cv9-special-batch12-lr4-small / trainer_state.json
TheRains's picture
End of training
9f92cd0
raw
history blame
25.8 kB
{
"best_metric": 0.3947650718729886,
"best_model_checkpoint": "./output/small/cv9-special-batch12-lr4-small/checkpoint-5000",
"epoch": 11.876484560570072,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.06,
"learning_rate": 4.4e-06,
"loss": 1.2864,
"step": 25
},
{
"epoch": 0.12,
"learning_rate": 9.4e-06,
"loss": 0.6622,
"step": 50
},
{
"epoch": 0.18,
"learning_rate": 1.44e-05,
"loss": 0.4779,
"step": 75
},
{
"epoch": 0.24,
"learning_rate": 1.94e-05,
"loss": 0.4205,
"step": 100
},
{
"epoch": 0.3,
"learning_rate": 2.44e-05,
"loss": 0.3893,
"step": 125
},
{
"epoch": 0.36,
"learning_rate": 2.94e-05,
"loss": 0.4268,
"step": 150
},
{
"epoch": 0.42,
"learning_rate": 3.4399999999999996e-05,
"loss": 0.4469,
"step": 175
},
{
"epoch": 0.48,
"learning_rate": 3.94e-05,
"loss": 0.4307,
"step": 200
},
{
"epoch": 0.53,
"learning_rate": 4.44e-05,
"loss": 0.4571,
"step": 225
},
{
"epoch": 0.59,
"learning_rate": 4.94e-05,
"loss": 0.4735,
"step": 250
},
{
"epoch": 0.65,
"learning_rate": 5.440000000000001e-05,
"loss": 0.5237,
"step": 275
},
{
"epoch": 0.71,
"learning_rate": 5.94e-05,
"loss": 0.5129,
"step": 300
},
{
"epoch": 0.77,
"learning_rate": 6.440000000000001e-05,
"loss": 0.5336,
"step": 325
},
{
"epoch": 0.83,
"learning_rate": 6.939999999999999e-05,
"loss": 0.536,
"step": 350
},
{
"epoch": 0.89,
"learning_rate": 7.42e-05,
"loss": 0.5525,
"step": 375
},
{
"epoch": 0.95,
"learning_rate": 7.920000000000001e-05,
"loss": 0.5774,
"step": 400
},
{
"epoch": 1.01,
"learning_rate": 8.42e-05,
"loss": 0.5515,
"step": 425
},
{
"epoch": 1.07,
"learning_rate": 8.92e-05,
"loss": 0.3206,
"step": 450
},
{
"epoch": 1.13,
"learning_rate": 9.42e-05,
"loss": 0.3541,
"step": 475
},
{
"epoch": 1.19,
"learning_rate": 9.92e-05,
"loss": 0.3557,
"step": 500
},
{
"epoch": 1.25,
"learning_rate": 9.953333333333333e-05,
"loss": 0.3749,
"step": 525
},
{
"epoch": 1.31,
"learning_rate": 9.897777777777778e-05,
"loss": 0.4073,
"step": 550
},
{
"epoch": 1.37,
"learning_rate": 9.842222222222223e-05,
"loss": 0.4469,
"step": 575
},
{
"epoch": 1.43,
"learning_rate": 9.786666666666667e-05,
"loss": 0.4036,
"step": 600
},
{
"epoch": 1.48,
"learning_rate": 9.731111111111112e-05,
"loss": 0.3851,
"step": 625
},
{
"epoch": 1.54,
"learning_rate": 9.675555555555556e-05,
"loss": 0.41,
"step": 650
},
{
"epoch": 1.6,
"learning_rate": 9.620000000000001e-05,
"loss": 0.4313,
"step": 675
},
{
"epoch": 1.66,
"learning_rate": 9.564444444444444e-05,
"loss": 0.3967,
"step": 700
},
{
"epoch": 1.72,
"learning_rate": 9.50888888888889e-05,
"loss": 0.4012,
"step": 725
},
{
"epoch": 1.78,
"learning_rate": 9.453333333333335e-05,
