whisper-medium-basque-lr1e-5 / trainer_state.json
Ansu's picture
End of training
7cd7fe6
Raw
History Blame Contribute Delete
51.2 kB
{
"best_metric": 5.957287373548145,
"best_model_checkpoint": "/scratch/andoni.sudupe/checkpoints/sweep/medium/lr1e-5-freezeFalse/checkpoint-10000",
"epoch": 2.4777006937561943,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.0000000000000003e-07,
"loss": 5.1041,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.000000000000001e-07,
"loss": 3.204,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 1.4000000000000001e-06,
"loss": 2.1581,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.9000000000000002e-06,
"loss": 1.4646,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.1151,
"step": 125
},
{
"epoch": 0.04,
"learning_rate": 2.9e-06,
"loss": 0.775,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.497,
"step": 175
},
{
"epoch": 0.05,
"learning_rate": 3.900000000000001e-06,
"loss": 0.4112,
"step": 200
},
{
"epoch": 0.06,
"learning_rate": 4.4e-06,
"loss": 0.3493,
"step": 225
},
{
"epoch": 0.06,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.3292,
"step": 250
},
{
"epoch": 0.07,
"learning_rate": 5.400000000000001e-06,
"loss": 0.2943,
"step": 275
},
{
"epoch": 0.07,
"learning_rate": 5.9e-06,
"loss": 0.2794,
"step": 300
},
{
"epoch": 0.08,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.268,
"step": 325
},
{
"epoch": 0.09,
"learning_rate": 6.9e-06,
"loss": 0.2394,
"step": 350
},
{
"epoch": 0.09,
"learning_rate": 7.4e-06,
"loss": 0.2407,
"step": 375
},
{
"epoch": 0.1,
"learning_rate": 7.9e-06,
"loss": 0.2148,
"step": 400
},
{
"epoch": 0.11,
"learning_rate": 8.400000000000001e-06,
"loss": 0.2109,
"step": 425
},
{
"epoch": 0.11,
"learning_rate": 8.900000000000001e-06,
"loss": 0.2037,
"step": 450
},
{
"epoch": 0.12,
"learning_rate": 9.4e-06,
"loss": 0.1969,
"step": 475
},
{
"epoch": 0.12,
"learning_rate": 9.9e-06,
"loss": 0.1891,
"step": 500
},
{
"epoch": 0.13,
"learning_rate": 9.978947368421053e-06,
"loss": 0.1858,
"step": 525
},
{
"epoch": 0.14,
"learning_rate": 9.95263157894737e-06,
"loss": 0.1814,
"step": 550
},
{
"epoch": 0.14,
"learning_rate": 9.926315789473685e-06,
"loss": 0.1774,
"step": 575
},
{
"epoch": 0.15,
"learning_rate": 9.9e-06,
"loss": 0.159,
"step": 600
},
{
"epoch": 0.15,
"learning_rate": 9.873684210526317e-06,
"loss": 0.169,
"step": 625
},
{
"epoch": 0.16,
"learning_rate": 9.847368421052632e-06,
"loss": 0.1533,
"step": 650
},
{
"epoch": 0.17,
"learning_rate": 9.821052631578948e-06,
"loss": 0.154,
"step": 675
},
{
"epoch": 0.17,
"learning_rate": 9.794736842105263e-06,
"loss": 0.152,
"step": 700
},
{
"epoch": 0.18,
"learning_rate": 9.76842105263158e-06,
"loss": 0.1441,
"step": 725
},
{
"epoch": 0.19,
"learning_rate": 9.742105263157897e-06,
"loss": 0.1441,
"step": 750
},
{
"epoch": 0.19,
"learning_rate": 9.715789473684212e-06,
"loss": 0.133,
"step": 775
},
{
"epoch": 0.2,
"learning_rate": 9.689473684210527e-06,
"loss": 0.1403,
"step": 800
},
{
"epoch": 0.2,
"learning_rate": 9.663157894736843e-06,
"loss": 0.1389,
"step": 825
},
{
"epoch": 0.21,
"learning_rate": 9.636842105263158e-06,
"loss": 0.1313,
"step": 850
},
{
"epoch": 0.22,
"learning_rate": 9.610526315789475e-06,
"loss": 0.1295,
"step": 875
},
{
"epoch": 0.22,
"learning_rate": 9.58421052631579e-06,
"loss": 0.1347,
"step": 900
},
{
"epoch": 0.23,
"learning_rate": 9.557894736842107e-06,
"loss": 0.1291,
"step": 925
},
{
"epoch": 0.24,
"learning_rate": 9.531578947368422e-06,
"loss": 0.1254,
"step": 950
},
{
"epoch": 0.24,
"learning_rate": 9.505263157894738e-06,
"loss": 0.1247,
"step": 975
},
{
"epoch": 0.25,
"learning_rate": 9.478947368421053e-06,
"loss": 0.119,
"step": 1000
},
{
"epoch": 0.25,
"eval_loss": 0.20477597415447235,
"eval_runtime": 475.6452,
"eval_samples_per_second": 3.555,
"eval_steps_per_second": 0.076,
"eval_wer": 11.040339702760086,
"step": 1000
},
{
"epoch": 0.25,
"learning_rate": 9.452631578947368e-06,
"loss": 0.1213,
"step": 1025
},
{
"epoch": 0.26,
"learning_rate": 9.426315789473685e-06,
"loss": 0.1231,
"step": 1050
},
{
"epoch": 0.27,
"learning_rate": 9.4e-06,
"loss": 0.1157,
"step": 1075
},
{
"epoch": 0.27,
"learning_rate": 9.373684210526316e-06,
"loss": 0.118,
"step": 1100
},
{
"epoch": 0.28,
"learning_rate": 9.347368421052633e-06,
"loss": 0.1207,
"step": 1125
},
{
"epoch": 0.28,
"learning_rate": 9.321052631578948e-06,
"loss": 0.1054,
"step": 1150
},
{
"epoch": 0.29,
"learning_rate": 9.294736842105265e-06,
"loss": 0.1139,
"step": 1175
},
{
"epoch": 0.3,
"learning_rate": 9.26842105263158e-06,
"loss": 0.1155,
"step": 1200
},
{
"epoch": 0.3,
"learning_rate": 9.242105263157896e-06,
"loss": 0.1164,
"step": 1225
},
{
"epoch": 0.31,
"learning_rate": 9.215789473684211e-06,
"loss": 0.1015,
"step": 1250
},
{
