whisper-small-sl / trainer_state.json
Umut
End of training
7a2f7ba
raw
history blame
25.5 kB
{
"best_metric": 27.026239067055393,
"best_model_checkpoint": "./checkpoint-2000",
"epoch": 65.012,
"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": 5.000000000000001e-07,
"loss": 1.8066,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 1.0000000000000002e-06,
"loss": 1.2682,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.5e-06,
"loss": 0.807,
"step": 75
},
{
"epoch": 1.0,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.7803,
"step": 100
},
{
"epoch": 1.01,
"learning_rate": 2.5e-06,
"loss": 0.4995,
"step": 125
},
{
"epoch": 1.01,
"learning_rate": 3e-06,
"loss": 0.372,
"step": 150
},
{
"epoch": 2.0,
"learning_rate": 3.5e-06,
"loss": 0.4778,
"step": 175
},
{
"epoch": 2.01,
"learning_rate": 4.000000000000001e-06,
"loss": 0.3357,
"step": 200
},
{
"epoch": 2.01,
"learning_rate": 4.5e-06,
"loss": 0.2368,
"step": 225
},
{
"epoch": 3.0,
"learning_rate": 5e-06,
"loss": 0.2898,
"step": 250
},
{
"epoch": 3.01,
"learning_rate": 5.500000000000001e-06,
"loss": 0.2053,
"step": 275
},
{
"epoch": 3.01,
"learning_rate": 6e-06,
"loss": 0.137,
"step": 300
},
{
"epoch": 4.0,
"learning_rate": 6.5000000000000004e-06,
"loss": 0.1592,
"step": 325
},
{
"epoch": 4.01,
"learning_rate": 7e-06,
"loss": 0.1006,
"step": 350
},
{
"epoch": 4.01,
"learning_rate": 7.500000000000001e-06,
"loss": 0.0663,
"step": 375
},
{
"epoch": 5.0,
"learning_rate": 8.000000000000001e-06,
"loss": 0.0711,
"step": 400
},
{
"epoch": 5.01,
"learning_rate": 8.5e-06,
"loss": 0.0492,
"step": 425
},
{
"epoch": 5.01,
"learning_rate": 9e-06,
"loss": 0.0316,
"step": 450
},
{
"epoch": 6.0,
"learning_rate": 9.5e-06,
"loss": 0.0414,
"step": 475
},
{
"epoch": 6.01,
"learning_rate": 1e-05,
"loss": 0.029,
"step": 500
},
{
"epoch": 6.01,
"learning_rate": 9.944444444444445e-06,
"loss": 0.0198,
"step": 525
},
{
"epoch": 7.0,
"learning_rate": 9.88888888888889e-06,
"loss": 0.0267,
"step": 550
},
{
"epoch": 7.01,
"learning_rate": 9.833333333333333e-06,
"loss": 0.0185,
"step": 575
},
{
"epoch": 7.01,
"learning_rate": 9.777777777777779e-06,
"loss": 0.0125,
"step": 600
},
{
"epoch": 8.0,
"learning_rate": 9.722222222222223e-06,
"loss": 0.0148,
"step": 625
},
{
"epoch": 8.01,
"learning_rate": 9.666666666666667e-06,
"loss": 0.0143,
"step": 650
},
{
"epoch": 8.01,
"learning_rate": 9.611111111111112e-06,
"loss": 0.0094,
"step": 675
},
{
"epoch": 9.0,
"learning_rate": 9.555555555555556e-06,
"loss": 0.011,
"step": 700
},
{
"epoch": 9.01,
"learning_rate": 9.5e-06,
"loss": 0.0083,
"step": 725
},
{
"epoch": 9.01,
"learning_rate": 9.444444444444445e-06,
"loss": 0.006,
"step": 750
},
{
"epoch": 10.0,
