whisper-large-ar / trainer_state.json
lorenzoncina's picture
End of training
6cdfa64
raw
history blame contribute delete
No virus
51.1 kB
{
"best_metric": 49.431999999999995,
"best_model_checkpoint": "./checkpoint-3000",
"epoch": 3.217,
"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": 4.4e-07,
"loss": 0.8731,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 0.5943,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 0.3422,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.94e-06,
"loss": 0.2422,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.2541,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.255,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.44e-06,
"loss": 0.2586,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.94e-06,
"loss": 0.282,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.440000000000001e-06,
"loss": 0.3017,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 4.94e-06,
"loss": 0.2541,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.2638,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.94e-06,
"loss": 0.2375,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.440000000000001e-06,
"loss": 0.2612,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.2622,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 7.440000000000001e-06,
"loss": 0.2281,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 7.94e-06,
"loss": 0.2826,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.44e-06,
"loss": 0.28,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.94e-06,
"loss": 0.2756,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 9.440000000000001e-06,
"loss": 0.2957,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 9.940000000000001e-06,
"loss": 0.2812,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 9.976842105263158e-06,
"loss": 0.2708,
"step": 525
},
{
"epoch": 0.06,
"learning_rate": 9.950526315789475e-06,
"loss": 0.2471,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 9.92421052631579e-06,
"loss": 0.2752,
"step": 575
},
{
"epoch": 0.06,
"learning_rate": 9.897894736842107e-06,
"loss": 0.2201,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 9.871578947368422e-06,
"loss": 0.3024,
"step": 625
},
{
"epoch": 0.07,
"learning_rate": 9.845263157894738e-06,
"loss": 0.2986,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 9.818947368421053e-06,
"loss": 0.2732,
"step": 675
},
{
"epoch": 0.07,
"learning_rate": 9.79263157894737e-06,
"loss": 0.2738,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 9.766315789473685e-06,
"loss": 0.2624,
"step": 725
},
{
"epoch": 0.07,
"learning_rate": 9.74e-06,
"loss": 0.2741,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 9.713684210526317e-06,
"loss": 0.2742,
"step": 775
},
{
"epoch": 0.08,
"learning_rate": 9.687368421052632e-06,
"loss": 0.2914,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 9.661052631578948e-06,
"loss": 0.343,
"step": 825
},
{
"epoch": 0.09,
"learning_rate": 9.634736842105265e-06,
"loss": 0.2789,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 9.60842105263158e-06,
"loss": 0.259,
"step": 875
},
{
"epoch": 0.09,
"learning_rate": 9.582105263157897e-06,
"loss": 0.347,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 9.555789473684211e-06,
"loss": 0.3105,
"step": 925
},
{
"epoch": 0.1,
"learning_rate": 9.529473684210528e-06,
"loss": 0.2564,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 9.503157894736843e-06,
"loss": 0.282,
"step": 975
},
{
"epoch": 0.1,
"learning_rate": 9.476842105263158e-06,
"loss": 0.2472,
"step": 1000
},
{
"epoch": 0.1,
"eval_loss": 0.3718579411506653,
"eval_runtime": 5114.1069,
"eval_samples_per_second": 2.041,
"eval_steps_per_second": 1.021,
"eval_wer": 58.955999999999996,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 9.450526315789475e-06,
"loss": 0.2387,
"step": 1025
},
{
"epoch": 0.1,
"learning_rate": 9.42421052631579e-06,
"loss": 0.2111,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 9.397894736842106e-06,
"loss": 0.2293,
"step": 1075
},
{
"epoch": 0.11,
"learning_rate": 9.371578947368421e-06,
"loss": 0.2512,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 9.345263157894738e-06,
"loss": 0.2043,
"step": 1125
},
{
"epoch": 0.12,
"learning_rate": 9.318947368421053e-06,
"loss": 0.2481,
"step": 1150
},
{
"epoch": 0.12,
"learning_rate": 9.292631578947368e-06,
"loss": 0.2144,
"step": 1175
},
{
"epoch": 0.12,
"learning_rate": 9.266315789473685e-06,
"loss": 0.1477,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 9.240000000000001e-06,
"loss": 0.1886,
"step": 1225
},
{
"epoch": 0.12,
"learning_rate": 9.213684210526316e-06,
"loss": 0.218,
"step": 1250
},
{
"epoch": 0.13,
"learning_rate": 9.187368421052633e-06,
"loss": 0.2195,
"step": 1275
