TheRains's picture
End of training
80e60ba
raw
history blame
No virus
25.8 kB
{
"best_metric": 12.680009201748332,
"best_model_checkpoint": "./output/small/cv9-special-batch8-adamw_torch-small/checkpoint-4000",
"epoch": 4.84027105517909,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 4.4e-07,
"loss": 1.291,
"step": 25
},
{
"epoch": 0.05,
"learning_rate": 9.400000000000001e-07,
"loss": 1.2727,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 1.44e-06,
"loss": 1.0164,
"step": 75
},
{
"epoch": 0.1,
"learning_rate": 1.94e-06,
"loss": 0.7644,
"step": 100
},
{
"epoch": 0.12,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.5502,
"step": 125
},
{
"epoch": 0.15,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.489,
"step": 150
},
{
"epoch": 0.17,
"learning_rate": 3.44e-06,
"loss": 0.4795,
"step": 175
},
{
"epoch": 0.19,
"learning_rate": 3.94e-06,
"loss": 0.4612,
"step": 200
},
{
"epoch": 0.22,
"learning_rate": 4.440000000000001e-06,
"loss": 0.385,
"step": 225
},
{
"epoch": 0.24,
"learning_rate": 4.94e-06,
"loss": 0.3829,
"step": 250
},
{
"epoch": 0.27,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.4168,
"step": 275
},
{
"epoch": 0.29,
"learning_rate": 5.94e-06,
"loss": 0.4124,
"step": 300
},
{
"epoch": 0.31,
"learning_rate": 6.440000000000001e-06,
"loss": 0.3616,
"step": 325
},
{
"epoch": 0.34,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.3877,
"step": 350
},
{
"epoch": 0.36,
"learning_rate": 7.440000000000001e-06,
"loss": 0.419,
"step": 375
},
{
"epoch": 0.39,
"learning_rate": 7.94e-06,
"loss": 0.3886,
"step": 400
},
{
"epoch": 0.41,
"learning_rate": 8.44e-06,
"loss": 0.3654,
"step": 425
},
{
"epoch": 0.44,
"learning_rate": 8.94e-06,
"loss": 0.3912,
"step": 450
},
{
"epoch": 0.46,
"learning_rate": 9.440000000000001e-06,
"loss": 0.3196,
"step": 475
},
{
"epoch": 0.48,
"learning_rate": 9.940000000000001e-06,
"loss": 0.3483,
"step": 500
},
{
"epoch": 0.51,
"learning_rate": 9.951111111111111e-06,
"loss": 0.3825,
"step": 525
},
{
"epoch": 0.53,
"learning_rate": 9.895555555555557e-06,
"loss": 0.3291,
"step": 550
},
{
"epoch": 0.56,
"learning_rate": 9.84e-06,
"loss": 0.3547,
"step": 575
},
{
"epoch": 0.58,
"learning_rate": 9.784444444444445e-06,
"loss": 0.3867,
"step": 600
},
{
"epoch": 0.61,
"learning_rate": 9.72888888888889e-06,
"loss": 0.3419,
"step": 625
},
{
"epoch": 0.63,
"learning_rate": 9.673333333333334e-06,
"loss": 0.3171,
"step": 650
},
{
"epoch": 0.65,
"learning_rate": 9.617777777777778e-06,
"loss": 0.3471,
"step": 675
},
{
"epoch": 0.68,
"learning_rate": 9.562222222222223e-06,
"loss": 0.3578,
"step": 700
},
{
"epoch": 0.7,
"learning_rate": 9.506666666666667e-06,
"loss": 0.3824,
"step": 725
},
{
"epoch": 0.73,
"learning_rate": 9.451111111111112e-06,
"loss": 0.3349,
"step": 750
},
{
"epoch": 0.75,
"learning_rate": 9.395555555555556e-06,
"loss": 0.3565,
"step": 775
},
{
"epoch": 0.77,
"learning_rate": 9.340000000000002e-06,
"loss": 0.3017,
"step": 800
},
{
"epoch": 0.8,
"learning_rate": 9.284444444444444e-06,
"loss": 0.3166,
"step": 825
},
{
"epoch": 0.82,
"learning_rate": 9.22888888888889e-06,
"loss": 0.3195,
"step": 850
},
{
"epoch": 0.85,
"learning_rate": 9.173333333333334e-06,
"loss": 0.3065,
"step": 875
},
{
"epoch": 0.87,
"learning_rate": 9.117777777777778e-06,
"loss": 0.3197,
"step": 900
},
{
"epoch": 0.9,
"learning_rate": 9.062222222222224e-06,
"loss": 0.3438,
"step": 925
},
{
"epoch": 0.92,
"learning_rate": 9.006666666666666e-06,
"loss": 0.3153,
"step": 950
},
{
"epoch": 0.94,
"learning_rate": 8.951111111111112e-06,
"loss": 0.3428,
"step": 975
},
{
"epoch": 0.97,
"learning_rate": 8.895555555555556e-06,
"loss": 0.3042,
"step": 1000
},
{
"epoch": 0.97,
"eval_loss": 0.26070263981819153,
"eval_runtime": 596.4653,
"eval_samples_per_second": 6.072,
"eval_steps_per_second": 1.519,
"eval_wer": 14.667586841499885,
"step": 1000
},
{
"epoch": 0.99,
