_rate": 4.3968421052631585e-06,
"loss": 0.0514,
"step": 5825
},
{
"epoch": 11.02,
"learning_rate": 4.370526315789474e-06,
"loss": 0.0667,
"step": 5850
},
{
"epoch": 11.02,
"learning_rate": 4.344210526315789e-06,
"loss": 0.0553,
"step": 5875
},
{
"epoch": 11.03,
"learning_rate": 4.317894736842105e-06,
"loss": 0.054,
"step": 5900
},
{
"epoch": 11.03,
"learning_rate": 4.291578947368422e-06,
"loss": 0.0815,
"step": 5925
},
{
"epoch": 11.03,
"learning_rate": 4.2652631578947376e-06,
"loss": 0.0597,
"step": 5950
},
{
"epoch": 11.03,
"learning_rate": 4.238947368421053e-06,
"loss": 0.0649,
"step": 5975
},
{
"epoch": 11.04,
"learning_rate": 4.212631578947368e-06,
"loss": 0.0646,
"step": 6000
},
{
"epoch": 11.04,
"eval_loss": 0.15545900166034698,
"eval_runtime": 3850.4362,
"eval_samples_per_second": 4.254,
"eval_steps_per_second": 0.133,
"eval_wer": 6.3440983138039435,
"step": 6000
},
{
"epoch": 11.04,
"learning_rate": 4.1873684210526325e-06,
"loss": 0.0632,
"step": 6025
},
{
"epoch": 11.04,
"learning_rate": 4.1610526315789475e-06,
"loss": 0.0562,
"step": 6050
},
{
"epoch": 11.04,
"learning_rate": 4.134736842105263e-06,
"loss": 0.0489,
"step": 6075
},
{
"epoch": 11.05,
"learning_rate": 4.108421052631579e-06,
"loss": 0.0553,
"step": 6100
},
{
"epoch": 11.05,
"learning_rate": 4.082105263157895e-06,
"loss": 0.0642,
"step": 6125
},
{
"epoch": 12.0,
"learning_rate": 4.055789473684211e-06,
"loss": 0.0766,
"step": 6150
},
{
"epoch": 12.0,
"learning_rate": 4.029473684210527e-06,
"loss": 0.0817,
"step": 6175
},
{
"epoch": 12.01,
"learning_rate": 4.0031578947368424e-06,
"loss": 0.0823,
"step": 6200
},
{
"epoch": 12.01,
"learning_rate": 3.976842105263158e-06,
"loss": 0.0519,
"step": 6225
},
{
"epoch": 12.01,
"learning_rate": 3.950526315789474e-06,
"loss": 0.0391,
"step": 6250
},
{
"epoch": 12.01,
"learning_rate": 3.92421052631579e-06,
"loss": 0.0393,
"step": 6275
},
{
"epoch": 12.02,
"learning_rate": 3.897894736842106e-06,
"loss": 0.0494,
"step": 6300
},
{
"epoch": 12.02,
"learning_rate": 3.8715789473684215e-06,
"loss": 0.0404,
"step": 6325
},
{
"epoch": 12.02,
"learning_rate": 3.845263157894737e-06,
"loss": 0.0331,
"step": 6350
},
{
"epoch": 12.02,
"learning_rate": 3.818947368421053e-06,
"loss": 0.0362,
"step": 6375
},
{
"epoch": 12.03,
"learning_rate": 3.792631578947369e-06,
"loss": 0.0352,
"step": 6400
},
{
"epoch": 12.03,
"learning_rate": 3.766315789473685e-06,
"loss": 0.0477,
"step": 6425
},
{
"epoch": 12.03,
"learning_rate": 3.74e-06,
"loss": 0.0514,
"step": 6450
},
{
"epoch": 12.03,
"learning_rate": 3.713684210526316e-06,
"loss": 0.0678,
"step": 6475
},
{
"epoch": 12.04,
"learning_rate": 3.687368421052632e-06,
"loss": 0.0355,
"step": 6500
},
{
"epoch": 12.04,
"learning_rate": 3.6610526315789472e-06,