},
{
"epoch": 0.13,
"learning_rate": 9.161052631578948e-06,
"loss": 0.2277,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 9.134736842105265e-06,
"loss": 0.2173,
"step": 1325
},
{
"epoch": 0.14,
"learning_rate": 9.10842105263158e-06,
"loss": 0.2032,
"step": 1350
},
{
"epoch": 0.14,
"learning_rate": 9.082105263157896e-06,
"loss": 0.2126,
"step": 1375
},
{
"epoch": 0.14,
"learning_rate": 9.055789473684211e-06,
"loss": 0.2117,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 9.029473684210526e-06,
"loss": 0.2007,
"step": 1425
},
{
"epoch": 0.14,
"learning_rate": 9.003157894736843e-06,
"loss": 0.2033,
"step": 1450
},
{
"epoch": 0.15,
"learning_rate": 8.976842105263158e-06,
"loss": 0.2162,
"step": 1475
},
{
"epoch": 0.15,
"learning_rate": 8.950526315789474e-06,
"loss": 0.2146,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 8.92421052631579e-06,
"loss": 0.2034,
"step": 1525
},
{
"epoch": 0.15,
"learning_rate": 8.897894736842106e-06,
"loss": 0.2155,
"step": 1550
},
{
"epoch": 0.16,
"learning_rate": 8.871578947368423e-06,
"loss": 0.219,
"step": 1575
},
{
"epoch": 0.16,
"learning_rate": 8.845263157894738e-06,
"loss": 0.2211,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 8.818947368421054e-06,
"loss": 0.2251,
"step": 1625
},
{
"epoch": 0.17,
"learning_rate": 8.79263157894737e-06,
"loss": 0.2207,
"step": 1650
},
{
"epoch": 0.17,
"learning_rate": 8.766315789473684e-06,
"loss": 0.2878,
"step": 1675
},
{
"epoch": 0.17,
"learning_rate": 8.740000000000001e-06,
"loss": 0.2485,
"step": 1700
},
{
"epoch": 0.17,
"learning_rate": 8.713684210526316e-06,
"loss": 0.2863,
"step": 1725
},
{
"epoch": 0.17,
"learning_rate": 8.687368421052633e-06,
"loss": 0.2637,
"step": 1750
},
{
"epoch": 0.18,
"learning_rate": 8.661052631578948e-06,
"loss": 0.2477,
"step": 1775
},
{
"epoch": 0.18,
"learning_rate": 8.634736842105264e-06,
"loss": 0.2323,
"step": 1800
},
{
"epoch": 0.18,
"learning_rate": 8.60842105263158e-06,
"loss": 0.2755,
"step": 1825
},
{
"epoch": 0.18,
"learning_rate": 8.582105263157894e-06,
"loss": 0.2407,
"step": 1850
},
{
"epoch": 0.19,
"learning_rate": 8.555789473684213e-06,
"loss": 0.226,
"step": 1875
},
{
"epoch": 0.19,
"learning_rate": 8.529473684210528e-06,
"loss": 0.2367,
"step": 1900
},
{
"epoch": 0.19,
"learning_rate": 8.503157894736843e-06,
"loss": 0.2028,
"step": 1925
},
{
"epoch": 0.2,
"learning_rate": 8.47684210526316e-06,
"loss": 0.1869,
"step": 1950
},
{
"epoch": 0.2,
"learning_rate": 8.450526315789474e-06,
"loss": 0.2269,
"step": 1975
},
{
"epoch": 0.2,
"learning_rate": 8.42421052631579e-06,
"loss": 0.2015,
"step": 2000
},
{
"epoch": 0.2,
"eval_loss": 0.3487010896205902,
"eval_runtime": 5134.453,
"eval_samples_per_second": 2.033,
"eval_steps_per_second": 1.017,
"eval_wer": 53.52133333333333,
"step": 2000
},
{
"epoch": 0.2,
"learning_rate": 8.397894736842106e-06,
"loss": 0.2632,
"step": 2025
},
{
"epoch": 0.2,
"learning_rate": 8.371578947368422e-06,
"loss": 0.2344,
"step": 2050
},
{
"epoch": 0.21,
"learning_rate": 8.345263157894737e-06,
"loss": 0.2304,
"step": 2075
},
{
"epoch": 0.21,
"learning_rate": 8.318947368421052e-06,
"loss": 0.266,
"step": 2100
},
{
"epoch": 0.21,
"learning_rate": 8.292631578947369e-06,
"loss": 0.2169,
"step": 2125
},
{
"epoch": 0.21,
"learning_rate": 8.266315789473684e-06,
"loss": 0.1856,
"step": 2150
},
{
"epoch": 0.22,
"learning_rate": 8.24e-06,
"loss": 0.2056,
"step": 2175
},
{
"epoch": 0.22,
"learning_rate": 8.213684210526316e-06,
"loss": 0.2006,
"step": 2200
},
{
"epoch": 0.22,
"learning_rate": 8.187368421052632e-06,
"loss": 0.1852,
"step": 2225
},
{
"epoch": 0.23,
"learning_rate": 8.161052631578949e-06,
"loss": 0.1718,
"step": 2250
},
{
"epoch": 0.23,
"learning_rate": 8.134736842105264e-06,
"loss": 0.1882,
"step": 2275
},
{
"epoch": 0.23,
"learning_rate": 8.10842105263158e-06,
"loss": 0.155,
"step": 2300
},
{
"epoch": 0.23,
"learning_rate": 8.082105263157896e-06,
"loss": 0.1518,
"step": 2325
},
{
"epoch": 0.23,
"learning_rate": 8.05578947368421e-06,
"loss": 0.169,
"step": 2350
},
{
"epoch": 0.24,
"learning_rate": 8.029473684210527e-06,
"loss": 0.1704,
"step": 2375
},
{
"epoch": 0.24,
"learning_rate": 8.003157894736842e-06,
"loss": 0.1622,
"step": 2400
},
{
"epoch": 0.24,
"learning_rate": 7.976842105263159e-06,
"loss": 0.1638,
"step": 2425
},
{
"epoch": 0.24,
"learning_rate": 7.950526315789474e-06,
"loss": 0.1526,
"step": 2450
},
{
"epoch": 0.25,
"learning_rate": 7.92421052631579e-06,
"loss": 0.1792,
"step": 2475
},
{
"epoch": 0.25,
"learning_rate": 7.897894736842106e-06,
"loss": 0.2317,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 7.87157894736842e-06,
"loss": 0.2005,
"step": 2525
},
{
"epoch": 0.26,
"learning_rate": 7.845263157894739e-06,
"loss": 0.2208,
"step": 2550