"epoch": 0.32,
"learning_rate": 9.189473684210526e-06,
"loss": 0.1061,
"step": 1275
},
{
"epoch": 0.32,
"learning_rate": 9.163157894736843e-06,
"loss": 0.1082,
"step": 1300
},
{
"epoch": 0.33,
"learning_rate": 9.136842105263158e-06,
"loss": 0.1095,
"step": 1325
},
{
"epoch": 0.33,
"learning_rate": 9.110526315789475e-06,
"loss": 0.1096,
"step": 1350
},
{
"epoch": 0.34,
"learning_rate": 9.08421052631579e-06,
"loss": 0.1004,
"step": 1375
},
{
"epoch": 0.35,
"learning_rate": 9.057894736842106e-06,
"loss": 0.1034,
"step": 1400
},
{
"epoch": 0.35,
"learning_rate": 9.031578947368423e-06,
"loss": 0.1,
"step": 1425
},
{
"epoch": 0.36,
"learning_rate": 9.005263157894738e-06,
"loss": 0.0981,
"step": 1450
},
{
"epoch": 0.37,
"learning_rate": 8.978947368421055e-06,
"loss": 0.1059,
"step": 1475
},
{
"epoch": 0.37,
"learning_rate": 8.95263157894737e-06,
"loss": 0.0946,
"step": 1500
},
{
"epoch": 0.38,
"learning_rate": 8.926315789473685e-06,
"loss": 0.097,
"step": 1525
},
{
"epoch": 0.38,
"learning_rate": 8.900000000000001e-06,
"loss": 0.0984,
"step": 1550
},
{
"epoch": 0.39,
"learning_rate": 8.873684210526316e-06,
"loss": 0.0947,
"step": 1575
},
{
"epoch": 0.4,
"learning_rate": 8.847368421052633e-06,
"loss": 0.094,
"step": 1600
},
{
"epoch": 0.4,
"learning_rate": 8.821052631578948e-06,
"loss": 0.0995,
"step": 1625
},
{
"epoch": 0.41,
"learning_rate": 8.794736842105264e-06,
"loss": 0.096,
"step": 1650
},
{
"epoch": 0.42,
"learning_rate": 8.76842105263158e-06,
"loss": 0.1006,
"step": 1675
},
{
"epoch": 0.42,
"learning_rate": 8.742105263157894e-06,
"loss": 0.099,
"step": 1700
},
{
"epoch": 0.43,
"learning_rate": 8.715789473684211e-06,
"loss": 0.0946,
"step": 1725
},
{
"epoch": 0.43,
"learning_rate": 8.689473684210526e-06,
"loss": 0.091,
"step": 1750
},
{
"epoch": 0.44,
"learning_rate": 8.663157894736843e-06,
"loss": 0.0917,
"step": 1775
},
{
"epoch": 0.45,
"learning_rate": 8.63684210526316e-06,
"loss": 0.0901,
"step": 1800
},
{
"epoch": 0.45,
"learning_rate": 8.610526315789474e-06,
"loss": 0.091,
"step": 1825
},
{
"epoch": 0.46,
"learning_rate": 8.584210526315791e-06,
"loss": 0.0929,
"step": 1850
},
{
"epoch": 0.46,
"learning_rate": 8.557894736842106e-06,
"loss": 0.087,
"step": 1875
},
{
"epoch": 0.47,
"learning_rate": 8.531578947368423e-06,
"loss": 0.0904,
"step": 1900
},
{
"epoch": 0.48,
"learning_rate": 8.505263157894738e-06,
"loss": 0.0926,
"step": 1925
},
{
"epoch": 0.48,
"learning_rate": 8.478947368421053e-06,
"loss": 0.0877,
"step": 1950
},
{
"epoch": 0.49,
"learning_rate": 8.45263157894737e-06,
"loss": 0.0888,
"step": 1975
},
{
"epoch": 0.5,
"learning_rate": 8.426315789473684e-06,
"loss": 0.0851,
"step": 2000
},
{
"epoch": 0.5,
"eval_loss": 0.17117592692375183,
"eval_runtime": 474.7147,
"eval_samples_per_second": 3.562,
"eval_steps_per_second": 0.076,
"eval_wer": 8.754839515424004,
"step": 2000
},
{
"epoch": 0.5,
"learning_rate": 8.400000000000001e-06,
"loss": 0.0849,
"step": 2025
},
{
"epoch": 0.51,
"learning_rate": 8.373684210526316e-06,
"loss": 0.092,
"step": 2050
},
{
"epoch": 0.51,
"learning_rate": 8.347368421052633e-06,
"loss": 0.0879,
"step": 2075
},
{
"epoch": 0.52,
"learning_rate": 8.32105263157895e-06,
"loss": 0.0879,
"step": 2100
},
{
"epoch": 0.53,
"learning_rate": 8.294736842105264e-06,
"loss": 0.0861,
"step": 2125
},
{
"epoch": 0.53,
"learning_rate": 8.26842105263158e-06,
"loss": 0.0859,
"step": 2150
},
{
"epoch": 0.54,
"learning_rate": 8.242105263157896e-06,
"loss": 0.0847,
"step": 2175
},
{
"epoch": 0.55,
"learning_rate": 8.21578947368421e-06,
"loss": 0.0836,
"step": 2200
},
{
"epoch": 0.55,
"learning_rate": 8.189473684210527e-06,
"loss": 0.0858,
"step": 2225
},
{
"epoch": 0.56,
"learning_rate": 8.163157894736842e-06,
"loss": 0.0821,
"step": 2250
},
{
"epoch": 0.56,
"learning_rate": 8.136842105263159e-06,
"loss": 0.0827,
"step": 2275
},
{
"epoch": 0.57,
"learning_rate": 8.110526315789474e-06,
"loss": 0.0813,
"step": 2300
},
{
"epoch": 0.58,
"learning_rate": 8.08421052631579e-06,
"loss": 0.0855,
"step": 2325
},
{
"epoch": 0.58,
"learning_rate": 8.057894736842106e-06,
"loss": 0.0813,
"step": 2350
},
{
"epoch": 0.59,
"learning_rate": 8.03157894736842e-06,
"loss": 0.0823,
"step": 2375
},
{
"epoch": 0.59,
"learning_rate": 8.005263157894737e-06,
"loss": 0.0794,
"step": 2400
},
{
"epoch": 0.6,
"learning_rate": 7.978947368421052e-06,
"loss": 0.0807,
"step": 2425
},
{
"epoch": 0.61,
"learning_rate": 7.952631578947369e-06,
"loss": 0.0792,
"step": 2450
},
{
"epoch": 0.61,
"learning_rate": 7.926315789473686e-06,
"loss": 0.0824,
"step": 2475
},
{
"epoch": 0.62,
"learning_rate": 7.9e-06,
"loss": 0.0803,
"step": 2500
},
{
"epoch": 0.63,
"learning_rate": 7.873684210526317e-06,
"loss": 0.0792,
"step": 2525
},
{
"epoch": 0.63,
"learning_rate": 7.847368421052632e-06,
