whisper-medium-eu / checkpoint-4000 /trainer_state.json
Xabi Ezpeleta
First trial
e94d61e
raw
history blame
20.5 kB
{
"best_metric": 15.300336182105392,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 1.1408,
"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": 4.6000000000000004e-07,
"loss": 1.4182,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 1.292,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 1.0018,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.94e-06,
"loss": 0.7765,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.7103,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.6597,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.44e-06,
"loss": 0.6657,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.94e-06,
"loss": 0.5853,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.440000000000001e-06,
"loss": 0.5273,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 4.94e-06,
"loss": 0.5979,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.5861,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 5.94e-06,
"loss": 0.5085,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 6.440000000000001e-06,
"loss": 0.4827,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.4909,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 7.440000000000001e-06,
"loss": 0.4651,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 7.94e-06,
"loss": 0.494,
"step": 400
},
{
"epoch": 0.09,
"learning_rate": 8.44e-06,
"loss": 0.4188,
"step": 425
},
{
"epoch": 0.09,
"learning_rate": 8.94e-06,
"loss": 0.3849,
"step": 450
},
{
"epoch": 0.1,
"learning_rate": 9.440000000000001e-06,
"loss": 0.4577,
"step": 475
},
{
"epoch": 0.1,
"learning_rate": 9.940000000000001e-06,
"loss": 0.4415,
"step": 500
},
{
"epoch": 0.1,
"learning_rate": 9.951111111111111e-06,
"loss": 0.4615,
"step": 525
},
{
"epoch": 0.11,
"learning_rate": 9.895555555555557e-06,
"loss": 0.4282,
"step": 550
},
{
"epoch": 0.12,
"learning_rate": 9.842222222222223e-06,
"loss": 0.4481,
"step": 575
},
{
"epoch": 0.12,
"learning_rate": 9.786666666666667e-06,
"loss": 0.4441,
"step": 600
},
{
"epoch": 0.12,
"learning_rate": 9.731111111111113e-06,
"loss": 0.4238,
"step": 625
},
{
"epoch": 0.13,
"learning_rate": 9.675555555555555e-06,
"loss": 0.4245,
"step": 650
},
{
"epoch": 0.14,
"learning_rate": 9.620000000000001e-06,
"loss": 0.4118,
"step": 675
},
{
"epoch": 0.14,
"learning_rate": 9.564444444444445e-06,
"loss": 0.4111,
"step": 700
},
{
"epoch": 0.14,
"learning_rate": 9.508888888888889e-06,
"loss": 0.3642,
"step": 725
},
{
"epoch": 0.15,
"learning_rate": 9.453333333333335e-06,
"loss": 0.401,
"step": 750
},
{
"epoch": 0.15,
"learning_rate": 9.397777777777779e-06,
"loss": 0.3855,
"step": 775
},
{
"epoch": 0.16,
"learning_rate": 9.342222222222223e-06,
"loss": 0.3668,
"step": 800
},
{
"epoch": 0.17,
"learning_rate": 9.286666666666667e-06,
"loss": 0.3794,
"step": 825
},
{
"epoch": 0.17,
"learning_rate": 9.231111111111111e-06,
"loss": 0.4296,
"step": 850
},
{
"epoch": 0.17,