},
{
"epoch": 0.26,
"learning_rate": 7.818947368421054e-06,
"loss": 0.2185,
"step": 2575
},
{
"epoch": 0.26,
"learning_rate": 7.792631578947369e-06,
"loss": 0.2263,
"step": 2600
},
{
"epoch": 1.0,
"learning_rate": 7.766315789473685e-06,
"loss": 0.1724,
"step": 2625
},
{
"epoch": 1.0,
"learning_rate": 7.74e-06,
"loss": 0.1691,
"step": 2650
},
{
"epoch": 1.01,
"learning_rate": 7.713684210526317e-06,
"loss": 0.1614,
"step": 2675
},
{
"epoch": 1.01,
"learning_rate": 7.687368421052632e-06,
"loss": 0.1423,
"step": 2700
},
{
"epoch": 1.01,
"learning_rate": 7.661052631578949e-06,
"loss": 0.155,
"step": 2725
},
{
"epoch": 1.01,
"learning_rate": 7.634736842105264e-06,
"loss": 0.1632,
"step": 2750
},
{
"epoch": 1.02,
"learning_rate": 7.6084210526315795e-06,
"loss": 0.1843,
"step": 2775
},
{
"epoch": 1.02,
"learning_rate": 7.582105263157895e-06,
"loss": 0.143,
"step": 2800
},
{
"epoch": 1.02,
"learning_rate": 7.555789473684211e-06,
"loss": 0.1746,
"step": 2825
},
{
"epoch": 1.02,
"learning_rate": 7.529473684210526e-06,
"loss": 0.1752,
"step": 2850
},
{
"epoch": 1.03,
"learning_rate": 7.503157894736842e-06,
"loss": 0.1427,
"step": 2875
},
{
"epoch": 1.03,
"learning_rate": 7.476842105263159e-06,
"loss": 0.1229,
"step": 2900
},
{
"epoch": 1.03,
"learning_rate": 7.4505263157894744e-06,
"loss": 0.1433,
"step": 2925
},
{
"epoch": 1.03,
"learning_rate": 7.42421052631579e-06,
"loss": 0.1384,
"step": 2950
},
{
"epoch": 1.04,
"learning_rate": 7.397894736842106e-06,
"loss": 0.1351,
"step": 2975
},
{
"epoch": 1.04,
"learning_rate": 7.371578947368422e-06,
"loss": 0.1418,
"step": 3000
},
{
"epoch": 1.04,
"eval_loss": 0.32313862442970276,
"eval_runtime": 4966.2504,
"eval_samples_per_second": 2.102,
"eval_steps_per_second": 1.051,
"eval_wer": 49.431999999999995,
"step": 3000
},
{
"epoch": 1.04,
"learning_rate": 7.345263157894738e-06,
"loss": 0.1378,
"step": 3025
},
{
"epoch": 1.04,
"learning_rate": 7.3189473684210535e-06,
"loss": 0.1429,
"step": 3050
},
{
"epoch": 1.05,
"learning_rate": 7.292631578947369e-06,
"loss": 0.12,
"step": 3075
},
{
"epoch": 1.05,
"learning_rate": 7.266315789473684e-06,
"loss": 0.1201,
"step": 3100
},
{
"epoch": 1.05,
"learning_rate": 7.24e-06,
"loss": 0.1221,
"step": 3125
},
{
"epoch": 1.05,
"learning_rate": 7.213684210526316e-06,
"loss": 0.12,
"step": 3150
},
{
"epoch": 1.06,
"learning_rate": 7.187368421052632e-06,
"loss": 0.1113,
"step": 3175
},
{
"epoch": 1.06,
"learning_rate": 7.161052631578948e-06,
"loss": 0.1134,
"step": 3200
},
{
"epoch": 1.06,
"learning_rate": 7.134736842105264e-06,
"loss": 0.1163,
"step": 3225
},
{
"epoch": 1.06,
"learning_rate": 7.10842105263158e-06,
"loss": 0.1163,
"step": 3250
},
{
"epoch": 1.07,
"learning_rate": 7.082105263157896e-06,
"loss": 0.1164,
"step": 3275
},
{
"epoch": 1.07,
"learning_rate": 7.055789473684212e-06,
"loss": 0.1184,
"step": 3300
},
{
"epoch": 1.07,
"learning_rate": 7.0294736842105275e-06,
"loss": 0.1352,
"step": 3325
},
{
"epoch": 1.07,
"learning_rate": 7.0031578947368425e-06,
"loss": 0.1365,
"step": 3350
},
{
"epoch": 1.08,
"learning_rate": 6.976842105263158e-06,
"loss": 0.121,
"step": 3375
},
{
"epoch": 1.08,
"learning_rate": 6.950526315789474e-06,
"loss": 0.1314,
"step": 3400
},
{
"epoch": 1.08,
"learning_rate": 6.92421052631579e-06,
"loss": 0.1198,
"step": 3425
},
{
"epoch": 1.08,
"learning_rate": 6.897894736842106e-06,
"loss": 0.1341,
"step": 3450
},
{
"epoch": 1.09,
"learning_rate": 6.871578947368422e-06,
"loss": 0.125,
"step": 3475
},
{
"epoch": 1.09,
"learning_rate": 6.845263157894737e-06,
"loss": 0.1219,
"step": 3500
},
{
"epoch": 1.09,
"learning_rate": 6.820000000000001e-06,
"loss": 0.1205,
"step": 3525
},
{
"epoch": 1.09,
"learning_rate": 6.793684210526317e-06,
"loss": 0.1063,
"step": 3550
},
{
"epoch": 1.1,
"learning_rate": 6.767368421052632e-06,
"loss": 0.1195,
"step": 3575
},
{
"epoch": 1.1,
"learning_rate": 6.741052631578948e-06,
"loss": 0.0936,
"step": 3600
},
{
"epoch": 1.1,
"learning_rate": 6.714736842105264e-06,
"loss": 0.1209,
"step": 3625
},
{
"epoch": 1.1,
"learning_rate": 6.68842105263158e-06,
"loss": 0.1093,
"step": 3650
},
{
"epoch": 1.11,
"learning_rate": 6.662105263157896e-06,
"loss": 0.107,
"step": 3675
},
{
"epoch": 1.11,
"learning_rate": 6.635789473684211e-06,
"loss": 0.0957,
"step": 3700
},
{
"epoch": 1.11,
"learning_rate": 6.6094736842105265e-06,
"loss": 0.1162,
"step": 3725
},
{
"epoch": 1.11,
"learning_rate": 6.583157894736842e-06,
"loss": 0.0953,
"step": 3750
},
{
"epoch": 1.12,
"learning_rate": 6.556842105263158e-06,
"loss": 0.0859,
"step": 3775
},
{
"epoch": 1.12,
"learning_rate": 6.530526315789474e-06,
"loss": 0.0913,
"step": 3800
},
{
"epoch": 1.12,
"learning_rate": 6.50421052631579e-06,
"loss": 0.0851,
"step": 3825
},
{
"epoch": 1.12,