"learning_rate": 8.84e-06,
"loss": 0.3678,
"step": 1025
},
{
"epoch": 1.02,
"learning_rate": 8.784444444444446e-06,
"loss": 0.2241,
"step": 1050
},
{
"epoch": 1.04,
"learning_rate": 8.72888888888889e-06,
"loss": 0.1568,
"step": 1075
},
{
"epoch": 1.06,
"learning_rate": 8.673333333333334e-06,
"loss": 0.1624,
"step": 1100
},
{
"epoch": 1.09,
"learning_rate": 8.617777777777778e-06,
"loss": 0.1814,
"step": 1125
},
{
"epoch": 1.11,
"learning_rate": 8.562222222222224e-06,
"loss": 0.1527,
"step": 1150
},
{
"epoch": 1.14,
"learning_rate": 8.506666666666668e-06,
"loss": 0.1669,
"step": 1175
},
{
"epoch": 1.16,
"learning_rate": 8.451111111111112e-06,
"loss": 0.1631,
"step": 1200
},
{
"epoch": 1.19,
"learning_rate": 8.395555555555557e-06,
"loss": 0.1403,
"step": 1225
},
{
"epoch": 1.21,
"learning_rate": 8.34e-06,
"loss": 0.1646,
"step": 1250
},
{
"epoch": 1.23,
"learning_rate": 8.284444444444446e-06,
"loss": 0.1599,
"step": 1275
},
{
"epoch": 1.26,
"learning_rate": 8.22888888888889e-06,
"loss": 0.1516,
"step": 1300
},
{
"epoch": 1.28,
"learning_rate": 8.173333333333334e-06,
"loss": 0.1632,
"step": 1325
},
{
"epoch": 1.31,
"learning_rate": 8.11777777777778e-06,
"loss": 0.1468,
"step": 1350
},
{
"epoch": 1.33,
"learning_rate": 8.062222222222222e-06,
"loss": 0.1676,
"step": 1375
},
{
"epoch": 1.36,
"learning_rate": 8.006666666666667e-06,
"loss": 0.1562,
"step": 1400
},
{
"epoch": 1.38,
"learning_rate": 7.951111111111111e-06,
"loss": 0.1596,
"step": 1425
},
{
"epoch": 1.4,
"learning_rate": 7.895555555555557e-06,
"loss": 0.1568,
"step": 1450
},
{
"epoch": 1.43,
"learning_rate": 7.840000000000001e-06,
"loss": 0.1564,
"step": 1475
},
{
"epoch": 1.45,
"learning_rate": 7.784444444444445e-06,
"loss": 0.1842,
"step": 1500
},
{
"epoch": 1.48,
"learning_rate": 7.72888888888889e-06,
"loss": 0.1356,
"step": 1525
},
{
"epoch": 1.5,
"learning_rate": 7.673333333333333e-06,
"loss": 0.1716,
"step": 1550
},
{
"epoch": 1.52,
"learning_rate": 7.617777777777778e-06,
"loss": 0.1527,
"step": 1575
},
{
"epoch": 1.55,
"learning_rate": 7.562222222222223e-06,
"loss": 0.1571,
"step": 1600
},
{
"epoch": 1.57,
"learning_rate": 7.506666666666668e-06,
"loss": 0.1495,
"step": 1625
},
{
"epoch": 1.6,
"learning_rate": 7.451111111111111e-06,
"loss": 0.1484,
"step": 1650
},
{
"epoch": 1.62,
"learning_rate": 7.395555555555556e-06,
"loss": 0.1811,
"step": 1675
},
{
"epoch": 1.65,
"learning_rate": 7.340000000000001e-06,
"loss": 0.1809,
"step": 1700
},
{
"epoch": 1.67,
"learning_rate": 7.284444444444445e-06,
"loss": 0.1649,
"step": 1725
},
{
"epoch": 1.69,
"learning_rate": 7.22888888888889e-06,
"loss": 0.1419,
"step": 1750
},
{
"epoch": 1.72,
"learning_rate": 7.173333333333335e-06,
"loss": 0.168,
"step": 1775
},
{
"epoch": 1.74,
"learning_rate": 7.117777777777778e-06,
"loss": 0.1728,
"step": 1800
},
{
"epoch": 1.77,
"learning_rate": 7.062222222222223e-06,
"loss": 0.1586,
"step": 1825
},
{
"epoch": 1.79,
"learning_rate": 7.006666666666667e-06,
"loss": 0.1427,
"step": 1850
},
{
"epoch": 1.82,
"learning_rate": 6.951111111111112e-06,
"loss": 0.1625,
"step": 1875
},
{
"epoch": 1.84,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.1391,
"step": 1900
},
{
"epoch": 1.86,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.1471,
"step": 1925
},
{
"epoch": 1.89,
"learning_rate": 6.784444444444445e-06,
"loss": 0.1687,
"step": 1950
},
{
"epoch": 1.91,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.1573,
"step": 1975
},
{
"epoch": 1.94,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.1439,
"step": 2000
},
{
"epoch": 1.94,
"eval_loss": 0.2489384263753891,
"eval_runtime": 599.2383,
"eval_samples_per_second": 6.044,
"eval_steps_per_second": 1.512,
"eval_wer": 13.76581550494594,
"step": 2000
},
{
"epoch": 1.96,
"learning_rate": 6.617777777777778e-06,
"loss": 0.1524,
"step": 2025
},
{
"epoch": 1.98,
"learning_rate": 6.562222222222223e-06,