"learning_rate": 9.38888888888889e-06,
"loss": 0.0072,
"step": 775
},
{
"epoch": 10.01,
"learning_rate": 9.333333333333334e-06,
"loss": 0.0087,
"step": 800
},
{
"epoch": 10.01,
"learning_rate": 9.277777777777778e-06,
"loss": 0.0047,
"step": 825
},
{
"epoch": 11.0,
"learning_rate": 9.222222222222224e-06,
"loss": 0.0077,
"step": 850
},
{
"epoch": 11.01,
"learning_rate": 9.166666666666666e-06,
"loss": 0.0046,
"step": 875
},
{
"epoch": 11.01,
"learning_rate": 9.111111111111112e-06,
"loss": 0.0052,
"step": 900
},
{
"epoch": 12.0,
"learning_rate": 9.055555555555556e-06,
"loss": 0.0036,
"step": 925
},
{
"epoch": 12.01,
"learning_rate": 9e-06,
"loss": 0.0036,
"step": 950
},
{
"epoch": 12.01,
"learning_rate": 8.944444444444446e-06,
"loss": 0.003,
"step": 975
},
{
"epoch": 13.0,
"learning_rate": 8.888888888888888e-06,
"loss": 0.006,
"step": 1000
},
{
"epoch": 13.0,
"eval_loss": 0.5071619749069214,
"eval_runtime": 229.6348,
"eval_samples_per_second": 5.256,
"eval_steps_per_second": 0.165,
"eval_wer": 28.338192419825077,
"step": 1000
},
{
"epoch": 13.01,
"learning_rate": 8.833333333333334e-06,
"loss": 0.0042,
"step": 1025
},
{
"epoch": 13.01,
"learning_rate": 8.777777777777778e-06,
"loss": 0.0017,
"step": 1050
},
{
"epoch": 14.0,
"learning_rate": 8.722222222222224e-06,
"loss": 0.0024,
"step": 1075
},
{
"epoch": 14.01,
"learning_rate": 8.666666666666668e-06,
"loss": 0.0012,
"step": 1100
},
{
"epoch": 14.01,
"learning_rate": 8.611111111111112e-06,
"loss": 0.0027,
"step": 1125
},
{
"epoch": 15.0,
"learning_rate": 8.555555555555556e-06,
"loss": 0.0016,
"step": 1150
},
{
"epoch": 15.01,
"learning_rate": 8.5e-06,
"loss": 0.002,
"step": 1175
},
{
"epoch": 15.01,
"learning_rate": 8.444444444444446e-06,
"loss": 0.0007,
"step": 1200
},
{
"epoch": 16.0,
"learning_rate": 8.38888888888889e-06,
"loss": 0.0012,
"step": 1225
},
{
"epoch": 16.01,
"learning_rate": 8.333333333333334e-06,
"loss": 0.0015,
"step": 1250
},
{
"epoch": 16.01,
"learning_rate": 8.277777777777778e-06,
"loss": 0.0015,
"step": 1275
},
{
"epoch": 17.0,
"learning_rate": 8.222222222222222e-06,
"loss": 0.001,
"step": 1300
},
{
"epoch": 17.01,
"learning_rate": 8.166666666666668e-06,
"loss": 0.0017,
"step": 1325
},
{
"epoch": 17.01,
"learning_rate": 8.111111111111112e-06,
"loss": 0.0009,
"step": 1350
},
{
"epoch": 18.0,
"learning_rate": 8.055555555555557e-06,
"loss": 0.0014,
"step": 1375
},
{
"epoch": 18.01,
"learning_rate": 8.000000000000001e-06,
"loss": 0.0005,
"step": 1400
},
{
"epoch": 18.01,
"learning_rate": 7.944444444444445e-06,
"loss": 0.0004,
"step": 1425
},
{
"epoch": 19.0,
"learning_rate": 7.88888888888889e-06,
"loss": 0.0002,
"step": 1450
},
{
"epoch": 19.01,