"loss": 0.4157,
"step": 750
},
{
"epoch": 1.84,
"learning_rate": 9.397777777777778e-05,
"loss": 0.442,
"step": 775
},
{
"epoch": 1.9,
"learning_rate": 9.342222222222222e-05,
"loss": 0.4496,
"step": 800
},
{
"epoch": 1.96,
"learning_rate": 9.286666666666667e-05,
"loss": 0.4323,
"step": 825
},
{
"epoch": 2.02,
"learning_rate": 9.231111111111112e-05,
"loss": 0.3541,
"step": 850
},
{
"epoch": 2.08,
"learning_rate": 9.175555555555556e-05,
"loss": 0.1938,
"step": 875
},
{
"epoch": 2.14,
"learning_rate": 9.120000000000001e-05,
"loss": 0.198,
"step": 900
},
{
"epoch": 2.2,
"learning_rate": 9.064444444444446e-05,
"loss": 0.1856,
"step": 925
},
{
"epoch": 2.26,
"learning_rate": 9.00888888888889e-05,
"loss": 0.2461,
"step": 950
},
{
"epoch": 2.32,
"learning_rate": 8.953333333333333e-05,
"loss": 0.2167,
"step": 975
},
{
"epoch": 2.38,
"learning_rate": 8.897777777777778e-05,
"loss": 0.2264,
"step": 1000
},
{
"epoch": 2.38,
"eval_loss": 0.1835644692182541,
"eval_runtime": 3841.5239,
"eval_samples_per_second": 1.313,
"eval_steps_per_second": 0.219,
"eval_wer": 14.998927268826431,
"step": 1000
},
{
"epoch": 2.43,
"learning_rate": 8.842222222222223e-05,
"loss": 0.2213,
"step": 1025
},
{
"epoch": 2.49,
"learning_rate": 8.786666666666667e-05,
"loss": 0.2369,
"step": 1050
},
{
"epoch": 2.55,
"learning_rate": 8.731111111111111e-05,
"loss": 0.2331,
"step": 1075
},
{
"epoch": 2.61,
"learning_rate": 8.675555555555556e-05,
"loss": 0.242,
"step": 1100
},
{
"epoch": 2.67,
"learning_rate": 8.620000000000001e-05,
"loss": 0.2202,
"step": 1125
},
{
"epoch": 2.73,
"learning_rate": 8.564444444444445e-05,
"loss": 0.2433,
"step": 1150
},
{
"epoch": 2.79,
"learning_rate": 8.50888888888889e-05,
"loss": 0.2322,
"step": 1175
},
{
"epoch": 2.85,
"learning_rate": 8.453333333333335e-05,
"loss": 0.2293,
"step": 1200
},
{
"epoch": 2.91,
"learning_rate": 8.397777777777778e-05,
"loss": 0.2536,
"step": 1225
},
{
"epoch": 2.97,
"learning_rate": 8.342222222222222e-05,
"loss": 0.2033,
"step": 1250
},
{
"epoch": 3.03,
"learning_rate": 8.286666666666667e-05,
"loss": 0.1874,
"step": 1275
},
{
"epoch": 3.09,
"learning_rate": 8.231111111111112e-05,
"loss": 0.1299,
"step": 1300
},
{
"epoch": 3.15,
"learning_rate": 8.175555555555556e-05,
"loss": 0.121,
"step": 1325
},
{
"epoch": 3.21,
"learning_rate": 8.120000000000001e-05,
"loss": 0.1583,
"step": 1350
},
{
"epoch": 3.27,
"learning_rate": 8.064444444444445e-05,
"loss": 0.1572,
"step": 1375
},
{
"epoch": 3.33,
"learning_rate": 8.00888888888889e-05,
"loss": 0.143,
"step": 1400
},
{
"epoch": 3.38,
"learning_rate": 7.953333333333333e-05,
"loss": 0.1455,
"step": 1425
},
{
"epoch": 3.44,
"learning_rate": 7.897777777777778e-05,
"loss": 0.1275,
"step": 1450
},
{
"epoch": 3.5,
"learning_rate": 7.842222222222223e-05,
