whisper-medium-el / checkpoint-4000 /trainer_state.json
emilios's picture
12
13b93a0
raw
history blame
20.6 kB
{
"best_metric": 12.574294205052006,
"best_model_checkpoint": "./whisper-medium-el/checkpoint-4000",
"epoch": 19.0362,
"global_step": 4000,
"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": 8.8645,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.000000000000001e-07,
"loss": 5.8694,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.4000000000000001e-06,
"loss": 3.1568,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.8800000000000002e-06,
"loss": 1.9833,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.38e-06,
"loss": 1.4934,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.88e-06,
"loss": 1.1115,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.3800000000000007e-06,
"loss": 0.9083,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.88e-06,
"loss": 0.6967,
"step": 200
},
{
"epoch": 1.0,
"learning_rate": 4.38e-06,
"loss": 0.5981,
"step": 225
},
{
"epoch": 1.01,
"learning_rate": 4.880000000000001e-06,
"loss": 0.5506,
"step": 250
},
{
"epoch": 1.01,
"learning_rate": 5.380000000000001e-06,
"loss": 0.453,
"step": 275
},
{
"epoch": 1.02,
"learning_rate": 5.8800000000000005e-06,
"loss": 0.3967,
"step": 300
},
{
"epoch": 1.02,
"learning_rate": 6.380000000000001e-06,
"loss": 0.4095,
"step": 325
},
{
"epoch": 1.03,
"learning_rate": 6.88e-06,
"loss": 0.3197,
"step": 350
},
{
"epoch": 1.03,
"learning_rate": 7.3800000000000005e-06,
"loss": 0.3176,
"step": 375
},
{
"epoch": 1.04,
"learning_rate": 7.88e-06,
"loss": 0.278,
"step": 400
},
{
"epoch": 2.0,
"learning_rate": 8.380000000000001e-06,
"loss": 0.2536,
"step": 425
},
{
"epoch": 2.01,
"learning_rate": 8.880000000000001e-06,
"loss": 0.2529,
"step": 450
},
{
"epoch": 2.01,
"learning_rate": 9.38e-06,
"loss": 0.2158,
"step": 475
},
{
"epoch": 2.02,
"learning_rate": 9.88e-06,
"loss": 0.204,
"step": 500
},
{
"epoch": 2.02,
"learning_rate": 9.957777777777779e-06,
"loss": 0.21,
"step": 525
},
{
"epoch": 2.03,
"learning_rate": 9.902222222222223e-06,
"loss": 0.1758,
"step": 550
},
{
"epoch": 2.03,
"learning_rate": 9.846666666666668e-06,
"loss": 0.1668,
"step": 575
},
{
"epoch": 2.04,
"learning_rate": 9.791111111111112e-06,
"loss": 0.1475,
"step": 600
},
{
"epoch": 3.0,
"learning_rate": 9.735555555555556e-06,
"loss": 0.1416,
"step": 625
},
{
"epoch": 3.01,
"learning_rate": 9.68e-06,
"loss": 0.138,
"step": 650
},
{
"epoch": 3.01,
"learning_rate": 9.624444444444445e-06,
"loss": 0.1225,
"step": 675
},
{
"epoch": 3.02,
"learning_rate": 9.56888888888889e-06,
"loss": 0.1156,
"step": 700
},
{
"epoch": 3.02,
"learning_rate": 9.513333333333334e-06,
"loss": 0.1093,
"step": 725
},
{
"epoch": 3.03,
"learning_rate": 9.457777777777778e-06,
"loss": 0.1041,
"step": 750
},
{
"epoch": 3.03,
"learning_rate": 9.402222222222222e-06,
"loss": 0.0971,
"step": 775
},
{
"epoch": 3.04,
"learning_rate": 9.346666666666666e-06,
"loss": 0.0875,
"step": 800
},
{
"epoch": 4.0,
"learning_rate": 9.291111111111112e-06,
"loss": 0.0837,
"step": 825
},