"loss": 0.0452,
"step": 6525
},
{
"epoch": 12.04,
"learning_rate": 3.6347368421052635e-06,
"loss": 0.0684,
"step": 6550
},
{
"epoch": 12.04,
"learning_rate": 3.6084210526315793e-06,
"loss": 0.1064,
"step": 6575
},
{
"epoch": 12.05,
"learning_rate": 3.582105263157895e-06,
"loss": 0.1047,
"step": 6600
},
{
"epoch": 12.05,
"learning_rate": 3.555789473684211e-06,
"loss": 0.111,
"step": 6625
},
{
"epoch": 12.05,
"learning_rate": 3.5294736842105263e-06,
"loss": 0.1216,
"step": 6650
},
{
"epoch": 13.0,
"learning_rate": 3.503157894736842e-06,
"loss": 0.0468,
"step": 6675
},
{
"epoch": 13.0,
"learning_rate": 3.476842105263158e-06,
"loss": 0.038,
"step": 6700
},
{
"epoch": 13.01,
"learning_rate": 3.450526315789474e-06,
"loss": 0.0388,
"step": 6725
},
{
"epoch": 13.01,
"learning_rate": 3.42421052631579e-06,
"loss": 0.0354,
"step": 6750
},
{
"epoch": 13.01,
"learning_rate": 3.3978947368421054e-06,
"loss": 0.0393,
"step": 6775
},
{
"epoch": 13.01,
"learning_rate": 3.3715789473684212e-06,
"loss": 0.0335,
"step": 6800
},
{
"epoch": 13.02,
"learning_rate": 3.345263157894737e-06,
"loss": 0.0353,
"step": 6825
},
{
"epoch": 13.02,
"learning_rate": 3.318947368421053e-06,
"loss": 0.0404,
"step": 6850
},
{
"epoch": 13.02,
"learning_rate": 3.292631578947369e-06,
"loss": 0.038,
"step": 6875
},
{
"epoch": 13.02,
"learning_rate": 3.2663157894736845e-06,
"loss": 0.0335,
"step": 6900
},
{
"epoch": 13.03,
"learning_rate": 3.2400000000000003e-06,
"loss": 0.0357,
"step": 6925
},
{
"epoch": 13.03,
"learning_rate": 3.213684210526316e-06,
"loss": 0.0375,
"step": 6950
},
{
"epoch": 13.03,
"learning_rate": 3.187368421052632e-06,
"loss": 0.0581,
"step": 6975
},
{
"epoch": 13.03,
"learning_rate": 3.1610526315789474e-06,
"loss": 0.0825,
"step": 7000
},
{
"epoch": 13.03,
"eval_loss": 0.18097521364688873,
"eval_runtime": 3975.6035,
"eval_samples_per_second": 4.12,
"eval_steps_per_second": 0.129,
"eval_wer": 6.484138325235782,
"step": 7000
},
{
"epoch": 13.04,
"learning_rate": 3.134736842105263e-06,
"loss": 0.069,
"step": 7025
},
{
"epoch": 13.04,
"learning_rate": 3.1084210526315794e-06,
"loss": 0.0701,
"step": 7050
},
{
"epoch": 13.04,
"learning_rate": 3.0821052631578952e-06,
"loss": 0.074,
"step": 7075
},
{
"epoch": 13.04,
"learning_rate": 3.055789473684211e-06,
"loss": 0.0737,
"step": 7100
},
{
"epoch": 13.05,
"learning_rate": 3.0294736842105264e-06,
"loss": 0.0647,
"step": 7125
},
{
"epoch": 13.05,
"learning_rate": 3.0031578947368423e-06,
"loss": 0.0699,
"step": 7150
},
{
"epoch": 14.0,
"learning_rate": 2.976842105263158e-06,
"loss": 0.0722,
"step": 7175
},
{
"epoch": 14.0,
"learning_rate": 2.9505263157894735e-06,
"loss": 0.0532,
"step": 7200
},
{
"epoch": 14.01,
"learning_rate": 2.92421052631579e-06,
"loss": 0.0657,
"step": 7225
