1271 lines
25 KiB
JSON
1271 lines
25 KiB
JSON
{
|
|
"best_metric": 0.02823920265780731,
|
|
"best_model_checkpoint": "./whisper-large-v2-Ko/checkpoint-4000",
|
|
"epoch": 50.0,
|
|
"global_step": 5000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.6000000000000004e-07,
|
|
"loss": 0.5495,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 9.600000000000001e-07,
|
|
"loss": 0.3767,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1.46e-06,
|
|
"loss": 0.1742,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.9600000000000003e-06,
|
|
"loss": 0.1477,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 2.4400000000000004e-06,
|
|
"loss": 0.1342,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 2.9400000000000002e-06,
|
|
"loss": 0.1284,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 3.4200000000000007e-06,
|
|
"loss": 0.121,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 3.920000000000001e-06,
|
|
"loss": 0.1195,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 4.42e-06,
|
|
"loss": 0.1046,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 4.92e-06,
|
|
"loss": 0.1032,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 5.420000000000001e-06,
|
|
"loss": 0.0975,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 5.92e-06,
|
|
"loss": 0.094,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 3.25,
|
|
"learning_rate": 6.42e-06,
|
|
"loss": 0.0852,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 3.5,
|
|
"learning_rate": 6.92e-06,
|
|
"loss": 0.0826,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 3.75,
|
|
"learning_rate": 7.420000000000001e-06,
|
|
"loss": 0.0835,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"learning_rate": 7.92e-06,
|
|
"loss": 0.0854,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 4.25,
|
|
"learning_rate": 8.42e-06,
|
|
"loss": 0.0731,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 4.5,
|
|
"learning_rate": 8.920000000000001e-06,
|
|
"loss": 0.0714,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 4.75,
|
|
"learning_rate": 9.42e-06,
|
|
"loss": 0.0738,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"learning_rate": 9.920000000000002e-06,
|
|
"loss": 0.0679,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 5.25,
|
|
"learning_rate": 9.953333333333333e-06,
|
|
"loss": 0.0655,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 5.5,
|
|
"learning_rate": 9.89777777777778e-06,
|
|
"loss": 0.0622,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 5.75,
|
|
"learning_rate": 9.842222222222223e-06,
|
|
"loss": 0.0601,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"learning_rate": 9.786666666666667e-06,
|
|
"loss": 0.059,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 6.25,
|
|
"learning_rate": 9.731111111111113e-06,
|
|
"loss": 0.0552,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 6.5,
|
|
"learning_rate": 9.675555555555555e-06,
|
|
"loss": 0.0515,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 6.75,
|
|
"learning_rate": 9.620000000000001e-06,
|
|
"loss": 0.0501,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"learning_rate": 9.564444444444445e-06,
