whisper-small-zh_tw / trainer_state.json
Jingmiao's picture
End of training
b926eb1
raw
history blame
25.9 kB
{
"best_metric": 41.96519959058342,
"best_model_checkpoint": "./checkpoint-1000",
"epoch": 33.0232,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.2000000000000006e-07,
"loss": 0.475,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.200000000000001e-07,
"loss": 0.3924,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.42e-06,
"loss": 0.2843,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.9200000000000003e-06,
"loss": 0.2428,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.42e-06,
"loss": 0.2466,
"step": 125
},
{
"epoch": 1.0,
"learning_rate": 2.92e-06,
"loss": 0.229,
"step": 150
},
{
"epoch": 1.01,
"learning_rate": 3.4200000000000007e-06,
"loss": 0.1966,
"step": 175
},
{
"epoch": 1.01,
"learning_rate": 3.920000000000001e-06,
"loss": 0.18,
"step": 200
},
{
"epoch": 1.02,
"learning_rate": 4.42e-06,
"loss": 0.1614,
"step": 225
},
{
"epoch": 1.02,
"learning_rate": 4.92e-06,
"loss": 0.1515,
"step": 250
},
{
"epoch": 1.03,
"learning_rate": 5.420000000000001e-06,
"loss": 0.1644,
"step": 275
},
{
"epoch": 2.0,
"learning_rate": 5.92e-06,
"loss": 0.1457,
"step": 300
},
{
"epoch": 2.01,
"learning_rate": 6.42e-06,
"loss": 0.1163,
"step": 325
},
{
"epoch": 2.01,
"learning_rate": 6.92e-06,
"loss": 0.106,
"step": 350
},
{
"epoch": 2.02,
"learning_rate": 7.420000000000001e-06,
"loss": 0.0945,
"step": 375
},
{
"epoch": 2.02,
"learning_rate": 7.92e-06,
"loss": 0.0906,
"step": 400
},
{
"epoch": 2.03,
"learning_rate": 8.42e-06,
"loss": 0.0901,
"step": 425
},
{
"epoch": 3.0,
"learning_rate": 8.920000000000001e-06,
"loss": 0.0784,
"step": 450
},
{
"epoch": 3.01,
"learning_rate": 9.42e-06,
"loss": 0.0559,
"step": 475
},
{
"epoch": 3.01,
"learning_rate": 9.920000000000002e-06,
"loss": 0.0509,
"step": 500
},
{
"epoch": 3.02,
"learning_rate": 9.953333333333333e-06,
"loss": 0.0441,
"step": 525
},
{
"epoch": 3.02,
"learning_rate": 9.89777777777778e-06,
"loss": 0.0422,
"step": 550
},
{
"epoch": 3.03,
"learning_rate": 9.842222222222223e-06,
"loss": 0.0426,
"step": 575
},
{
"epoch": 4.0,
"learning_rate": 9.786666666666667e-06,
"loss": 0.0342,
"step": 600
},
{
"epoch": 4.01,
"learning_rate": 9.731111111111113e-06,
"loss": 0.0261,
"step": 625
},
{
"epoch": 4.01,
"learning_rate": 9.675555555555555e-06,
"loss": 0.0208,
"step": 650
},
{
"epoch": 4.02,
"learning_rate": 9.620000000000001e-06,
"loss": 0.0195,
"step": 675
},
{
"epoch": 4.02,
"learning_rate": 9.564444444444445e-06,
"loss": 0.0173,
"step": 700
},
{
"epoch": 4.03,
"learning_rate": 9.508888888888889e-06,
"loss": 0.019,
"step": 725
},
{
"epoch": 5.0,
"learning_rate": 9.453333333333335e-06,
"loss": 0.0151,
"step": 750
},
{
"epoch": 5.01,