},
{
"epoch": 14.01,
"learning_rate": 2.8978947368421055e-06,
"loss": 0.0471,
"step": 7250
},
{
"epoch": 14.01,
"learning_rate": 2.8715789473684214e-06,
"loss": 0.0624,
"step": 7275
},
{
"epoch": 14.01,
"learning_rate": 2.845263157894737e-06,
"loss": 0.0644,
"step": 7300
},
{
"epoch": 14.02,
"learning_rate": 2.8189473684210526e-06,
"loss": 0.0617,
"step": 7325
},
{
"epoch": 14.02,
"learning_rate": 2.7926315789473684e-06,
"loss": 0.0608,
"step": 7350
},
{
"epoch": 14.02,
"learning_rate": 2.766315789473684e-06,
"loss": 0.0633,
"step": 7375
},
{
"epoch": 14.02,
"learning_rate": 2.7400000000000004e-06,
"loss": 0.0515,
"step": 7400
},
{
"epoch": 14.03,
"learning_rate": 2.7136842105263163e-06,
"loss": 0.0497,
"step": 7425
},
{
"epoch": 14.03,
"learning_rate": 2.6873684210526317e-06,
"loss": 0.0718,
"step": 7450
},
{
"epoch": 14.03,
"learning_rate": 2.6610526315789475e-06,
"loss": 0.056,
"step": 7475
},
{
"epoch": 14.03,
"learning_rate": 2.6347368421052633e-06,
"loss": 0.0561,
"step": 7500
},
{
"epoch": 14.04,
"learning_rate": 2.608421052631579e-06,
"loss": 0.0522,
"step": 7525
},
{
"epoch": 14.04,
"learning_rate": 2.5821052631578954e-06,
"loss": 0.0473,
"step": 7550
},
{
"epoch": 14.04,
"learning_rate": 2.5557894736842108e-06,
"loss": 0.0643,
"step": 7575
},
{
"epoch": 14.04,
"learning_rate": 2.5294736842105266e-06,
"loss": 0.0652,
"step": 7600
},
{
"epoch": 14.05,
"learning_rate": 2.5031578947368424e-06,
"loss": 0.0684,
"step": 7625
},
{
"epoch": 14.05,
"learning_rate": 2.476842105263158e-06,
"loss": 0.0622,
"step": 7650
},
{
"epoch": 14.05,
"learning_rate": 2.4505263157894736e-06,
"loss": 0.0666,
"step": 7675
},
{
"epoch": 15.0,
"learning_rate": 2.42421052631579e-06,
"loss": 0.0355,
"step": 7700
},
{
"epoch": 15.0,
"learning_rate": 2.3978947368421052e-06,
"loss": 0.0306,
"step": 7725
},
{
"epoch": 15.01,
"learning_rate": 2.371578947368421e-06,
"loss": 0.0392,
"step": 7750
},
{
"epoch": 15.01,
"learning_rate": 2.3452631578947373e-06,
"loss": 0.0324,
"step": 7775
},
{
"epoch": 15.01,
"learning_rate": 2.3189473684210527e-06,
"loss": 0.0379,
"step": 7800
},
{
"epoch": 15.01,
"learning_rate": 2.2926315789473685e-06,
"loss": 0.0248,
"step": 7825
},
{
"epoch": 15.02,
"learning_rate": 2.2663157894736843e-06,
"loss": 0.023,
"step": 7850
},
{
"epoch": 15.02,
"learning_rate": 2.24e-06,
"loss": 0.0226,
"step": 7875
},
{
"epoch": 15.02,
"learning_rate": 2.213684210526316e-06,
"loss": 0.0328,
"step": 7900
},
{
"epoch": 15.02,
"learning_rate": 2.187368421052632e-06,
"loss": 0.0294,
"step": 7925
},
{
"epoch": 15.03,
"learning_rate": 2.1610526315789476e-06,
"loss": 0.0334,
"step": 7950
},
{
"epoch": 15.03,
"learning_rate": 2.1347368421052634e-06,
"loss": 0.0384,
"step": 7975
},
{
"epoch": 15.03,
