whisper-small-ja / trainer_state.json
jakeyoo's picture
End of training
e755e23
raw history blame
No virus
25.8 kB
{
"best_metric": 68.94594978011065,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 8.1024,
"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": 4.4e-07,
"loss": 0.4853,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 0.3865,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 0.3522,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.94e-06,
"loss": 0.3163,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.2694,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.2778,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.44e-06,
"loss": 0.2661,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.94e-06,
"loss": 0.2548,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.440000000000001e-06,
"loss": 0.2881,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 4.94e-06,
"loss": 0.3555,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.3511,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 5.94e-06,
"loss": 0.3755,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 6.440000000000001e-06,
"loss": 0.4138,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.4444,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 7.440000000000001e-06,
"loss": 0.4261,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 7.94e-06,
"loss": 0.3986,
"step": 400
},
{
"epoch": 0.09,
"learning_rate": 8.44e-06,
"loss": 0.4023,
"step": 425
},
{
"epoch": 0.09,
"learning_rate": 8.94e-06,
"loss": 0.3332,
"step": 450
},
{
"epoch": 0.1,
"learning_rate": 9.440000000000001e-06,
"loss": 0.3155,
"step": 475
},
{
"epoch": 0.1,
"learning_rate": 9.940000000000001e-06,
"loss": 0.3369,
"step": 500
},
{
"epoch": 0.1,
"learning_rate": 9.951111111111111e-06,
"loss": 0.3411,
"step": 525
},
{
"epoch": 0.11,
"learning_rate": 9.895555555555557e-06,
"loss": 0.3097,
"step": 550
},
{
"epoch": 1.0,
"learning_rate": 9.84e-06,
"loss": 0.2865,
"step": 575
},
{
"epoch": 1.01,
"learning_rate": 9.784444444444445e-06,
"loss": 0.2581,
"step": 600
},
{
"epoch": 1.01,
"learning_rate": 9.72888888888889e-06,
"loss": 0.2263,
"step": 625
},
{
"epoch": 1.02,
"learning_rate": 9.673333333333334e-06,
"loss": 0.2226,
"step": 650
},
{
"epoch": 1.02,
"learning_rate": 9.617777777777778e-06,
"loss": 0.2168,
"step": 675
},
{
"epoch": 1.03,
"learning_rate": 9.562222222222223e-06,
"loss": 0.1895,
"step": 700
},
{
"epoch": 1.03,
"learning_rate": 9.506666666666667e-06,
"loss": 0.173,
"step": 725
},
{
"epoch": 1.04,
"learning_rate": 9.451111111111112e-06,
"loss": 0.1555,
"step": 750
},
{
"epoch": 1.04,
"learning_rate": 9.395555555555556e-06,
"loss": 0.1849,
"step": 775
},
{
"epoch": 1.05,
"learning_rate": 9.340000000000002e-06,
"loss": 0.205,
"step": 800
},
{
"epoch": 1.05,
"learning_rate": 9.284444444444444e-06,
"loss": 0.2026,
"step": 825
},
{
"epoch": 1.06,
"learning_rate": 9.22888888888889e-06,
"loss": 0.2183,
"step": 850
},
{
"epoch": 1.06,
"learning_rate": 9.173333333333334e-06,
"loss": 0.2295,
"step": 875
},
{
"epoch": 1.07,
"learning_rate": 9.117777777777778e-06,
"loss": 0.2396,
"step": 900
},
{
"epoch": 1.07,
"learning_rate": 9.062222222222224e-06,
"loss": 0.2419,
"step": 925
},
{
"epoch": 1.08,
"learning_rate": 9.006666666666666e-06,
"loss": 0.2356,
"step": 950
},
{
"epoch": 1.08,
"learning_rate": 8.951111111111112e-06,
"loss": 0.1935,
"step": 975
},
{
"epoch": 1.09,
"learning_rate": 8.895555555555556e-06,
"loss": 0.1938,
"step": 1000
},
{
"epoch": 1.09,
"eval_loss": 0.2840821146965027,
"eval_runtime": 1882.7876,
"eval_samples_per_second": 2.445,
"eval_steps_per_second": 0.306,
"eval_wer": 74.66307277628033,
"step": 1000
},
{
"epoch": 1.09,
"learning_rate": 8.84e-06,
"loss": 0.1567,
"step": 1025
},
{
"epoch": 1.1,
"learning_rate": 8.784444444444446e-06,
"loss": 0.1469,