"learning_rate": 9.397777777777779e-06,
"loss": 0.0108,
"step": 775
},
{
"epoch": 5.01,
"learning_rate": 9.342222222222223e-06,
"loss": 0.0112,
"step": 800
},
{
"epoch": 5.02,
"learning_rate": 9.286666666666667e-06,
"loss": 0.0098,
"step": 825
},
{
"epoch": 5.02,
"learning_rate": 9.231111111111111e-06,
"loss": 0.0104,
"step": 850
},
{
"epoch": 5.03,
"learning_rate": 9.175555555555557e-06,
"loss": 0.0101,
"step": 875
},
{
"epoch": 6.0,
"learning_rate": 9.12e-06,
"loss": 0.007,
"step": 900
},
{
"epoch": 6.01,
"learning_rate": 9.064444444444447e-06,
"loss": 0.005,
"step": 925
},
{
"epoch": 6.01,
"learning_rate": 9.008888888888889e-06,
"loss": 0.006,
"step": 950
},
{
"epoch": 6.02,
"learning_rate": 8.953333333333335e-06,
"loss": 0.0059,
"step": 975
},
{
"epoch": 6.02,
"learning_rate": 8.897777777777779e-06,
"loss": 0.0049,
"step": 1000
},
{
"epoch": 6.02,
"eval_loss": 0.2282676249742508,
"eval_runtime": 949.6353,
"eval_samples_per_second": 4.959,
"eval_steps_per_second": 0.156,
"eval_wer": 41.96519959058342,
"step": 1000
},
{
"epoch": 6.03,
"learning_rate": 8.842222222222223e-06,
"loss": 0.0062,
"step": 1025
},
{
"epoch": 7.0,
"learning_rate": 8.786666666666668e-06,
"loss": 0.0043,
"step": 1050
},
{
"epoch": 7.01,
"learning_rate": 8.73111111111111e-06,
"loss": 0.0046,
"step": 1075
},
{
"epoch": 7.01,
"learning_rate": 8.675555555555556e-06,
"loss": 0.0029,
"step": 1100
},
{
"epoch": 7.02,
"learning_rate": 8.62e-06,
"loss": 0.0049,
"step": 1125
},
{
"epoch": 7.02,
"learning_rate": 8.564444444444445e-06,
"loss": 0.0036,
"step": 1150
},
{
"epoch": 7.03,
"learning_rate": 8.50888888888889e-06,
"loss": 0.0045,
"step": 1175
},
{
"epoch": 8.0,
"learning_rate": 8.453333333333334e-06,
"loss": 0.0031,
"step": 1200
},
{
"epoch": 8.01,
"learning_rate": 8.397777777777778e-06,
"loss": 0.0028,
"step": 1225
},
{
"epoch": 8.01,
"learning_rate": 8.342222222222222e-06,
"loss": 0.0024,
"step": 1250
},
{
"epoch": 8.02,
"learning_rate": 8.286666666666668e-06,
"loss": 0.0036,
"step": 1275
},
{
"epoch": 8.02,
"learning_rate": 8.231111111111112e-06,
"loss": 0.003,
"step": 1300
},
{
"epoch": 8.03,
"learning_rate": 8.175555555555556e-06,
"loss": 0.003,
"step": 1325
},
{
"epoch": 9.0,
"learning_rate": 8.120000000000002e-06,
"loss": 0.0018,
"step": 1350
},
{
"epoch": 9.01,
"learning_rate": 8.064444444444444e-06,
"loss": 0.0029,
"step": 1375
},
{
"epoch": 9.01,
"learning_rate": 8.00888888888889e-06,
"loss": 0.0017,
"step": 1400
},
{
"epoch": 9.02,
"learning_rate": 7.953333333333334e-06,
"loss": 0.0017,
"step": 1425
},
{
"epoch": 9.02,
"learning_rate": 7.897777777777778e-06,
"loss": 0.0018,
"step": 1450
},
{
"epoch": 9.03,
"learning_rate": 7.842222222222224e-06,
"loss": 0.0022,
"step": 1475
},
{
"epoch": 10.0,
