Ansu's picture
End of training
b5d4fd4
Raw
History Blame Contribute Delete
51.2 kB
{
"best_metric": 15.274135131759712,
"best_model_checkpoint": "/scratch/andoni.sudupe/checkpoints/sweep/tiny/lr1e-5-freezeFalse/checkpoint-10000",
"epoch": 6.605019815059445,
"global_step": 10000,
"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": 4.2873,
"step": 25
},
{
"epoch": 0.03,
"learning_rate": 9.400000000000001e-07,
"loss": 3.9829,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 1.44e-06,
"loss": 3.4255,
"step": 75
},
{
"epoch": 0.07,
"learning_rate": 1.94e-06,
"loss": 2.8206,
"step": 100
},
{
"epoch": 0.08,
"learning_rate": 2.4400000000000004e-06,
"loss": 2.275,
"step": 125
},
{
"epoch": 0.1,
"learning_rate": 2.9400000000000002e-06,
"loss": 1.9621,
"step": 150
},
{
"epoch": 0.12,
"learning_rate": 3.44e-06,
"loss": 1.7578,
"step": 175
},
{
"epoch": 0.13,
"learning_rate": 3.94e-06,
"loss": 1.5926,
"step": 200
},
{
"epoch": 0.15,
"learning_rate": 4.440000000000001e-06,
"loss": 1.4555,
"step": 225
},
{
"epoch": 0.17,
"learning_rate": 4.94e-06,
"loss": 1.336,
"step": 250
},
{
"epoch": 0.18,
"learning_rate": 5.4400000000000004e-06,
"loss": 1.2103,
"step": 275
},
{
"epoch": 0.2,
"learning_rate": 5.94e-06,
"loss": 1.1009,
"step": 300
},
{
"epoch": 0.21,
"learning_rate": 6.440000000000001e-06,
"loss": 0.9782,
"step": 325
},
{
"epoch": 0.23,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.8044,
"step": 350
},
{
"epoch": 0.25,
"learning_rate": 7.440000000000001e-06,
"loss": 0.6977,
"step": 375
},
{
"epoch": 0.26,
"learning_rate": 7.94e-06,
"loss": 0.6428,
"step": 400
},
{
"epoch": 0.28,
"learning_rate": 8.44e-06,
"loss": 0.6024,
"step": 425
},
{
"epoch": 0.3,
"learning_rate": 8.94e-06,
"loss": 0.5664,
"step": 450
},
{
"epoch": 0.31,
"learning_rate": 9.440000000000001e-06,
"loss": 0.5333,
"step": 475
},
{
"epoch": 0.33,
"learning_rate": 9.940000000000001e-06,
"loss": 0.5129,
"step": 500
},
{
"epoch": 0.35,
"learning_rate": 9.976842105263158e-06,
"loss": 0.4892,
"step": 525
},
{
"epoch": 0.36,
"learning_rate": 9.950526315789475e-06,
"loss": 0.4702,
"step": 550
},
{
"epoch": 0.38,
"learning_rate": 9.92421052631579e-06,
"loss": 0.4512,
"step": 575
},
{
"epoch": 0.4,
"learning_rate": 9.897894736842107e-06,
"loss": 0.4349,
"step": 600
},
{
"epoch": 0.41,
"learning_rate": 9.871578947368422e-06,
"loss": 0.4282,
"step": 625
},
{
"epoch": 0.43,
"learning_rate": 9.845263157894738e-06,
"loss": 0.418,
"step": 650
},
{
"epoch": 0.45,
"learning_rate": 9.818947368421053e-06,
"loss": 0.397,
"step": 675
},
{
"epoch": 0.46,
"learning_rate": 9.79263157894737e-06,
"loss": 0.389,
"step": 700
},
{
"epoch": 0.48,
"learning_rate": 9.766315789473685e-06,
"loss": 0.3886,
"step": 725
},
{
"epoch": 0.5,
"learning_rate": 9.74e-06,
"loss": 0.3704,
"step": 750
},
{
"epoch": 0.51,
"learning_rate": 9.713684210526317e-06,
"loss": 0.3692,
"step": 775
},
{
"epoch": 0.53,
"learning_rate": 9.687368421052632e-06,
"loss": 0.358,
"step": 800
},
{
"epoch": 0.54,
"learning_rate": 9.661052631578948e-06,
"loss": 0.3494,
"step": 825