"step": 1050
},
{
"epoch": 1.1,
"learning_rate": 8.72888888888889e-06,
"loss": 0.1598,
"step": 1075
},
{
"epoch": 1.11,
"learning_rate": 8.673333333333334e-06,
"loss": 0.1409,
"step": 1100
},
{
"epoch": 2.0,
"learning_rate": 8.617777777777778e-06,
"loss": 0.1483,
"step": 1125
},
{
"epoch": 2.01,
"learning_rate": 8.562222222222224e-06,
"loss": 0.1105,
"step": 1150
},
{
"epoch": 2.01,
"learning_rate": 8.506666666666668e-06,
"loss": 0.1066,
"step": 1175
},
{
"epoch": 2.02,
"learning_rate": 8.451111111111112e-06,
"loss": 0.117,
"step": 1200
},
{
"epoch": 2.02,
"learning_rate": 8.395555555555557e-06,
"loss": 0.0923,
"step": 1225
},
{
"epoch": 2.03,
"learning_rate": 8.34e-06,
"loss": 0.0896,
"step": 1250
},
{
"epoch": 2.03,
"learning_rate": 8.284444444444446e-06,
"loss": 0.0815,
"step": 1275
},
{
"epoch": 2.04,
"learning_rate": 8.22888888888889e-06,
"loss": 0.0745,
"step": 1300
},
{
"epoch": 2.04,
"learning_rate": 8.173333333333334e-06,
"loss": 0.0682,
"step": 1325
},
{
"epoch": 2.05,
"learning_rate": 8.11777777777778e-06,
"loss": 0.0874,
"step": 1350
},
{
"epoch": 2.05,
"learning_rate": 8.062222222222222e-06,
"loss": 0.1075,
"step": 1375
},
{
"epoch": 2.06,
"learning_rate": 8.006666666666667e-06,
"loss": 0.1029,
"step": 1400
},
{
"epoch": 2.06,
"learning_rate": 7.951111111111111e-06,
"loss": 0.0912,
"step": 1425
},
{
"epoch": 2.07,
"learning_rate": 7.895555555555557e-06,
"loss": 0.0998,
"step": 1450
},
{
"epoch": 2.07,
"learning_rate": 7.840000000000001e-06,
"loss": 0.1104,
"step": 1475
},
{
"epoch": 2.08,
"learning_rate": 7.784444444444445e-06,
"loss": 0.11,
"step": 1500
},
{
"epoch": 2.08,
"learning_rate": 7.72888888888889e-06,
"loss": 0.1007,
"step": 1525
},
{
"epoch": 2.09,
"learning_rate": 7.673333333333333e-06,
"loss": 0.0844,
"step": 1550
},
{
"epoch": 2.09,
"learning_rate": 7.617777777777778e-06,
"loss": 0.0763,
"step": 1575
},
{
"epoch": 2.1,
"learning_rate": 7.562222222222223e-06,
"loss": 0.0722,
"step": 1600
},
{
"epoch": 2.1,
"learning_rate": 7.506666666666668e-06,
"loss": 0.0727,
"step": 1625
},
{
"epoch": 2.11,
"learning_rate": 7.451111111111111e-06,
"loss": 0.0701,
"step": 1650
},
{
"epoch": 2.11,
"learning_rate": 7.395555555555556e-06,
"loss": 0.0634,
"step": 1675
},
{
"epoch": 3.0,
"learning_rate": 7.340000000000001e-06,
"loss": 0.0564,
"step": 1700
},
{
"epoch": 3.01,
"learning_rate": 7.284444444444445e-06,
"loss": 0.0456,
"step": 1725
},
{
"epoch": 3.01,
"learning_rate": 7.22888888888889e-06,
"loss": 0.044,
"step": 1750
},
{
"epoch": 3.02,
"learning_rate": 7.173333333333335e-06,
"loss": 0.0393,
"step": 1775
},
{
"epoch": 3.02,
"learning_rate": 7.117777777777778e-06,
"loss": 0.0363,
"step": 1800
},
{
"epoch": 3.03,
"learning_rate": 7.062222222222223e-06,
"loss": 0.0385,
"step": 1825
},
{
"epoch": 3.03,
"learning_rate": 7.006666666666667e-06,
"loss": 0.0329,
"step": 1850
},
{
"epoch": 3.04,
"learning_rate": 6.951111111111112e-06,
"loss": 0.0252,
"step": 1875
},
{
"epoch": 3.04,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.0346,
"step": 1900
},
{
"epoch": 3.05,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.0431,
"step": 1925
},
{
"epoch": 3.05,
"learning_rate": 6.784444444444445e-06,
"loss": 0.043,
"step": 1950
},
{
"epoch": 3.06,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.0424,
"step": 1975
},
{
"epoch": 3.06,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.0466,
"step": 2000
},
{
"epoch": 3.06,
"eval_loss": 0.2996249198913574,
"eval_runtime": 1904.3889,
"eval_samples_per_second": 2.418,
"eval_steps_per_second": 0.302,
"eval_wer": 72.0953326713009,
"step": 2000
},
{
"epoch": 3.07,
"learning_rate": 6.617777777777778e-06,
"loss": 0.0444,
"step": 2025
},
{
"epoch": 3.07,
"learning_rate": 6.562222222222223e-06,
"loss": 0.0476,
"step": 2050
},
{
"epoch": 3.08,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.0368,