|
|
"loss": 0.0493,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 7.25,
|
|
"learning_rate": 9.508888888888889e-06,
|
|
"loss": 0.0439,
|
|
"step": 725
|
|
},
|
|
{
|
|
"epoch": 7.5,
|
|
"learning_rate": 9.453333333333335e-06,
|
|
"loss": 0.0446,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 7.75,
|
|
"learning_rate": 9.397777777777779e-06,
|
|
"loss": 0.0399,
|
|
"step": 775
|
|
},
|
|
{
|
|
"epoch": 8.0,
|
|
"learning_rate": 9.342222222222223e-06,
|
|
"loss": 0.0422,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 8.25,
|
|
"learning_rate": 9.286666666666667e-06,
|
|
"loss": 0.0385,
|
|
"step": 825
|
|
},
|
|
{
|
|
"epoch": 8.5,
|
|
"learning_rate": 9.231111111111111e-06,
|
|
"loss": 0.0384,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 8.75,
|
|
"learning_rate": 9.175555555555557e-06,
|
|
"loss": 0.0371,
|
|
"step": 875
|
|
},
|
|
{
|
|
"epoch": 9.0,
|
|
"learning_rate": 9.12e-06,
|
|
"loss": 0.0342,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 9.25,
|
|
"learning_rate": 9.064444444444447e-06,
|
|
"loss": 0.0312,
|
|
"step": 925
|
|
},
|
|
{
|
|
"epoch": 9.5,
|
|
"learning_rate": 9.008888888888889e-06,
|
|
"loss": 0.0329,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 9.75,
|
|
"learning_rate": 8.953333333333335e-06,
|
|
"loss": 0.0329,
|
|
"step": 975
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"learning_rate": 8.897777777777779e-06,
|
|
"loss": 0.0299,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"eval_loss": 0.07447270303964615,
|
|
"eval_runtime": 37.8186,
|
|
"eval_samples_per_second": 12.058,
|
|
"eval_steps_per_second": 0.132,
|
|
"eval_wer": 0.04469948655995168,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 10.25,
|
|
"learning_rate": 8.842222222222223e-06,
|
|
"loss": 0.0298,
|
|
"step": 1025
|
|
},
|
|
{
|
|
"epoch": 10.5,
|
|
"learning_rate": 8.786666666666668e-06,
|
|
"loss": 0.0283,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 10.75,
|
|
"learning_rate": 8.73111111111111e-06,
|
|
"loss": 0.026,
|
|
"step": 1075
|
|
},
|
|
{
|
|
"epoch": 11.0,
|
|
"learning_rate": 8.675555555555556e-06,
|
|
"loss": 0.0267,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 11.25,
|
|
"learning_rate": 8.62e-06,
|
|
"loss": 0.0232,
|
|
"step": 1125
|
|
},
|
|
{
|
|
"epoch": 11.5,
|
|
"learning_rate": 8.564444444444445e-06,
|
|
"loss": 0.0247,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 11.75,
|
|
"learning_rate": 8.50888888888889e-06,
|
|
"loss": 0.024,
|
|
"step": 1175
|
|
},
|
|
{
|
|
"epoch": 12.0,
|
|
"learning_rate": 8.453333333333334e-06,
|
|
"loss": 0.0248,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 12.25,
|
|
"learning_rate": 8.397777777777778e-06,
|
|
"loss": 0.0224,
|
|
"step": 1225
|
|
},
|
|
{
|
|
"epoch": 12.5,
|
|
"learning_rate": 8.342222222222222e-06,
|
|
"loss": 0.0214,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 12.75,
|
|
"learning_rate": 8.286666666666668e-06,
|
|
"loss": 0.021,
|
|
"step": 1275
|
|
},
|
|
{
|
|
"epoch": 13.0,
|
|
"learning_rate": 8.231111111111112e-06,