"loss": 0.1783,
"step": 2050
},
{
"epoch": 2.01,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.1271,
"step": 2075
},
{
"epoch": 2.03,
"learning_rate": 6.451111111111111e-06,
"loss": 0.0575,
"step": 2100
},
{
"epoch": 2.06,
"learning_rate": 6.395555555555556e-06,
"loss": 0.0509,
"step": 2125
},
{
"epoch": 2.08,
"learning_rate": 6.34e-06,
"loss": 0.0583,
"step": 2150
},
{
"epoch": 2.11,
"learning_rate": 6.284444444444445e-06,
"loss": 0.0588,
"step": 2175
},
{
"epoch": 2.13,
"learning_rate": 6.22888888888889e-06,
"loss": 0.0554,
"step": 2200
},
{
"epoch": 2.15,
"learning_rate": 6.173333333333333e-06,
"loss": 0.0569,
"step": 2225
},
{
"epoch": 2.18,
"learning_rate": 6.117777777777778e-06,
"loss": 0.0693,
"step": 2250
},
{
"epoch": 2.2,
"learning_rate": 6.062222222222223e-06,
"loss": 0.0554,
"step": 2275
},
{
"epoch": 2.23,
"learning_rate": 6.006666666666667e-06,
"loss": 0.0611,
"step": 2300
},
{
"epoch": 2.25,
"learning_rate": 5.951111111111112e-06,
"loss": 0.0561,
"step": 2325
},
{
"epoch": 2.27,
"learning_rate": 5.895555555555557e-06,
"loss": 0.0573,
"step": 2350
},
{
"epoch": 2.3,
"learning_rate": 5.84e-06,
"loss": 0.0542,
"step": 2375
},
{
"epoch": 2.32,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0711,
"step": 2400
},
{
"epoch": 2.35,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0687,
"step": 2425
},
{
"epoch": 2.37,
"learning_rate": 5.673333333333334e-06,
"loss": 0.0594,
"step": 2450
},
{
"epoch": 2.4,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0721,
"step": 2475
},
{
"epoch": 2.42,
"learning_rate": 5.562222222222222e-06,
"loss": 0.052,
"step": 2500
},
{
"epoch": 2.44,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0552,
"step": 2525
},
{
"epoch": 2.47,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0572,
"step": 2550
},
{
"epoch": 2.49,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0551,
"step": 2575
},
{
"epoch": 2.52,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.0604,
"step": 2600
},
{
"epoch": 2.54,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0596,
"step": 2625
},
{
"epoch": 2.57,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0707,
"step": 2650
},
{
"epoch": 2.59,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0554,
"step": 2675
},
{
"epoch": 2.61,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0701,
"step": 2700
},
{
"epoch": 2.64,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0539,
"step": 2725
},
{
"epoch": 2.66,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0669,
"step": 2750
},
{
"epoch": 2.69,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0619,
"step": 2775
},
{
"epoch": 2.71,
"learning_rate": 4.895555555555556e-06,
"loss": 0.059,
"step": 2800
},
{
"epoch": 2.73,
"learning_rate": 4.84e-06,
"loss": 0.0578,
"step": 2825
},
{
"epoch": 2.76,
"learning_rate": 4.784444444444445e-06,
"loss": 0.0625,
"step": 2850
},
{
"epoch": 2.78,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0471,
"step": 2875
},
{
"epoch": 2.81,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0598,
"step": 2900
},
{
"epoch": 2.83,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0604,
"step": 2925
},
{
"epoch": 2.86,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0603,
"step": 2950
},
{
"epoch": 2.88,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0613,
"step": 2975
},
{
"epoch": 2.9,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0561,
"step": 3000
},
{
"epoch": 2.9,
"eval_loss": 0.2530084550380707,
"eval_runtime": 590.0569,
"eval_samples_per_second": 6.138,
"eval_steps_per_second": 1.535,
"eval_wer": 12.776627559236257,
"step": 3000
},
{
"epoch": 2.93,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0453,
"step": 3025
},
{
"epoch": 2.95,
"learning_rate": 4.34e-06,
"loss": 0.0683,
"step": 3050
},
{
"epoch": 2.98,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0521,