"learning_rate": 7.786666666666666e-06,
"loss": 0.0021,
"step": 1500
},
{
"epoch": 10.01,
"learning_rate": 7.731111111111112e-06,
"loss": 0.0018,
"step": 1525
},
{
"epoch": 10.01,
"learning_rate": 7.675555555555556e-06,
"loss": 0.0012,
"step": 1550
},
{
"epoch": 10.02,
"learning_rate": 7.620000000000001e-06,
"loss": 0.0011,
"step": 1575
},
{
"epoch": 10.02,
"learning_rate": 7.564444444444446e-06,
"loss": 0.0012,
"step": 1600
},
{
"epoch": 10.03,
"learning_rate": 7.50888888888889e-06,
"loss": 0.0014,
"step": 1625
},
{
"epoch": 11.0,
"learning_rate": 7.453333333333334e-06,
"loss": 0.0011,
"step": 1650
},
{
"epoch": 11.01,
"learning_rate": 7.3977777777777786e-06,
"loss": 0.0015,
"step": 1675
},
{
"epoch": 11.01,
"learning_rate": 7.342222222222223e-06,
"loss": 0.0015,
"step": 1700
},
{
"epoch": 11.02,
"learning_rate": 7.2866666666666675e-06,
"loss": 0.001,
"step": 1725
},
{
"epoch": 11.02,
"learning_rate": 7.231111111111112e-06,
"loss": 0.0009,
"step": 1750
},
{
"epoch": 11.03,
"learning_rate": 7.1755555555555556e-06,
"loss": 0.001,
"step": 1775
},
{
"epoch": 12.0,
"learning_rate": 7.1200000000000004e-06,
"loss": 0.0009,
"step": 1800
},
{
"epoch": 12.01,
"learning_rate": 7.0644444444444445e-06,
"loss": 0.0015,
"step": 1825
},
{
"epoch": 12.01,
"learning_rate": 7.008888888888889e-06,
"loss": 0.0008,
"step": 1850
},
{
"epoch": 12.02,
"learning_rate": 6.953333333333334e-06,
"loss": 0.0008,
"step": 1875
},
{
"epoch": 12.02,
"learning_rate": 6.897777777777779e-06,
"loss": 0.0008,
"step": 1900
},
{
"epoch": 13.0,
"learning_rate": 6.842222222222222e-06,
"loss": 0.0008,
"step": 1925
},
{
"epoch": 13.01,
"learning_rate": 6.786666666666667e-06,
"loss": 0.0007,
"step": 1950
},
{
"epoch": 13.01,
"learning_rate": 6.731111111111111e-06,
"loss": 0.0007,
"step": 1975
},
{
"epoch": 13.02,
"learning_rate": 6.675555555555556e-06,
"loss": 0.0008,
"step": 2000
},
{
"epoch": 13.02,
"eval_loss": 0.2556239664554596,
"eval_runtime": 960.6526,
"eval_samples_per_second": 4.902,
"eval_steps_per_second": 0.154,
"eval_wer": 42.02661207778915,
"step": 2000
},
{
"epoch": 13.02,
"learning_rate": 6.620000000000001e-06,
"loss": 0.0007,
"step": 2025
},
{
"epoch": 13.03,
"learning_rate": 6.564444444444446e-06,
"loss": 0.0007,
"step": 2050
},
{
"epoch": 14.0,
"learning_rate": 6.508888888888889e-06,
"loss": 0.0007,
"step": 2075
},
{
"epoch": 14.01,
"learning_rate": 6.453333333333334e-06,
"loss": 0.0006,
"step": 2100
},
{
"epoch": 14.01,
"learning_rate": 6.397777777777778e-06,
"loss": 0.0011,
"step": 2125
},
{
"epoch": 14.02,
"learning_rate": 6.342222222222223e-06,
"loss": 0.0006,
"step": 2150
},
{
"epoch": 14.02,
"learning_rate": 6.286666666666668e-06,
"loss": 0.0006,
"step": 2175
},
{
"epoch": 14.03,
"learning_rate": 6.231111111111111e-06,