"learning_rate": 7.833333333333333e-06,
"loss": 0.0015,
"step": 1475
},
{
"epoch": 19.01,
"learning_rate": 7.77777777777778e-06,
"loss": 0.0007,
"step": 1500
},
{
"epoch": 20.0,
"learning_rate": 7.722222222222223e-06,
"loss": 0.0002,
"step": 1525
},
{
"epoch": 20.01,
"learning_rate": 7.666666666666667e-06,
"loss": 0.0003,
"step": 1550
},
{
"epoch": 20.01,
"learning_rate": 7.611111111111111e-06,
"loss": 0.0003,
"step": 1575
},
{
"epoch": 21.0,
"learning_rate": 7.555555555555556e-06,
"loss": 0.0002,
"step": 1600
},
{
"epoch": 21.01,
"learning_rate": 7.500000000000001e-06,
"loss": 0.0006,
"step": 1625
},
{
"epoch": 21.01,
"learning_rate": 7.444444444444445e-06,
"loss": 0.0002,
"step": 1650
},
{
"epoch": 22.0,
"learning_rate": 7.38888888888889e-06,
"loss": 0.0001,
"step": 1675
},
{
"epoch": 22.01,
"learning_rate": 7.333333333333333e-06,
"loss": 0.0006,
"step": 1700
},
{
"epoch": 22.01,
"learning_rate": 7.277777777777778e-06,
"loss": 0.0015,
"step": 1725
},
{
"epoch": 23.0,
"learning_rate": 7.222222222222223e-06,
"loss": 0.0,
"step": 1750
},
{
"epoch": 23.01,
"learning_rate": 7.166666666666667e-06,
"loss": 0.0011,
"step": 1775
},
{
"epoch": 23.01,
"learning_rate": 7.111111111111112e-06,
"loss": 0.0007,
"step": 1800
},
{
"epoch": 24.0,
"learning_rate": 7.055555555555557e-06,
"loss": 0.0007,
"step": 1825
},
{
"epoch": 24.01,
"learning_rate": 7e-06,
"loss": 0.0004,
"step": 1850
},
{
"epoch": 24.01,
"learning_rate": 6.944444444444445e-06,
"loss": 0.0001,
"step": 1875
},
{
"epoch": 24.02,
"learning_rate": 6.88888888888889e-06,
"loss": 0.0001,
"step": 1900
},
{
"epoch": 25.0,
"learning_rate": 6.833333333333334e-06,
"loss": 0.0001,
"step": 1925
},
{
"epoch": 25.01,
"learning_rate": 6.777777777777779e-06,
"loss": 0.0008,
"step": 1950
},
{
"epoch": 25.02,
"learning_rate": 6.7222222222222235e-06,
"loss": 0.0,
"step": 1975
},
{
"epoch": 26.0,
"learning_rate": 6.666666666666667e-06,
"loss": 0.0011,
"step": 2000
},
{
"epoch": 26.0,
"eval_loss": 0.7109041213989258,
"eval_runtime": 215.3397,
"eval_samples_per_second": 5.605,
"eval_steps_per_second": 0.176,
"eval_wer": 27.026239067055393,
"step": 2000
},
{
"epoch": 26.01,
"learning_rate": 6.6111111111111115e-06,
"loss": 0.0,
"step": 2025
},
{
"epoch": 26.01,
"learning_rate": 6.555555555555556e-06,
"loss": 0.0,
"step": 2050
},
{
"epoch": 27.0,
"learning_rate": 6.5000000000000004e-06,
"loss": 0.0004,
"step": 2075
},
{
"epoch": 27.01,
"learning_rate": 6.444444444444445e-06,
"loss": 0.0003,
"step": 2100
},
{
"epoch": 27.01,
"learning_rate": 6.3888888888888885e-06,
"loss": 0.0003,
"step": 2125
},
{
"epoch": 28.0,
"learning_rate": 6.333333333333333e-06,
"loss": 0.0005,
"step": 2150
},
{
"epoch": 28.01,