"learning_rate": 2.1084210526315792e-06,
"loss": 0.0309,
"step": 8000
},
{
"epoch": 15.03,
"eval_loss": 0.14641128480434418,
"eval_runtime": 3910.3674,
"eval_samples_per_second": 4.189,
"eval_steps_per_second": 0.131,
"eval_wer": 6.354386967705058,
"step": 8000
},
{
"epoch": 15.03,
"learning_rate": 2.082105263157895e-06,
"loss": 0.0371,
"step": 8025
},
{
"epoch": 15.04,
"learning_rate": 2.055789473684211e-06,
"loss": 0.0483,
"step": 8050
},
{
"epoch": 15.04,
"learning_rate": 2.0294736842105263e-06,
"loss": 0.0306,
"step": 8075
},
{
"epoch": 15.04,
"learning_rate": 2.003157894736842e-06,
"loss": 0.0215,
"step": 8100
},
{
"epoch": 15.04,
"learning_rate": 1.976842105263158e-06,
"loss": 0.0281,
"step": 8125
},
{
"epoch": 15.05,
"learning_rate": 1.9505263157894737e-06,
"loss": 0.0262,
"step": 8150
},
{
"epoch": 15.05,
"learning_rate": 1.9242105263157896e-06,
"loss": 0.0317,
"step": 8175
},
{
"epoch": 16.0,
"learning_rate": 1.8978947368421056e-06,
"loss": 0.025,
"step": 8200
},
{
"epoch": 16.0,
"learning_rate": 1.8715789473684212e-06,
"loss": 0.0321,
"step": 8225
},
{
"epoch": 16.01,
"learning_rate": 1.845263157894737e-06,
"loss": 0.0194,
"step": 8250
},
{
"epoch": 16.01,
"learning_rate": 1.8189473684210528e-06,
"loss": 0.0165,
"step": 8275
},
{
"epoch": 16.01,
"learning_rate": 1.7926315789473686e-06,
"loss": 0.0207,
"step": 8300
},
{
"epoch": 16.01,
"learning_rate": 1.7663157894736843e-06,
"loss": 0.0213,
"step": 8325
},
{
"epoch": 16.02,
"learning_rate": 1.74e-06,
"loss": 0.0221,
"step": 8350
},
{
"epoch": 16.02,
"learning_rate": 1.713684210526316e-06,
"loss": 0.0249,
"step": 8375
},
{
"epoch": 16.02,
"learning_rate": 1.6873684210526317e-06,
"loss": 0.0199,
"step": 8400
},
{
"epoch": 16.02,
"learning_rate": 1.6621052631578948e-06,
"loss": 0.0251,
"step": 8425
},
{
"epoch": 16.03,
"learning_rate": 1.6357894736842107e-06,
"loss": 0.0386,
"step": 8450
},
{
"epoch": 16.03,
"learning_rate": 1.6094736842105265e-06,
"loss": 0.0328,
"step": 8475
},
{
"epoch": 16.03,
"learning_rate": 1.5831578947368423e-06,
"loss": 0.0301,
"step": 8500
},
{
"epoch": 16.03,
"learning_rate": 1.556842105263158e-06,
"loss": 0.0357,
"step": 8525
},
{
"epoch": 16.04,
"learning_rate": 1.5305263157894737e-06,
"loss": 0.037,
"step": 8550
},
{
"epoch": 16.04,
"learning_rate": 1.5042105263157898e-06,
"loss": 0.0397,
"step": 8575
},
{
"epoch": 16.04,
"learning_rate": 1.4778947368421054e-06,
"loss": 0.0391,
"step": 8600
},
{
"epoch": 16.04,
"learning_rate": 1.4515789473684212e-06,
"loss": 0.0385,
"step": 8625
},
{
"epoch": 16.05,
"learning_rate": 1.425263157894737e-06,
"loss": 0.041,
"step": 8650
},
{
"epoch": 16.05,
"learning_rate": 1.3989473684210528e-06,
"loss": 0.0455,
"step": 8675
},
{
"epoch": 16.05,
"learning_rate": 1.3726315789473684e-06,