"step": 3075
},
{
"epoch": 3.0,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0502,
"step": 3100
},
{
"epoch": 3.03,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0233,
"step": 3125
},
{
"epoch": 3.05,
"learning_rate": 4.117777777777779e-06,
"loss": 0.0248,
"step": 3150
},
{
"epoch": 3.07,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0283,
"step": 3175
},
{
"epoch": 3.1,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0311,
"step": 3200
},
{
"epoch": 3.12,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0279,
"step": 3225
},
{
"epoch": 3.15,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0217,
"step": 3250
},
{
"epoch": 3.17,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0237,
"step": 3275
},
{
"epoch": 3.19,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0237,
"step": 3300
},
{
"epoch": 3.22,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0193,
"step": 3325
},
{
"epoch": 3.24,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0196,
"step": 3350
},
{
"epoch": 3.27,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0391,
"step": 3375
},
{
"epoch": 3.29,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0208,
"step": 3400
},
{
"epoch": 3.32,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.019,
"step": 3425
},
{
"epoch": 3.34,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0209,
"step": 3450
},
{
"epoch": 3.36,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0207,
"step": 3475
},
{
"epoch": 3.39,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0227,
"step": 3500
},
{
"epoch": 3.41,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0176,
"step": 3525
},
{
"epoch": 3.44,
"learning_rate": 3.228888888888889e-06,
"loss": 0.0209,
"step": 3550
},
{
"epoch": 3.46,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0242,
"step": 3575
},
{
"epoch": 3.48,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0242,
"step": 3600
},
{
"epoch": 3.51,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.022,
"step": 3625
},
{
"epoch": 3.53,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0254,
"step": 3650
},
{
"epoch": 3.56,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0252,
"step": 3675
},
{
"epoch": 3.58,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0238,
"step": 3700
},
{
"epoch": 3.61,
"learning_rate": 2.84e-06,
"loss": 0.0175,
"step": 3725
},
{
"epoch": 3.63,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0201,
"step": 3750
},
{
"epoch": 3.65,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0206,
"step": 3775
},
{
"epoch": 3.68,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0238,
"step": 3800
},
{
"epoch": 3.7,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0206,
"step": 3825
},
{
"epoch": 3.73,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.0319,
"step": 3850
},
{
"epoch": 3.75,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0179,
"step": 3875
},
{
"epoch": 3.78,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0169,
"step": 3900
},
{
"epoch": 3.8,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0185,
"step": 3925
},
{
"epoch": 3.82,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0201,
"step": 3950
},
{
"epoch": 3.85,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0237,
"step": 3975
},
{
"epoch": 3.87,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0185,
"step": 4000
},
{
"epoch": 3.87,
"eval_loss": 0.2687680125236511,
"eval_runtime": 590.1609,
"eval_samples_per_second": 6.137,
"eval_steps_per_second": 1.535,
"eval_wer": 12.680009201748332,
"step": 4000
},
{
"epoch": 3.9,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0204,
"step": 4025
},
{
"epoch": 3.92,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0203,
"step": 4050
},
{
"epoch": 3.94,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0172,
"step": 4075
},
{