"learning_rate": 6.277777777777778e-06,
"loss": 0.0001,
"step": 2175
},
{
"epoch": 28.01,
"learning_rate": 6.222222222222223e-06,
"loss": 0.0,
"step": 2200
},
{
"epoch": 29.0,
"learning_rate": 6.166666666666667e-06,
"loss": 0.0004,
"step": 2225
},
{
"epoch": 29.01,
"learning_rate": 6.111111111111112e-06,
"loss": 0.0,
"step": 2250
},
{
"epoch": 29.01,
"learning_rate": 6.055555555555555e-06,
"loss": 0.0,
"step": 2275
},
{
"epoch": 30.0,
"learning_rate": 6e-06,
"loss": 0.0004,
"step": 2300
},
{
"epoch": 30.01,
"learning_rate": 5.944444444444445e-06,
"loss": 0.0,
"step": 2325
},
{
"epoch": 30.01,
"learning_rate": 5.88888888888889e-06,
"loss": 0.0,
"step": 2350
},
{
"epoch": 31.0,
"learning_rate": 5.833333333333334e-06,
"loss": 0.0,
"step": 2375
},
{
"epoch": 31.01,
"learning_rate": 5.777777777777778e-06,
"loss": 0.0,
"step": 2400
},
{
"epoch": 31.01,
"learning_rate": 5.722222222222222e-06,
"loss": 0.0,
"step": 2425
},
{
"epoch": 32.0,
"learning_rate": 5.666666666666667e-06,
"loss": 0.0006,
"step": 2450
},
{
"epoch": 32.01,
"learning_rate": 5.611111111111112e-06,
"loss": 0.0,
"step": 2475
},
{
"epoch": 32.01,
"learning_rate": 5.555555555555557e-06,
"loss": 0.0,
"step": 2500
},
{
"epoch": 33.0,
"learning_rate": 5.500000000000001e-06,
"loss": 0.0001,
"step": 2525
},
{
"epoch": 33.01,
"learning_rate": 5.444444444444445e-06,
"loss": 0.0002,
"step": 2550
},
{
"epoch": 33.01,
"learning_rate": 5.388888888888889e-06,
"loss": 0.0,
"step": 2575
},
{
"epoch": 34.0,
"learning_rate": 5.333333333333334e-06,
"loss": 0.0003,
"step": 2600
},
{
"epoch": 34.01,
"learning_rate": 5.2777777777777785e-06,
"loss": 0.0,
"step": 2625
},
{
"epoch": 34.01,
"learning_rate": 5.2222222222222226e-06,
"loss": 0.0001,
"step": 2650
},
{
"epoch": 35.0,
"learning_rate": 5.1666666666666675e-06,
"loss": 0.0,
"step": 2675
},
{
"epoch": 35.01,
"learning_rate": 5.1111111111111115e-06,
"loss": 0.0,
"step": 2700
},
{
"epoch": 35.01,
"learning_rate": 5.0555555555555555e-06,
"loss": 0.0001,
"step": 2725
},
{
"epoch": 36.0,
"learning_rate": 5e-06,
"loss": 0.0,
"step": 2750
},
{
"epoch": 36.01,
"learning_rate": 4.944444444444445e-06,
"loss": 0.0,
"step": 2775
},
{
"epoch": 36.01,
"learning_rate": 4.888888888888889e-06,
"loss": 0.0,
"step": 2800
},
{
"epoch": 37.0,
"learning_rate": 4.833333333333333e-06,
"loss": 0.0,
"step": 2825
},
{
"epoch": 37.01,
"learning_rate": 4.777777777777778e-06,
"loss": 0.0,
"step": 2850
},
{
"epoch": 37.01,
"learning_rate": 4.722222222222222e-06,
"loss": 0.0,
"step": 2875
},
{
"epoch": 38.0,
"learning_rate": 4.666666666666667e-06,
"loss": 0.0,
"step": 2900
},
{
"epoch": 38.01,
"learning_rate": 4.611111111111112e-06,
"loss": 0.0,
"step": 2925
},
{
"epoch": 38.01,
