whisper-large-pt-cv11 / trainer_state.json
jonatasgrosman's picture
End of training
7a725b0
raw
history blame
20.7 kB
{
"best_metric": 5.192249470178625,
"best_model_checkpoint": "data/model/whisper-large-pt-cv11/checkpoint-1000",
"epoch": 4.9437924644842495,
"global_step": 4000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 5.7500000000000005e-08,
"loss": 1.0816,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 1.1500000000000001e-07,
"loss": 1.0477,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 1.775e-07,
"loss": 0.9898,
"step": 75
},
{
"epoch": 0.12,
"learning_rate": 2.4000000000000003e-07,
"loss": 0.8425,
"step": 100
},
{
"epoch": 0.15,
"learning_rate": 3.025e-07,
"loss": 0.5596,
"step": 125
},
{
"epoch": 0.19,
"learning_rate": 3.65e-07,
"loss": 0.3455,
"step": 150
},
{
"epoch": 0.22,
"learning_rate": 4.275000000000001e-07,
"loss": 0.3087,
"step": 175
},
{
"epoch": 0.25,
"learning_rate": 4.900000000000001e-07,
"loss": 0.2832,
"step": 200
},
{
"epoch": 0.28,
"learning_rate": 5.525e-07,
"loss": 0.2673,
"step": 225
},
{
"epoch": 0.31,
"learning_rate": 6.15e-07,
"loss": 0.2447,
"step": 250
},
{
"epoch": 0.34,
"learning_rate": 6.775000000000001e-07,
"loss": 0.2114,
"step": 275
},
{
"epoch": 0.37,
"learning_rate": 7.4e-07,
"loss": 0.2151,
"step": 300
},
{
"epoch": 0.4,
"learning_rate": 8.025e-07,
"loss": 0.1882,
"step": 325
},
{
"epoch": 0.43,
"learning_rate": 8.65e-07,
"loss": 0.1968,
"step": 350
},
{
"epoch": 0.46,
"learning_rate": 9.275000000000001e-07,
"loss": 0.1756,
"step": 375
},
{
"epoch": 0.49,
"learning_rate": 9.9e-07,
"loss": 0.1739,
"step": 400
},
{
"epoch": 0.53,
"learning_rate": 1.0525e-06,
"loss": 0.1861,
"step": 425
},
{
"epoch": 0.56,
"learning_rate": 1.1150000000000001e-06,
"loss": 0.1675,
"step": 450
},
{
"epoch": 0.59,
"learning_rate": 1.1775e-06,
"loss": 0.1706,
"step": 475
},
{
"epoch": 0.62,
"learning_rate": 1.2400000000000002e-06,
"loss": 0.1555,
"step": 500
},
{
"epoch": 0.65,
"learning_rate": 1.3025000000000002e-06,
"loss": 0.1488,
"step": 525
},
{
"epoch": 0.68,
"learning_rate": 1.3650000000000003e-06,
"loss": 0.1534,
"step": 550
},
{
"epoch": 0.71,
"learning_rate": 1.4275e-06,
"loss": 0.1366,
"step": 575
},
{
"epoch": 0.74,
"learning_rate": 1.4900000000000001e-06,
"loss": 0.1398,
"step": 600
},
{
"epoch": 0.77,
"learning_rate": 1.5525000000000002e-06,
"loss": 0.1494,
"step": 625
},
{
"epoch": 0.8,
"learning_rate": 1.6150000000000002e-06,
"loss": 0.1487,
"step": 650
},
{
"epoch": 0.83,
"learning_rate": 1.6775000000000002e-06,
"loss": 0.1461,
"step": 675
},
{
"epoch": 0.86,
"learning_rate": 1.74e-06,
"loss": 0.146,
"step": 700
},
{
"epoch": 0.9,
"learning_rate": 1.8025000000000001e-06,
"loss": 0.1425,
"step": 725
},
{
"epoch": 0.93,
"learning_rate": 1.8650000000000001e-06,
"loss": 0.1532,
"step": 750
},
{
"epoch": 0.96,
"learning_rate": 1.9275e-06,
"loss": 0.1332,
"step": 775
},
{
"epoch": 0.99,
"learning_rate": 1.9900000000000004e-06,
"loss": 0.1346,
"step": 800
},
{
"epoch": 1.02,
"learning_rate": 2.0525000000000003e-06,
"loss": 0.1127,
"step": 825
},
{