"epoch": 3.97,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0164,
"step": 4100
},
{
"epoch": 3.99,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0175,
"step": 4125
},
{
"epoch": 4.02,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0107,
"step": 4150
},
{
"epoch": 4.04,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0095,
"step": 4175
},
{
"epoch": 4.07,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.009,
"step": 4200
},
{
"epoch": 4.09,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0106,
"step": 4225
},
{
"epoch": 4.11,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0073,
"step": 4250
},
{
"epoch": 4.14,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0102,
"step": 4275
},
{
"epoch": 4.16,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0074,
"step": 4300
},
{
"epoch": 4.19,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0072,
"step": 4325
},
{
"epoch": 4.21,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0089,
"step": 4350
},
{
"epoch": 4.24,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0098,
"step": 4375
},
{
"epoch": 4.26,
"learning_rate": 1.34e-06,
"loss": 0.0082,
"step": 4400
},
{
"epoch": 4.28,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0069,
"step": 4425
},
{
"epoch": 4.31,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0094,
"step": 4450
},
{
"epoch": 4.33,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.0098,
"step": 4475
},
{
"epoch": 4.36,
"learning_rate": 1.117777777777778e-06,
"loss": 0.0105,
"step": 4500
},
{
"epoch": 4.38,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.007,
"step": 4525
},
{
"epoch": 4.4,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0071,
"step": 4550
},
{
"epoch": 4.43,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0062,
"step": 4575
},
{
"epoch": 4.45,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0079,
"step": 4600
},
{
"epoch": 4.48,
"learning_rate": 8.400000000000001e-07,
"loss": 0.008,
"step": 4625
},
{
"epoch": 4.5,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0075,
"step": 4650
},
{
"epoch": 4.53,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0131,
"step": 4675
},
{
"epoch": 4.55,
"learning_rate": 6.733333333333334e-07,
"loss": 0.0118,
"step": 4700
},
{
"epoch": 4.57,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0063,
"step": 4725
},
{
"epoch": 4.6,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0062,
"step": 4750
},
{
"epoch": 4.62,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0146,
"step": 4775
},
{
"epoch": 4.65,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0076,
"step": 4800
},
{
"epoch": 4.67,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.0085,
"step": 4825
},
{
"epoch": 4.7,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0073,
"step": 4850
},
{
"epoch": 4.72,
"learning_rate": 2.844444444444445e-07,
"loss": 0.0071,
"step": 4875
},
{
"epoch": 4.74,
"learning_rate": 2.2888888888888892e-07,
"loss": 0.0071,
"step": 4900
},
{
"epoch": 4.77,
"learning_rate": 1.7333333333333335e-07,
"loss": 0.0068,
"step": 4925
},
{
"epoch": 4.79,
"learning_rate": 1.1777777777777778e-07,
"loss": 0.007,
"step": 4950
},
{
"epoch": 4.82,
"learning_rate": 6.222222222222223e-08,
"loss": 0.0075,
"step": 4975
},
{
"epoch": 4.84,
"learning_rate": 6.666666666666667e-09,
"loss": 0.0074,
"step": 5000
},
{
"epoch": 4.84,
"eval_loss": 0.2866577208042145,
"eval_runtime": 598.1353,
"eval_samples_per_second": 6.055,
"eval_steps_per_second": 1.515,
"eval_wer": 12.781228433402347,
"step": 5000
},
{
"epoch": 4.84,
"step": 5000,
"total_flos": 1.153995298799616e+19,
"train_loss": 0.14189173013567924,
"train_runtime": 7808.1605,
"train_samples_per_second": 5.123,
"train_steps_per_second": 0.64
}
],
"max_steps": 5000,
"num_train_epochs": 5,
"total_flos": 1.153995298799616e+19,
"trial_name": null,
"trial_params": null
}