"loss": 0.1438,
"step": 1475
},
{
"epoch": 3.56,
"learning_rate": 7.786666666666667e-05,
"loss": 0.1506,
"step": 1500
},
{
"epoch": 3.62,
"learning_rate": 7.731111111111111e-05,
"loss": 0.1527,
"step": 1525
},
{
"epoch": 3.68,
"learning_rate": 7.675555555555556e-05,
"loss": 0.1321,
"step": 1550
},
{
"epoch": 3.74,
"learning_rate": 7.620000000000001e-05,
"loss": 0.1424,
"step": 1575
},
{
"epoch": 3.8,
"learning_rate": 7.564444444444445e-05,
"loss": 0.151,
"step": 1600
},
{
"epoch": 3.86,
"learning_rate": 7.50888888888889e-05,
"loss": 0.1436,
"step": 1625
},
{
"epoch": 3.92,
"learning_rate": 7.453333333333333e-05,
"loss": 0.1442,
"step": 1650
},
{
"epoch": 3.98,
"learning_rate": 7.397777777777778e-05,
"loss": 0.1365,
"step": 1675
},
{
"epoch": 4.04,
"learning_rate": 7.342222222222222e-05,
"loss": 0.1192,
"step": 1700
},
{
"epoch": 4.1,
"learning_rate": 7.286666666666667e-05,
"loss": 0.0831,
"step": 1725
},
{
"epoch": 4.16,
"learning_rate": 7.231111111111112e-05,
"loss": 0.0916,
"step": 1750
},
{
"epoch": 4.22,
"learning_rate": 7.175555555555556e-05,
"loss": 0.1008,
"step": 1775
},
{
"epoch": 4.28,
"learning_rate": 7.12e-05,
"loss": 0.0885,
"step": 1800
},
{
"epoch": 4.33,
"learning_rate": 7.064444444444445e-05,
"loss": 0.1035,
"step": 1825
},
{
"epoch": 4.39,
"learning_rate": 7.00888888888889e-05,
"loss": 0.0881,
"step": 1850
},
{
"epoch": 4.45,
"learning_rate": 6.953333333333333e-05,
"loss": 0.0956,
"step": 1875
},
{
"epoch": 4.51,
"learning_rate": 6.897777777777778e-05,
"loss": 0.0928,
"step": 1900
},
{
"epoch": 4.57,
"learning_rate": 6.842222222222222e-05,
"loss": 0.0819,
"step": 1925
},
{
"epoch": 4.63,
"learning_rate": 6.786666666666667e-05,
"loss": 0.0943,
"step": 1950
},
{
"epoch": 4.69,
"learning_rate": 6.731111111111111e-05,
"loss": 0.0999,
"step": 1975
},
{
"epoch": 4.75,
"learning_rate": 6.675555555555556e-05,
"loss": 0.0873,
"step": 2000
},
{
"epoch": 4.75,
"eval_loss": 0.0633193850517273,
"eval_runtime": 3766.183,
"eval_samples_per_second": 1.339,
"eval_steps_per_second": 0.223,
"eval_wer": 5.086891225059,
"step": 2000
},
{
"epoch": 4.81,
"learning_rate": 6.620000000000001e-05,
"loss": 0.0905,
"step": 2025
},
{
"epoch": 4.87,
"learning_rate": 6.564444444444445e-05,
"loss": 0.0869,
"step": 2050
},
{
"epoch": 4.93,
"learning_rate": 6.508888888888888e-05,
"loss": 0.1147,
"step": 2075
},
{
"epoch": 4.99,
"learning_rate": 6.453333333333333e-05,
"loss": 0.0957,
"step": 2100
},
{
"epoch": 5.05,
"learning_rate": 6.397777777777778e-05,
"loss": 0.0705,
"step": 2125
},
{
"epoch": 5.11,
"learning_rate": 6.342222222222222e-05,
"loss": 0.0548,
"step": 2150
},
{
"epoch": 5.17,
"learning_rate": 6.286666666666667e-05,
"loss": 0.0564,
"step": 2175
},
{
"epoch": 5.23,