"epoch": 1.05,
"learning_rate": 2.115e-06,
"loss": 0.1124,
"step": 850
},
{
"epoch": 1.08,
"learning_rate": 2.1775000000000003e-06,
"loss": 0.1121,
"step": 875
},
{
"epoch": 1.11,
"learning_rate": 2.24e-06,
"loss": 0.1035,
"step": 900
},
{
"epoch": 1.14,
"learning_rate": 2.3025000000000004e-06,
"loss": 0.1104,
"step": 925
},
{
"epoch": 1.17,
"learning_rate": 2.3650000000000002e-06,
"loss": 0.1102,
"step": 950
},
{
"epoch": 1.21,
"learning_rate": 2.4275e-06,
"loss": 0.1109,
"step": 975
},
{
"epoch": 1.24,
"learning_rate": 2.4900000000000003e-06,
"loss": 0.1051,
"step": 1000
},
{
"epoch": 1.24,
"eval_cer": 1.5979199956666574,
"eval_loss": 0.15005148947238922,
"eval_runtime": 824.7668,
"eval_samples_per_second": 1.212,
"eval_steps_per_second": 0.152,
"eval_wer": 5.192249470178625,
"step": 1000
},
{
"epoch": 1.27,
"learning_rate": 2.5525e-06,
"loss": 0.1065,
"step": 1025
},
{
"epoch": 1.3,
"learning_rate": 2.6150000000000004e-06,
"loss": 0.103,
"step": 1050
},
{
"epoch": 1.33,
"learning_rate": 2.6775e-06,
"loss": 0.1171,
"step": 1075
},
{
"epoch": 1.36,
"learning_rate": 2.7400000000000004e-06,
"loss": 0.1097,
"step": 1100
},
{
"epoch": 1.39,
"learning_rate": 2.8025000000000003e-06,
"loss": 0.1113,
"step": 1125
},
{
"epoch": 1.42,
"learning_rate": 2.865e-06,
"loss": 0.1125,
"step": 1150
},
{
"epoch": 1.45,
"learning_rate": 2.9275000000000003e-06,
"loss": 0.1174,
"step": 1175
},
{
"epoch": 1.48,
"learning_rate": 2.99e-06,
"loss": 0.1073,
"step": 1200
},
{
"epoch": 1.51,
"learning_rate": 3.0525000000000004e-06,
"loss": 0.1189,
"step": 1225
},
{
"epoch": 1.54,
"learning_rate": 3.1150000000000002e-06,
"loss": 0.1103,
"step": 1250
},
{
"epoch": 1.58,
"learning_rate": 3.1775e-06,
"loss": 0.1127,
"step": 1275
},
{
"epoch": 1.61,
"learning_rate": 3.2400000000000003e-06,
"loss": 0.1049,
"step": 1300
},
{
"epoch": 1.64,
"learning_rate": 3.3000000000000006e-06,
"loss": 0.1138,
"step": 1325
},
{
"epoch": 1.67,
"learning_rate": 3.3625000000000004e-06,
"loss": 0.1062,
"step": 1350
},
{
"epoch": 1.7,
"learning_rate": 3.4250000000000007e-06,
"loss": 0.1154,
"step": 1375
},
{
"epoch": 1.73,
"learning_rate": 3.4875000000000005e-06,
"loss": 0.1133,
"step": 1400
},
{
"epoch": 1.76,
"learning_rate": 3.5500000000000003e-06,
"loss": 0.1152,
"step": 1425
},
{
"epoch": 1.79,
"learning_rate": 3.6125000000000006e-06,
"loss": 0.1182,
"step": 1450
},
{
"epoch": 1.82,
"learning_rate": 3.6750000000000004e-06,
"loss": 0.1137,
"step": 1475
},
{
"epoch": 1.85,
"learning_rate": 3.7375000000000006e-06,
"loss": 0.1112,
"step": 1500
},
{
"epoch": 1.88,
"learning_rate": 3.8000000000000005e-06,
"loss": 0.1078,
"step": 1525
},
{
"epoch": 1.92,
"learning_rate": 3.8625e-06,
"loss": 0.1108,
"step": 1550
},
{
"epoch": 1.95,
"learning_rate": 3.9250000000000005e-06,
"loss": 0.1055,
"step": 1575
},
{
"epoch": 1.98,
"learning_rate": 3.9875e-06,
"loss": 0.1173,
"step": 1600
},
{
"epoch": 2.01,
"learning_rate": 4.05e-06,
"loss": 0.0997,
"step": 1625
},
{
"epoch": 2.04,
"learning_rate": 4.1125e-06,
"loss": 0.0735,
"step": 1650
},
{
"epoch": 2.07,
"learning_rate": 4.175e-06,