"loss": 0.0007,
"step": 2200
},
{
"epoch": 15.0,
"learning_rate": 6.175555555555556e-06,
"loss": 0.0006,
"step": 2225
},
{
"epoch": 15.01,
"learning_rate": 6.120000000000001e-06,
"loss": 0.0006,
"step": 2250
},
{
"epoch": 15.01,
"learning_rate": 6.064444444444445e-06,
"loss": 0.001,
"step": 2275
},
{
"epoch": 15.02,
"learning_rate": 6.00888888888889e-06,
"loss": 0.0006,
"step": 2300
},
{
"epoch": 15.02,
"learning_rate": 5.9533333333333345e-06,
"loss": 0.0006,
"step": 2325
},
{
"epoch": 15.03,
"learning_rate": 5.897777777777778e-06,
"loss": 0.0006,
"step": 2350
},
{
"epoch": 16.0,
"learning_rate": 5.8422222222222226e-06,
"loss": 0.0006,
"step": 2375
},
{
"epoch": 16.01,
"learning_rate": 5.7866666666666674e-06,
"loss": 0.0005,
"step": 2400
},
{
"epoch": 16.01,
"learning_rate": 5.7311111111111115e-06,
"loss": 0.0006,
"step": 2425
},
{
"epoch": 16.02,
"learning_rate": 5.675555555555556e-06,
"loss": 0.0005,
"step": 2450
},
{
"epoch": 16.02,
"learning_rate": 5.620000000000001e-06,
"loss": 0.0005,
"step": 2475
},
{
"epoch": 16.03,
"learning_rate": 5.5644444444444444e-06,
"loss": 0.0005,
"step": 2500
},
{
"epoch": 17.0,
"learning_rate": 5.508888888888889e-06,
"loss": 0.0005,
"step": 2525
},
{
"epoch": 17.01,
"learning_rate": 5.453333333333334e-06,
"loss": 0.0005,
"step": 2550
},
{
"epoch": 17.01,
"learning_rate": 5.397777777777778e-06,
"loss": 0.0005,
"step": 2575
},
{
"epoch": 17.02,
"learning_rate": 5.342222222222223e-06,
"loss": 0.0005,
"step": 2600
},
{
"epoch": 17.02,
"learning_rate": 5.286666666666666e-06,
"loss": 0.0005,
"step": 2625
},
{
"epoch": 17.03,
"learning_rate": 5.231111111111111e-06,
"loss": 0.0005,
"step": 2650
},
{
"epoch": 18.0,
"learning_rate": 5.175555555555556e-06,
"loss": 0.0005,
"step": 2675
},
{
"epoch": 18.01,
"learning_rate": 5.12e-06,
"loss": 0.0005,
"step": 2700
},
{
"epoch": 18.01,
"learning_rate": 5.064444444444445e-06,
"loss": 0.0004,
"step": 2725
},
{
"epoch": 18.02,
"learning_rate": 5.00888888888889e-06,
"loss": 0.0004,
"step": 2750
},
{
"epoch": 18.02,
"learning_rate": 4.953333333333334e-06,
"loss": 0.0004,
"step": 2775
},
{
"epoch": 18.03,
"learning_rate": 4.897777777777778e-06,
"loss": 0.0005,
"step": 2800
},
{
"epoch": 19.0,
"learning_rate": 4.842222222222223e-06,
"loss": 0.0004,
"step": 2825
},
{
"epoch": 19.01,
"learning_rate": 4.786666666666667e-06,
"loss": 0.0004,
"step": 2850
},
{
"epoch": 19.01,
"learning_rate": 4.731111111111112e-06,
"loss": 0.0004,
"step": 2875
},
{
"epoch": 19.02,
"learning_rate": 4.675555555555556e-06,
"loss": 0.0004,
"step": 2900
},
{
"epoch": 19.02,
"learning_rate": 4.620000000000001e-06,
"loss": 0.0004,
"step": 2925
},
{
"epoch": 19.03,
"learning_rate": 4.564444444444445e-06,
"loss": 0.0004,
"step": 2950
},
{
"epoch": 20.0,