"loss": 0.0824,
"step": 2550
},
{
"epoch": 0.64,
"learning_rate": 7.821052631578949e-06,
"loss": 0.0757,
"step": 2575
},
{
"epoch": 0.64,
"learning_rate": 7.794736842105264e-06,
"loss": 0.0857,
"step": 2600
},
{
"epoch": 0.65,
"learning_rate": 7.768421052631579e-06,
"loss": 0.0762,
"step": 2625
},
{
"epoch": 0.66,
"learning_rate": 7.742105263157896e-06,
"loss": 0.0755,
"step": 2650
},
{
"epoch": 0.66,
"learning_rate": 7.71578947368421e-06,
"loss": 0.0801,
"step": 2675
},
{
"epoch": 0.67,
"learning_rate": 7.689473684210527e-06,
"loss": 0.0787,
"step": 2700
},
{
"epoch": 0.68,
"learning_rate": 7.663157894736842e-06,
"loss": 0.0785,
"step": 2725
},
{
"epoch": 0.68,
"learning_rate": 7.636842105263159e-06,
"loss": 0.0736,
"step": 2750
},
{
"epoch": 0.69,
"learning_rate": 7.610526315789474e-06,
"loss": 0.0811,
"step": 2775
},
{
"epoch": 0.69,
"learning_rate": 7.58421052631579e-06,
"loss": 0.0806,
"step": 2800
},
{
"epoch": 0.7,
"learning_rate": 7.557894736842106e-06,
"loss": 0.0782,
"step": 2825
},
{
"epoch": 0.71,
"learning_rate": 7.531578947368422e-06,
"loss": 0.0759,
"step": 2850
},
{
"epoch": 0.71,
"learning_rate": 7.505263157894738e-06,
"loss": 0.0762,
"step": 2875
},
{
"epoch": 0.72,
"learning_rate": 7.478947368421054e-06,
"loss": 0.0717,
"step": 2900
},
{
"epoch": 0.72,
"learning_rate": 7.4526315789473695e-06,
"loss": 0.0715,
"step": 2925
},
{
"epoch": 0.73,
"learning_rate": 7.4263157894736845e-06,
"loss": 0.0707,
"step": 2950
},
{
"epoch": 0.74,
"learning_rate": 7.4e-06,
"loss": 0.0752,
"step": 2975
},
{
"epoch": 0.74,
"learning_rate": 7.373684210526316e-06,
"loss": 0.0721,
"step": 3000
},
{
"epoch": 0.74,
"eval_loss": 0.15383455157279968,
"eval_runtime": 475.1252,
"eval_samples_per_second": 3.559,
"eval_steps_per_second": 0.076,
"eval_wer": 7.562133133508181,
"step": 3000
},
{
"epoch": 0.75,
"learning_rate": 7.347368421052632e-06,
"loss": 0.0763,
"step": 3025
},
{
"epoch": 0.76,
"learning_rate": 7.321052631578948e-06,
"loss": 0.0742,
"step": 3050
},
{
"epoch": 0.76,
"learning_rate": 7.2947368421052636e-06,
"loss": 0.0728,
"step": 3075
},
{
"epoch": 0.77,
"learning_rate": 7.268421052631579e-06,
"loss": 0.0755,
"step": 3100
},
{
"epoch": 0.77,
"learning_rate": 7.242105263157896e-06,
"loss": 0.0743,
"step": 3125
},
{
"epoch": 0.78,
"learning_rate": 7.215789473684212e-06,
"loss": 0.0695,
"step": 3150
},
{
"epoch": 0.79,
"learning_rate": 7.189473684210527e-06,
"loss": 0.0731,
"step": 3175
},
{
"epoch": 0.79,
"learning_rate": 7.163157894736843e-06,
"loss": 0.0741,
"step": 3200
},
{
"epoch": 0.8,
"learning_rate": 7.1368421052631585e-06,
"loss": 0.0762,
"step": 3225
},
{
"epoch": 0.81,
"learning_rate": 7.110526315789474e-06,
"loss": 0.0706,
"step": 3250
},
{
"epoch": 0.81,
"learning_rate": 7.08421052631579e-06,
"loss": 0.0728,
"step": 3275
},
{
"epoch": 0.82,
"learning_rate": 7.057894736842106e-06,
"loss": 0.0698,
"step": 3300
},
{
"epoch": 0.82,
"learning_rate": 7.031578947368422e-06,
"loss": 0.0715,
"step": 3325
},
{
"epoch": 0.83,
"learning_rate": 7.005263157894738e-06,
"loss": 0.0744,
"step": 3350
},
{
"epoch": 0.84,
"learning_rate": 6.9789473684210525e-06,
"loss": 0.0733,
"step": 3375
},
{
"epoch": 0.84,
"learning_rate": 6.952631578947368e-06,
"loss": 0.0735,
"step": 3400
},
{
"epoch": 0.85,
"learning_rate": 6.926315789473684e-06,
"loss": 0.0746,
"step": 3425
},
{
"epoch": 0.85,
"learning_rate": 6.9e-06,
"loss": 0.0697,
"step": 3450
},
{
"epoch": 0.86,
"learning_rate": 6.873684210526317e-06,
"loss": 0.0751,
"step": 3475
},
{
"epoch": 0.87,
"learning_rate": 6.8473684210526325e-06,
"loss": 0.0686,
"step": 3500
},
{
"epoch": 0.87,
"learning_rate": 6.821052631578948e-06,
"loss": 0.0699,
"step": 3525
},
{
"epoch": 0.88,
"learning_rate": 6.794736842105264e-06,
"loss": 0.0682,
"step": 3550
},
{
"epoch": 0.89,
"learning_rate": 6.76842105263158e-06,
"loss": 0.07,
"step": 3575
},
{
"epoch": 0.89,
"learning_rate": 6.742105263157896e-06,
"loss": 0.0644,
"step": 3600
},
{
"epoch": 0.9,
"learning_rate": 6.715789473684211e-06,
"loss": 0.0691,
"step": 3625
},
{
"epoch": 0.9,
"learning_rate": 6.6894736842105266e-06,
"loss": 0.0695,
"step": 3650
},
{
"epoch": 0.91,
"learning_rate": 6.663157894736842e-06,
"loss": 0.0706,
"step": 3675
},
{
"epoch": 0.92,
"learning_rate": 6.636842105263158e-06,
"loss": 0.0658,
"step": 3700
},
{
"epoch": 0.92,
"learning_rate": 6.610526315789474e-06,
"loss": 0.0649,
"step": 3725
},
{
"epoch": 0.93,
"learning_rate": 6.58421052631579e-06,
"loss": 0.0648,
"step": 3750
},
{
"epoch": 0.94,
"learning_rate": 6.557894736842106e-06,
"loss": 0.0691,
"step": 3775
},
{
"epoch": 0.94,
"learning_rate": 6.531578947368422e-06,
"loss": 0.0659,
"step": 3800
},
{
"epoch": 0.95,
"learning_rate": 6.505263157894738e-06,
"loss": 0.066,
"step": 3825