"learning_rate": 6.231111111111111e-05,
"loss": 0.0648,
"step": 2200
},
{
"epoch": 5.29,
"learning_rate": 6.175555555555556e-05,
"loss": 0.0606,
"step": 2225
},
{
"epoch": 5.34,
"learning_rate": 6.12e-05,
"loss": 0.0521,
"step": 2250
},
{
"epoch": 5.4,
"learning_rate": 6.064444444444445e-05,
"loss": 0.0618,
"step": 2275
},
{
"epoch": 5.46,
"learning_rate": 6.00888888888889e-05,
"loss": 0.051,
"step": 2300
},
{
"epoch": 5.52,
"learning_rate": 5.953333333333334e-05,
"loss": 0.0607,
"step": 2325
},
{
"epoch": 5.58,
"learning_rate": 5.897777777777778e-05,
"loss": 0.0649,
"step": 2350
},
{
"epoch": 5.64,
"learning_rate": 5.842222222222222e-05,
"loss": 0.0601,
"step": 2375
},
{
"epoch": 5.7,
"learning_rate": 5.7866666666666666e-05,
"loss": 0.0679,
"step": 2400
},
{
"epoch": 5.76,
"learning_rate": 5.7311111111111117e-05,
"loss": 0.0571,
"step": 2425
},
{
"epoch": 5.82,
"learning_rate": 5.675555555555556e-05,
"loss": 0.0628,
"step": 2450
},
{
"epoch": 5.88,
"learning_rate": 5.620000000000001e-05,
"loss": 0.0643,
"step": 2475
},
{
"epoch": 5.94,
"learning_rate": 5.564444444444444e-05,
"loss": 0.0519,
"step": 2500
},
{
"epoch": 6.0,
"learning_rate": 5.508888888888889e-05,
"loss": 0.0535,
"step": 2525
},
{
"epoch": 6.06,
"learning_rate": 5.4533333333333335e-05,
"loss": 0.0367,
"step": 2550
},
{
"epoch": 6.12,
"learning_rate": 5.3977777777777786e-05,
"loss": 0.0286,
"step": 2575
},
{
"epoch": 6.18,
"learning_rate": 5.342222222222223e-05,
"loss": 0.0392,
"step": 2600
},
{
"epoch": 6.24,
"learning_rate": 5.2866666666666666e-05,
"loss": 0.0435,
"step": 2625
},
{
"epoch": 6.29,
"learning_rate": 5.231111111111111e-05,
"loss": 0.0448,
"step": 2650
},
{
"epoch": 6.35,
"learning_rate": 5.1755555555555554e-05,
"loss": 0.0368,
"step": 2675
},
{
"epoch": 6.41,
"learning_rate": 5.1200000000000004e-05,
"loss": 0.0413,
"step": 2700
},
{
"epoch": 6.47,
"learning_rate": 5.064444444444445e-05,
"loss": 0.0446,
"step": 2725
},
{
"epoch": 6.53,
"learning_rate": 5.00888888888889e-05,
"loss": 0.0434,
"step": 2750
},
{
"epoch": 6.59,
"learning_rate": 4.9533333333333336e-05,
"loss": 0.042,
"step": 2775
},
{
"epoch": 6.65,
"learning_rate": 4.897777777777778e-05,
"loss": 0.032,
"step": 2800
},
{
"epoch": 6.71,
"learning_rate": 4.842222222222222e-05,
"loss": 0.0425,
"step": 2825
},
{
"epoch": 6.77,
"learning_rate": 4.7866666666666674e-05,
"loss": 0.0437,
"step": 2850
},
{
"epoch": 6.83,
"learning_rate": 4.731111111111111e-05,
"loss": 0.033,
"step": 2875
},
{
"epoch": 6.89,
"learning_rate": 4.675555555555556e-05,
"loss": 0.0401,
"step": 2900
},
{
"epoch": 6.95,
"learning_rate": 4.6200000000000005e-05,
"loss": 0.0306,
"step": 2925
},
{
"epoch": 7.01,
"learning_rate": 4.564444444444444e-05,
"loss": 0.0302,
"step": 2950