"learning_rate": 4.50888888888889e-06,
"loss": 0.0004,
"step": 2975
},
{
"epoch": 20.01,
"learning_rate": 4.453333333333334e-06,
"loss": 0.0004,
"step": 3000
},
{
"epoch": 20.01,
"eval_loss": 0.26901087164878845,
"eval_runtime": 957.9477,
"eval_samples_per_second": 4.916,
"eval_steps_per_second": 0.154,
"eval_wer": 42.41555783009212,
"step": 3000
},
{
"epoch": 20.01,
"learning_rate": 4.397777777777778e-06,
"loss": 0.0004,
"step": 3025
},
{
"epoch": 20.02,
"learning_rate": 4.3422222222222225e-06,
"loss": 0.0004,
"step": 3050
},
{
"epoch": 20.02,
"learning_rate": 4.2866666666666666e-06,
"loss": 0.0004,
"step": 3075
},
{
"epoch": 20.03,
"learning_rate": 4.2311111111111114e-06,
"loss": 0.0004,
"step": 3100
},
{
"epoch": 21.0,
"learning_rate": 4.175555555555556e-06,
"loss": 0.0004,
"step": 3125
},
{
"epoch": 21.01,
"learning_rate": 4.12e-06,
"loss": 0.0004,
"step": 3150
},
{
"epoch": 21.01,
"learning_rate": 4.064444444444444e-06,
"loss": 0.0004,
"step": 3175
},
{
"epoch": 21.02,
"learning_rate": 4.008888888888889e-06,
"loss": 0.0004,
"step": 3200
},
{
"epoch": 21.02,
"learning_rate": 3.953333333333333e-06,
"loss": 0.0004,
"step": 3225
},
{
"epoch": 21.03,
"learning_rate": 3.897777777777778e-06,
"loss": 0.0004,
"step": 3250
},
{
"epoch": 22.0,
"learning_rate": 3.842222222222223e-06,
"loss": 0.0003,
"step": 3275
},
{
"epoch": 22.01,
"learning_rate": 3.7866666666666667e-06,
"loss": 0.0003,
"step": 3300
},
{
"epoch": 22.01,
"learning_rate": 3.7311111111111116e-06,
"loss": 0.0003,
"step": 3325
},
{
"epoch": 22.02,
"learning_rate": 3.675555555555556e-06,
"loss": 0.0003,
"step": 3350
},
{
"epoch": 22.02,
"learning_rate": 3.62e-06,
"loss": 0.0003,
"step": 3375
},
{
"epoch": 22.03,
"learning_rate": 3.564444444444445e-06,
"loss": 0.0003,
"step": 3400
},
{
"epoch": 23.0,
"learning_rate": 3.508888888888889e-06,
"loss": 0.0003,
"step": 3425
},
{
"epoch": 23.01,
"learning_rate": 3.4533333333333334e-06,
"loss": 0.0003,
"step": 3450
},
{
"epoch": 23.01,
"learning_rate": 3.3977777777777783e-06,
"loss": 0.0003,
"step": 3475
},
{
"epoch": 23.02,
"learning_rate": 3.3422222222222224e-06,
"loss": 0.0003,
"step": 3500
},
{
"epoch": 23.02,
"learning_rate": 3.286666666666667e-06,
"loss": 0.0003,
"step": 3525
},
{
"epoch": 23.03,
"learning_rate": 3.2311111111111117e-06,
"loss": 0.0003,
"step": 3550
},
{
"epoch": 24.0,
"learning_rate": 3.1755555555555557e-06,
"loss": 0.0003,
"step": 3575
},
{
"epoch": 24.01,
"learning_rate": 3.12e-06,
"loss": 0.0003,
"step": 3600
},
{
"epoch": 24.01,
"learning_rate": 3.064444444444445e-06,
"loss": 0.0003,
"step": 3625
},
{
"epoch": 24.02,
"learning_rate": 3.008888888888889e-06,
"loss": 0.0003,
"step": 3650
},
{
"epoch": 24.02,
"learning_rate": 2.9533333333333336e-06,
"loss": 0.0003,