"learning_rate": 6.477894736842106e-06,
"loss": 0.1086,
"step": 3850
},
{
"epoch": 1.13,
"learning_rate": 6.451578947368422e-06,
"loss": 0.0857,
"step": 3875
},
{
"epoch": 1.13,
"learning_rate": 6.425263157894738e-06,
"loss": 0.079,
"step": 3900
},
{
"epoch": 1.13,
"learning_rate": 6.398947368421054e-06,
"loss": 0.1039,
"step": 3925
},
{
"epoch": 1.13,
"learning_rate": 6.372631578947369e-06,
"loss": 0.0977,
"step": 3950
},
{
"epoch": 1.14,
"learning_rate": 6.346315789473685e-06,
"loss": 0.0923,
"step": 3975
},
{
"epoch": 1.14,
"learning_rate": 6.3200000000000005e-06,
"loss": 0.0921,
"step": 4000
},
{
"epoch": 1.14,
"eval_loss": 0.3284112215042114,
"eval_runtime": 5147.1835,
"eval_samples_per_second": 2.028,
"eval_steps_per_second": 1.014,
"eval_wer": 56.11066666666667,
"step": 4000
},
{
"epoch": 1.14,
"learning_rate": 6.293684210526316e-06,
"loss": 0.0948,
"step": 4025
},
{
"epoch": 1.14,
"learning_rate": 6.267368421052632e-06,
"loss": 0.0805,
"step": 4050
},
{
"epoch": 1.15,
"learning_rate": 6.241052631578948e-06,
"loss": 0.081,
"step": 4075
},
{
"epoch": 1.15,
"learning_rate": 6.214736842105264e-06,
"loss": 0.0883,
"step": 4100
},
{
"epoch": 1.15,
"learning_rate": 6.188421052631579e-06,
"loss": 0.089,
"step": 4125
},
{
"epoch": 1.15,
"learning_rate": 6.1621052631578945e-06,
"loss": 0.0731,
"step": 4150
},
{
"epoch": 1.16,
"learning_rate": 6.13578947368421e-06,
"loss": 0.0937,
"step": 4175
},
{
"epoch": 1.16,
"learning_rate": 6.109473684210527e-06,
"loss": 0.0927,
"step": 4200
},
{
"epoch": 1.16,
"learning_rate": 6.083157894736843e-06,
"loss": 0.0892,
"step": 4225
},
{
"epoch": 1.16,
"learning_rate": 6.056842105263159e-06,
"loss": 0.091,
"step": 4250
},
{
"epoch": 1.17,
"learning_rate": 6.0305263157894745e-06,
"loss": 0.1257,
"step": 4275
},
{
"epoch": 1.17,
"learning_rate": 6.00421052631579e-06,
"loss": 0.1128,
"step": 4300
},
{
"epoch": 1.17,
"learning_rate": 5.977894736842106e-06,
"loss": 0.1248,
"step": 4325
},
{
"epoch": 1.17,
"learning_rate": 5.951578947368422e-06,
"loss": 0.1104,
"step": 4350
},
{
"epoch": 1.18,
"learning_rate": 5.925263157894737e-06,
"loss": 0.1141,
"step": 4375
},
{
"epoch": 1.18,
"learning_rate": 5.898947368421053e-06,
"loss": 0.1129,
"step": 4400
},
{
"epoch": 1.18,
"learning_rate": 5.8726315789473685e-06,
"loss": 0.113,
"step": 4425
},
{
"epoch": 1.18,
"learning_rate": 5.846315789473684e-06,
"loss": 0.1012,
"step": 4450
},
{
"epoch": 1.19,
"learning_rate": 5.82e-06,
"loss": 0.0889,
"step": 4475
},
{
"epoch": 1.19,
"learning_rate": 5.793684210526316e-06,
"loss": 0.1044,
"step": 4500
},
{
"epoch": 1.19,
"learning_rate": 5.767368421052632e-06,
"loss": 0.0869,
"step": 4525
},
{
"epoch": 1.19,
"learning_rate": 5.7410526315789485e-06,
"loss": 0.0924,
"step": 4550
},
{
"epoch": 1.2,
"learning_rate": 5.714736842105264e-06,
"loss": 0.0807,
"step": 4575
},
{
"epoch": 1.2,
"learning_rate": 5.68842105263158e-06,
"loss": 0.0853,
"step": 4600
},
{
"epoch": 1.2,
"learning_rate": 5.662105263157895e-06,
"loss": 0.0885,
"step": 4625
},
{
"epoch": 1.2,
"learning_rate": 5.635789473684211e-06,
"loss": 0.1236,
"step": 4650
},
{
"epoch": 1.21,
"learning_rate": 5.609473684210527e-06,
"loss": 0.1326,
"step": 4675
},
{
"epoch": 1.21,
"learning_rate": 5.5831578947368425e-06,
"loss": 0.1103,
"step": 4700
},
{
"epoch": 1.21,
"learning_rate": 5.556842105263158e-06,
"loss": 0.0857,
"step": 4725
},
{
"epoch": 1.21,
"learning_rate": 5.530526315789474e-06,
"loss": 0.117,
"step": 4750
},
{
"epoch": 1.22,
"learning_rate": 5.50421052631579e-06,
"loss": 0.0878,
"step": 4775
},
{
"epoch": 1.22,
"learning_rate": 5.477894736842105e-06,
"loss": 0.0963,
"step": 4800
},
{
"epoch": 1.22,
"learning_rate": 5.451578947368421e-06,
"loss": 0.066,
"step": 4825
},
{
"epoch": 1.22,
"learning_rate": 5.425263157894737e-06,
"loss": 0.0736,
"step": 4850
},
{
"epoch": 1.23,
"learning_rate": 5.398947368421053e-06,
"loss": 0.0752,
"step": 4875
},
{
"epoch": 1.23,
"learning_rate": 5.372631578947369e-06,
"loss": 0.0615,
"step": 4900
},
{
"epoch": 1.23,
"learning_rate": 5.346315789473685e-06,
"loss": 0.0635,
"step": 4925
},
{
"epoch": 1.23,
"learning_rate": 5.320000000000001e-06,
"loss": 0.0757,
"step": 4950
},
{
"epoch": 1.24,
"learning_rate": 5.2936842105263166e-06,
"loss": 0.0728,
"step": 4975
},
{
"epoch": 1.24,
"learning_rate": 5.267368421052632e-06,
"loss": 0.0923,
"step": 5000
},
{
"epoch": 1.24,
"eval_loss": 0.3304431736469269,
"eval_runtime": 5163.7943,
"eval_samples_per_second": 2.022,
"eval_steps_per_second": 1.011,
"eval_wer": 61.42266666666667,
"step": 5000
},
{
"epoch": 1.24,
"learning_rate": 5.241052631578948e-06,
"loss": 0.0593,
"step": 5025
},
{
"epoch": 1.24,
"learning_rate": 5.214736842105263e-06,
"loss": 0.0745,
"step": 5050
},
{