"loss": 0.0715,
"step": 1675
},
{
"epoch": 2.1,
"learning_rate": 4.2375000000000005e-06,
"loss": 0.0691,
"step": 1700
},
{
"epoch": 2.13,
"learning_rate": 4.3e-06,
"loss": 0.0743,
"step": 1725
},
{
"epoch": 2.16,
"learning_rate": 4.362500000000001e-06,
"loss": 0.0728,
"step": 1750
},
{
"epoch": 2.19,
"learning_rate": 4.425e-06,
"loss": 0.0745,
"step": 1775
},
{
"epoch": 2.22,
"learning_rate": 4.4875e-06,
"loss": 0.0657,
"step": 1800
},
{
"epoch": 2.26,
"learning_rate": 4.5500000000000005e-06,
"loss": 0.0679,
"step": 1825
},
{
"epoch": 2.29,
"learning_rate": 4.6125e-06,
"loss": 0.0837,
"step": 1850
},
{
"epoch": 2.32,
"learning_rate": 4.675000000000001e-06,
"loss": 0.0633,
"step": 1875
},
{
"epoch": 2.35,
"learning_rate": 4.737500000000001e-06,
"loss": 0.07,
"step": 1900
},
{
"epoch": 2.38,
"learning_rate": 4.800000000000001e-06,
"loss": 0.0702,
"step": 1925
},
{
"epoch": 2.41,
"learning_rate": 4.8625000000000005e-06,
"loss": 0.088,
"step": 1950
},
{
"epoch": 2.44,
"learning_rate": 4.925e-06,
"loss": 0.0737,
"step": 1975
},
{
"epoch": 2.47,
"learning_rate": 4.987500000000001e-06,
"loss": 0.0682,
"step": 2000
},
{
"epoch": 2.47,
"eval_cer": 1.8633372152858652,
"eval_loss": 0.15889793634414673,
"eval_runtime": 828.4944,
"eval_samples_per_second": 1.207,
"eval_steps_per_second": 0.151,
"eval_wer": 5.752346351801393,
"step": 2000
},
{
"epoch": 2.5,
"learning_rate": 4.994444444444445e-06,
"loss": 0.0764,
"step": 2025
},
{
"epoch": 2.53,
"learning_rate": 4.987500000000001e-06,
"loss": 0.0777,
"step": 2050
},
{
"epoch": 2.56,
"learning_rate": 4.980555555555555e-06,
"loss": 0.0777,
"step": 2075
},
{
"epoch": 2.6,
"learning_rate": 4.9736111111111115e-06,
"loss": 0.0849,
"step": 2100
},
{
"epoch": 2.63,
"learning_rate": 4.966666666666667e-06,
"loss": 0.0786,
"step": 2125
},
{
"epoch": 2.66,
"learning_rate": 4.959722222222222e-06,
"loss": 0.0746,
"step": 2150
},
{
"epoch": 2.69,
"learning_rate": 4.952777777777778e-06,
"loss": 0.0714,
"step": 2175
},
{
"epoch": 2.72,
"learning_rate": 4.9458333333333336e-06,
"loss": 0.0846,
"step": 2200
},
{
"epoch": 2.75,
"learning_rate": 4.938888888888889e-06,
"loss": 0.0805,
"step": 2225
},
{
"epoch": 2.78,
"learning_rate": 4.931944444444445e-06,
"loss": 0.0792,
"step": 2250
},
{
"epoch": 2.81,
"learning_rate": 4.925e-06,
"loss": 0.0854,
"step": 2275
},
{
"epoch": 2.84,
"learning_rate": 4.9180555555555556e-06,
"loss": 0.0867,
"step": 2300
},
{
"epoch": 2.87,
"learning_rate": 4.911111111111112e-06,
"loss": 0.0801,
"step": 2325
},
{
"epoch": 2.9,
"learning_rate": 4.904166666666667e-06,
"loss": 0.0776,
"step": 2350
},
{
"epoch": 2.94,
"learning_rate": 4.897222222222222e-06,
"loss": 0.0727,
"step": 2375
},
{
"epoch": 2.97,
"learning_rate": 4.8902777777777784e-06,
"loss": 0.0803,
"step": 2400
},
{
"epoch": 3.0,
"learning_rate": 4.883333333333334e-06,
"loss": 0.0726,
"step": 2425
},
{
"epoch": 3.03,
"learning_rate": 4.876388888888889e-06,
"loss": 0.0487,
"step": 2450
},
{
"epoch": 3.06,
"learning_rate": 4.869444444444445e-06,
"loss": 0.051,
"step": 2475
},
{
"epoch": 3.09,