{
"epoch": 4.01,
"learning_rate": 9.235555555555556e-06,
"loss": 0.0826,
"step": 850
},
{
"epoch": 4.01,
"learning_rate": 9.180000000000002e-06,
"loss": 0.0737,
"step": 875
},
{
"epoch": 4.02,
"learning_rate": 9.124444444444444e-06,
"loss": 0.0717,
"step": 900
},
{
"epoch": 4.02,
"learning_rate": 9.06888888888889e-06,
"loss": 0.0707,
"step": 925
},
{
"epoch": 4.03,
"learning_rate": 9.013333333333334e-06,
"loss": 0.0647,
"step": 950
},
{
"epoch": 4.03,
"learning_rate": 8.957777777777778e-06,
"loss": 0.0593,
"step": 975
},
{
"epoch": 4.04,
"learning_rate": 8.902222222222224e-06,
"loss": 0.0582,
"step": 1000
},
{
"epoch": 4.04,
"eval_loss": 0.3256979286670685,
"eval_runtime": 1207.5418,
"eval_samples_per_second": 1.405,
"eval_steps_per_second": 0.088,
"eval_wer": 13.085066864784547,
"step": 1000
},
{
"epoch": 5.0,
"learning_rate": 8.846666666666668e-06,
"loss": 0.0559,
"step": 1025
},
{
"epoch": 5.01,
"learning_rate": 8.791111111111112e-06,
"loss": 0.0549,
"step": 1050
},
{
"epoch": 5.01,
"learning_rate": 8.735555555555556e-06,
"loss": 0.0474,
"step": 1075
},
{
"epoch": 5.02,
"learning_rate": 8.68e-06,
"loss": 0.046,
"step": 1100
},
{
"epoch": 5.02,
"learning_rate": 8.624444444444446e-06,
"loss": 0.0518,
"step": 1125
},
{
"epoch": 5.03,
"learning_rate": 8.56888888888889e-06,
"loss": 0.042,
"step": 1150
},
{
"epoch": 5.03,
"learning_rate": 8.513333333333335e-06,
"loss": 0.0464,
"step": 1175
},
{
"epoch": 5.04,
"learning_rate": 8.457777777777778e-06,
"loss": 0.0391,
"step": 1200
},
{
"epoch": 6.0,
"learning_rate": 8.402222222222223e-06,
"loss": 0.0383,
"step": 1225
},
{
"epoch": 6.01,
"learning_rate": 8.346666666666668e-06,
"loss": 0.036,
"step": 1250
},
{
"epoch": 6.01,
"learning_rate": 8.291111111111112e-06,
"loss": 0.0326,
"step": 1275
},
{
"epoch": 6.02,
"learning_rate": 8.235555555555557e-06,
"loss": 0.033,
"step": 1300
},
{
"epoch": 6.02,
"learning_rate": 8.18e-06,
"loss": 0.0357,
"step": 1325
},
{
"epoch": 6.03,
"learning_rate": 8.124444444444445e-06,
"loss": 0.0321,
"step": 1350
},
{
"epoch": 6.03,
"learning_rate": 8.06888888888889e-06,
"loss": 0.0278,
"step": 1375
},
{
"epoch": 6.04,
"learning_rate": 8.013333333333333e-06,
"loss": 0.0283,
"step": 1400
},
{
"epoch": 7.0,
"learning_rate": 7.957777777777779e-06,
"loss": 0.0295,
"step": 1425
},
{
"epoch": 7.01,
"learning_rate": 7.902222222222223e-06,
"loss": 0.0245,
"step": 1450
},
{
"epoch": 7.01,
"learning_rate": 7.846666666666667e-06,
"loss": 0.0271,
"step": 1475
},
{
"epoch": 7.02,
"learning_rate": 7.791111111111111e-06,
"loss": 0.028,
"step": 1500
},
{
"epoch": 7.02,
"learning_rate": 7.735555555555557e-06,
"loss": 0.0271,
"step": 1525
},
{
"epoch": 7.03,
"learning_rate": 7.680000000000001e-06,
"loss": 0.0238,
"step": 1550
},
{
"epoch": 7.03,
"learning_rate": 7.624444444444445e-06,
"loss": 0.0237,
"step": 1575
},
{
"epoch": 7.04,
"learning_rate": 7.56888888888889e-06,
"loss": 0.023,
"step": 1600
},
{
"epoch": 8.0,
"learning_rate": 7.513333333333334e-06,
"loss": 0.0207,
"step": 1625
},
{
"epoch": 8.01,
"learning_rate": 7.457777777777778e-06,