},
{
"epoch": 0.95,
"learning_rate": 6.478947368421054e-06,
"loss": 0.0651,
"step": 3850
},
{
"epoch": 0.96,
"learning_rate": 6.452631578947369e-06,
"loss": 0.0657,
"step": 3875
},
{
"epoch": 0.97,
"learning_rate": 6.426315789473685e-06,
"loss": 0.0665,
"step": 3900
},
{
"epoch": 0.97,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.0624,
"step": 3925
},
{
"epoch": 0.98,
"learning_rate": 6.373684210526316e-06,
"loss": 0.0694,
"step": 3950
},
{
"epoch": 0.98,
"learning_rate": 6.347368421052632e-06,
"loss": 0.0598,
"step": 3975
},
{
"epoch": 0.99,
"learning_rate": 6.321052631578948e-06,
"loss": 0.066,
"step": 4000
},
{
"epoch": 0.99,
"eval_loss": 0.14321504533290863,
"eval_runtime": 474.0103,
"eval_samples_per_second": 3.567,
"eval_steps_per_second": 0.076,
"eval_wer": 6.812788809791432,
"step": 4000
},
{
"epoch": 1.0,
"learning_rate": 6.294736842105264e-06,
"loss": 0.067,
"step": 4025
},
{
"epoch": 1.0,
"learning_rate": 6.268421052631579e-06,
"loss": 0.0513,
"step": 4050
},
{
"epoch": 1.01,
"learning_rate": 6.242105263157895e-06,
"loss": 0.0385,
"step": 4075
},
{
"epoch": 1.02,
"learning_rate": 6.2157894736842104e-06,
"loss": 0.0386,
"step": 4100
},
{
"epoch": 1.02,
"learning_rate": 6.189473684210526e-06,
"loss": 0.0451,
"step": 4125
},
{
"epoch": 1.03,
"learning_rate": 6.163157894736843e-06,
"loss": 0.0397,
"step": 4150
},
{
"epoch": 1.03,
"learning_rate": 6.136842105263159e-06,
"loss": 0.0404,
"step": 4175
},
{
"epoch": 1.04,
"learning_rate": 6.1105263157894746e-06,
"loss": 0.0427,
"step": 4200
},
{
"epoch": 1.05,
"learning_rate": 6.08421052631579e-06,
"loss": 0.0462,
"step": 4225
},
{
"epoch": 1.05,
"learning_rate": 6.057894736842106e-06,
"loss": 0.0412,
"step": 4250
},
{
"epoch": 1.06,
"learning_rate": 6.031578947368422e-06,
"loss": 0.0426,
"step": 4275
},
{
"epoch": 1.07,
"learning_rate": 6.005263157894737e-06,
"loss": 0.0387,
"step": 4300
},
{
"epoch": 1.07,
"learning_rate": 5.978947368421053e-06,
"loss": 0.0402,
"step": 4325
},
{
"epoch": 1.08,
"learning_rate": 5.952631578947369e-06,
"loss": 0.0428,
"step": 4350
},
{
"epoch": 1.08,
"learning_rate": 5.9263157894736844e-06,
"loss": 0.0428,
"step": 4375
},
{
"epoch": 1.09,
"learning_rate": 5.9e-06,
"loss": 0.0417,
"step": 4400
},
{
"epoch": 1.1,
"learning_rate": 5.873684210526316e-06,
"loss": 0.041,
"step": 4425
},
{
"epoch": 1.1,
"learning_rate": 5.847368421052632e-06,
"loss": 0.0434,
"step": 4450
},
{
"epoch": 1.11,
"learning_rate": 5.8210526315789486e-06,
"loss": 0.043,
"step": 4475
},
{
"epoch": 1.11,
"learning_rate": 5.794736842105264e-06,
"loss": 0.0505,
"step": 4500
},
{
"epoch": 1.12,
"learning_rate": 5.76842105263158e-06,
"loss": 0.0397,
"step": 4525
},
{
"epoch": 1.13,
"learning_rate": 5.742105263157895e-06,
"loss": 0.0425,
"step": 4550
},
{
"epoch": 1.13,
"learning_rate": 5.715789473684211e-06,
"loss": 0.0406,
"step": 4575
},
{
"epoch": 1.14,
"learning_rate": 5.689473684210527e-06,
"loss": 0.0387,
"step": 4600
},
{
"epoch": 1.15,
"learning_rate": 5.663157894736843e-06,
"loss": 0.0425,
"step": 4625
},
{
"epoch": 1.15,
"learning_rate": 5.6368421052631584e-06,
"loss": 0.0435,
"step": 4650
},
{
"epoch": 1.16,
"learning_rate": 5.610526315789474e-06,
"loss": 0.039,
"step": 4675
},
{
"epoch": 1.16,
"learning_rate": 5.58421052631579e-06,
"loss": 0.041,
"step": 4700
},
{
"epoch": 1.17,
"learning_rate": 5.557894736842105e-06,
"loss": 0.0422,
"step": 4725
},
{
"epoch": 1.18,
"learning_rate": 5.531578947368421e-06,
"loss": 0.0396,
"step": 4750
},
{
"epoch": 1.18,
"learning_rate": 5.505263157894737e-06,
"loss": 0.0439,
"step": 4775
},
{
"epoch": 1.19,
"learning_rate": 5.4789473684210525e-06,
"loss": 0.0453,
"step": 4800
},
{
"epoch": 1.2,
"learning_rate": 5.452631578947369e-06,
"loss": 0.0399,
"step": 4825
},
{
"epoch": 1.2,
"learning_rate": 5.426315789473685e-06,
"loss": 0.0447,
"step": 4850
},
{
"epoch": 1.21,
"learning_rate": 5.400000000000001e-06,
"loss": 0.0409,
"step": 4875
},
{
"epoch": 1.21,
"learning_rate": 5.373684210526317e-06,
"loss": 0.0406,
"step": 4900
},
{
"epoch": 1.22,
"learning_rate": 5.3473684210526325e-06,
"loss": 0.0407,
"step": 4925
},
{
"epoch": 1.23,
"learning_rate": 5.321052631578948e-06,
"loss": 0.0414,
"step": 4950
},
{
"epoch": 1.23,
"learning_rate": 5.294736842105263e-06,
"loss": 0.0407,
"step": 4975
},
{
"epoch": 1.24,
"learning_rate": 5.268421052631579e-06,
"loss": 0.0416,
"step": 5000
},
{
"epoch": 1.24,
"eval_loss": 0.13942231237888336,
"eval_runtime": 474.6679,
"eval_samples_per_second": 3.562,
"eval_steps_per_second": 0.076,
"eval_wer": 6.625452728862245,
"step": 5000
},
{
"epoch": 1.25,
"learning_rate": 5.242105263157895e-06,
"loss": 0.0412,
"step": 5025
},
{
"epoch": 1.25,
"learning_rate": 5.215789473684211e-06,
"loss": 0.045,
"step": 5050
},
{