|
|
"loss": 0.022,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 13.25,
|
|
"learning_rate": 8.175555555555556e-06,
|
|
"loss": 0.0189,
|
|
"step": 1325
|
|
},
|
|
{
|
|
"epoch": 13.5,
|
|
"learning_rate": 8.120000000000002e-06,
|
|
"loss": 0.0197,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 13.75,
|
|
"learning_rate": 8.064444444444444e-06,
|
|
"loss": 0.0193,
|
|
"step": 1375
|
|
},
|
|
{
|
|
"epoch": 14.0,
|
|
"learning_rate": 8.00888888888889e-06,
|
|
"loss": 0.0186,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 14.25,
|
|
"learning_rate": 7.953333333333334e-06,
|
|
"loss": 0.0167,
|
|
"step": 1425
|
|
},
|
|
{
|
|
"epoch": 14.5,
|
|
"learning_rate": 7.897777777777778e-06,
|
|
"loss": 0.0168,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 14.75,
|
|
"learning_rate": 7.842222222222224e-06,
|
|
"loss": 0.0157,
|
|
"step": 1475
|
|
},
|
|
{
|
|
"epoch": 15.0,
|
|
"learning_rate": 7.786666666666666e-06,
|
|
"loss": 0.0169,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 15.25,
|
|
"learning_rate": 7.731111111111112e-06,
|
|
"loss": 0.0144,
|
|
"step": 1525
|
|
},
|
|
{
|
|
"epoch": 15.5,
|
|
"learning_rate": 7.675555555555556e-06,
|
|
"loss": 0.0145,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 15.75,
|
|
"learning_rate": 7.620000000000001e-06,
|
|
"loss": 0.0145,
|
|
"step": 1575
|
|
},
|
|
{
|
|
"epoch": 16.0,
|
|
"learning_rate": 7.564444444444446e-06,
|
|
"loss": 0.014,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 16.25,
|
|
"learning_rate": 7.50888888888889e-06,
|
|
"loss": 0.014,
|
|
"step": 1625
|
|
},
|
|
{
|
|
"epoch": 16.5,
|
|
"learning_rate": 7.453333333333334e-06,
|
|
"loss": 0.0122,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 16.75,
|
|
"learning_rate": 7.3977777777777786e-06,
|
|
"loss": 0.0136,
|
|
"step": 1675
|
|
},
|
|
{
|
|
"epoch": 17.0,
|
|
"learning_rate": 7.342222222222223e-06,
|
|
"loss": 0.0127,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 17.25,
|
|
"learning_rate": 7.2866666666666675e-06,
|
|
"loss": 0.0122,
|
|
"step": 1725
|
|
},
|
|
{
|
|
"epoch": 17.5,
|
|
"learning_rate": 7.231111111111112e-06,
|
|
"loss": 0.0108,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 17.75,
|
|
"learning_rate": 7.1755555555555556e-06,
|
|
"loss": 0.0121,
|
|
"step": 1775
|
|
},
|
|
{
|
|
"epoch": 18.0,
|
|
"learning_rate": 7.1200000000000004e-06,
|
|
"loss": 0.0107,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 18.25,
|
|
"learning_rate": 7.0644444444444445e-06,
|
|
"loss": 0.01,
|
|
"step": 1825
|
|
},
|
|
{
|
|
"epoch": 18.5,
|
|
"learning_rate": 7.008888888888889e-06,
|
|
"loss": 0.0106,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 18.75,
|
|
"learning_rate": 6.953333333333334e-06,
|
|
"loss": 0.0107,
|
|
"step": 1875
|
|
},
|
|
{
|
|
"epoch": 19.0,
|
|
"learning_rate": 6.897777777777779e-06,
|
|
"loss": 0.0105,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 19.25,
|
|
"learning_rate": 6.842222222222222e-06,
|
|
"loss": 0.0085,
|
|
"step": 1925
|
|
},
|
|
{
|
|
"epoch": 19.5,
|
|
"learning_rate": 6.786666666666667e-06,