"step": 2075
},
{
"epoch": 3.08,
"learning_rate": 6.451111111111111e-06,
"loss": 0.0386,
"step": 2100
},
{
"epoch": 3.09,
"learning_rate": 6.395555555555556e-06,
"loss": 0.0366,
"step": 2125
},
{
"epoch": 3.09,
"learning_rate": 6.34e-06,
"loss": 0.0335,
"step": 2150
},
{
"epoch": 3.1,
"learning_rate": 6.284444444444445e-06,
"loss": 0.0302,
"step": 2175
},
{
"epoch": 3.1,
"learning_rate": 6.22888888888889e-06,
"loss": 0.0261,
"step": 2200
},
{
"epoch": 3.11,
"learning_rate": 6.173333333333333e-06,
"loss": 0.0312,
"step": 2225
},
{
"epoch": 4.0,
"learning_rate": 6.117777777777778e-06,
"loss": 0.0249,
"step": 2250
},
{
"epoch": 4.01,
"learning_rate": 6.062222222222223e-06,
"loss": 0.022,
"step": 2275
},
{
"epoch": 4.01,
"learning_rate": 6.006666666666667e-06,
"loss": 0.0216,
"step": 2300
},
{
"epoch": 4.02,
"learning_rate": 5.951111111111112e-06,
"loss": 0.0168,
"step": 2325
},
{
"epoch": 4.02,
"learning_rate": 5.895555555555557e-06,
"loss": 0.0149,
"step": 2350
},
{
"epoch": 4.03,
"learning_rate": 5.84e-06,
"loss": 0.0145,
"step": 2375
},
{
"epoch": 4.03,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0122,
"step": 2400
},
{
"epoch": 4.04,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0123,
"step": 2425
},
{
"epoch": 4.04,
"learning_rate": 5.673333333333334e-06,
"loss": 0.013,
"step": 2450
},
{
"epoch": 4.05,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0178,
"step": 2475
},
{
"epoch": 4.05,
"learning_rate": 5.562222222222222e-06,
"loss": 0.0172,
"step": 2500
},
{
"epoch": 4.06,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0188,
"step": 2525
},
{
"epoch": 4.06,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0202,
"step": 2550
},
{
"epoch": 4.07,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0214,
"step": 2575
},
{
"epoch": 4.07,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.0182,
"step": 2600
},
{
"epoch": 4.08,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0176,
"step": 2625
},
{
"epoch": 4.08,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0186,
"step": 2650
},
{
"epoch": 4.09,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0175,
"step": 2675
},
{
"epoch": 4.09,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0152,
"step": 2700
},
{
"epoch": 4.1,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0124,
"step": 2725
},
{
"epoch": 4.1,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0109,
"step": 2750
},
{
"epoch": 4.11,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0135,
"step": 2775
},
{
"epoch": 4.11,
"learning_rate": 4.895555555555556e-06,
"loss": 0.0101,
"step": 2800
},
{
"epoch": 5.0,
"learning_rate": 4.84e-06,
"loss": 0.009,
"step": 2825
},
{
"epoch": 5.01,
"learning_rate": 4.784444444444445e-06,
"loss": 0.0086,
"step": 2850
},
{
"epoch": 5.01,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0072,
"step": 2875
},
{
"epoch": 5.02,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0062,
"step": 2900
},
{
"epoch": 5.02,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0057,
"step": 2925
},
{
"epoch": 5.03,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0056,
"step": 2950
},
{
"epoch": 5.03,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0047,
"step": 2975
},
{
"epoch": 5.04,
"learning_rate": 4.451111111111112e-06,
"loss": 0.005,
"step": 3000
},
{
"epoch": 5.04,
"eval_loss": 0.3375891149044037,
"eval_runtime": 1894.1398,
"eval_samples_per_second": 2.431,
"eval_steps_per_second": 0.304,
"eval_wer": 70.43552276918712,
"step": 3000
},
{
"epoch": 5.04,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0081,
"step": 3025
},
{
"epoch": 5.05,
"learning_rate": 4.34e-06,
"loss": 0.0077,
"step": 3050
},
{
"epoch": 5.05,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0082,
"step": 3075
},
{