"epoch": 1.26,
"learning_rate": 5.1894736842105265e-06,
"loss": 0.0425,
"step": 5075
},
{
"epoch": 1.26,
"learning_rate": 5.163157894736842e-06,
"loss": 0.0391,
"step": 5100
},
{
"epoch": 1.27,
"learning_rate": 5.136842105263158e-06,
"loss": 0.0391,
"step": 5125
},
{
"epoch": 1.28,
"learning_rate": 5.110526315789474e-06,
"loss": 0.0388,
"step": 5150
},
{
"epoch": 1.28,
"learning_rate": 5.084210526315791e-06,
"loss": 0.0447,
"step": 5175
},
{
"epoch": 1.29,
"learning_rate": 5.0578947368421065e-06,
"loss": 0.0406,
"step": 5200
},
{
"epoch": 1.29,
"learning_rate": 5.0315789473684214e-06,
"loss": 0.0391,
"step": 5225
},
{
"epoch": 1.3,
"learning_rate": 5.005263157894737e-06,
"loss": 0.0425,
"step": 5250
},
{
"epoch": 1.31,
"learning_rate": 4.978947368421053e-06,
"loss": 0.0433,
"step": 5275
},
{
"epoch": 1.31,
"learning_rate": 4.952631578947369e-06,
"loss": 0.0387,
"step": 5300
},
{
"epoch": 1.32,
"learning_rate": 4.926315789473685e-06,
"loss": 0.0413,
"step": 5325
},
{
"epoch": 1.33,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.038,
"step": 5350
},
{
"epoch": 1.33,
"learning_rate": 4.873684210526316e-06,
"loss": 0.045,
"step": 5375
},
{
"epoch": 1.34,
"learning_rate": 4.847368421052631e-06,
"loss": 0.0417,
"step": 5400
},
{
"epoch": 1.34,
"learning_rate": 4.821052631578948e-06,
"loss": 0.038,
"step": 5425
},
{
"epoch": 1.35,
"learning_rate": 4.794736842105264e-06,
"loss": 0.0386,
"step": 5450
},
{
"epoch": 1.36,
"learning_rate": 4.76842105263158e-06,
"loss": 0.0356,
"step": 5475
},
{
"epoch": 1.36,
"learning_rate": 4.7421052631578954e-06,
"loss": 0.0396,
"step": 5500
},
{
"epoch": 1.37,
"learning_rate": 4.71578947368421e-06,
"loss": 0.0372,
"step": 5525
},
{
"epoch": 1.38,
"learning_rate": 4.689473684210526e-06,
"loss": 0.039,
"step": 5550
},
{
"epoch": 1.38,
"learning_rate": 4.663157894736842e-06,
"loss": 0.0403,
"step": 5575
},
{
"epoch": 1.39,
"learning_rate": 4.636842105263159e-06,
"loss": 0.0455,
"step": 5600
},
{
"epoch": 1.39,
"learning_rate": 4.6105263157894745e-06,
"loss": 0.038,
"step": 5625
},
{
"epoch": 1.4,
"learning_rate": 4.5842105263157895e-06,
"loss": 0.0395,
"step": 5650
},
{
"epoch": 1.41,
"learning_rate": 4.557894736842105e-06,
"loss": 0.0385,
"step": 5675
},
{
"epoch": 1.41,
"learning_rate": 4.531578947368421e-06,
"loss": 0.0374,
"step": 5700
},
{
"epoch": 1.42,
"learning_rate": 4.505263157894737e-06,
"loss": 0.038,
"step": 5725
},
{
"epoch": 1.42,
"learning_rate": 4.478947368421054e-06,
"loss": 0.0397,
"step": 5750
},
{
"epoch": 1.43,
"learning_rate": 4.452631578947369e-06,
"loss": 0.0369,
"step": 5775
},
{
"epoch": 1.44,
"learning_rate": 4.426315789473684e-06,
"loss": 0.0429,
"step": 5800
},
{
"epoch": 1.44,
"learning_rate": 4.4e-06,
"loss": 0.0392,
"step": 5825
},
{
"epoch": 1.45,
"learning_rate": 4.373684210526316e-06,
"loss": 0.044,
"step": 5850
},
{
"epoch": 1.46,
"learning_rate": 4.347368421052632e-06,
"loss": 0.0409,
"step": 5875
},
{
"epoch": 1.46,
"learning_rate": 4.321052631578948e-06,
"loss": 0.0389,
"step": 5900
},
{
"epoch": 1.47,
"learning_rate": 4.2947368421052635e-06,
"loss": 0.0417,
"step": 5925
},
{
"epoch": 1.47,
"learning_rate": 4.268421052631579e-06,
"loss": 0.0376,
"step": 5950
},
{
"epoch": 1.48,
"learning_rate": 4.242105263157895e-06,
"loss": 0.0384,
"step": 5975
},
{
"epoch": 1.49,
"learning_rate": 4.215789473684211e-06,
"loss": 0.0355,
"step": 6000
},
{
"epoch": 1.49,
"eval_loss": 0.13929006457328796,
"eval_runtime": 473.3245,
"eval_samples_per_second": 3.573,
"eval_steps_per_second": 0.076,
"eval_wer": 6.450605719995004,
"step": 6000
},
{
"epoch": 1.49,
"learning_rate": 4.189473684210527e-06,
"loss": 0.0382,
"step": 6025
},
{
"epoch": 1.5,
"learning_rate": 4.163157894736843e-06,
"loss": 0.0402,
"step": 6050
},
{
"epoch": 1.51,
"learning_rate": 4.136842105263158e-06,
"loss": 0.0363,
"step": 6075
},
{
"epoch": 1.51,
"learning_rate": 4.110526315789474e-06,
"loss": 0.0359,
"step": 6100
},
{
"epoch": 1.52,
"learning_rate": 4.08421052631579e-06,
"loss": 0.0377,
"step": 6125
},
{
"epoch": 1.52,
"learning_rate": 4.057894736842106e-06,
"loss": 0.0403,
"step": 6150
},
{
"epoch": 1.53,
"learning_rate": 4.031578947368422e-06,
"loss": 0.0368,
"step": 6175
},
{
"epoch": 1.54,
"learning_rate": 4.005263157894737e-06,
"loss": 0.0411,
"step": 6200
},
{
"epoch": 1.54,
"learning_rate": 3.9789473684210525e-06,
"loss": 0.0397,
"step": 6225
},
{
"epoch": 1.55,
"learning_rate": 3.952631578947368e-06,
"loss": 0.0378,
"step": 6250
},
{
"epoch": 1.55,
"learning_rate": 3.926315789473685e-06,
"loss": 0.0394,
"step": 6275
},
{
"epoch": 1.56,
"learning_rate": 3.900000000000001e-06,
"loss": 0.0387,
"step": 6300
},
{
"epoch": 1.57,
"learning_rate": 3.873684210526316e-06,
"loss": 0.0383,
"step": 6325
},
{
"epoch": 1.57,