"epoch": 1.25,
"learning_rate": 5.188421052631579e-06,
"loss": 0.075,
"step": 5075
},
{
"epoch": 1.25,
"learning_rate": 5.162105263157895e-06,
"loss": 0.0956,
"step": 5100
},
{
"epoch": 1.25,
"learning_rate": 5.135789473684211e-06,
"loss": 0.0974,
"step": 5125
},
{
"epoch": 1.25,
"learning_rate": 5.1094736842105264e-06,
"loss": 0.0869,
"step": 5150
},
{
"epoch": 1.26,
"learning_rate": 5.083157894736842e-06,
"loss": 0.0931,
"step": 5175
},
{
"epoch": 1.26,
"learning_rate": 5.056842105263158e-06,
"loss": 0.0994,
"step": 5200
},
{
"epoch": 2.0,
"learning_rate": 5.030526315789475e-06,
"loss": 0.086,
"step": 5225
},
{
"epoch": 2.0,
"learning_rate": 5.0042105263157906e-06,
"loss": 0.0694,
"step": 5250
},
{
"epoch": 2.01,
"learning_rate": 4.9778947368421055e-06,
"loss": 0.073,
"step": 5275
},
{
"epoch": 2.01,
"learning_rate": 4.951578947368421e-06,
"loss": 0.0718,
"step": 5300
},
{
"epoch": 2.01,
"learning_rate": 4.925263157894737e-06,
"loss": 0.0683,
"step": 5325
},
{
"epoch": 2.01,
"learning_rate": 4.898947368421053e-06,
"loss": 0.076,
"step": 5350
},
{
"epoch": 2.02,
"learning_rate": 4.872631578947369e-06,
"loss": 0.0693,
"step": 5375
},
{
"epoch": 2.02,
"learning_rate": 4.846315789473685e-06,
"loss": 0.0748,
"step": 5400
},
{
"epoch": 2.02,
"learning_rate": 4.8200000000000004e-06,
"loss": 0.085,
"step": 5425
},
{
"epoch": 2.02,
"learning_rate": 4.793684210526316e-06,
"loss": 0.0693,
"step": 5450
},
{
"epoch": 2.03,
"learning_rate": 4.767368421052632e-06,
"loss": 0.0623,
"step": 5475
},
{
"epoch": 2.03,
"learning_rate": 4.741052631578948e-06,
"loss": 0.0584,
"step": 5500
},
{
"epoch": 2.03,
"learning_rate": 4.714736842105264e-06,
"loss": 0.0531,
"step": 5525
},
{
"epoch": 2.03,
"learning_rate": 4.6884210526315795e-06,
"loss": 0.067,
"step": 5550
},
{
"epoch": 2.04,
"learning_rate": 4.662105263157895e-06,
"loss": 0.0584,
"step": 5575
},
{
"epoch": 2.04,
"learning_rate": 4.63578947368421e-06,
"loss": 0.0688,
"step": 5600
},
{
"epoch": 2.04,
"learning_rate": 4.609473684210526e-06,
"loss": 0.0535,
"step": 5625
},
{
"epoch": 2.04,
"learning_rate": 4.583157894736843e-06,
"loss": 0.0529,
"step": 5650
},
{
"epoch": 2.05,
"learning_rate": 4.556842105263159e-06,
"loss": 0.0564,
"step": 5675
},
{
"epoch": 2.05,
"learning_rate": 4.5305263157894744e-06,
"loss": 0.0607,
"step": 5700
},
{
"epoch": 2.05,
"learning_rate": 4.504210526315789e-06,
"loss": 0.0521,
"step": 5725
},
{
"epoch": 2.05,
"learning_rate": 4.477894736842105e-06,
"loss": 0.0565,
"step": 5750
},
{
"epoch": 2.06,
"learning_rate": 4.451578947368421e-06,
"loss": 0.0567,
"step": 5775
},
{
"epoch": 2.06,
"learning_rate": 4.425263157894737e-06,
"loss": 0.0597,
"step": 5800
},
{
"epoch": 2.06,
"learning_rate": 4.3989473684210535e-06,
"loss": 0.0384,
"step": 5825
},
{
"epoch": 2.06,
"learning_rate": 4.3726315789473685e-06,
"loss": 0.048,
"step": 5850
},
{
"epoch": 2.07,
"learning_rate": 4.346315789473684e-06,
"loss": 0.0554,
"step": 5875
},
{
"epoch": 2.07,
"learning_rate": 4.32e-06,
"loss": 0.048,
"step": 5900
},
{
"epoch": 2.07,
"learning_rate": 4.293684210526316e-06,
"loss": 0.0546,
"step": 5925
},
{
"epoch": 2.07,
"learning_rate": 4.267368421052632e-06,
"loss": 0.0489,
"step": 5950
},
{
"epoch": 2.08,
"learning_rate": 4.241052631578948e-06,
"loss": 0.0518,
"step": 5975
},
{
"epoch": 2.08,
"learning_rate": 4.214736842105263e-06,
"loss": 0.0483,
"step": 6000
},
{
"epoch": 2.08,
"eval_loss": 0.3459864556789398,
"eval_runtime": 5173.4895,
"eval_samples_per_second": 2.018,
"eval_steps_per_second": 1.009,
"eval_wer": 55.952,
"step": 6000
},
{
"epoch": 2.08,
"learning_rate": 4.188421052631579e-06,
"loss": 0.0552,
"step": 6025
},
{
"epoch": 2.08,
"learning_rate": 4.162105263157895e-06,
"loss": 0.0538,
"step": 6050
},
{
"epoch": 2.09,
"learning_rate": 4.135789473684211e-06,
"loss": 0.0567,
"step": 6075
},
{
"epoch": 2.09,
"learning_rate": 4.109473684210527e-06,
"loss": 0.0424,
"step": 6100
},
{
"epoch": 2.09,
"learning_rate": 4.0831578947368425e-06,
"loss": 0.0541,
"step": 6125
},
{
"epoch": 2.09,
"learning_rate": 4.056842105263158e-06,
"loss": 0.0502,
"step": 6150
},
{
"epoch": 2.1,
"learning_rate": 4.030526315789474e-06,
"loss": 0.0472,
"step": 6175
},
{
"epoch": 2.1,
"learning_rate": 4.00421052631579e-06,
"loss": 0.0457,
"step": 6200
},
{
"epoch": 2.1,
"learning_rate": 3.977894736842106e-06,
"loss": 0.04,
"step": 6225
},
{
"epoch": 2.1,
"learning_rate": 3.951578947368422e-06,
"loss": 0.0454,
"step": 6250
},
{
"epoch": 2.11,
"learning_rate": 3.9252631578947366e-06,
"loss": 0.0429,
"step": 6275
},
{
"epoch": 2.11,
"learning_rate": 3.898947368421052e-06,
"loss": 0.0509,
"step": 6300
},
{
"epoch": 2.11,
"learning_rate": 3.872631578947369e-06,
"loss": 0.0428,
"step": 6325
},
{
"epoch": 2.11,