"learning_rate": 4.555555555555556e-06,
"loss": 0.0,
"step": 2950
},
{
"epoch": 39.0,
"learning_rate": 4.5e-06,
"loss": 0.0,
"step": 2975
},
{
"epoch": 39.01,
"learning_rate": 4.444444444444444e-06,
"loss": 0.0005,
"step": 3000
},
{
"epoch": 39.01,
"eval_loss": 0.8864887952804565,
"eval_runtime": 216.196,
"eval_samples_per_second": 5.583,
"eval_steps_per_second": 0.176,
"eval_wer": 27.1865889212828,
"step": 3000
},
{
"epoch": 39.01,
"learning_rate": 4.388888888888889e-06,
"loss": 0.0,
"step": 3025
},
{
"epoch": 40.0,
"learning_rate": 4.333333333333334e-06,
"loss": 0.0,
"step": 3050
},
{
"epoch": 40.01,
"learning_rate": 4.277777777777778e-06,
"loss": 0.0,
"step": 3075
},
{
"epoch": 40.01,
"learning_rate": 4.222222222222223e-06,
"loss": 0.0,
"step": 3100
},
{
"epoch": 41.0,
"learning_rate": 4.166666666666667e-06,
"loss": 0.0,
"step": 3125
},
{
"epoch": 41.01,
"learning_rate": 4.111111111111111e-06,
"loss": 0.0,
"step": 3150
},
{
"epoch": 41.01,
"learning_rate": 4.055555555555556e-06,
"loss": 0.0,
"step": 3175
},
{
"epoch": 42.0,
"learning_rate": 4.000000000000001e-06,
"loss": 0.0,
"step": 3200
},
{
"epoch": 42.01,
"learning_rate": 3.944444444444445e-06,
"loss": 0.0,
"step": 3225
},
{
"epoch": 42.01,
"learning_rate": 3.88888888888889e-06,
"loss": 0.0,
"step": 3250
},
{
"epoch": 43.0,
"learning_rate": 3.833333333333334e-06,
"loss": 0.0,
"step": 3275
},
{
"epoch": 43.01,
"learning_rate": 3.777777777777778e-06,
"loss": 0.0,
"step": 3300
},
{
"epoch": 43.01,
"learning_rate": 3.7222222222222225e-06,
"loss": 0.0,
"step": 3325
},
{
"epoch": 44.0,
"learning_rate": 3.6666666666666666e-06,
"loss": 0.0,
"step": 3350
},
{
"epoch": 44.01,
"learning_rate": 3.6111111111111115e-06,
"loss": 0.0,
"step": 3375
},
{
"epoch": 44.01,
"learning_rate": 3.555555555555556e-06,
"loss": 0.0,
"step": 3400
},
{
"epoch": 45.0,
"learning_rate": 3.5e-06,
"loss": 0.0,
"step": 3425
},
{
"epoch": 45.01,
"learning_rate": 3.444444444444445e-06,
"loss": 0.0,
"step": 3450
},
{
"epoch": 45.01,
"learning_rate": 3.3888888888888893e-06,
"loss": 0.0,
"step": 3475
},
{
"epoch": 46.0,
"learning_rate": 3.3333333333333333e-06,
"loss": 0.0,
"step": 3500
},
{
"epoch": 46.01,
"learning_rate": 3.277777777777778e-06,
"loss": 0.0,
"step": 3525
},
{
"epoch": 46.01,
"learning_rate": 3.2222222222222227e-06,
"loss": 0.0,
"step": 3550
},
{
"epoch": 47.0,
"learning_rate": 3.1666666666666667e-06,
"loss": 0.0,
"step": 3575
},
{
"epoch": 47.01,
"learning_rate": 3.1111111111111116e-06,
"loss": 0.0,
"step": 3600
},
{
"epoch": 47.01,
"learning_rate": 3.055555555555556e-06,
"loss": 0.0,
"step": 3625
},
{
"epoch": 48.0,
"learning_rate": 3e-06,
"loss": 0.0,
"step": 3650
},
{
"epoch": 48.01,