"learning_rate": 3.8473684210526316e-06,
"loss": 0.0363,
"step": 6350
},
{
"epoch": 1.58,
"learning_rate": 3.821052631578947e-06,
"loss": 0.038,
"step": 6375
},
{
"epoch": 1.59,
"learning_rate": 3.794736842105263e-06,
"loss": 0.0405,
"step": 6400
},
{
"epoch": 1.59,
"learning_rate": 3.768421052631579e-06,
"loss": 0.0382,
"step": 6425
},
{
"epoch": 1.6,
"learning_rate": 3.7421052631578953e-06,
"loss": 0.037,
"step": 6450
},
{
"epoch": 1.6,
"learning_rate": 3.715789473684211e-06,
"loss": 0.0367,
"step": 6475
},
{
"epoch": 1.61,
"learning_rate": 3.6894736842105265e-06,
"loss": 0.037,
"step": 6500
},
{
"epoch": 1.62,
"learning_rate": 3.6631578947368423e-06,
"loss": 0.0358,
"step": 6525
},
{
"epoch": 1.62,
"learning_rate": 3.636842105263158e-06,
"loss": 0.0367,
"step": 6550
},
{
"epoch": 1.63,
"learning_rate": 3.610526315789474e-06,
"loss": 0.0373,
"step": 6575
},
{
"epoch": 1.64,
"learning_rate": 3.58421052631579e-06,
"loss": 0.0382,
"step": 6600
},
{
"epoch": 1.64,
"learning_rate": 3.5578947368421056e-06,
"loss": 0.0348,
"step": 6625
},
{
"epoch": 1.65,
"learning_rate": 3.5315789473684214e-06,
"loss": 0.0384,
"step": 6650
},
{
"epoch": 1.65,
"learning_rate": 3.505263157894737e-06,
"loss": 0.0358,
"step": 6675
},
{
"epoch": 1.66,
"learning_rate": 3.478947368421053e-06,
"loss": 0.0353,
"step": 6700
},
{
"epoch": 1.67,
"learning_rate": 3.4526315789473684e-06,
"loss": 0.0357,
"step": 6725
},
{
"epoch": 1.67,
"learning_rate": 3.4263157894736842e-06,
"loss": 0.0404,
"step": 6750
},
{
"epoch": 1.68,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.0369,
"step": 6775
},
{
"epoch": 1.68,
"learning_rate": 3.3736842105263163e-06,
"loss": 0.0374,
"step": 6800
},
{
"epoch": 1.69,
"learning_rate": 3.347368421052632e-06,
"loss": 0.0358,
"step": 6825
},
{
"epoch": 1.7,
"learning_rate": 3.3210526315789475e-06,
"loss": 0.0377,
"step": 6850
},
{
"epoch": 1.7,
"learning_rate": 3.2947368421052633e-06,
"loss": 0.0361,
"step": 6875
},
{
"epoch": 1.71,
"learning_rate": 3.268421052631579e-06,
"loss": 0.0364,
"step": 6900
},
{
"epoch": 1.72,
"learning_rate": 3.2421052631578945e-06,
"loss": 0.039,
"step": 6925
},
{
"epoch": 1.72,
"learning_rate": 3.215789473684211e-06,
"loss": 0.0366,
"step": 6950
},
{
"epoch": 1.73,
"learning_rate": 3.1894736842105266e-06,
"loss": 0.0376,
"step": 6975
},
{
"epoch": 1.73,
"learning_rate": 3.1631578947368424e-06,
"loss": 0.0382,
"step": 7000
},
{
"epoch": 1.73,
"eval_loss": 0.1361781358718872,
"eval_runtime": 472.4079,
"eval_samples_per_second": 3.58,
"eval_steps_per_second": 0.076,
"eval_wer": 6.369426751592357,
"step": 7000
},
{
"epoch": 1.74,
"learning_rate": 3.1368421052631582e-06,
"loss": 0.0362,
"step": 7025
},
{
"epoch": 1.75,
"learning_rate": 3.1105263157894736e-06,
"loss": 0.038,
"step": 7050
},
{
"epoch": 1.75,
"learning_rate": 3.0842105263157895e-06,
"loss": 0.0358,
"step": 7075
},
{
"epoch": 1.76,
"learning_rate": 3.0578947368421053e-06,
"loss": 0.0381,
"step": 7100
},
{
"epoch": 1.77,
"learning_rate": 3.0315789473684215e-06,
"loss": 0.0364,
"step": 7125
},
{
"epoch": 1.77,
"learning_rate": 3.0052631578947373e-06,
"loss": 0.0384,
"step": 7150
},
{
"epoch": 1.78,
"learning_rate": 2.9789473684210527e-06,
"loss": 0.0381,
"step": 7175
},
{
"epoch": 1.78,
"learning_rate": 2.9526315789473685e-06,
"loss": 0.0359,
"step": 7200
},
{
"epoch": 1.79,
"learning_rate": 2.9263157894736844e-06,
"loss": 0.0357,
"step": 7225
},
{
"epoch": 1.8,
"learning_rate": 2.9e-06,
"loss": 0.0358,
"step": 7250
},
{
"epoch": 1.8,
"learning_rate": 2.8736842105263164e-06,
"loss": 0.0355,
"step": 7275
},
{
"epoch": 1.81,
"learning_rate": 2.847368421052632e-06,
"loss": 0.037,
"step": 7300
},
{
"epoch": 1.81,
"learning_rate": 2.8210526315789476e-06,
"loss": 0.0349,
"step": 7325
},
{
"epoch": 1.82,
"learning_rate": 2.7947368421052635e-06,
"loss": 0.0377,
"step": 7350
},
{
"epoch": 1.83,
"learning_rate": 2.7684210526315793e-06,
"loss": 0.0382,
"step": 7375
},
{
"epoch": 1.83,
"learning_rate": 2.7421052631578947e-06,
"loss": 0.0386,
"step": 7400
},
{
"epoch": 1.84,
"learning_rate": 2.7157894736842105e-06,
"loss": 0.0359,
"step": 7425
},
{
"epoch": 1.85,
"learning_rate": 2.6894736842105267e-06,
"loss": 0.0327,
"step": 7450
},
{
"epoch": 1.85,
"learning_rate": 2.6631578947368426e-06,
"loss": 0.0347,
"step": 7475
},
{
"epoch": 1.86,
"learning_rate": 2.6368421052631584e-06,
"loss": 0.0358,
"step": 7500
},
{
"epoch": 1.86,
"learning_rate": 2.6105263157894738e-06,
"loss": 0.0328,
"step": 7525
},
{
"epoch": 1.87,
"learning_rate": 2.5842105263157896e-06,
"loss": 0.0369,
"step": 7550
},
{
"epoch": 1.88,
"learning_rate": 2.5578947368421054e-06,
"loss": 0.0355,
"step": 7575
},
{
"epoch": 1.88,
"learning_rate": 2.5315789473684212e-06,
"loss": 0.0352,
"step": 7600
},
{
"epoch": 1.89,