"learning_rate": 3.846315789473685e-06,
"loss": 0.0304,
"step": 6350
},
{
"epoch": 2.12,
"learning_rate": 3.820000000000001e-06,
"loss": 0.0482,
"step": 6375
},
{
"epoch": 2.12,
"learning_rate": 3.793684210526316e-06,
"loss": 0.0385,
"step": 6400
},
{
"epoch": 2.12,
"learning_rate": 3.767368421052632e-06,
"loss": 0.0339,
"step": 6425
},
{
"epoch": 2.12,
"learning_rate": 3.7410526315789473e-06,
"loss": 0.0449,
"step": 6450
},
{
"epoch": 2.13,
"learning_rate": 3.714736842105263e-06,
"loss": 0.0345,
"step": 6475
},
{
"epoch": 2.13,
"learning_rate": 3.6884210526315794e-06,
"loss": 0.0469,
"step": 6500
},
{
"epoch": 2.13,
"learning_rate": 3.662105263157895e-06,
"loss": 0.0401,
"step": 6525
},
{
"epoch": 2.13,
"learning_rate": 3.635789473684211e-06,
"loss": 0.0444,
"step": 6550
},
{
"epoch": 2.14,
"learning_rate": 3.6094736842105264e-06,
"loss": 0.0366,
"step": 6575
},
{
"epoch": 2.14,
"learning_rate": 3.5831578947368422e-06,
"loss": 0.0284,
"step": 6600
},
{
"epoch": 2.14,
"learning_rate": 3.556842105263158e-06,
"loss": 0.0315,
"step": 6625
},
{
"epoch": 2.14,
"learning_rate": 3.5305263157894743e-06,
"loss": 0.0335,
"step": 6650
},
{
"epoch": 2.15,
"learning_rate": 3.50421052631579e-06,
"loss": 0.0344,
"step": 6675
},
{
"epoch": 2.15,
"learning_rate": 3.4778947368421055e-06,
"loss": 0.0313,
"step": 6700
},
{
"epoch": 2.15,
"learning_rate": 3.4515789473684213e-06,
"loss": 0.0269,
"step": 6725
},
{
"epoch": 2.15,
"learning_rate": 3.425263157894737e-06,
"loss": 0.0394,
"step": 6750
},
{
"epoch": 2.16,
"learning_rate": 3.398947368421053e-06,
"loss": 0.03,
"step": 6775
},
{
"epoch": 2.16,
"learning_rate": 3.3726315789473683e-06,
"loss": 0.0292,
"step": 6800
},
{
"epoch": 2.16,
"learning_rate": 3.3463157894736846e-06,
"loss": 0.0359,
"step": 6825
},
{
"epoch": 2.16,
"learning_rate": 3.3200000000000004e-06,
"loss": 0.0402,
"step": 6850
},
{
"epoch": 2.17,
"learning_rate": 3.2936842105263162e-06,
"loss": 0.0516,
"step": 6875
},
{
"epoch": 2.17,
"learning_rate": 3.267368421052632e-06,
"loss": 0.0524,
"step": 6900
},
{
"epoch": 2.17,
"learning_rate": 3.2410526315789474e-06,
"loss": 0.0503,
"step": 6925
},
{
"epoch": 2.17,
"learning_rate": 3.2147368421052633e-06,
"loss": 0.0538,
"step": 6950
},
{
"epoch": 2.18,
"learning_rate": 3.188421052631579e-06,
"loss": 0.042,
"step": 6975
},
{
"epoch": 2.18,
"learning_rate": 3.1621052631578953e-06,
"loss": 0.0391,
"step": 7000
},
{
"epoch": 2.18,
"eval_loss": 0.3538491427898407,
"eval_runtime": 5142.0777,
"eval_samples_per_second": 2.03,
"eval_steps_per_second": 1.015,
"eval_wer": 51.10666666666667,
"step": 7000
},
{
"epoch": 2.18,
"learning_rate": 3.135789473684211e-06,
"loss": 0.0475,
"step": 7025
},
{
"epoch": 2.18,
"learning_rate": 3.1094736842105265e-06,
"loss": 0.0385,
"step": 7050
},
{
"epoch": 2.19,
"learning_rate": 3.0831578947368423e-06,
"loss": 0.0469,
"step": 7075
},
{
"epoch": 2.19,
"learning_rate": 3.056842105263158e-06,
"loss": 0.0337,
"step": 7100
},
{
"epoch": 2.19,
"learning_rate": 3.0305263157894736e-06,
"loss": 0.0457,
"step": 7125
},
{
"epoch": 2.19,
"learning_rate": 3.0042105263157894e-06,
"loss": 0.0414,
"step": 7150
},
{
"epoch": 2.2,
"learning_rate": 2.9778947368421056e-06,
"loss": 0.0334,
"step": 7175
},
{
"epoch": 2.2,
"learning_rate": 2.9515789473684214e-06,
"loss": 0.0347,
"step": 7200
},
{
"epoch": 2.2,
"learning_rate": 2.9252631578947373e-06,
"loss": 0.0395,
"step": 7225
},
{
"epoch": 2.2,
"learning_rate": 2.8989473684210526e-06,
"loss": 0.0399,
"step": 7250
},
{
"epoch": 2.21,
"learning_rate": 2.8726315789473685e-06,
"loss": 0.0521,
"step": 7275
},
{
"epoch": 2.21,
"learning_rate": 2.8463157894736843e-06,
"loss": 0.0407,
"step": 7300
},
{
"epoch": 2.21,
"learning_rate": 2.82e-06,
"loss": 0.04,
"step": 7325
},
{
"epoch": 2.21,
"learning_rate": 2.7936842105263163e-06,
"loss": 0.0441,
"step": 7350
},
{
"epoch": 2.22,
"learning_rate": 2.7673684210526317e-06,
"loss": 0.0378,
"step": 7375
},
{
"epoch": 2.22,
"learning_rate": 2.7410526315789476e-06,
"loss": 0.0448,
"step": 7400
},
{
"epoch": 2.22,
"learning_rate": 2.7147368421052634e-06,
"loss": 0.0301,
"step": 7425
},
{
"epoch": 2.22,
"learning_rate": 2.688421052631579e-06,
"loss": 0.0297,
"step": 7450
},
{
"epoch": 2.23,
"learning_rate": 2.6621052631578946e-06,
"loss": 0.0381,
"step": 7475
},
{
"epoch": 2.23,
"learning_rate": 2.635789473684211e-06,
"loss": 0.023,
"step": 7500
},
{
"epoch": 2.23,
"learning_rate": 2.6094736842105267e-06,
"loss": 0.0269,
"step": 7525
},
{
"epoch": 2.23,
"learning_rate": 2.5831578947368425e-06,
"loss": 0.0263,
"step": 7550
},
{
"epoch": 2.24,
"learning_rate": 2.5568421052631583e-06,
"loss": 0.0275,
"step": 7575
},
{
"epoch": 2.24,
"learning_rate": 2.5305263157894737e-06,
"loss": 0.0312,
"step": 7600