"loss": 0.0193,
"step": 1650
},
{
"epoch": 8.01,
"learning_rate": 7.402222222222223e-06,
"loss": 0.0224,
"step": 1675
},
{
"epoch": 8.02,
"learning_rate": 7.346666666666668e-06,
"loss": 0.0184,
"step": 1700
},
{
"epoch": 8.02,
"learning_rate": 7.291111111111112e-06,
"loss": 0.021,
"step": 1725
},
{
"epoch": 8.03,
"learning_rate": 7.235555555555556e-06,
"loss": 0.0197,
"step": 1750
},
{
"epoch": 8.03,
"learning_rate": 7.180000000000001e-06,
"loss": 0.0167,
"step": 1775
},
{
"epoch": 8.04,
"learning_rate": 7.124444444444445e-06,
"loss": 0.0155,
"step": 1800
},
{
"epoch": 9.0,
"learning_rate": 7.06888888888889e-06,
"loss": 0.0161,
"step": 1825
},
{
"epoch": 9.01,
"learning_rate": 7.0133333333333345e-06,
"loss": 0.0181,
"step": 1850
},
{
"epoch": 9.01,
"learning_rate": 6.9577777777777785e-06,
"loss": 0.0166,
"step": 1875
},
{
"epoch": 9.02,
"learning_rate": 6.902222222222223e-06,
"loss": 0.0149,
"step": 1900
},
{
"epoch": 9.02,
"learning_rate": 6.846666666666667e-06,
"loss": 0.0161,
"step": 1925
},
{
"epoch": 9.03,
"learning_rate": 6.7911111111111115e-06,
"loss": 0.0163,
"step": 1950
},
{
"epoch": 9.03,
"learning_rate": 6.735555555555556e-06,
"loss": 0.0155,
"step": 1975
},
{
"epoch": 9.04,
"learning_rate": 6.680000000000001e-06,
"loss": 0.0121,
"step": 2000
},
{
"epoch": 9.04,
"eval_loss": 0.3753364384174347,
"eval_runtime": 1217.4277,
"eval_samples_per_second": 1.393,
"eval_steps_per_second": 0.087,
"eval_wer": 13.187221396731056,
"step": 2000
},
{
"epoch": 10.0,
"learning_rate": 6.6244444444444445e-06,
"loss": 0.015,
"step": 2025
},
{
"epoch": 10.01,
"learning_rate": 6.568888888888889e-06,
"loss": 0.013,
"step": 2050
},
{
"epoch": 10.01,
"learning_rate": 6.513333333333333e-06,
"loss": 0.0125,
"step": 2075
},
{
"epoch": 10.02,
"learning_rate": 6.457777777777778e-06,
"loss": 0.0128,
"step": 2100
},
{
"epoch": 10.02,
"learning_rate": 6.402222222222223e-06,
"loss": 0.0134,
"step": 2125
},
{
"epoch": 10.03,
"learning_rate": 6.346666666666668e-06,
"loss": 0.0133,
"step": 2150
},
{
"epoch": 10.03,
"learning_rate": 6.291111111111111e-06,
"loss": 0.0102,
"step": 2175
},
{
"epoch": 10.04,
"learning_rate": 6.235555555555556e-06,
"loss": 0.0096,
"step": 2200
},
{
"epoch": 11.0,
"learning_rate": 6.18e-06,
"loss": 0.0124,
"step": 2225
},
{
"epoch": 11.01,
"learning_rate": 6.124444444444445e-06,
"loss": 0.0104,
"step": 2250
},
{
"epoch": 11.01,
"learning_rate": 6.06888888888889e-06,
"loss": 0.011,
"step": 2275
},
{
"epoch": 11.02,
"learning_rate": 6.013333333333335e-06,
"loss": 0.0097,
"step": 2300
},
{
"epoch": 11.02,
"learning_rate": 5.957777777777778e-06,
"loss": 0.0112,
"step": 2325
},
{
"epoch": 11.03,
"learning_rate": 5.902222222222223e-06,
"loss": 0.0101,
"step": 2350
},
{
"epoch": 11.03,
"learning_rate": 5.846666666666667e-06,
"loss": 0.0099,
"step": 2375
},
{
"epoch": 11.04,
"learning_rate": 5.791111111111112e-06,
"loss": 0.0088,
"step": 2400
},
{
"epoch": 12.0,
"learning_rate": 5.735555555555557e-06,
"loss": 0.0092,
"step": 2425
},
{
"epoch": 12.01,
"learning_rate": 5.68e-06,