|
|
"loss": 0.0102,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 19.75,
|
|
"learning_rate": 6.731111111111111e-06,
|
|
"loss": 0.009,
|
|
"step": 1975
|
|
},
|
|
{
|
|
"epoch": 20.0,
|
|
"learning_rate": 6.675555555555556e-06,
|
|
"loss": 0.0085,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 20.0,
|
|
"eval_loss": 0.06077822297811508,
|
|
"eval_runtime": 36.8719,
|
|
"eval_samples_per_second": 12.367,
|
|
"eval_steps_per_second": 0.136,
|
|
"eval_wer": 0.035336756266988824,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 20.25,
|
|
"learning_rate": 6.620000000000001e-06,
|
|
"loss": 0.0091,
|
|
"step": 2025
|
|
},
|
|
{
|
|
"epoch": 20.5,
|
|
"learning_rate": 6.564444444444446e-06,
|
|
"loss": 0.0093,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 20.75,
|
|
"learning_rate": 6.508888888888889e-06,
|
|
"loss": 0.0081,
|
|
"step": 2075
|
|
},
|
|
{
|
|
"epoch": 21.0,
|
|
"learning_rate": 6.453333333333334e-06,
|
|
"loss": 0.0089,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 21.25,
|
|
"learning_rate": 6.397777777777778e-06,
|
|
"loss": 0.0078,
|
|
"step": 2125
|
|
},
|
|
{
|
|
"epoch": 21.5,
|
|
"learning_rate": 6.342222222222223e-06,
|
|
"loss": 0.0086,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 21.75,
|
|
"learning_rate": 6.286666666666668e-06,
|
|
"loss": 0.008,
|
|
"step": 2175
|
|
},
|
|
{
|
|
"epoch": 22.0,
|
|
"learning_rate": 6.231111111111111e-06,
|
|
"loss": 0.0076,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 22.25,
|
|
"learning_rate": 6.175555555555556e-06,
|
|
"loss": 0.006,
|
|
"step": 2225
|
|
},
|
|
{
|
|
"epoch": 22.5,
|
|
"learning_rate": 6.120000000000001e-06,
|
|
"loss": 0.0063,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 22.75,
|
|
"learning_rate": 6.064444444444445e-06,
|
|
"loss": 0.0068,
|
|
"step": 2275
|
|
},
|
|
{
|
|
"epoch": 23.0,
|
|
"learning_rate": 6.00888888888889e-06,
|
|
"loss": 0.0062,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 23.25,
|
|
"learning_rate": 5.9533333333333345e-06,
|
|
"loss": 0.0057,
|
|
"step": 2325
|
|
},
|
|
{
|
|
"epoch": 23.5,
|
|
"learning_rate": 5.897777777777778e-06,
|
|
"loss": 0.0053,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 23.75,
|
|
"learning_rate": 5.8422222222222226e-06,
|
|
"loss": 0.0062,
|
|
"step": 2375
|
|
},
|
|
{
|
|
"epoch": 24.0,
|
|
"learning_rate": 5.7866666666666674e-06,
|
|
"loss": 0.0066,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 24.25,
|
|
"learning_rate": 5.7311111111111115e-06,
|
|
"loss": 0.0051,
|
|
"step": 2425
|
|
},
|
|
{
|
|
"epoch": 24.5,
|
|
"learning_rate": 5.675555555555556e-06,
|
|
"loss": 0.0053,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 24.75,
|
|
"learning_rate": 5.620000000000001e-06,
|
|
"loss": 0.0051,
|
|
"step": 2475
|
|
},
|
|
{
|
|
"epoch": 25.0,
|
|
"learning_rate": 5.5644444444444444e-06,
|
|
"loss": 0.0055,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 25.25,
|
|
"learning_rate": 5.508888888888889e-06,
|
|
"loss": 0.0047,
|
|
"step": 2525
|
|
},
|
|
{
|
|
"epoch": 25.5,
|
|
"learning_rate": 5.453333333333334e-06,