"learning_rate": 2.5052631578947375e-06,
"loss": 0.0362,
"step": 7625
},
{
"epoch": 1.9,
"learning_rate": 2.478947368421053e-06,
"loss": 0.0362,
"step": 7650
},
{
"epoch": 1.9,
"learning_rate": 2.4526315789473687e-06,
"loss": 0.0371,
"step": 7675
},
{
"epoch": 1.91,
"learning_rate": 2.4263157894736845e-06,
"loss": 0.0353,
"step": 7700
},
{
"epoch": 1.91,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.0354,
"step": 7725
},
{
"epoch": 1.92,
"learning_rate": 2.373684210526316e-06,
"loss": 0.0376,
"step": 7750
},
{
"epoch": 1.93,
"learning_rate": 2.347368421052632e-06,
"loss": 0.0366,
"step": 7775
},
{
"epoch": 1.93,
"learning_rate": 2.3210526315789473e-06,
"loss": 0.0358,
"step": 7800
},
{
"epoch": 1.94,
"learning_rate": 2.294736842105263e-06,
"loss": 0.0364,
"step": 7825
},
{
"epoch": 1.94,
"learning_rate": 2.268421052631579e-06,
"loss": 0.0348,
"step": 7850
},
{
"epoch": 1.95,
"learning_rate": 2.242105263157895e-06,
"loss": 0.0341,
"step": 7875
},
{
"epoch": 1.96,
"learning_rate": 2.2157894736842106e-06,
"loss": 0.0386,
"step": 7900
},
{
"epoch": 1.96,
"learning_rate": 2.1894736842105264e-06,
"loss": 0.0358,
"step": 7925
},
{
"epoch": 1.97,
"learning_rate": 2.1631578947368423e-06,
"loss": 0.0361,
"step": 7950
},
{
"epoch": 1.98,
"learning_rate": 2.136842105263158e-06,
"loss": 0.0336,
"step": 7975
},
{
"epoch": 1.98,
"learning_rate": 2.110526315789474e-06,
"loss": 0.0344,
"step": 8000
},
{
"epoch": 1.98,
"eval_loss": 0.13453030586242676,
"eval_runtime": 468.6645,
"eval_samples_per_second": 3.608,
"eval_steps_per_second": 0.077,
"eval_wer": 6.257025103034844,
"step": 8000
},
{
"epoch": 1.99,
"learning_rate": 2.0842105263157897e-06,
"loss": 0.0375,
"step": 8025
},
{
"epoch": 1.99,
"learning_rate": 2.0578947368421055e-06,
"loss": 0.0319,
"step": 8050
},
{
"epoch": 2.0,
"learning_rate": 2.031578947368421e-06,
"loss": 0.0307,
"step": 8075
},
{
"epoch": 2.01,
"learning_rate": 2.005263157894737e-06,
"loss": 0.0209,
"step": 8100
},
{
"epoch": 2.01,
"learning_rate": 1.978947368421053e-06,
"loss": 0.0191,
"step": 8125
},
{
"epoch": 2.02,
"learning_rate": 1.9526315789473684e-06,
"loss": 0.0209,
"step": 8150
},
{
"epoch": 2.03,
"learning_rate": 1.9263157894736846e-06,
"loss": 0.023,
"step": 8175
},
{
"epoch": 2.03,
"learning_rate": 1.9000000000000002e-06,
"loss": 0.0199,
"step": 8200
},
{
"epoch": 2.04,
"learning_rate": 1.8736842105263158e-06,
"loss": 0.0198,
"step": 8225
},
{
"epoch": 2.04,
"learning_rate": 1.8473684210526319e-06,
"loss": 0.0198,
"step": 8250
},
{
"epoch": 2.05,
"learning_rate": 1.8210526315789475e-06,
"loss": 0.0213,
"step": 8275
},
{
"epoch": 2.06,
"learning_rate": 1.7947368421052633e-06,
"loss": 0.0196,
"step": 8300
},
{
"epoch": 2.06,
"learning_rate": 1.768421052631579e-06,
"loss": 0.0206,
"step": 8325
},
{
"epoch": 2.07,
"learning_rate": 1.742105263157895e-06,
"loss": 0.023,
"step": 8350
},
{
"epoch": 2.08,
"learning_rate": 1.7157894736842107e-06,
"loss": 0.0254,
"step": 8375
},
{
"epoch": 2.08,
"learning_rate": 1.6894736842105264e-06,
"loss": 0.0204,
"step": 8400
},
{
"epoch": 2.09,
"learning_rate": 1.6631578947368424e-06,
"loss": 0.0197,
"step": 8425
},
{
"epoch": 2.09,
"learning_rate": 1.636842105263158e-06,
"loss": 0.0197,
"step": 8450
},
{
"epoch": 2.1,
"learning_rate": 1.6105263157894738e-06,
"loss": 0.0179,
"step": 8475
},
{
"epoch": 2.11,
"learning_rate": 1.5842105263157894e-06,
"loss": 0.0188,
"step": 8500
},
{
"epoch": 2.11,
"learning_rate": 1.5578947368421054e-06,
"loss": 0.0207,
"step": 8525
},
{
"epoch": 2.12,
"learning_rate": 1.5315789473684213e-06,
"loss": 0.0202,
"step": 8550
},
{
"epoch": 2.12,
"learning_rate": 1.5052631578947369e-06,
"loss": 0.0185,
"step": 8575
},
{
"epoch": 2.13,
"learning_rate": 1.478947368421053e-06,
"loss": 0.0208,
"step": 8600
},
{
"epoch": 2.14,
"learning_rate": 1.4526315789473685e-06,
"loss": 0.02,
"step": 8625
},
{
"epoch": 2.14,
"learning_rate": 1.4263157894736843e-06,
"loss": 0.0186,
"step": 8650
},
{
"epoch": 2.15,
"learning_rate": 1.4000000000000001e-06,
"loss": 0.0192,
"step": 8675
},
{
"epoch": 2.16,
"learning_rate": 1.373684210526316e-06,
"loss": 0.0186,
"step": 8700
},
{
"epoch": 2.16,
"learning_rate": 1.3473684210526316e-06,
"loss": 0.0194,
"step": 8725
},
{
"epoch": 2.17,
"learning_rate": 1.3210526315789474e-06,
"loss": 0.0201,
"step": 8750
},
{
"epoch": 2.17,
"learning_rate": 1.2947368421052634e-06,
"loss": 0.022,
"step": 8775
},
{
"epoch": 2.18,
"learning_rate": 1.268421052631579e-06,
"loss": 0.0193,
"step": 8800
},
{
"epoch": 2.19,
"learning_rate": 1.2421052631578948e-06,
"loss": 0.0165,
"step": 8825
},
{
"epoch": 2.19,
"learning_rate": 1.2157894736842107e-06,
"loss": 0.0206,
"step": 8850
},
{
"epoch": 2.2,
"learning_rate": 1.1894736842105265e-06,
"loss": 0.0166,