"learning_rate": 2.944444444444445e-06,
"loss": 0.0,
"step": 3675
},
{
"epoch": 48.01,
"learning_rate": 2.888888888888889e-06,
"loss": 0.0,
"step": 3700
},
{
"epoch": 49.0,
"learning_rate": 2.8333333333333335e-06,
"loss": 0.0,
"step": 3725
},
{
"epoch": 49.01,
"learning_rate": 2.7777777777777783e-06,
"loss": 0.0,
"step": 3750
},
{
"epoch": 49.01,
"learning_rate": 2.7222222222222224e-06,
"loss": 0.0,
"step": 3775
},
{
"epoch": 49.02,
"learning_rate": 2.666666666666667e-06,
"loss": 0.0,
"step": 3800
},
{
"epoch": 50.01,
"learning_rate": 2.6111111111111113e-06,
"loss": 0.0,
"step": 3825
},
{
"epoch": 50.01,
"learning_rate": 2.5555555555555557e-06,
"loss": 0.0,
"step": 3850
},
{
"epoch": 50.02,
"learning_rate": 2.5e-06,
"loss": 0.0,
"step": 3875
},
{
"epoch": 51.0,
"learning_rate": 2.4444444444444447e-06,
"loss": 0.0,
"step": 3900
},
{
"epoch": 51.01,
"learning_rate": 2.388888888888889e-06,
"loss": 0.0,
"step": 3925
},
{
"epoch": 51.01,
"learning_rate": 2.3333333333333336e-06,
"loss": 0.0,
"step": 3950
},
{
"epoch": 52.0,
"learning_rate": 2.277777777777778e-06,
"loss": 0.0,
"step": 3975
},
{
"epoch": 52.01,
"learning_rate": 2.222222222222222e-06,
"loss": 0.0,
"step": 4000
},
{
"epoch": 52.01,
"eval_loss": 1.026085615158081,
"eval_runtime": 219.1966,
"eval_samples_per_second": 5.506,
"eval_steps_per_second": 0.173,
"eval_wer": 29.008746355685133,
"step": 4000
},
{
"epoch": 52.01,
"learning_rate": 2.166666666666667e-06,
"loss": 0.0,
"step": 4025
},
{
"epoch": 53.0,
"learning_rate": 2.1111111111111114e-06,
"loss": 0.0,
"step": 4050
},
{
"epoch": 53.01,
"learning_rate": 2.0555555555555555e-06,
"loss": 0.0,
"step": 4075
},
{
"epoch": 53.01,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.0,
"step": 4100
},
{
"epoch": 54.0,
"learning_rate": 1.944444444444445e-06,
"loss": 0.0,
"step": 4125
},
{
"epoch": 54.01,
"learning_rate": 1.888888888888889e-06,
"loss": 0.0,
"step": 4150
},
{
"epoch": 54.01,
"learning_rate": 1.8333333333333333e-06,
"loss": 0.0,
"step": 4175
},
{
"epoch": 55.0,
"learning_rate": 1.777777777777778e-06,
"loss": 0.0,
"step": 4200
},
{
"epoch": 55.01,
"learning_rate": 1.7222222222222224e-06,
"loss": 0.0,
"step": 4225
},
{
"epoch": 55.01,
"learning_rate": 1.6666666666666667e-06,
"loss": 0.0,
"step": 4250
},
{
"epoch": 56.0,
"learning_rate": 1.6111111111111113e-06,
"loss": 0.0,
"step": 4275
},
{
"epoch": 56.01,
"learning_rate": 1.5555555555555558e-06,
"loss": 0.0,
"step": 4300
},
{
"epoch": 56.01,
"learning_rate": 1.5e-06,
"loss": 0.0,
"step": 4325
},
{
"epoch": 57.0,
"learning_rate": 1.4444444444444445e-06,
"loss": 0.0,
"step": 4350
},
{
"epoch": 57.01,
"learning_rate": 1.3888888888888892e-06,
"loss": 0.0,
"step": 4375