"step": 3675
},
{
"epoch": 24.03,
"learning_rate": 2.8977777777777785e-06,
"loss": 0.0003,
"step": 3700
},
{
"epoch": 25.0,
"learning_rate": 2.8422222222222225e-06,
"loss": 0.0003,
"step": 3725
},
{
"epoch": 25.01,
"learning_rate": 2.786666666666667e-06,
"loss": 0.0003,
"step": 3750
},
{
"epoch": 25.02,
"learning_rate": 2.7311111111111114e-06,
"loss": 0.0003,
"step": 3775
},
{
"epoch": 25.02,
"learning_rate": 2.675555555555556e-06,
"loss": 0.0003,
"step": 3800
},
{
"epoch": 25.02,
"learning_rate": 2.6200000000000003e-06,
"loss": 0.0003,
"step": 3825
},
{
"epoch": 26.0,
"learning_rate": 2.5644444444444444e-06,
"loss": 0.0003,
"step": 3850
},
{
"epoch": 26.01,
"learning_rate": 2.5088888888888892e-06,
"loss": 0.0003,
"step": 3875
},
{
"epoch": 26.01,
"learning_rate": 2.4533333333333333e-06,
"loss": 0.0003,
"step": 3900
},
{
"epoch": 26.02,
"learning_rate": 2.397777777777778e-06,
"loss": 0.0003,
"step": 3925
},
{
"epoch": 26.02,
"learning_rate": 2.342222222222222e-06,
"loss": 0.0003,
"step": 3950
},
{
"epoch": 26.03,
"learning_rate": 2.2866666666666667e-06,
"loss": 0.0003,
"step": 3975
},
{
"epoch": 27.0,
"learning_rate": 2.2311111111111115e-06,
"loss": 0.0003,
"step": 4000
},
{
"epoch": 27.0,
"eval_loss": 0.2788228988647461,
"eval_runtime": 946.3161,
"eval_samples_per_second": 4.976,
"eval_steps_per_second": 0.156,
"eval_wer": 42.784032753326514,
"step": 4000
},
{
"epoch": 27.01,
"learning_rate": 2.1755555555555556e-06,
"loss": 0.0003,
"step": 4025
},
{
"epoch": 27.01,
"learning_rate": 2.12e-06,
"loss": 0.0003,
"step": 4050
},
{
"epoch": 27.02,
"learning_rate": 2.064444444444445e-06,
"loss": 0.0003,
"step": 4075
},
{
"epoch": 27.02,
"learning_rate": 2.008888888888889e-06,
"loss": 0.0003,
"step": 4100
},
{
"epoch": 27.03,
"learning_rate": 1.9533333333333334e-06,
"loss": 0.0003,
"step": 4125
},
{
"epoch": 28.0,
"learning_rate": 1.8977777777777779e-06,
"loss": 0.0003,
"step": 4150
},
{
"epoch": 28.01,
"learning_rate": 1.8422222222222225e-06,
"loss": 0.0003,
"step": 4175
},
{
"epoch": 28.01,
"learning_rate": 1.7866666666666668e-06,
"loss": 0.0003,
"step": 4200
},
{
"epoch": 28.02,
"learning_rate": 1.7311111111111112e-06,
"loss": 0.0003,
"step": 4225
},
{
"epoch": 28.02,
"learning_rate": 1.675555555555556e-06,
"loss": 0.0003,
"step": 4250
},
{
"epoch": 28.03,
"learning_rate": 1.6200000000000002e-06,
"loss": 0.0003,
"step": 4275
},
{
"epoch": 29.0,
"learning_rate": 1.5644444444444446e-06,
"loss": 0.0003,
"step": 4300
},
{
"epoch": 29.01,
"learning_rate": 1.5088888888888889e-06,
"loss": 0.0002,
"step": 4325
},
{
"epoch": 29.01,
"learning_rate": 1.4533333333333335e-06,
"loss": 0.0002,
"step": 4350
},
{
"epoch": 29.02,
"learning_rate": 1.397777777777778e-06,
"loss": 0.0002,
"step": 4375
},
{
"epoch": 29.02,