"epoch": 5.06,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0079,
"step": 3100
},
{
"epoch": 5.06,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0092,
"step": 3125
},
{
"epoch": 5.07,
"learning_rate": 4.117777777777779e-06,
"loss": 0.009,
"step": 3150
},
{
"epoch": 5.07,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0088,
"step": 3175
},
{
"epoch": 5.08,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0083,
"step": 3200
},
{
"epoch": 5.08,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0087,
"step": 3225
},
{
"epoch": 5.09,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0068,
"step": 3250
},
{
"epoch": 5.09,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0054,
"step": 3275
},
{
"epoch": 5.1,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0065,
"step": 3300
},
{
"epoch": 5.1,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0056,
"step": 3325
},
{
"epoch": 5.11,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0052,
"step": 3350
},
{
"epoch": 6.0,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0046,
"step": 3375
},
{
"epoch": 6.01,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0037,
"step": 3400
},
{
"epoch": 6.01,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.0034,
"step": 3425
},
{
"epoch": 6.02,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0033,
"step": 3450
},
{
"epoch": 6.02,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0031,
"step": 3475
},
{
"epoch": 6.03,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0041,
"step": 3500
},
{
"epoch": 6.03,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0033,
"step": 3525
},
{
"epoch": 6.04,
"learning_rate": 3.228888888888889e-06,
"loss": 0.0024,
"step": 3550
},
{
"epoch": 6.04,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0026,
"step": 3575
},
{
"epoch": 6.05,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0034,
"step": 3600
},
{
"epoch": 6.05,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.0047,
"step": 3625
},
{
"epoch": 6.06,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0043,
"step": 3650
},
{
"epoch": 6.06,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0038,
"step": 3675
},
{
"epoch": 6.07,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0038,
"step": 3700
},
{
"epoch": 6.07,
"learning_rate": 2.84e-06,
"loss": 0.0044,
"step": 3725
},
{
"epoch": 6.08,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0042,
"step": 3750
},
{
"epoch": 6.08,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0043,
"step": 3775
},
{
"epoch": 6.09,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0038,
"step": 3800
},
{
"epoch": 6.09,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0031,
"step": 3825
},
{
"epoch": 6.1,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.003,
"step": 3850
},
{
"epoch": 6.1,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0032,
"step": 3875
},
{
"epoch": 6.11,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0028,
"step": 3900
},
{
"epoch": 6.11,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0027,
"step": 3925
},
{
"epoch": 7.0,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0027,
"step": 3950
},
{
"epoch": 7.01,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0022,
"step": 3975
},
{
"epoch": 7.01,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0021,
"step": 4000
},
{
"epoch": 7.01,
"eval_loss": 0.36166784167289734,
"eval_runtime": 1873.4041,
"eval_samples_per_second": 2.458,
"eval_steps_per_second": 0.307,
"eval_wer": 68.94594978011065,
"step": 4000
},
{
"epoch": 7.02,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0021,
"step": 4025
},
{
"epoch": 7.02,
"learning_rate": 2.117777777777778e-06,
"loss": 0.002,
"step": 4050
},
{
"epoch": 7.03,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.002,
"step": 4075
},
{