},
{
"epoch": 57.01,
"learning_rate": 1.3333333333333334e-06,
"loss": 0.0,
"step": 4400
},
{
"epoch": 58.0,
"learning_rate": 1.2777777777777779e-06,
"loss": 0.0,
"step": 4425
},
{
"epoch": 58.01,
"learning_rate": 1.2222222222222223e-06,
"loss": 0.0,
"step": 4450
},
{
"epoch": 58.01,
"learning_rate": 1.1666666666666668e-06,
"loss": 0.0,
"step": 4475
},
{
"epoch": 59.0,
"learning_rate": 1.111111111111111e-06,
"loss": 0.0,
"step": 4500
},
{
"epoch": 59.01,
"learning_rate": 1.0555555555555557e-06,
"loss": 0.0,
"step": 4525
},
{
"epoch": 59.01,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.0,
"step": 4550
},
{
"epoch": 60.0,
"learning_rate": 9.444444444444445e-07,
"loss": 0.0,
"step": 4575
},
{
"epoch": 60.01,
"learning_rate": 8.88888888888889e-07,
"loss": 0.0,
"step": 4600
},
{
"epoch": 60.01,
"learning_rate": 8.333333333333333e-07,
"loss": 0.0,
"step": 4625
},
{
"epoch": 61.0,
"learning_rate": 7.777777777777779e-07,
"loss": 0.0,
"step": 4650
},
{
"epoch": 61.01,
"learning_rate": 7.222222222222222e-07,
"loss": 0.0,
"step": 4675
},
{
"epoch": 61.01,
"learning_rate": 6.666666666666667e-07,
"loss": 0.0,
"step": 4700
},
{
"epoch": 62.0,
"learning_rate": 6.111111111111112e-07,
"loss": 0.0,
"step": 4725
},
{
"epoch": 62.01,
"learning_rate": 5.555555555555555e-07,
"loss": 0.0,
"step": 4750
},
{
"epoch": 62.01,
"learning_rate": 5.000000000000001e-07,
"loss": 0.0,
"step": 4775
},
{
"epoch": 63.0,
"learning_rate": 4.444444444444445e-07,
"loss": 0.0,
"step": 4800
},
{
"epoch": 63.01,
"learning_rate": 3.8888888888888895e-07,
"loss": 0.0,
"step": 4825
},
{
"epoch": 63.01,
"learning_rate": 3.3333333333333335e-07,
"loss": 0.0,
"step": 4850
},
{
"epoch": 64.0,
"learning_rate": 2.7777777777777776e-07,
"loss": 0.0,
"step": 4875
},
{
"epoch": 64.01,
"learning_rate": 2.2222222222222224e-07,
"loss": 0.0,
"step": 4900
},
{
"epoch": 64.01,
"learning_rate": 1.6666666666666668e-07,
"loss": 0.0,
"step": 4925
},
{
"epoch": 65.0,
"learning_rate": 1.1111111111111112e-07,
"loss": 0.0,
"step": 4950
},
{
"epoch": 65.01,
"learning_rate": 5.555555555555556e-08,
"loss": 0.0,
"step": 4975
},
{
"epoch": 65.01,
"learning_rate": 0.0,
"loss": 0.0,
"step": 5000
},
{
"epoch": 65.01,
"eval_loss": 1.0871795415878296,
"eval_runtime": 216.3632,
"eval_samples_per_second": 5.579,
"eval_steps_per_second": 0.176,
"eval_wer": 28.965014577259474,
"step": 5000
},
{
"epoch": 65.01,
"step": 5000,
"total_flos": 4.5798503030784e+19,
"train_loss": 0.040012699496316056,
"train_runtime": 59816.3885,
"train_samples_per_second": 2.675,
"train_steps_per_second": 0.084
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 4.5798503030784e+19,
"trial_name": null,
"trial_params": null
}