"learning_rate": 1.3422222222222222e-06,
"loss": 0.0002,
"step": 4400
},
{
"epoch": 29.03,
"learning_rate": 1.286666666666667e-06,
"loss": 0.0003,
"step": 4425
},
{
"epoch": 30.0,
"learning_rate": 1.2311111111111112e-06,
"loss": 0.0002,
"step": 4450
},
{
"epoch": 30.01,
"learning_rate": 1.1755555555555556e-06,
"loss": 0.0002,
"step": 4475
},
{
"epoch": 30.01,
"learning_rate": 1.12e-06,
"loss": 0.0002,
"step": 4500
},
{
"epoch": 30.02,
"learning_rate": 1.0644444444444445e-06,
"loss": 0.0002,
"step": 4525
},
{
"epoch": 30.02,
"learning_rate": 1.008888888888889e-06,
"loss": 0.0002,
"step": 4550
},
{
"epoch": 30.03,
"learning_rate": 9.533333333333335e-07,
"loss": 0.0003,
"step": 4575
},
{
"epoch": 31.0,
"learning_rate": 8.977777777777778e-07,
"loss": 0.0002,
"step": 4600
},
{
"epoch": 31.01,
"learning_rate": 8.422222222222224e-07,
"loss": 0.0002,
"step": 4625
},
{
"epoch": 31.01,
"learning_rate": 7.866666666666667e-07,
"loss": 0.0002,
"step": 4650
},
{
"epoch": 31.02,
"learning_rate": 7.311111111111112e-07,
"loss": 0.0002,
"step": 4675
},
{
"epoch": 31.02,
"learning_rate": 6.755555555555555e-07,
"loss": 0.0002,
"step": 4700
},
{
"epoch": 31.03,
"learning_rate": 6.200000000000001e-07,
"loss": 0.0002,
"step": 4725
},
{
"epoch": 32.0,
"learning_rate": 5.644444444444445e-07,
"loss": 0.0002,
"step": 4750
},
{
"epoch": 32.01,
"learning_rate": 5.088888888888889e-07,
"loss": 0.0002,
"step": 4775
},
{
"epoch": 32.01,
"learning_rate": 4.533333333333334e-07,
"loss": 0.0002,
"step": 4800
},
{
"epoch": 32.02,
"learning_rate": 3.9777777777777783e-07,
"loss": 0.0002,
"step": 4825
},
{
"epoch": 32.02,
"learning_rate": 3.422222222222223e-07,
"loss": 0.0002,
"step": 4850
},
{
"epoch": 32.03,
"learning_rate": 2.866666666666667e-07,
"loss": 0.0002,
"step": 4875
},
{
"epoch": 33.0,
"learning_rate": 2.3111111111111112e-07,
"loss": 0.0002,
"step": 4900
},
{
"epoch": 33.01,
"learning_rate": 1.7555555555555558e-07,
"loss": 0.0002,
"step": 4925
},
{
"epoch": 33.01,
"learning_rate": 1.2000000000000002e-07,
"loss": 0.0002,
"step": 4950
},
{
"epoch": 33.02,
"learning_rate": 6.444444444444445e-08,
"loss": 0.0002,
"step": 4975
},
{
"epoch": 33.02,
"learning_rate": 8.88888888888889e-09,
"loss": 0.0002,
"step": 5000
},
{
"epoch": 33.02,
"eval_loss": 0.2825862765312195,
"eval_runtime": 944.9056,
"eval_samples_per_second": 4.984,
"eval_steps_per_second": 0.157,
"eval_wer": 43.02968270214944,
"step": 5000
},
{
"epoch": 33.02,
"step": 5000,
"total_flos": 9.183306891902976e+19,
"train_loss": 0.020219367493875325,
"train_runtime": 37243.6262,
"train_samples_per_second": 8.592,
"train_steps_per_second": 0.134
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 9.183306891902976e+19,
"trial_name": null,
"trial_params": null
}