"loss": 0.065,
"step": 8700
},
{
"epoch": 17.0,
"learning_rate": 1.3463157894736842e-06,
"loss": 0.0796,
"step": 8725
},
{
"epoch": 17.0,
"learning_rate": 1.32e-06,
"loss": 0.0716,
"step": 8750
},
{
"epoch": 17.01,
"learning_rate": 1.2936842105263159e-06,
"loss": 0.0343,
"step": 8775
},
{
"epoch": 17.01,
"learning_rate": 1.2673684210526315e-06,
"loss": 0.0458,
"step": 8800
},
{
"epoch": 17.01,
"learning_rate": 1.2410526315789475e-06,
"loss": 0.0488,
"step": 8825
},
{
"epoch": 17.01,
"learning_rate": 1.2147368421052633e-06,
"loss": 0.031,
"step": 8850
},
{
"epoch": 17.02,
"learning_rate": 1.188421052631579e-06,
"loss": 0.0204,
"step": 8875
},
{
"epoch": 17.02,
"learning_rate": 1.1621052631578948e-06,
"loss": 0.0359,
"step": 8900
},
{
"epoch": 17.02,
"learning_rate": 1.1357894736842106e-06,
"loss": 0.0479,
"step": 8925
},
{
"epoch": 17.02,
"learning_rate": 1.1094736842105264e-06,
"loss": 0.0442,
"step": 8950
},
{
"epoch": 17.03,
"learning_rate": 1.0831578947368422e-06,
"loss": 0.0632,
"step": 8975
},
{
"epoch": 17.03,
"learning_rate": 1.0568421052631578e-06,
"loss": 0.0695,
"step": 9000
},
{
"epoch": 17.03,
"eval_loss": 0.14336933195590973,
"eval_runtime": 3882.9989,
"eval_samples_per_second": 4.218,
"eval_steps_per_second": 0.132,
"eval_wer": 5.995427264932838,
"step": 9000
},
{
"epoch": 17.03,
"learning_rate": 1.0305263157894739e-06,
"loss": 0.0432,
"step": 9025
},
{
"epoch": 17.03,
"learning_rate": 1.0042105263157897e-06,
"loss": 0.0511,
"step": 9050
},
{
"epoch": 17.04,
"learning_rate": 9.778947368421053e-07,
"loss": 0.0538,
"step": 9075
},
{
"epoch": 17.04,
"learning_rate": 9.515789473684212e-07,
"loss": 0.0615,
"step": 9100
},
{
"epoch": 17.04,
"learning_rate": 9.252631578947368e-07,
"loss": 0.0447,
"step": 9125
},
{
"epoch": 17.04,
"learning_rate": 8.989473684210527e-07,
"loss": 0.0466,
"step": 9150
},
{
"epoch": 17.05,
"learning_rate": 8.726315789473686e-07,
"loss": 0.047,
"step": 9175
},
{
"epoch": 17.05,
"learning_rate": 8.463157894736843e-07,
"loss": 0.0568,
"step": 9200
},
{
"epoch": 18.0,
"learning_rate": 8.200000000000001e-07,
"loss": 0.0532,
"step": 9225
},
{
"epoch": 18.0,
"learning_rate": 7.936842105263158e-07,
"loss": 0.0361,
"step": 9250
},
{
"epoch": 18.01,
"learning_rate": 7.673684210526316e-07,
"loss": 0.0321,
"step": 9275
},
{
"epoch": 18.01,
"learning_rate": 7.410526315789475e-07,
"loss": 0.0338,
"step": 9300
},
{
"epoch": 18.01,
"learning_rate": 7.147368421052632e-07,
"loss": 0.0315,
"step": 9325
},
{
"epoch": 18.01,
"learning_rate": 6.884210526315791e-07,
"loss": 0.0278,
"step": 9350
},
{
"epoch": 18.02,
"learning_rate": 6.621052631578948e-07,
"loss": 0.0305,
"step": 9375
},
{
"epoch": 18.02,
"learning_rate": 6.357894736842106e-07,
"loss": 0.0296,