"epoch": 7.03,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0022,
"step": 4100
},
{
"epoch": 7.04,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0018,
"step": 4125
},
{
"epoch": 7.04,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0021,
"step": 4150
},
{
"epoch": 7.05,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0023,
"step": 4175
},
{
"epoch": 7.05,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0023,
"step": 4200
},
{
"epoch": 7.06,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0026,
"step": 4225
},
{
"epoch": 7.06,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0024,
"step": 4250
},
{
"epoch": 7.07,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0026,
"step": 4275
},
{
"epoch": 7.07,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0029,
"step": 4300
},
{
"epoch": 7.08,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0025,
"step": 4325
},
{
"epoch": 7.08,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0025,
"step": 4350
},
{
"epoch": 7.09,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0022,
"step": 4375
},
{
"epoch": 7.09,
"learning_rate": 1.34e-06,
"loss": 0.0022,
"step": 4400
},
{
"epoch": 7.1,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0021,
"step": 4425
},
{
"epoch": 7.1,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0021,
"step": 4450
},
{
"epoch": 7.11,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.002,
"step": 4475
},
{
"epoch": 8.0,
"learning_rate": 1.117777777777778e-06,
"loss": 0.0019,
"step": 4500
},
{
"epoch": 8.01,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.0017,
"step": 4525
},
{
"epoch": 8.01,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0017,
"step": 4550
},
{
"epoch": 8.02,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0017,
"step": 4575
},
{
"epoch": 8.02,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0015,
"step": 4600
},
{
"epoch": 8.03,
"learning_rate": 8.400000000000001e-07,
"loss": 0.0015,
"step": 4625
},
{
"epoch": 8.03,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0015,
"step": 4650
},
{
"epoch": 8.04,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0015,
"step": 4675
},
{
"epoch": 8.04,
"learning_rate": 6.733333333333334e-07,
"loss": 0.0015,
"step": 4700
},
{
"epoch": 8.05,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0018,
"step": 4725
},
{
"epoch": 8.05,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0019,
"step": 4750
},
{
"epoch": 8.06,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0019,
"step": 4775
},
{
"epoch": 8.06,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0023,
"step": 4800
},
{
"epoch": 8.07,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.0022,
"step": 4825
},
{
"epoch": 8.07,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0022,
"step": 4850
},
{
"epoch": 8.08,
"learning_rate": 2.844444444444445e-07,
"loss": 0.0022,
"step": 4875
},
{
"epoch": 8.08,
"learning_rate": 2.2888888888888892e-07,
"loss": 0.0022,
"step": 4900
},
{
"epoch": 8.09,
"learning_rate": 1.7333333333333335e-07,
"loss": 0.002,
"step": 4925
},
{
"epoch": 8.09,
"learning_rate": 1.1777777777777778e-07,
"loss": 0.0019,
"step": 4950
},
{
"epoch": 8.1,
"learning_rate": 6.222222222222223e-08,
"loss": 0.0019,
"step": 4975
},
{
"epoch": 8.1,
"learning_rate": 6.666666666666667e-09,
"loss": 0.002,
"step": 5000
},
{
"epoch": 8.1,
"eval_loss": 0.37349361181259155,
"eval_runtime": 1909.1983,
"eval_samples_per_second": 2.411,
"eval_steps_per_second": 0.302,
"eval_wer": 71.47113065683077,
"step": 5000
},
{
"epoch": 8.1,
"step": 5000,
"total_flos": 2.307297992638464e+19,
"train_loss": 0.07939068283215166,
"train_runtime": 25230.8329,
"train_samples_per_second": 3.171,
"train_steps_per_second": 0.198
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 2.307297992638464e+19,
"trial_name": null,
"trial_params": null
}