"learning_rate": 9.175555555555557e-06,
"loss": 0.4003,
"step": 875
},
{
"epoch": 0.18,
"learning_rate": 9.12e-06,
"loss": 0.374,
"step": 900
},
{
"epoch": 0.18,
"learning_rate": 9.064444444444447e-06,
"loss": 0.4051,
"step": 925
},
{
"epoch": 0.19,
"learning_rate": 9.008888888888889e-06,
"loss": 0.3806,
"step": 950
},
{
"epoch": 0.2,
"learning_rate": 8.953333333333335e-06,
"loss": 0.4161,
"step": 975
},
{
"epoch": 0.2,
"learning_rate": 8.897777777777779e-06,
"loss": 0.4198,
"step": 1000
},
{
"epoch": 0.2,
"eval_loss": 0.41016528010368347,
"eval_runtime": 1788.2777,
"eval_samples_per_second": 3.686,
"eval_steps_per_second": 0.461,
"eval_wer": 28.34865729677184,
"step": 1000
},
{
"epoch": 0.2,
"learning_rate": 8.842222222222223e-06,
"loss": 0.409,
"step": 1025
},
{
"epoch": 0.21,
"learning_rate": 8.786666666666668e-06,
"loss": 0.3674,
"step": 1050
},
{
"epoch": 0.21,
"learning_rate": 8.73111111111111e-06,
"loss": 0.3591,
"step": 1075
},
{
"epoch": 0.22,
"learning_rate": 8.675555555555556e-06,
"loss": 0.3892,
"step": 1100
},
{
"epoch": 0.23,
"learning_rate": 8.62e-06,
"loss": 0.3843,
"step": 1125
},
{
"epoch": 0.23,
"learning_rate": 8.564444444444445e-06,
"loss": 0.3605,
"step": 1150
},
{
"epoch": 0.23,
"learning_rate": 8.50888888888889e-06,
"loss": 0.326,
"step": 1175
},
{
"epoch": 0.24,
"learning_rate": 8.453333333333334e-06,
"loss": 0.3103,
"step": 1200
},
{
"epoch": 0.24,
"learning_rate": 8.397777777777778e-06,
"loss": 0.2766,
"step": 1225
},
{
"epoch": 0.25,
"learning_rate": 8.342222222222222e-06,
"loss": 0.3204,
"step": 1250
},
{
"epoch": 0.26,
"learning_rate": 8.286666666666668e-06,
"loss": 0.3426,
"step": 1275
},
{
"epoch": 0.26,
"learning_rate": 8.231111111111112e-06,
"loss": 0.3417,
"step": 1300
},
{
"epoch": 0.27,
"learning_rate": 8.175555555555556e-06,
"loss": 0.3179,
"step": 1325
},
{
"epoch": 0.27,
"learning_rate": 8.120000000000002e-06,
"loss": 0.2598,
"step": 1350
},
{
"epoch": 0.28,
"learning_rate": 8.064444444444444e-06,
"loss": 0.3453,
"step": 1375
},
{
"epoch": 0.28,
"learning_rate": 8.00888888888889e-06,
"loss": 0.2752,
"step": 1400
},
{
"epoch": 0.28,
"learning_rate": 7.953333333333334e-06,
"loss": 0.2927,
"step": 1425
},
{
"epoch": 0.29,
"learning_rate": 7.897777777777778e-06,
"loss": 0.3859,
"step": 1450
},
{
"epoch": 0.29,
"learning_rate": 7.842222222222224e-06,
"loss": 0.3137,
"step": 1475
},
{
"epoch": 0.3,
"learning_rate": 7.786666666666666e-06,
"loss": 0.2678,
"step": 1500
},
{
"epoch": 0.3,
"learning_rate": 7.731111111111112e-06,
"loss": 0.2803,
"step": 1525
},
{
"epoch": 0.31,
"learning_rate": 7.675555555555556e-06,
"loss": 0.2828,
"step": 1550
},
{
"epoch": 0.32,
"learning_rate": 7.620000000000001e-06,
"loss": 0.3655,
"step": 1575
},
{
"epoch": 0.32,
"learning_rate": 7.564444444444446e-06,
"loss": 0.3321,
"step": 1600
},
{
"epoch": 0.33,
"learning_rate": 7.50888888888889e-06,
"loss": 0.3649,
"step": 1625
},
{
"epoch": 0.33,
"learning_rate": 7.453333333333334e-06,
"loss": 0.3229,
"step": 1650
},
{
"epoch": 0.34,