|
|
"loss": 0.0054,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 25.75,
|
|
"learning_rate": 5.397777777777778e-06,
|
|
"loss": 0.0047,
|
|
"step": 2575
|
|
},
|
|
{
|
|
"epoch": 26.0,
|
|
"learning_rate": 5.342222222222223e-06,
|
|
"loss": 0.0047,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 26.25,
|
|
"learning_rate": 5.286666666666666e-06,
|
|
"loss": 0.0046,
|
|
"step": 2625
|
|
},
|
|
{
|
|
"epoch": 26.5,
|
|
"learning_rate": 5.231111111111111e-06,
|
|
"loss": 0.0046,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 26.75,
|
|
"learning_rate": 5.175555555555556e-06,
|
|
"loss": 0.0044,
|
|
"step": 2675
|
|
},
|
|
{
|
|
"epoch": 27.0,
|
|
"learning_rate": 5.12e-06,
|
|
"loss": 0.0056,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 27.25,
|
|
"learning_rate": 5.064444444444445e-06,
|
|
"loss": 0.0036,
|
|
"step": 2725
|
|
},
|
|
{
|
|
"epoch": 27.5,
|
|
"learning_rate": 5.00888888888889e-06,
|
|
"loss": 0.0038,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 27.75,
|
|
"learning_rate": 4.953333333333334e-06,
|
|
"loss": 0.004,
|
|
"step": 2775
|
|
},
|
|
{
|
|
"epoch": 28.0,
|
|
"learning_rate": 4.897777777777778e-06,
|
|
"loss": 0.0041,
|
|
"step": 2800
|
|
},
|
|
{
|
|
"epoch": 28.25,
|
|
"learning_rate": 4.842222222222223e-06,
|
|
"loss": 0.0037,
|
|
"step": 2825
|
|
},
|
|
{
|
|
"epoch": 28.5,
|
|
"learning_rate": 4.786666666666667e-06,
|
|
"loss": 0.0032,
|
|
"step": 2850
|
|
},
|
|
{
|
|
"epoch": 28.75,
|
|
"learning_rate": 4.731111111111112e-06,
|
|
"loss": 0.0036,
|
|
"step": 2875
|
|
},
|
|
{
|
|
"epoch": 29.0,
|
|
"learning_rate": 4.675555555555556e-06,
|
|
"loss": 0.0035,
|
|
"step": 2900
|
|
},
|
|
{
|
|
"epoch": 29.25,
|
|
"learning_rate": 4.620000000000001e-06,
|
|
"loss": 0.0033,
|
|
"step": 2925
|
|
},
|
|
{
|
|
"epoch": 29.5,
|
|
"learning_rate": 4.564444444444445e-06,
|
|
"loss": 0.0032,
|
|
"step": 2950
|
|
},
|
|
{
|
|
"epoch": 29.75,
|
|
"learning_rate": 4.50888888888889e-06,
|
|
"loss": 0.003,
|
|
"step": 2975
|
|
},
|
|
{
|
|
"epoch": 30.0,
|
|
"learning_rate": 4.453333333333334e-06,
|
|
"loss": 0.0036,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 30.0,
|
|
"eval_loss": 0.05926226079463959,
|
|
"eval_runtime": 43.1384,
|
|
"eval_samples_per_second": 10.571,
|
|
"eval_steps_per_second": 0.116,
|
|
"eval_wer": 0.030202355783751134,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 30.25,
|
|
"learning_rate": 4.397777777777778e-06,
|
|
"loss": 0.0031,
|
|
"step": 3025
|
|
},
|
|
{
|
|
"epoch": 30.5,
|
|
"learning_rate": 4.3422222222222225e-06,
|
|
"loss": 0.0029,
|
|
"step": 3050
|
|
},
|
|
{
|
|
"epoch": 30.75,
|
|
"learning_rate": 4.2866666666666666e-06,
|
|
"loss": 0.0031,
|
|
"step": 3075
|
|
},
|
|
{
|
|
"epoch": 31.0,
|
|
"learning_rate": 4.2311111111111114e-06,
|
|
"loss": 0.0029,
|
|
"step": 3100
|
|
},
|
|
{
|
|
"epoch": 31.25,
|
|
"learning_rate": 4.175555555555556e-06,
|
|
"loss": 0.0041,
|
|
"step": 3125
|
|
},
|
|
{
|
|
"epoch": 31.5,
|
|
"learning_rate": 4.12e-06,
|
|
"loss": 0.003,
|
|
"step": 3150
|
|
},
|
|
{
|
|
"epoch": 31.75,
|