"learning_rate": 4.8625000000000005e-06,
"loss": 0.0461,
"step": 2500
},
{
"epoch": 3.12,
"learning_rate": 4.855555555555556e-06,
"loss": 0.0445,
"step": 2525
},
{
"epoch": 3.15,
"learning_rate": 4.848611111111112e-06,
"loss": 0.0415,
"step": 2550
},
{
"epoch": 3.18,
"learning_rate": 4.841666666666667e-06,
"loss": 0.0429,
"step": 2575
},
{
"epoch": 3.21,
"learning_rate": 4.835e-06,
"loss": 0.0551,
"step": 2600
},
{
"epoch": 3.24,
"learning_rate": 4.828055555555556e-06,
"loss": 0.0471,
"step": 2625
},
{
"epoch": 3.28,
"learning_rate": 4.821111111111112e-06,
"loss": 0.0452,
"step": 2650
},
{
"epoch": 3.31,
"learning_rate": 4.814166666666667e-06,
"loss": 0.0515,
"step": 2675
},
{
"epoch": 3.34,
"learning_rate": 4.807222222222222e-06,
"loss": 0.0478,
"step": 2700
},
{
"epoch": 3.37,
"learning_rate": 4.8002777777777785e-06,
"loss": 0.0474,
"step": 2725
},
{
"epoch": 3.4,
"learning_rate": 4.793333333333334e-06,
"loss": 0.0486,
"step": 2750
},
{
"epoch": 3.43,
"learning_rate": 4.786388888888889e-06,
"loss": 0.0466,
"step": 2775
},
{
"epoch": 3.46,
"learning_rate": 4.779444444444445e-06,
"loss": 0.0477,
"step": 2800
},
{
"epoch": 3.49,
"learning_rate": 4.7725000000000005e-06,
"loss": 0.0487,
"step": 2825
},
{
"epoch": 3.52,
"learning_rate": 4.765555555555556e-06,
"loss": 0.0567,
"step": 2850
},
{
"epoch": 3.55,
"learning_rate": 4.758611111111112e-06,
"loss": 0.0494,
"step": 2875
},
{
"epoch": 3.58,
"learning_rate": 4.751666666666667e-06,
"loss": 0.0474,
"step": 2900
},
{
"epoch": 3.62,
"learning_rate": 4.7447222222222225e-06,
"loss": 0.0492,
"step": 2925
},
{
"epoch": 3.65,
"learning_rate": 4.737777777777779e-06,
"loss": 0.043,
"step": 2950
},
{
"epoch": 3.68,
"learning_rate": 4.730833333333334e-06,
"loss": 0.0484,
"step": 2975
},
{
"epoch": 3.71,
"learning_rate": 4.723888888888889e-06,
"loss": 0.0489,
"step": 3000
},
{
"epoch": 3.71,
"eval_cer": 1.68187850391355,
"eval_loss": 0.16307751834392548,
"eval_runtime": 823.056,
"eval_samples_per_second": 1.215,
"eval_steps_per_second": 0.152,
"eval_wer": 5.358764759309718,
"step": 3000
},
{
"epoch": 3.74,
"learning_rate": 4.716944444444445e-06,
"loss": 0.0459,
"step": 3025
},
{
"epoch": 3.77,
"learning_rate": 4.71e-06,
"loss": 0.05,
"step": 3050
},
{
"epoch": 3.8,
"learning_rate": 4.703055555555556e-06,
"loss": 0.0521,
"step": 3075
},
{
"epoch": 3.83,
"learning_rate": 4.696111111111111e-06,
"loss": 0.0477,
"step": 3100
},
{
"epoch": 3.86,
"learning_rate": 4.6891666666666665e-06,
"loss": 0.0487,
"step": 3125
},
{
"epoch": 3.89,
"learning_rate": 4.682222222222223e-06,
"loss": 0.0461,
"step": 3150
},
{
"epoch": 3.92,
"learning_rate": 4.675277777777778e-06,
"loss": 0.0512,
"step": 3175
},
{
"epoch": 3.95,
"learning_rate": 4.668333333333333e-06,
"loss": 0.0497,
"step": 3200
},
{
"epoch": 3.99,
"learning_rate": 4.661388888888889e-06,
"loss": 0.0471,
"step": 3225
},
{
"epoch": 4.02,
"learning_rate": 4.654444444444445e-06,
"loss": 0.0366,
"step": 3250
},
{
"epoch": 4.05,
"learning_rate": 4.6475e-06,
"loss": 0.0265,
"step": 3275
},
{
"epoch": 4.08,
"learning_rate": 4.640555555555556e-06,