"learning_rate": 7.3977777777777786e-06,
"loss": 0.3115,
"step": 1675
},
{
"epoch": 0.34,
"learning_rate": 7.342222222222223e-06,
"loss": 0.2925,
"step": 1700
},
{
"epoch": 0.34,
"learning_rate": 7.2866666666666675e-06,
"loss": 0.3014,
"step": 1725
},
{
"epoch": 0.35,
"learning_rate": 7.231111111111112e-06,
"loss": 0.3303,
"step": 1750
},
{
"epoch": 0.35,
"learning_rate": 7.1755555555555556e-06,
"loss": 0.3174,
"step": 1775
},
{
"epoch": 0.36,
"learning_rate": 7.1200000000000004e-06,
"loss": 0.3249,
"step": 1800
},
{
"epoch": 0.36,
"learning_rate": 7.0644444444444445e-06,
"loss": 0.2678,
"step": 1825
},
{
"epoch": 0.37,
"learning_rate": 7.008888888888889e-06,
"loss": 0.3088,
"step": 1850
},
{
"epoch": 0.38,
"learning_rate": 6.953333333333334e-06,
"loss": 0.2515,
"step": 1875
},
{
"epoch": 0.38,
"learning_rate": 6.897777777777779e-06,
"loss": 0.2838,
"step": 1900
},
{
"epoch": 0.39,
"learning_rate": 6.842222222222222e-06,
"loss": 0.2494,
"step": 1925
},
{
"epoch": 0.39,
"learning_rate": 6.786666666666667e-06,
"loss": 0.205,
"step": 1950
},
{
"epoch": 0.4,
"learning_rate": 6.731111111111111e-06,
"loss": 0.2439,
"step": 1975
},
{
"epoch": 0.4,
"learning_rate": 6.675555555555556e-06,
"loss": 0.2547,
"step": 2000
},
{
"epoch": 0.4,
"eval_loss": 0.31417879462242126,
"eval_runtime": 1808.8984,
"eval_samples_per_second": 3.644,
"eval_steps_per_second": 0.456,
"eval_wer": 21.643241929604276,
"step": 2000
},
{
"epoch": 0.41,
"learning_rate": 6.620000000000001e-06,
"loss": 0.212,
"step": 2025
},
{
"epoch": 0.41,
"learning_rate": 6.564444444444446e-06,
"loss": 0.2386,
"step": 2050
},
{
"epoch": 0.41,
"learning_rate": 6.508888888888889e-06,
"loss": 0.2429,
"step": 2075
},
{
"epoch": 0.42,
"learning_rate": 6.453333333333334e-06,
"loss": 0.3079,
"step": 2100
},
{
"epoch": 0.42,
"learning_rate": 6.397777777777778e-06,
"loss": 0.2576,
"step": 2125
},
{
"epoch": 0.43,
"learning_rate": 6.342222222222223e-06,
"loss": 0.2558,
"step": 2150
},
{
"epoch": 0.43,
"learning_rate": 6.286666666666668e-06,
"loss": 0.2904,
"step": 2175
},
{
"epoch": 0.44,
"learning_rate": 6.231111111111111e-06,
"loss": 0.2423,
"step": 2200
},
{
"epoch": 0.45,
"learning_rate": 6.175555555555556e-06,
"loss": 0.255,
"step": 2225
},
{
"epoch": 0.45,
"learning_rate": 6.120000000000001e-06,
"loss": 0.2142,
"step": 2250
},
{
"epoch": 0.46,
"learning_rate": 6.064444444444445e-06,
"loss": 0.2687,
"step": 2275
},
{
"epoch": 0.46,
"learning_rate": 6.00888888888889e-06,
"loss": 0.2617,
"step": 2300
},
{
"epoch": 0.47,
"learning_rate": 5.9533333333333345e-06,
"loss": 0.2414,
"step": 2325
},
{
"epoch": 0.47,
"learning_rate": 5.897777777777778e-06,
"loss": 0.2048,
"step": 2350
},
{
"epoch": 0.47,
"learning_rate": 5.8422222222222226e-06,
"loss": 0.222,
"step": 2375
},
{
"epoch": 0.48,
"learning_rate": 5.7866666666666674e-06,
"loss": 0.2453,
"step": 2400
},
{
"epoch": 0.48,
"learning_rate": 5.7311111111111115e-06,
"loss": 0.2099,
"step": 2425
},
{
"epoch": 0.49,
"learning_rate": 5.675555555555556e-06,
"loss": 0.2515,