|
"learning_rate": 4.064444444444444e-06,
|
|
"loss": 0.0023,
|
|
"step": 3175
|
|
},
|
|
{
|
|
"epoch": 32.0,
|
|
"learning_rate": 4.008888888888889e-06,
|
|
"loss": 0.0027,
|
|
"step": 3200
|
|
},
|
|
{
|
|
"epoch": 32.25,
|
|
"learning_rate": 3.953333333333333e-06,
|
|
"loss": 0.0027,
|
|
"step": 3225
|
|
},
|
|
{
|
|
"epoch": 32.5,
|
|
"learning_rate": 3.897777777777778e-06,
|
|
"loss": 0.003,
|
|
"step": 3250
|
|
},
|
|
{
|
|
"epoch": 32.75,
|
|
"learning_rate": 3.842222222222223e-06,
|
|
"loss": 0.0027,
|
|
"step": 3275
|
|
},
|
|
{
|
|
"epoch": 33.0,
|
|
"learning_rate": 3.7866666666666667e-06,
|
|
"loss": 0.0021,
|
|
"step": 3300
|
|
},
|
|
{
|
|
"epoch": 33.25,
|
|
"learning_rate": 3.7311111111111116e-06,
|
|
"loss": 0.0025,
|
|
"step": 3325
|
|
},
|
|
{
|
|
"epoch": 33.5,
|
|
"learning_rate": 3.675555555555556e-06,
|
|
"loss": 0.0023,
|
|
"step": 3350
|
|
},
|
|
{
|
|
"epoch": 33.75,
|
|
"learning_rate": 3.62e-06,
|
|
"loss": 0.0023,
|
|
"step": 3375
|
|
},
|
|
{
|
|
"epoch": 34.0,
|
|
"learning_rate": 3.564444444444445e-06,
|
|
"loss": 0.0021,
|
|
"step": 3400
|
|
},
|
|
{
|
|
"epoch": 34.25,
|
|
"learning_rate": 3.508888888888889e-06,
|
|
"loss": 0.0021,
|
|
"step": 3425
|
|
},
|
|
{
|
|
"epoch": 34.5,
|
|
"learning_rate": 3.4533333333333334e-06,
|
|
"loss": 0.0021,
|
|
"step": 3450
|
|
},
|
|
{
|
|
"epoch": 34.75,
|
|
"learning_rate": 3.3977777777777783e-06,
|
|
"loss": 0.002,
|
|
"step": 3475
|
|
},
|
|
{
|
|
"epoch": 35.0,
|
|
"learning_rate": 3.3422222222222224e-06,
|
|
"loss": 0.0027,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 35.25,
|
|
"learning_rate": 3.286666666666667e-06,
|
|
"loss": 0.0018,
|
|
"step": 3525
|
|
},
|
|
{
|
|
"epoch": 35.5,
|
|
"learning_rate": 3.2311111111111117e-06,
|
|
"loss": 0.002,
|
|
"step": 3550
|
|
},
|
|
{
|
|
"epoch": 35.75,
|
|
"learning_rate": 3.1755555555555557e-06,
|
|
"loss": 0.0017,
|
|
"step": 3575
|
|
},
|
|
{
|
|
"epoch": 36.0,
|
|
"learning_rate": 3.12e-06,
|
|
"loss": 0.0019,
|
|
"step": 3600
|
|
},
|
|
{
|
|
"epoch": 36.25,
|
|
"learning_rate": 3.064444444444445e-06,
|
|
"loss": 0.0019,
|
|
"step": 3625
|
|
},
|
|
{
|
|
"epoch": 36.5,
|
|
"learning_rate": 3.008888888888889e-06,
|
|
"loss": 0.0016,
|
|
"step": 3650
|
|
},
|
|
{
|
|
"epoch": 36.75,
|
|
"learning_rate": 2.9533333333333336e-06,
|
|
"loss": 0.0017,
|
|
"step": 3675
|
|
},
|
|
{
|
|
"epoch": 37.0,
|
|
"learning_rate": 2.8977777777777785e-06,
|
|
"loss": 0.0015,
|
|
"step": 3700
|
|
},
|
|
{
|
|
"epoch": 37.25,
|
|
"learning_rate": 2.8422222222222225e-06,
|
|
"loss": 0.0018,
|
|
"step": 3725
|
|
},
|
|
{
|
|
"epoch": 37.5,
|
|
"learning_rate": 2.786666666666667e-06,
|
|
"loss": 0.0015,
|
|
"step": 3750
|
|
},
|
|
{
|
|
"epoch": 37.75,
|
|
"learning_rate": 2.7311111111111114e-06,
|
|
"loss": 0.0017,
|
|
"step": 3775
|
|
},
|
|
{
|
|
"epoch": 38.0,
|
|
"learning_rate": 2.675555555555556e-06,
|
|
"loss": 0.0015,
|
|
"step": 3800
|
|
},
|
|
{
|
|
"epoch": 38.25,
|
|
"learning_rate": 2.6200000000000003e-06,
|
|
"loss": 0.0013,
|
|
"step": 3825
|
|
},
|
|
{
|
|
"epoch": 38.5,
|
|