"loss": 0.0097,
"step": 2450
},
{
"epoch": 12.01,
"learning_rate": 5.624444444444445e-06,
"loss": 0.009,
"step": 2475
},
{
"epoch": 12.02,
"learning_rate": 5.56888888888889e-06,
"loss": 0.0081,
"step": 2500
},
{
"epoch": 12.02,
"learning_rate": 5.513333333333334e-06,
"loss": 0.0084,
"step": 2525
},
{
"epoch": 12.03,
"learning_rate": 5.4577777777777785e-06,
"loss": 0.009,
"step": 2550
},
{
"epoch": 12.03,
"learning_rate": 5.402222222222223e-06,
"loss": 0.0082,
"step": 2575
},
{
"epoch": 12.04,
"learning_rate": 5.346666666666667e-06,
"loss": 0.0076,
"step": 2600
},
{
"epoch": 13.0,
"learning_rate": 5.2911111111111115e-06,
"loss": 0.0072,
"step": 2625
},
{
"epoch": 13.01,
"learning_rate": 5.235555555555556e-06,
"loss": 0.0077,
"step": 2650
},
{
"epoch": 13.01,
"learning_rate": 5.18e-06,
"loss": 0.0073,
"step": 2675
},
{
"epoch": 13.02,
"learning_rate": 5.124444444444445e-06,
"loss": 0.0072,
"step": 2700
},
{
"epoch": 13.02,
"learning_rate": 5.06888888888889e-06,
"loss": 0.0074,
"step": 2725
},
{
"epoch": 13.03,
"learning_rate": 5.013333333333333e-06,
"loss": 0.0079,
"step": 2750
},
{
"epoch": 13.03,
"learning_rate": 4.957777777777778e-06,
"loss": 0.0057,
"step": 2775
},
{
"epoch": 13.04,
"learning_rate": 4.902222222222222e-06,
"loss": 0.0082,
"step": 2800
},
{
"epoch": 14.0,
"learning_rate": 4.846666666666667e-06,
"loss": 0.0063,
"step": 2825
},
{
"epoch": 14.01,
"learning_rate": 4.791111111111111e-06,
"loss": 0.0064,
"step": 2850
},
{
"epoch": 14.01,
"learning_rate": 4.735555555555556e-06,
"loss": 0.0066,
"step": 2875
},
{
"epoch": 14.02,
"learning_rate": 4.680000000000001e-06,
"loss": 0.0066,
"step": 2900
},
{
"epoch": 14.02,
"learning_rate": 4.624444444444445e-06,
"loss": 0.0069,
"step": 2925
},
{
"epoch": 14.03,
"learning_rate": 4.568888888888889e-06,
"loss": 0.0053,
"step": 2950
},
{
"epoch": 14.03,
"learning_rate": 4.513333333333333e-06,
"loss": 0.0058,
"step": 2975
},
{
"epoch": 14.04,
"learning_rate": 4.457777777777778e-06,
"loss": 0.0067,
"step": 3000
},
{
"epoch": 14.04,
"eval_loss": 0.37926045060157776,
"eval_runtime": 1203.9399,
"eval_samples_per_second": 1.409,
"eval_steps_per_second": 0.088,
"eval_wer": 12.862184249628527,
"step": 3000
},
{
"epoch": 15.0,
"learning_rate": 4.402222222222223e-06,
"loss": 0.005,
"step": 3025
},
{
"epoch": 15.01,
"learning_rate": 4.346666666666667e-06,
"loss": 0.0051,
"step": 3050
},
{
"epoch": 15.01,
"learning_rate": 4.291111111111112e-06,
"loss": 0.0053,
"step": 3075
},
{
"epoch": 15.02,
"learning_rate": 4.235555555555556e-06,
"loss": 0.0053,
"step": 3100
},
{
"epoch": 15.02,
"learning_rate": 4.18e-06,
"loss": 0.0059,
"step": 3125
},
{
"epoch": 15.03,
"learning_rate": 4.124444444444445e-06,
"loss": 0.0053,
"step": 3150
},
{
"epoch": 15.03,
"learning_rate": 4.0688888888888896e-06,
"loss": 0.0058,
"step": 3175
},
{
"epoch": 15.04,
"learning_rate": 4.013333333333334e-06,
"loss": 0.0045,
"step": 3200
},
{
"epoch": 16.0,
"learning_rate": 3.9577777777777785e-06,
"loss": 0.0038,
"step": 3225
},
{
"epoch": 16.01,
"learning_rate": 3.9022222222222225e-06,