"step": 8875
},
{
"epoch": 2.21,
"learning_rate": 1.163157894736842e-06,
"loss": 0.0204,
"step": 8900
},
{
"epoch": 2.21,
"learning_rate": 1.136842105263158e-06,
"loss": 0.0202,
"step": 8925
},
{
"epoch": 2.22,
"learning_rate": 1.110526315789474e-06,
"loss": 0.0192,
"step": 8950
},
{
"epoch": 2.22,
"learning_rate": 1.0842105263157895e-06,
"loss": 0.0184,
"step": 8975
},
{
"epoch": 2.23,
"learning_rate": 1.0578947368421054e-06,
"loss": 0.0188,
"step": 9000
},
{
"epoch": 2.23,
"eval_loss": 0.1395842432975769,
"eval_runtime": 473.9745,
"eval_samples_per_second": 3.568,
"eval_steps_per_second": 0.076,
"eval_wer": 6.113400774322468,
"step": 9000
},
{
"epoch": 2.24,
"learning_rate": 1.0315789473684212e-06,
"loss": 0.0193,
"step": 9025
},
{
"epoch": 2.24,
"learning_rate": 1.005263157894737e-06,
"loss": 0.0201,
"step": 9050
},
{
"epoch": 2.25,
"learning_rate": 9.789473684210526e-07,
"loss": 0.0193,
"step": 9075
},
{
"epoch": 2.25,
"learning_rate": 9.526315789473685e-07,
"loss": 0.0178,
"step": 9100
},
{
"epoch": 2.26,
"learning_rate": 9.263157894736844e-07,
"loss": 0.0191,
"step": 9125
},
{
"epoch": 2.27,
"learning_rate": 9.000000000000001e-07,
"loss": 0.0186,
"step": 9150
},
{
"epoch": 2.27,
"learning_rate": 8.736842105263159e-07,
"loss": 0.0212,
"step": 9175
},
{
"epoch": 2.28,
"learning_rate": 8.473684210526316e-07,
"loss": 0.0201,
"step": 9200
},
{
"epoch": 2.29,
"learning_rate": 8.210526315789474e-07,
"loss": 0.0221,
"step": 9225
},
{
"epoch": 2.29,
"learning_rate": 7.947368421052632e-07,
"loss": 0.0183,
"step": 9250
},
{
"epoch": 2.3,
"learning_rate": 7.684210526315789e-07,
"loss": 0.0189,
"step": 9275
},
{
"epoch": 2.3,
"learning_rate": 7.421052631578949e-07,
"loss": 0.0193,
"step": 9300
},
{
"epoch": 2.31,
"learning_rate": 7.157894736842106e-07,
"loss": 0.0192,
"step": 9325
},
{
"epoch": 2.32,
"learning_rate": 6.894736842105264e-07,
"loss": 0.0181,
"step": 9350
},
{
"epoch": 2.32,
"learning_rate": 6.631578947368422e-07,
"loss": 0.0185,
"step": 9375
},
{
"epoch": 2.33,
"learning_rate": 6.368421052631579e-07,
"loss": 0.0174,
"step": 9400
},
{
"epoch": 2.34,
"learning_rate": 6.105263157894738e-07,
"loss": 0.0188,
"step": 9425
},
{
"epoch": 2.34,
"learning_rate": 5.842105263157896e-07,
"loss": 0.0202,
"step": 9450
},
{
"epoch": 2.35,
"learning_rate": 5.578947368421053e-07,
"loss": 0.0185,
"step": 9475
},
{
"epoch": 2.35,
"learning_rate": 5.315789473684211e-07,
"loss": 0.0177,
"step": 9500
},
{
"epoch": 2.36,
"learning_rate": 5.052631578947369e-07,
"loss": 0.0183,
"step": 9525
},
{
"epoch": 2.37,
"learning_rate": 4.789473684210526e-07,
"loss": 0.0205,
"step": 9550
},
{
"epoch": 2.37,
"learning_rate": 4.526315789473685e-07,
"loss": 0.0185,
"step": 9575
},
{
"epoch": 2.38,
"learning_rate": 4.2631578947368427e-07,
"loss": 0.0186,
"step": 9600
},
{
"epoch": 2.38,
"learning_rate": 4.0000000000000003e-07,
"loss": 0.0197,
"step": 9625
},
{
"epoch": 2.39,
"learning_rate": 3.736842105263158e-07,
"loss": 0.0188,
"step": 9650
},
{
"epoch": 2.4,
"learning_rate": 3.4736842105263157e-07,
"loss": 0.0185,
"step": 9675
},
{
"epoch": 2.4,
"learning_rate": 3.210526315789474e-07,
"loss": 0.0208,
"step": 9700
},
{
"epoch": 2.41,
"learning_rate": 2.9473684210526315e-07,
"loss": 0.0197,
"step": 9725
},
{
"epoch": 2.42,
"learning_rate": 2.6842105263157897e-07,
"loss": 0.0197,
"step": 9750
},
{
"epoch": 2.42,
"learning_rate": 2.4210526315789473e-07,
"loss": 0.0182,
"step": 9775
},
{
"epoch": 2.43,
"learning_rate": 2.1578947368421053e-07,
"loss": 0.0197,
"step": 9800
},
{
"epoch": 2.43,
"learning_rate": 1.8947368421052634e-07,
"loss": 0.0177,
"step": 9825
},
{
"epoch": 2.44,
"learning_rate": 1.631578947368421e-07,
"loss": 0.0186,
"step": 9850
},
{
"epoch": 2.45,
"learning_rate": 1.368421052631579e-07,
"loss": 0.0197,
"step": 9875
},
{
"epoch": 2.45,
"learning_rate": 1.1052631578947368e-07,
"loss": 0.0193,
"step": 9900
},
{
"epoch": 2.46,
"learning_rate": 8.421052631578947e-08,
"loss": 0.0207,
"step": 9925
},
{
"epoch": 2.47,
"learning_rate": 5.789473684210527e-08,
"loss": 0.019,
"step": 9950
},
{
"epoch": 2.47,
"learning_rate": 3.157894736842106e-08,
"loss": 0.0198,
"step": 9975
},
{
"epoch": 2.48,
"learning_rate": 5.263157894736842e-09,
"loss": 0.0204,
"step": 10000
},
{
"epoch": 2.48,
"eval_loss": 0.13874408602714539,
"eval_runtime": 474.3434,
"eval_samples_per_second": 3.565,
"eval_steps_per_second": 0.076,
"eval_wer": 5.957287373548145,
"step": 10000
},
{
"epoch": 2.48,
"step": 10000,
"total_flos": 9.797194749247488e+20,
"train_loss": 0.09647957161068917,
"train_runtime": 71880.1897,
"train_samples_per_second": 13.356,
"train_steps_per_second": 0.139
}
],
"max_steps": 10000,
"num_train_epochs": 3,
"total_flos": 9.797194749247488e+20,
"trial_name": null,
"trial_params": null
}