"step": 2450
},
{
"epoch": 0.49,
"learning_rate": 5.620000000000001e-06,
"loss": 0.2232,
"step": 2475
},
{
"epoch": 0.5,
"learning_rate": 5.5644444444444444e-06,
"loss": 0.1946,
"step": 2500
},
{
"epoch": 0.51,
"learning_rate": 5.508888888888889e-06,
"loss": 0.2176,
"step": 2525
},
{
"epoch": 0.51,
"learning_rate": 5.453333333333334e-06,
"loss": 0.2565,
"step": 2550
},
{
"epoch": 0.52,
"learning_rate": 5.397777777777778e-06,
"loss": 0.2452,
"step": 2575
},
{
"epoch": 0.52,
"learning_rate": 5.342222222222223e-06,
"loss": 0.2851,
"step": 2600
},
{
"epoch": 0.53,
"learning_rate": 5.286666666666666e-06,
"loss": 0.1891,
"step": 2625
},
{
"epoch": 0.53,
"learning_rate": 5.231111111111111e-06,
"loss": 0.2404,
"step": 2650
},
{
"epoch": 0.54,
"learning_rate": 5.175555555555556e-06,
"loss": 0.2037,
"step": 2675
},
{
"epoch": 0.54,
"learning_rate": 5.12e-06,
"loss": 0.215,
"step": 2700
},
{
"epoch": 0.55,
"learning_rate": 5.064444444444445e-06,
"loss": 0.2115,
"step": 2725
},
{
"epoch": 0.55,
"learning_rate": 5.00888888888889e-06,
"loss": 0.2491,
"step": 2750
},
{
"epoch": 0.56,
"learning_rate": 4.953333333333334e-06,
"loss": 0.1979,
"step": 2775
},
{
"epoch": 0.56,
"learning_rate": 4.897777777777778e-06,
"loss": 0.224,
"step": 2800
},
{
"epoch": 0.56,
"learning_rate": 4.842222222222223e-06,
"loss": 0.2065,
"step": 2825
},
{
"epoch": 0.57,
"learning_rate": 4.786666666666667e-06,
"loss": 0.2144,
"step": 2850
},
{
"epoch": 0.57,
"learning_rate": 4.731111111111112e-06,
"loss": 0.2141,
"step": 2875
},
{
"epoch": 0.58,
"learning_rate": 4.675555555555556e-06,
"loss": 0.1953,
"step": 2900
},
{
"epoch": 0.58,
"learning_rate": 4.620000000000001e-06,
"loss": 0.1907,
"step": 2925
},
{
"epoch": 0.59,
"learning_rate": 4.564444444444445e-06,
"loss": 0.2292,
"step": 2950
},
{
"epoch": 0.59,
"learning_rate": 4.50888888888889e-06,
"loss": 0.2164,
"step": 2975
},
{
"epoch": 0.6,
"learning_rate": 4.453333333333334e-06,
"loss": 0.2145,
"step": 3000
},
{
"epoch": 0.6,
"eval_loss": 0.260960191488266,
"eval_runtime": 1803.5499,
"eval_samples_per_second": 3.654,
"eval_steps_per_second": 0.457,
"eval_wer": 17.515897768236865,
"step": 3000
},
{
"epoch": 0.6,
"learning_rate": 4.397777777777778e-06,
"loss": 0.1934,
"step": 3025
},
{
"epoch": 0.61,
"learning_rate": 4.3422222222222225e-06,
"loss": 0.1973,
"step": 3050
},
{
"epoch": 0.61,
"learning_rate": 4.2866666666666666e-06,
"loss": 0.1767,
"step": 3075
},
{
"epoch": 0.62,
"learning_rate": 4.2311111111111114e-06,
"loss": 0.1918,
"step": 3100
},
{
"epoch": 0.62,
"learning_rate": 4.175555555555556e-06,
"loss": 0.1946,
"step": 3125
},
{
"epoch": 0.63,
"learning_rate": 4.12e-06,
"loss": 0.1897,
"step": 3150
},
{
"epoch": 0.64,
"learning_rate": 4.064444444444444e-06,
"loss": 0.2185,
"step": 3175
},
{
"epoch": 0.64,
"learning_rate": 4.008888888888889e-06,
"loss": 0.1954,
"step": 3200
},
{
"epoch": 0.65,
"learning_rate": 3.953333333333333e-06,
"loss": 0.2318,
"step": 3225
},
{
"epoch": 0.65,
"learning_rate": 3.897777777777778e-06,