"loss": 0.0046,
"step": 3250
},
{
"epoch": 16.01,
"learning_rate": 3.8466666666666665e-06,
"loss": 0.0055,
"step": 3275
},
{
"epoch": 16.02,
"learning_rate": 3.7911111111111114e-06,
"loss": 0.005,
"step": 3300
},
{
"epoch": 16.02,
"learning_rate": 3.7355555555555555e-06,
"loss": 0.0051,
"step": 3325
},
{
"epoch": 16.03,
"learning_rate": 3.6800000000000003e-06,
"loss": 0.0047,
"step": 3350
},
{
"epoch": 16.03,
"learning_rate": 3.624444444444445e-06,
"loss": 0.0041,
"step": 3375
},
{
"epoch": 16.04,
"learning_rate": 3.568888888888889e-06,
"loss": 0.004,
"step": 3400
},
{
"epoch": 17.0,
"learning_rate": 3.5133333333333337e-06,
"loss": 0.0042,
"step": 3425
},
{
"epoch": 17.01,
"learning_rate": 3.457777777777778e-06,
"loss": 0.0043,
"step": 3450
},
{
"epoch": 17.01,
"learning_rate": 3.4022222222222222e-06,
"loss": 0.0047,
"step": 3475
},
{
"epoch": 17.02,
"learning_rate": 3.346666666666667e-06,
"loss": 0.0035,
"step": 3500
},
{
"epoch": 17.02,
"learning_rate": 3.2911111111111116e-06,
"loss": 0.0047,
"step": 3525
},
{
"epoch": 17.03,
"learning_rate": 3.2355555555555556e-06,
"loss": 0.0032,
"step": 3550
},
{
"epoch": 17.03,
"learning_rate": 3.1800000000000005e-06,
"loss": 0.0039,
"step": 3575
},
{
"epoch": 17.04,
"learning_rate": 3.124444444444445e-06,
"loss": 0.0039,
"step": 3600
},
{
"epoch": 18.0,
"learning_rate": 3.068888888888889e-06,
"loss": 0.0037,
"step": 3625
},
{
"epoch": 18.01,
"learning_rate": 3.013333333333334e-06,
"loss": 0.0041,
"step": 3650
},
{
"epoch": 18.01,
"learning_rate": 2.957777777777778e-06,
"loss": 0.0041,
"step": 3675
},
{
"epoch": 18.02,
"learning_rate": 2.9022222222222223e-06,
"loss": 0.0031,
"step": 3700
},
{
"epoch": 18.02,
"learning_rate": 2.8466666666666672e-06,
"loss": 0.0044,
"step": 3725
},
{
"epoch": 18.03,
"learning_rate": 2.7911111111111113e-06,
"loss": 0.0029,
"step": 3750
},
{
"epoch": 18.03,
"learning_rate": 2.7355555555555557e-06,
"loss": 0.0039,
"step": 3775
},
{
"epoch": 18.04,
"learning_rate": 2.68e-06,
"loss": 0.0034,
"step": 3800
},
{
"epoch": 19.0,
"learning_rate": 2.6244444444444446e-06,
"loss": 0.003,
"step": 3825
},
{
"epoch": 19.01,
"learning_rate": 2.568888888888889e-06,
"loss": 0.0035,
"step": 3850
},
{
"epoch": 19.01,
"learning_rate": 2.5133333333333336e-06,
"loss": 0.003,
"step": 3875
},
{
"epoch": 19.02,
"learning_rate": 2.457777777777778e-06,
"loss": 0.0032,
"step": 3900
},
{
"epoch": 19.02,
"learning_rate": 2.4022222222222225e-06,
"loss": 0.003,
"step": 3925
},
{
"epoch": 19.03,
"learning_rate": 2.346666666666667e-06,
"loss": 0.0036,
"step": 3950
},
{
"epoch": 19.03,
"learning_rate": 2.2911111111111114e-06,
"loss": 0.0036,
"step": 3975
},
{
"epoch": 19.04,
"learning_rate": 2.235555555555556e-06,
"loss": 0.0029,
"step": 4000
},
{
"epoch": 19.04,
"eval_loss": 0.3888511061668396,
"eval_runtime": 1204.5144,
"eval_samples_per_second": 1.408,
"eval_steps_per_second": 0.088,
"eval_wer": 12.574294205052006,
"step": 4000
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.3055986218958848e+20,
"trial_name": null,
"trial_params": null
}