"learning_rate": 2.5644444444444444e-06,
|
|
"loss": 0.0017,
|
|
"step": 3850
|
|
},
|
|
{
|
|
"epoch": 38.75,
|
|
"learning_rate": 2.5088888888888892e-06,
|
|
"loss": 0.0014,
|
|
"step": 3875
|
|
},
|
|
{
|
|
"epoch": 39.0,
|
|
"learning_rate": 2.4533333333333333e-06,
|
|
"loss": 0.0012,
|
|
"step": 3900
|
|
},
|
|
{
|
|
"epoch": 39.25,
|
|
"learning_rate": 2.397777777777778e-06,
|
|
"loss": 0.0012,
|
|
"step": 3925
|
|
},
|
|
{
|
|
"epoch": 39.5,
|
|
"learning_rate": 2.342222222222222e-06,
|
|
"loss": 0.0013,
|
|
"step": 3950
|
|
},
|
|
{
|
|
"epoch": 39.75,
|
|
"learning_rate": 2.2866666666666667e-06,
|
|
"loss": 0.0015,
|
|
"step": 3975
|
|
},
|
|
{
|
|
"epoch": 40.0,
|
|
"learning_rate": 2.2311111111111115e-06,
|
|
"loss": 0.0013,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 40.0,
|
|
"eval_loss": 0.06085631996393204,
|
|
"eval_runtime": 44.6517,
|
|
"eval_samples_per_second": 10.212,
|
|
"eval_steps_per_second": 0.112,
|
|
"eval_wer": 0.02823920265780731,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 40.25,
|
|
"learning_rate": 2.1755555555555556e-06,
|
|
"loss": 0.0012,
|
|
"step": 4025
|
|
},
|
|
{
|
|
"epoch": 40.5,
|
|
"learning_rate": 2.12e-06,
|
|
"loss": 0.0016,
|
|
"step": 4050
|
|
},
|
|
{
|
|
"epoch": 40.75,
|
|
"learning_rate": 2.064444444444445e-06,
|
|
"loss": 0.0011,
|
|
"step": 4075
|
|
},
|
|
{
|
|
"epoch": 41.0,
|
|
"learning_rate": 2.008888888888889e-06,
|
|
"loss": 0.0012,
|
|
"step": 4100
|
|
},
|
|
{
|
|
"epoch": 41.25,
|
|
"learning_rate": 1.9533333333333334e-06,
|
|
"loss": 0.0013,
|
|
"step": 4125
|
|
},
|
|
{
|
|
"epoch": 41.5,
|
|
"learning_rate": 1.8977777777777779e-06,
|
|
"loss": 0.001,
|
|
"step": 4150
|
|
},
|
|
{
|
|
"epoch": 41.75,
|
|
"learning_rate": 1.8422222222222225e-06,
|
|
"loss": 0.0012,
|
|
"step": 4175
|
|
},
|
|
{
|
|
"epoch": 42.0,
|
|
"learning_rate": 1.7866666666666668e-06,
|
|
"loss": 0.0012,
|
|
"step": 4200
|
|
},
|
|
{
|
|
"epoch": 42.25,
|
|
"learning_rate": 1.7311111111111112e-06,
|
|
"loss": 0.0012,
|
|
"step": 4225
|
|
},
|
|
{
|
|
"epoch": 42.5,
|
|
"learning_rate": 1.675555555555556e-06,
|
|
"loss": 0.001,
|
|
"step": 4250
|
|
},
|
|
{
|
|
"epoch": 42.75,
|
|
"learning_rate": 1.6200000000000002e-06,
|
|
"loss": 0.0012,
|
|
"step": 4275
|
|
},
|
|
{
|
|
"epoch": 43.0,
|
|
"learning_rate": 1.5644444444444446e-06,
|
|
"loss": 0.001,
|
|
"step": 4300
|
|
},
|
|
{
|
|
"epoch": 43.25,
|
|
"learning_rate": 1.5088888888888889e-06,
|
|
"loss": 0.0009,
|
|
"step": 4325
|
|
},
|
|
{
|
|
"epoch": 43.5,
|
|
"learning_rate": 1.4533333333333335e-06,
|
|
"loss": 0.0011,
|
|
"step": 4350
|
|
},
|
|
{
|
|
"epoch": 43.75,
|
|
"learning_rate": 1.397777777777778e-06,
|
|
"loss": 0.001,
|
|
"step": 4375
|
|
},
|
|
{
|
|
"epoch": 44.0,
|
|
"learning_rate": 1.3422222222222222e-06,
|
|
"loss": 0.0012,
|
|
"step": 4400
|
|
},
|
|
{
|
|
"epoch": 44.25,
|
|
"learning_rate": 1.286666666666667e-06,
|
|
"loss": 0.0012,
|
|
"step": 4425
|
|
},
|
|
{
|
|
"epoch": 44.5,
|
|
"learning_rate": 1.2311111111111112e-06,
|
|
"loss": 0.0009,
|
|
"step": 4450
|
|
},
|
|
{
|
|
"epoch": 44.75,
|
|