},
{
"epoch": 7.07,
"learning_rate": 4.508888888888889e-05,
"loss": 0.0234,
"step": 2975
},
{
"epoch": 7.13,
"learning_rate": 4.4533333333333336e-05,
"loss": 0.0236,
"step": 3000
},
{
"epoch": 7.13,
"eval_loss": 0.0261550173163414,
"eval_runtime": 3718.4566,
"eval_samples_per_second": 1.356,
"eval_steps_per_second": 0.226,
"eval_wer": 1.808624758635486,
"step": 3000
},
{
"epoch": 7.19,
"learning_rate": 4.397777777777778e-05,
"loss": 0.0321,
"step": 3025
},
{
"epoch": 7.24,
"learning_rate": 4.3422222222222224e-05,
"loss": 0.0212,
"step": 3050
},
{
"epoch": 7.3,
"learning_rate": 4.286666666666667e-05,
"loss": 0.0312,
"step": 3075
},
{
"epoch": 7.36,
"learning_rate": 4.231111111111111e-05,
"loss": 0.0311,
"step": 3100
},
{
"epoch": 7.42,
"learning_rate": 4.175555555555556e-05,
"loss": 0.0165,
"step": 3125
},
{
"epoch": 7.48,
"learning_rate": 4.12e-05,
"loss": 0.0224,
"step": 3150
},
{
"epoch": 7.54,
"learning_rate": 4.064444444444445e-05,
"loss": 0.0244,
"step": 3175
},
{
"epoch": 7.6,
"learning_rate": 4.008888888888889e-05,
"loss": 0.015,
"step": 3200
},
{
"epoch": 7.66,
"learning_rate": 3.9533333333333337e-05,
"loss": 0.0224,
"step": 3225
},
{
"epoch": 7.72,
"learning_rate": 3.897777777777778e-05,
"loss": 0.0188,
"step": 3250
},
{
"epoch": 7.78,
"learning_rate": 3.8422222222222224e-05,
"loss": 0.0259,
"step": 3275
},
{
"epoch": 7.84,
"learning_rate": 3.786666666666667e-05,
"loss": 0.022,
"step": 3300
},
{
"epoch": 7.9,
"learning_rate": 3.731111111111111e-05,
"loss": 0.0217,
"step": 3325
},
{
"epoch": 7.96,
"learning_rate": 3.675555555555556e-05,
"loss": 0.0262,
"step": 3350
},
{
"epoch": 8.02,
"learning_rate": 3.62e-05,
"loss": 0.0222,
"step": 3375
},
{
"epoch": 8.08,
"learning_rate": 3.564444444444445e-05,
"loss": 0.0161,
"step": 3400
},
{
"epoch": 8.14,
"learning_rate": 3.5088888888888886e-05,
"loss": 0.0188,
"step": 3425
},
{
"epoch": 8.19,
"learning_rate": 3.453333333333334e-05,
"loss": 0.0141,
"step": 3450
},
{
"epoch": 8.25,
"learning_rate": 3.397777777777778e-05,
"loss": 0.0104,
"step": 3475
},
{
"epoch": 8.31,
"learning_rate": 3.3422222222222224e-05,
"loss": 0.012,
"step": 3500
},
{
"epoch": 8.37,
"learning_rate": 3.286666666666667e-05,
"loss": 0.0144,
"step": 3525
},
{
"epoch": 8.43,
"learning_rate": 3.231111111111111e-05,
"loss": 0.0054,
"step": 3550
},
{
"epoch": 8.49,
"learning_rate": 3.1755555555555556e-05,
"loss": 0.0177,
"step": 3575
},
{
"epoch": 8.55,
"learning_rate": 3.12e-05,
"loss": 0.0105,
"step": 3600
},
{
"epoch": 8.61,
"learning_rate": 3.064444444444445e-05,
"loss": 0.0093,
"step": 3625
},
{
"epoch": 8.67,
"learning_rate": 3.008888888888889e-05,
"loss": 0.0162,
"step": 3650
},
{
"epoch": 8.73,
"learning_rate": 2.9533333333333334e-05,
"loss": 0.0144,