"step": 9400
},
{
"epoch": 18.02,
"learning_rate": 6.094736842105263e-07,
"loss": 0.0265,
"step": 9425
},
{
"epoch": 18.02,
"learning_rate": 5.831578947368421e-07,
"loss": 0.0287,
"step": 9450
},
{
"epoch": 18.03,
"learning_rate": 5.56842105263158e-07,
"loss": 0.029,
"step": 9475
},
{
"epoch": 18.03,
"learning_rate": 5.305263157894737e-07,
"loss": 0.0243,
"step": 9500
},
{
"epoch": 18.03,
"learning_rate": 5.042105263157895e-07,
"loss": 0.0247,
"step": 9525
},
{
"epoch": 18.03,
"learning_rate": 4.778947368421053e-07,
"loss": 0.0271,
"step": 9550
},
{
"epoch": 18.04,
"learning_rate": 4.5157894736842107e-07,
"loss": 0.0242,
"step": 9575
},
{
"epoch": 18.04,
"learning_rate": 4.2526315789473684e-07,
"loss": 0.0247,
"step": 9600
},
{
"epoch": 18.04,
"learning_rate": 3.9894736842105266e-07,
"loss": 0.0221,
"step": 9625
},
{
"epoch": 18.04,
"learning_rate": 3.726315789473685e-07,
"loss": 0.022,
"step": 9650
},
{
"epoch": 18.05,
"learning_rate": 3.4631578947368424e-07,
"loss": 0.0213,
"step": 9675
},
{
"epoch": 18.05,
"learning_rate": 3.2e-07,
"loss": 0.0206,
"step": 9700
},
{
"epoch": 18.05,
"learning_rate": 2.936842105263158e-07,
"loss": 0.0204,
"step": 9725
},
{
"epoch": 19.0,
"learning_rate": 2.6736842105263164e-07,
"loss": 0.0198,
"step": 9750
},
{
"epoch": 19.0,
"learning_rate": 2.410526315789474e-07,
"loss": 0.0196,
"step": 9775
},
{
"epoch": 19.01,
"learning_rate": 2.1473684210526317e-07,
"loss": 0.0195,
"step": 9800
},
{
"epoch": 19.01,
"learning_rate": 1.8842105263157897e-07,
"loss": 0.0169,
"step": 9825
},
{
"epoch": 19.01,
"learning_rate": 1.6210526315789476e-07,
"loss": 0.0186,
"step": 9850
},
{
"epoch": 19.01,
"learning_rate": 1.3578947368421055e-07,
"loss": 0.0184,
"step": 9875
},
{
"epoch": 19.02,
"learning_rate": 1.0947368421052632e-07,
"loss": 0.019,
"step": 9900
},
{
"epoch": 19.02,
"learning_rate": 8.315789473684211e-08,
"loss": 0.0161,
"step": 9925
},
{
"epoch": 19.02,
"learning_rate": 5.68421052631579e-08,
"loss": 0.0173,
"step": 9950
},
{
"epoch": 19.02,
"learning_rate": 3.0526315789473686e-08,
"loss": 0.017,
"step": 9975
},
{
"epoch": 19.03,
"learning_rate": 4.210526315789474e-09,
"loss": 0.0186,
"step": 10000
},
{
"epoch": 19.03,
"eval_loss": 0.17058740556240082,
"eval_runtime": 4107.3564,
"eval_samples_per_second": 3.988,
"eval_steps_per_second": 0.125,
"eval_wer": 6.109745641611889,
"step": 10000
},
{
"epoch": 19.03,
"step": 10000,
"total_flos": 6.53032008866857e+20,
"train_loss": 0.06488132039755583,
"train_runtime": 164129.7236,
"train_samples_per_second": 3.899,
"train_steps_per_second": 0.061
}
],
"logging_steps": 25,
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"save_steps": 1000,
"total_flos": 6.53032008866857e+20,
"trial_name": null,
"trial_params": null
}