"learning_rate": 1.1755555555555556e-06,
|
|
"loss": 0.001,
|
|
"step": 4475
|
|
},
|
|
{
|
|
"epoch": 45.0,
|
|
"learning_rate": 1.12e-06,
|
|
"loss": 0.001,
|
|
"step": 4500
|
|
},
|
|
{
|
|
"epoch": 45.25,
|
|
"learning_rate": 1.0644444444444445e-06,
|
|
"loss": 0.0009,
|
|
"step": 4525
|
|
},
|
|
{
|
|
"epoch": 45.5,
|
|
"learning_rate": 1.008888888888889e-06,
|
|
"loss": 0.0009,
|
|
"step": 4550
|
|
},
|
|
{
|
|
"epoch": 45.75,
|
|
"learning_rate": 9.533333333333335e-07,
|
|
"loss": 0.0008,
|
|
"step": 4575
|
|
},
|
|
{
|
|
"epoch": 46.0,
|
|
"learning_rate": 8.977777777777778e-07,
|
|
"loss": 0.0008,
|
|
"step": 4600
|
|
},
|
|
{
|
|
"epoch": 46.25,
|
|
"learning_rate": 8.422222222222224e-07,
|
|
"loss": 0.0008,
|
|
"step": 4625
|
|
},
|
|
{
|
|
"epoch": 46.5,
|
|
"learning_rate": 7.866666666666667e-07,
|
|
"loss": 0.0007,
|
|
"step": 4650
|
|
},
|
|
{
|
|
"epoch": 46.75,
|
|
"learning_rate": 7.311111111111112e-07,
|
|
"loss": 0.0007,
|
|
"step": 4675
|
|
},
|
|
{
|
|
"epoch": 47.0,
|
|
"learning_rate": 6.755555555555555e-07,
|
|
"loss": 0.0008,
|
|
"step": 4700
|
|
},
|
|
{
|
|
"epoch": 47.25,
|
|
"learning_rate": 6.200000000000001e-07,
|
|
"loss": 0.0008,
|
|
"step": 4725
|
|
},
|
|
{
|
|
"epoch": 47.5,
|
|
"learning_rate": 5.644444444444445e-07,
|
|
"loss": 0.0009,
|
|
"step": 4750
|
|
},
|
|
{
|
|
"epoch": 47.75,
|
|
"learning_rate": 5.088888888888889e-07,
|
|
"loss": 0.0008,
|
|
"step": 4775
|
|
},
|
|
{
|
|
"epoch": 48.0,
|
|
"learning_rate": 4.533333333333334e-07,
|
|
"loss": 0.0008,
|
|
"step": 4800
|
|
},
|
|
{
|
|
"epoch": 48.25,
|
|
"learning_rate": 3.9777777777777783e-07,
|
|
"loss": 0.0008,
|
|
"step": 4825
|
|
},
|
|
{
|
|
"epoch": 48.5,
|
|
"learning_rate": 3.422222222222223e-07,
|
|
"loss": 0.0008,
|
|
"step": 4850
|
|
},
|
|
{
|
|
"epoch": 48.75,
|
|
"learning_rate": 2.866666666666667e-07,
|
|
"loss": 0.0007,
|
|
"step": 4875
|
|
},
|
|
{
|
|
"epoch": 49.0,
|
|
"learning_rate": 2.3111111111111112e-07,
|
|
"loss": 0.0007,
|
|
"step": 4900
|
|
},
|
|
{
|
|
"epoch": 49.25,
|
|
"learning_rate": 1.7555555555555558e-07,
|
|
"loss": 0.0009,
|
|
"step": 4925
|
|
},
|
|
{
|
|
"epoch": 49.5,
|
|
"learning_rate": 1.2000000000000002e-07,
|
|
"loss": 0.0006,
|
|
"step": 4950
|
|
},
|
|
{
|
|
"epoch": 49.75,
|
|
"learning_rate": 6.444444444444445e-08,
|
|
"loss": 0.0007,
|
|
"step": 4975
|
|
},
|
|
{
|
|
"epoch": 50.0,
|
|
"learning_rate": 8.88888888888889e-09,
|
|
"loss": 0.0008,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 50.0,
|
|
"eval_loss": 0.06168894097208977,
|
|
"eval_runtime": 48.7713,
|
|
"eval_samples_per_second": 9.35,
|
|
"eval_steps_per_second": 0.103,
|
|
"eval_wer": 0.028994261552401087,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 50.0,
|
|
"step": 5000,
|
|
"total_flos": 2.363832516900972e+21,
|
|
"train_loss": 0.023422172606736423,
|
|
"train_runtime": 58818.0035,
|
|
"train_samples_per_second": 19.042,
|
|
"train_steps_per_second": 0.085
|
|
}
|
|
],
|
|
"max_steps": 5000,
|
|
"num_train_epochs": 50,
|
|
"total_flos": 2.363832516900972e+21,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|