},
{
"epoch": 2.24,
"learning_rate": 2.5042105263157895e-06,
"loss": 0.0303,
"step": 7625
},
{
"epoch": 2.24,
"learning_rate": 2.4778947368421053e-06,
"loss": 0.0339,
"step": 7650
},
{
"epoch": 2.25,
"learning_rate": 2.451578947368421e-06,
"loss": 0.0259,
"step": 7675
},
{
"epoch": 2.25,
"learning_rate": 2.425263157894737e-06,
"loss": 0.027,
"step": 7700
},
{
"epoch": 2.25,
"learning_rate": 2.3989473684210528e-06,
"loss": 0.0411,
"step": 7725
},
{
"epoch": 2.25,
"learning_rate": 2.3726315789473686e-06,
"loss": 0.037,
"step": 7750
},
{
"epoch": 2.26,
"learning_rate": 2.3463157894736844e-06,
"loss": 0.0443,
"step": 7775
},
{
"epoch": 2.26,
"learning_rate": 2.3200000000000002e-06,
"loss": 0.0384,
"step": 7800
},
{
"epoch": 2.26,
"learning_rate": 2.293684210526316e-06,
"loss": 0.039,
"step": 7825
},
{
"epoch": 3.0,
"learning_rate": 2.267368421052632e-06,
"loss": 0.0282,
"step": 7850
},
{
"epoch": 3.0,
"learning_rate": 2.2410526315789473e-06,
"loss": 0.0267,
"step": 7875
},
{
"epoch": 3.01,
"learning_rate": 2.2147368421052635e-06,
"loss": 0.0261,
"step": 7900
},
{
"epoch": 3.01,
"learning_rate": 2.188421052631579e-06,
"loss": 0.0255,
"step": 7925
},
{
"epoch": 3.01,
"learning_rate": 2.1621052631578947e-06,
"loss": 0.0333,
"step": 7950
},
{
"epoch": 3.01,
"learning_rate": 2.135789473684211e-06,
"loss": 0.0367,
"step": 7975
},
{
"epoch": 3.02,
"learning_rate": 2.1094736842105264e-06,
"loss": 0.0228,
"step": 8000
},
{
"epoch": 3.02,
"eval_loss": 0.34926050901412964,
"eval_runtime": 5016.1265,
"eval_samples_per_second": 2.081,
"eval_steps_per_second": 1.041,
"eval_wer": 51.82,
"step": 8000
},
{
"epoch": 3.02,
"learning_rate": 2.083157894736842e-06,
"loss": 0.0249,
"step": 8025
},
{
"epoch": 3.02,
"learning_rate": 2.056842105263158e-06,
"loss": 0.0264,
"step": 8050
},
{
"epoch": 3.02,
"learning_rate": 2.030526315789474e-06,
"loss": 0.02,
"step": 8075
},
{
"epoch": 3.03,
"learning_rate": 2.0042105263157896e-06,
"loss": 0.023,
"step": 8100
},
{
"epoch": 3.03,
"learning_rate": 1.9778947368421055e-06,
"loss": 0.0257,
"step": 8125
},
{
"epoch": 3.03,
"learning_rate": 1.9515789473684213e-06,
"loss": 0.0206,
"step": 8150
},
{
"epoch": 3.03,
"learning_rate": 1.925263157894737e-06,
"loss": 0.0188,
"step": 8175
},
{
"epoch": 3.04,
"learning_rate": 1.8989473684210527e-06,
"loss": 0.0208,
"step": 8200
},
{
"epoch": 3.04,
"learning_rate": 1.8726315789473687e-06,
"loss": 0.0138,
"step": 8225
},
{
"epoch": 3.04,
"learning_rate": 1.8463157894736843e-06,
"loss": 0.014,
"step": 8250
},
{
"epoch": 3.04,
"learning_rate": 1.8200000000000002e-06,
"loss": 0.0163,
"step": 8275
},
{
"epoch": 3.05,
"learning_rate": 1.7936842105263158e-06,
"loss": 0.0182,
"step": 8300
},
{
"epoch": 3.05,
"learning_rate": 1.7673684210526318e-06,
"loss": 0.0139,
"step": 8325
},
{
"epoch": 3.05,
"learning_rate": 1.7410526315789474e-06,
"loss": 0.022,
"step": 8350
},
{
"epoch": 3.05,
"learning_rate": 1.7147368421052632e-06,
"loss": 0.0228,
"step": 8375
},
{
"epoch": 3.06,
"learning_rate": 1.6884210526315792e-06,
"loss": 0.0244,
"step": 8400
},
{
"epoch": 3.06,
"learning_rate": 1.6621052631578948e-06,
"loss": 0.0187,
"step": 8425
},
{
"epoch": 3.06,
"learning_rate": 1.6357894736842107e-06,
"loss": 0.0178,
"step": 8450
},
{
"epoch": 3.06,
"learning_rate": 1.6094736842105265e-06,
"loss": 0.014,
"step": 8475
},
{
"epoch": 3.07,
"learning_rate": 1.5831578947368423e-06,
"loss": 0.0195,
"step": 8500
},
{
"epoch": 3.07,
"learning_rate": 1.556842105263158e-06,
"loss": 0.0191,
"step": 8525
},
{
"epoch": 3.07,
"learning_rate": 1.5305263157894737e-06,
"loss": 0.0209,
"step": 8550
},
{
"epoch": 3.07,
"learning_rate": 1.5042105263157898e-06,
"loss": 0.0252,
"step": 8575
},
{
"epoch": 3.08,
"learning_rate": 1.4778947368421054e-06,
"loss": 0.0179,
"step": 8600
},
{
"epoch": 3.08,
"learning_rate": 1.4515789473684212e-06,
"loss": 0.0141,
"step": 8625
},
{
"epoch": 3.08,
"learning_rate": 1.425263157894737e-06,
"loss": 0.0157,
"step": 8650
},
{
"epoch": 3.08,
"learning_rate": 1.3989473684210528e-06,
"loss": 0.0183,
"step": 8675
},
{
"epoch": 3.09,
"learning_rate": 1.3726315789473684e-06,
"loss": 0.014,
"step": 8700
},
{
"epoch": 3.09,
"learning_rate": 1.3463157894736842e-06,
"loss": 0.0177,
"step": 8725
},
{
"epoch": 3.09,
"learning_rate": 1.32e-06,
"loss": 0.0183,
"step": 8750
},
{
"epoch": 3.09,
"learning_rate": 1.2936842105263159e-06,
"loss": 0.0162,
"step": 8775
},
{
"epoch": 3.1,
"learning_rate": 1.2673684210526315e-06,
"loss": 0.0124,
"step": 8800
},
{
"epoch": 3.1,
"learning_rate": 1.2410526315789475e-06,
"loss": 0.0204,
"step": 8825
},
{
"epoch": 3.1,
"learning_rate": 1.2147368421052633e-06,
"loss": 0.0133,
"step": 8850
},
{
"epoch": 3.1,
"learning_rate": 1.188421052631579e-06,
"loss": 0.0179,