"loss": 0.0253,
"step": 3300
},
{
"epoch": 4.11,
"learning_rate": 4.633611111111111e-06,
"loss": 0.0254,
"step": 3325
},
{
"epoch": 4.14,
"learning_rate": 4.626666666666667e-06,
"loss": 0.0229,
"step": 3350
},
{
"epoch": 4.17,
"learning_rate": 4.619722222222223e-06,
"loss": 0.0307,
"step": 3375
},
{
"epoch": 4.2,
"learning_rate": 4.612777777777778e-06,
"loss": 0.0325,
"step": 3400
},
{
"epoch": 4.23,
"learning_rate": 4.605833333333333e-06,
"loss": 0.0298,
"step": 3425
},
{
"epoch": 4.26,
"learning_rate": 4.5988888888888896e-06,
"loss": 0.0267,
"step": 3450
},
{
"epoch": 4.3,
"learning_rate": 4.591944444444445e-06,
"loss": 0.0293,
"step": 3475
},
{
"epoch": 4.33,
"learning_rate": 4.585e-06,
"loss": 0.0245,
"step": 3500
},
{
"epoch": 4.36,
"learning_rate": 4.578055555555556e-06,
"loss": 0.0318,
"step": 3525
},
{
"epoch": 4.39,
"learning_rate": 4.571111111111112e-06,
"loss": 0.0301,
"step": 3550
},
{
"epoch": 4.42,
"learning_rate": 4.564166666666668e-06,
"loss": 0.0282,
"step": 3575
},
{
"epoch": 4.45,
"learning_rate": 4.557222222222222e-06,
"loss": 0.029,
"step": 3600
},
{
"epoch": 4.48,
"learning_rate": 4.550277777777778e-06,
"loss": 0.028,
"step": 3625
},
{
"epoch": 4.51,
"learning_rate": 4.543333333333334e-06,
"loss": 0.0262,
"step": 3650
},
{
"epoch": 4.54,
"learning_rate": 4.536388888888889e-06,
"loss": 0.0279,
"step": 3675
},
{
"epoch": 4.57,
"learning_rate": 4.529444444444445e-06,
"loss": 0.0302,
"step": 3700
},
{
"epoch": 4.6,
"learning_rate": 4.5225e-06,
"loss": 0.0286,
"step": 3725
},
{
"epoch": 4.63,
"learning_rate": 4.515555555555556e-06,
"loss": 0.0307,
"step": 3750
},
{
"epoch": 4.67,
"learning_rate": 4.508611111111112e-06,
"loss": 0.0294,
"step": 3775
},
{
"epoch": 4.7,
"learning_rate": 4.501666666666667e-06,
"loss": 0.0274,
"step": 3800
},
{
"epoch": 4.73,
"learning_rate": 4.494722222222222e-06,
"loss": 0.0288,
"step": 3825
},
{
"epoch": 4.76,
"learning_rate": 4.4877777777777785e-06,
"loss": 0.0288,
"step": 3850
},
{
"epoch": 4.79,
"learning_rate": 4.480833333333334e-06,
"loss": 0.0305,
"step": 3875
},
{
"epoch": 4.82,
"learning_rate": 4.473888888888889e-06,
"loss": 0.0311,
"step": 3900
},
{
"epoch": 4.85,
"learning_rate": 4.466944444444445e-06,
"loss": 0.029,
"step": 3925
},
{
"epoch": 4.88,
"learning_rate": 4.4600000000000005e-06,
"loss": 0.0326,
"step": 3950
},
{
"epoch": 4.91,
"learning_rate": 4.453055555555556e-06,
"loss": 0.0275,
"step": 3975
},
{
"epoch": 4.94,
"learning_rate": 4.446111111111112e-06,
"loss": 0.0309,
"step": 4000
},
{
"epoch": 4.94,
"eval_cer": 1.68187850391355,
"eval_loss": 0.17070834338665009,
"eval_runtime": 824.4112,
"eval_samples_per_second": 1.213,
"eval_steps_per_second": 0.152,
"eval_wer": 5.283075991522858,
"step": 4000
},
{
"epoch": 4.94,
"step": 4000,
"total_flos": 2.718593101873152e+20,
"train_loss": 0.11291856229305268,
"train_runtime": 31242.3684,
"train_samples_per_second": 20.485,
"train_steps_per_second": 0.64
}
],
"max_steps": 20000,
"num_train_epochs": 25,
"total_flos": 2.718593101873152e+20,
"trial_name": null,
"trial_params": null
}