"loss": 0.2615,
"step": 3250
},
{
"epoch": 0.66,
"learning_rate": 3.842222222222223e-06,
"loss": 0.1846,
"step": 3275
},
{
"epoch": 1.0,
"learning_rate": 3.7866666666666667e-06,
"loss": 0.222,
"step": 3300
},
{
"epoch": 1.01,
"learning_rate": 3.7311111111111116e-06,
"loss": 0.2224,
"step": 3325
},
{
"epoch": 1.01,
"learning_rate": 3.675555555555556e-06,
"loss": 0.2128,
"step": 3350
},
{
"epoch": 1.02,
"learning_rate": 3.62e-06,
"loss": 0.2002,
"step": 3375
},
{
"epoch": 1.02,
"learning_rate": 3.564444444444445e-06,
"loss": 0.1861,
"step": 3400
},
{
"epoch": 1.03,
"learning_rate": 3.508888888888889e-06,
"loss": 0.176,
"step": 3425
},
{
"epoch": 1.03,
"learning_rate": 3.4533333333333334e-06,
"loss": 0.1659,
"step": 3450
},
{
"epoch": 1.04,
"learning_rate": 3.3977777777777783e-06,
"loss": 0.1545,
"step": 3475
},
{
"epoch": 1.04,
"learning_rate": 3.3422222222222224e-06,
"loss": 0.1314,
"step": 3500
},
{
"epoch": 1.05,
"learning_rate": 3.286666666666667e-06,
"loss": 0.1573,
"step": 3525
},
{
"epoch": 1.05,
"learning_rate": 3.2311111111111117e-06,
"loss": 0.1696,
"step": 3550
},
{
"epoch": 1.06,
"learning_rate": 3.1755555555555557e-06,
"loss": 0.1348,
"step": 3575
},
{
"epoch": 1.06,
"learning_rate": 3.12e-06,
"loss": 0.1477,
"step": 3600
},
{
"epoch": 1.07,
"learning_rate": 3.064444444444445e-06,
"loss": 0.1464,
"step": 3625
},
{
"epoch": 1.07,
"learning_rate": 3.008888888888889e-06,
"loss": 0.1027,
"step": 3650
},
{
"epoch": 1.08,
"learning_rate": 2.9533333333333336e-06,
"loss": 0.1032,
"step": 3675
},
{
"epoch": 1.08,
"learning_rate": 2.8977777777777785e-06,
"loss": 0.0937,
"step": 3700
},
{
"epoch": 1.09,
"learning_rate": 2.8422222222222225e-06,
"loss": 0.0975,
"step": 3725
},
{
"epoch": 1.09,
"learning_rate": 2.786666666666667e-06,
"loss": 0.0922,
"step": 3750
},
{
"epoch": 1.1,
"learning_rate": 2.7311111111111114e-06,
"loss": 0.1045,
"step": 3775
},
{
"epoch": 1.1,
"learning_rate": 2.675555555555556e-06,
"loss": 0.0816,
"step": 3800
},
{
"epoch": 1.11,
"learning_rate": 2.6200000000000003e-06,
"loss": 0.1011,
"step": 3825
},
{
"epoch": 1.11,
"learning_rate": 2.5644444444444444e-06,
"loss": 0.0883,
"step": 3850
},
{
"epoch": 1.12,
"learning_rate": 2.5088888888888892e-06,
"loss": 0.0872,
"step": 3875
},
{
"epoch": 1.12,
"learning_rate": 2.4533333333333333e-06,
"loss": 0.1131,
"step": 3900
},
{
"epoch": 1.13,
"learning_rate": 2.397777777777778e-06,
"loss": 0.1031,
"step": 3925
},
{
"epoch": 1.13,
"learning_rate": 2.342222222222222e-06,
"loss": 0.0985,
"step": 3950
},
{
"epoch": 1.14,
"learning_rate": 2.2866666666666667e-06,
"loss": 0.1057,
"step": 3975
},
{
"epoch": 1.14,
"learning_rate": 2.2311111111111115e-06,
"loss": 0.0828,
"step": 4000
},
{
"epoch": 1.14,
"eval_loss": 0.23880085349082947,
"eval_runtime": 1813.5298,
"eval_samples_per_second": 3.634,
"eval_steps_per_second": 0.454,
"eval_wer": 15.300336182105392,
"step": 4000
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.632763731050496e+19,
"trial_name": null,
"trial_params": null
}