"step": 3675
},
{
"epoch": 8.79,
"learning_rate": 2.897777777777778e-05,
"loss": 0.0045,
"step": 3700
},
{
"epoch": 8.85,
"learning_rate": 2.842222222222222e-05,
"loss": 0.0119,
"step": 3725
},
{
"epoch": 8.91,
"learning_rate": 2.786666666666667e-05,
"loss": 0.0107,
"step": 3750
},
{
"epoch": 8.97,
"learning_rate": 2.7311111111111116e-05,
"loss": 0.0108,
"step": 3775
},
{
"epoch": 9.03,
"learning_rate": 2.6755555555555556e-05,
"loss": 0.0079,
"step": 3800
},
{
"epoch": 9.09,
"learning_rate": 2.6200000000000003e-05,
"loss": 0.0079,
"step": 3825
},
{
"epoch": 9.14,
"learning_rate": 2.5644444444444444e-05,
"loss": 0.0056,
"step": 3850
},
{
"epoch": 9.2,
"learning_rate": 2.508888888888889e-05,
"loss": 0.0051,
"step": 3875
},
{
"epoch": 9.26,
"learning_rate": 2.4533333333333334e-05,
"loss": 0.0065,
"step": 3900
},
{
"epoch": 9.32,
"learning_rate": 2.3977777777777778e-05,
"loss": 0.0046,
"step": 3925
},
{
"epoch": 9.38,
"learning_rate": 2.3422222222222222e-05,
"loss": 0.0062,
"step": 3950
},
{
"epoch": 9.44,
"learning_rate": 2.2866666666666666e-05,
"loss": 0.0059,
"step": 3975
},
{
"epoch": 9.5,
"learning_rate": 2.2311111111111113e-05,
"loss": 0.0068,
"step": 4000
},
{
"epoch": 9.5,
"eval_loss": 0.006168792489916086,
"eval_runtime": 3744.22,
"eval_samples_per_second": 1.347,
"eval_steps_per_second": 0.225,
"eval_wer": 0.6221840806693842,
"step": 4000
},
{
"epoch": 9.56,
"learning_rate": 2.1755555555555557e-05,
"loss": 0.0071,
"step": 4025
},
{
"epoch": 9.62,
"learning_rate": 2.12e-05,
"loss": 0.0096,
"step": 4050
},
{
"epoch": 9.68,
"learning_rate": 2.0644444444444447e-05,
"loss": 0.0059,
"step": 4075
},
{
"epoch": 9.74,
"learning_rate": 2.008888888888889e-05,
"loss": 0.0052,
"step": 4100
},
{
"epoch": 9.8,
"learning_rate": 1.9533333333333335e-05,
"loss": 0.0057,
"step": 4125
},
{
"epoch": 9.86,
"learning_rate": 1.897777777777778e-05,
"loss": 0.0065,
"step": 4150
},
{
"epoch": 9.92,
"learning_rate": 1.8422222222222222e-05,
"loss": 0.004,
"step": 4175
},
{
"epoch": 9.98,
"learning_rate": 1.7866666666666666e-05,
"loss": 0.0099,
"step": 4200
},
{
"epoch": 10.04,
"learning_rate": 1.731111111111111e-05,
"loss": 0.0038,
"step": 4225
},
{
"epoch": 10.1,
"learning_rate": 1.6755555555555557e-05,
"loss": 0.0035,
"step": 4250
},
{
"epoch": 10.15,
"learning_rate": 1.62e-05,
"loss": 0.0066,
"step": 4275
},
{
"epoch": 10.21,
"learning_rate": 1.5644444444444444e-05,
"loss": 0.0037,
"step": 4300
},
{
"epoch": 10.27,
"learning_rate": 1.5088888888888888e-05,
"loss": 0.0041,
"step": 4325
},
{
"epoch": 10.33,
"learning_rate": 1.4533333333333335e-05,
"loss": 0.0027,
"step": 4350
},
{
"epoch": 10.39,
"learning_rate": 1.3977777777777779e-05,
"loss": 0.0014,
"step": 4375
},
{
"epoch": 10.45,
"learning_rate": 1.3422222222222223e-05,