"step": 8875
},
{
"epoch": 3.11,
"learning_rate": 1.1621052631578948e-06,
"loss": 0.015,
"step": 8900
},
{
"epoch": 3.11,
"learning_rate": 1.1357894736842106e-06,
"loss": 0.0132,
"step": 8925
},
{
"epoch": 3.11,
"learning_rate": 1.1094736842105264e-06,
"loss": 0.0074,
"step": 8950
},
{
"epoch": 3.11,
"learning_rate": 1.0831578947368422e-06,
"loss": 0.0113,
"step": 8975
},
{
"epoch": 3.12,
"learning_rate": 1.0568421052631578e-06,
"loss": 0.0206,
"step": 9000
},
{
"epoch": 3.12,
"eval_loss": 0.3729253113269806,
"eval_runtime": 4842.0992,
"eval_samples_per_second": 2.156,
"eval_steps_per_second": 1.078,
"eval_wer": 52.400000000000006,
"step": 9000
},
{
"epoch": 3.12,
"learning_rate": 1.0305263157894739e-06,
"loss": 0.0195,
"step": 9025
},
{
"epoch": 3.12,
"learning_rate": 1.0042105263157897e-06,
"loss": 0.0138,
"step": 9050
},
{
"epoch": 3.12,
"learning_rate": 9.778947368421053e-07,
"loss": 0.0119,
"step": 9075
},
{
"epoch": 3.13,
"learning_rate": 9.515789473684212e-07,
"loss": 0.0146,
"step": 9100
},
{
"epoch": 3.13,
"learning_rate": 9.252631578947368e-07,
"loss": 0.0148,
"step": 9125
},
{
"epoch": 3.13,
"learning_rate": 8.989473684210527e-07,
"loss": 0.0129,
"step": 9150
},
{
"epoch": 3.13,
"learning_rate": 8.726315789473686e-07,
"loss": 0.0092,
"step": 9175
},
{
"epoch": 3.14,
"learning_rate": 8.463157894736843e-07,
"loss": 0.0097,
"step": 9200
},
{
"epoch": 3.14,
"learning_rate": 8.200000000000001e-07,
"loss": 0.0147,
"step": 9225
},
{
"epoch": 3.14,
"learning_rate": 7.936842105263158e-07,
"loss": 0.0074,
"step": 9250
},
{
"epoch": 3.14,
"learning_rate": 7.673684210526316e-07,
"loss": 0.0077,
"step": 9275
},
{
"epoch": 3.15,
"learning_rate": 7.410526315789475e-07,
"loss": 0.0093,
"step": 9300
},
{
"epoch": 3.15,
"learning_rate": 7.147368421052632e-07,
"loss": 0.0078,
"step": 9325
},
{
"epoch": 3.15,
"learning_rate": 6.884210526315791e-07,
"loss": 0.0148,
"step": 9350
},
{
"epoch": 3.15,
"learning_rate": 6.621052631578948e-07,
"loss": 0.0071,
"step": 9375
},
{
"epoch": 3.16,
"learning_rate": 6.357894736842106e-07,
"loss": 0.007,
"step": 9400
},
{
"epoch": 3.16,
"learning_rate": 6.094736842105263e-07,
"loss": 0.0154,
"step": 9425
},
{
"epoch": 3.16,
"learning_rate": 5.831578947368421e-07,
"loss": 0.0137,
"step": 9450
},
{
"epoch": 3.16,
"learning_rate": 5.56842105263158e-07,
"loss": 0.0119,
"step": 9475
},
{
"epoch": 3.17,
"learning_rate": 5.305263157894737e-07,
"loss": 0.0164,
"step": 9500
},
{
"epoch": 3.17,
"learning_rate": 5.042105263157895e-07,
"loss": 0.0157,
"step": 9525
},
{
"epoch": 3.17,
"learning_rate": 4.778947368421053e-07,
"loss": 0.0161,
"step": 9550
},
{
"epoch": 3.17,
"learning_rate": 4.5157894736842107e-07,
"loss": 0.0159,
"step": 9575
},
{
"epoch": 3.18,
"learning_rate": 4.2526315789473684e-07,
"loss": 0.0118,
"step": 9600
},
{
"epoch": 3.18,
"learning_rate": 3.9894736842105266e-07,
"loss": 0.0114,
"step": 9625
},
{
"epoch": 3.18,
"learning_rate": 3.726315789473685e-07,
"loss": 0.0118,
"step": 9650
},
{
"epoch": 3.18,
"learning_rate": 3.4631578947368424e-07,
"loss": 0.0163,
"step": 9675
},
{
"epoch": 3.19,
"learning_rate": 3.2e-07,
"loss": 0.0151,
"step": 9700
},
{
"epoch": 3.19,
"learning_rate": 2.936842105263158e-07,
"loss": 0.007,
"step": 9725
},
{
"epoch": 3.19,
"learning_rate": 2.6736842105263164e-07,
"loss": 0.0099,
"step": 9750
},
{
"epoch": 3.19,
"learning_rate": 2.410526315789474e-07,
"loss": 0.0119,
"step": 9775
},
{
"epoch": 3.2,
"learning_rate": 2.1473684210526317e-07,
"loss": 0.0133,
"step": 9800
},
{
"epoch": 3.2,
"learning_rate": 1.8842105263157897e-07,
"loss": 0.0178,
"step": 9825
},
{
"epoch": 3.2,
"learning_rate": 1.6210526315789476e-07,
"loss": 0.0156,
"step": 9850
},
{
"epoch": 3.2,
"learning_rate": 1.3578947368421055e-07,
"loss": 0.0173,
"step": 9875
},
{
"epoch": 3.21,
"learning_rate": 1.0947368421052632e-07,
"loss": 0.0155,
"step": 9900
},
{
"epoch": 3.21,
"learning_rate": 8.315789473684211e-08,
"loss": 0.0178,
"step": 9925
},
{
"epoch": 3.21,
"learning_rate": 5.68421052631579e-08,
"loss": 0.0142,
"step": 9950
},
{
"epoch": 3.21,
"learning_rate": 3.0526315789473686e-08,
"loss": 0.0207,
"step": 9975
},
{
"epoch": 3.22,
"learning_rate": 4.210526315789474e-09,
"loss": 0.018,
"step": 10000
},
{
"epoch": 3.22,
"eval_loss": 0.36761578917503357,
"eval_runtime": 4887.3963,
"eval_samples_per_second": 2.136,
"eval_steps_per_second": 1.068,
"eval_wer": 51.296,
"step": 10000
},
{
"epoch": 3.22,
"step": 10000,
"total_flos": 1.698282036117504e+20,
"train_loss": 0.10820973655730486,
"train_runtime": 82952.5549,
"train_samples_per_second": 0.964,
"train_steps_per_second": 0.121
}
],
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.698282036117504e+20,
"trial_name": null,
"trial_params": null
}