"loss": 0.0027,
"step": 4400
},
{
"epoch": 10.51,
"learning_rate": 1.2866666666666668e-05,
"loss": 0.0036,
"step": 4425
},
{
"epoch": 10.57,
"learning_rate": 1.2311111111111112e-05,
"loss": 0.0017,
"step": 4450
},
{
"epoch": 10.63,
"learning_rate": 1.1755555555555556e-05,
"loss": 0.0039,
"step": 4475
},
{
"epoch": 10.69,
"learning_rate": 1.1200000000000001e-05,
"loss": 0.0008,
"step": 4500
},
{
"epoch": 10.75,
"learning_rate": 1.0644444444444445e-05,
"loss": 0.002,
"step": 4525
},
{
"epoch": 10.81,
"learning_rate": 1.0088888888888889e-05,
"loss": 0.0025,
"step": 4550
},
{
"epoch": 10.87,
"learning_rate": 9.533333333333334e-06,
"loss": 0.0011,
"step": 4575
},
{
"epoch": 10.93,
"learning_rate": 8.977777777777778e-06,
"loss": 0.0022,
"step": 4600
},
{
"epoch": 10.99,
"learning_rate": 8.422222222222223e-06,
"loss": 0.0007,
"step": 4625
},
{
"epoch": 11.05,
"learning_rate": 7.866666666666667e-06,
"loss": 0.0003,
"step": 4650
},
{
"epoch": 11.1,
"learning_rate": 7.311111111111112e-06,
"loss": 0.0003,
"step": 4675
},
{
"epoch": 11.16,
"learning_rate": 6.755555555555555e-06,
"loss": 0.0007,
"step": 4700
},
{
"epoch": 11.22,
"learning_rate": 6.2e-06,
"loss": 0.0003,
"step": 4725
},
{
"epoch": 11.28,
"learning_rate": 5.6444444444444445e-06,
"loss": 0.0003,
"step": 4750
},
{
"epoch": 11.34,
"learning_rate": 5.088888888888889e-06,
"loss": 0.0006,
"step": 4775
},
{
"epoch": 11.4,
"learning_rate": 4.533333333333334e-06,
"loss": 0.0003,
"step": 4800
},
{
"epoch": 11.46,
"learning_rate": 3.977777777777778e-06,
"loss": 0.0003,
"step": 4825
},
{
"epoch": 11.52,
"learning_rate": 3.4222222222222224e-06,
"loss": 0.0003,
"step": 4850
},
{
"epoch": 11.58,
"learning_rate": 2.8666666666666666e-06,
"loss": 0.0016,
"step": 4875
},
{
"epoch": 11.64,
"learning_rate": 2.311111111111111e-06,
"loss": 0.0003,
"step": 4900
},
{
"epoch": 11.7,
"learning_rate": 1.7555555555555558e-06,
"loss": 0.0003,
"step": 4925
},
{
"epoch": 11.76,
"learning_rate": 1.2000000000000002e-06,
"loss": 0.0003,
"step": 4950
},
{
"epoch": 11.82,
"learning_rate": 6.444444444444444e-07,
"loss": 0.0027,
"step": 4975
},
{
"epoch": 11.88,
"learning_rate": 8.88888888888889e-08,
"loss": 0.0002,
"step": 5000
},
{
"epoch": 11.88,
"eval_loss": 0.00039999079308472574,
"eval_runtime": 3766.4449,
"eval_samples_per_second": 1.339,
"eval_steps_per_second": 0.223,
"eval_wer": 0.3947650718729886,
"step": 5000
},
{
"epoch": 11.88,
"step": 5000,
"total_flos": 1.728655406456832e+19,
"train_loss": 0.13066921601230277,
"train_runtime": 56461.7194,
"train_samples_per_second": 1.063,
"train_steps_per_second": 0.089
}
],
"max_steps": 5000,
"num_train_epochs": 12,
"total_flos": 1.728655406456832e+19,
"trial_name": null,
"trial_params": null
}