},
{
"epoch": 0.56,
"learning_rate": 9.634736842105265e-06,
"loss": 0.3499,
"step": 850
},
{
"epoch": 0.58,
"learning_rate": 9.60842105263158e-06,
"loss": 0.3374,
"step": 875
},
{
"epoch": 0.59,
"learning_rate": 9.582105263157897e-06,
"loss": 0.3306,
"step": 900
},
{
"epoch": 0.61,
"learning_rate": 9.555789473684211e-06,
"loss": 0.3309,
"step": 925
},
{
"epoch": 0.63,
"learning_rate": 9.529473684210528e-06,
"loss": 0.3173,
"step": 950
},
{
"epoch": 0.64,
"learning_rate": 9.503157894736843e-06,
"loss": 0.3186,
"step": 975
},
{
"epoch": 0.66,
"learning_rate": 9.476842105263158e-06,
"loss": 0.3108,
"step": 1000
},
{
"epoch": 0.66,
"eval_loss": 0.4898373484611511,
"eval_runtime": 58.5159,
"eval_samples_per_second": 28.898,
"eval_steps_per_second": 0.239,
"eval_wer": 27.413513175971026,
"step": 1000
},
{
"epoch": 0.68,
"learning_rate": 9.450526315789475e-06,
"loss": 0.3055,
"step": 1025
},
{
"epoch": 0.69,
"learning_rate": 9.42421052631579e-06,
"loss": 0.3077,
"step": 1050
},
{
"epoch": 0.71,
"learning_rate": 9.397894736842106e-06,
"loss": 0.3034,
"step": 1075
},
{
"epoch": 0.73,
"learning_rate": 9.371578947368421e-06,
"loss": 0.2906,
"step": 1100
},
{
"epoch": 0.74,
"learning_rate": 9.345263157894738e-06,
"loss": 0.2884,
"step": 1125
},
{
"epoch": 0.76,
"learning_rate": 9.318947368421053e-06,
"loss": 0.2869,
"step": 1150
},
{
"epoch": 0.78,
"learning_rate": 9.292631578947368e-06,
"loss": 0.2856,
"step": 1175
},
{
"epoch": 0.79,
"learning_rate": 9.266315789473685e-06,
"loss": 0.284,
"step": 1200
},
{
"epoch": 0.81,
"learning_rate": 9.240000000000001e-06,
"loss": 0.2799,
"step": 1225
},
{
"epoch": 0.83,
"learning_rate": 9.213684210526316e-06,
"loss": 0.276,
"step": 1250
},
{
"epoch": 0.84,
"learning_rate": 9.187368421052633e-06,
"loss": 0.2802,
"step": 1275
},
{
"epoch": 0.86,
"learning_rate": 9.161052631578948e-06,
"loss": 0.2741,
"step": 1300
},
{
"epoch": 0.88,
"learning_rate": 9.134736842105265e-06,
"loss": 0.2681,
"step": 1325
},
{
"epoch": 0.89,
"learning_rate": 9.10842105263158e-06,
"loss": 0.2647,
"step": 1350
},
{
"epoch": 0.91,
"learning_rate": 9.082105263157896e-06,
"loss": 0.263,
"step": 1375
},
{
"epoch": 0.92,
"learning_rate": 9.055789473684211e-06,
"loss": 0.2556,
"step": 1400
},
{
"epoch": 0.94,
"learning_rate": 9.029473684210526e-06,
"loss": 0.254,
"step": 1425
},
{
"epoch": 0.96,
"learning_rate": 9.003157894736843e-06,
"loss": 0.2526,
"step": 1450
},
{
"epoch": 0.97,
"learning_rate": 8.976842105263158e-06,
"loss": 0.2497,
"step": 1475
},
{
"epoch": 0.99,
"learning_rate": 8.950526315789474e-06,
"loss": 0.2493,
"step": 1500
},
{
"epoch": 1.01,
"learning_rate": 8.92421052631579e-06,
"loss": 0.2382,
"step": 1525
},
{
"epoch": 1.02,
"learning_rate": 8.897894736842106e-06,
"loss": 0.2341,
"step": 1550
},
{
"epoch": 1.04,
"learning_rate": 8.871578947368423e-06,
"loss": 0.2315,
"step": 1575
},
{
"epoch": 1.06,
"learning_rate": 8.845263157894738e-06,
"loss": 0.2326,
"step": 1600
},
{
"epoch": 1.07,
"learning_rate": 8.818947368421054e-06,
"loss": 0.2262,
"step": 1625
},
{
"epoch": 1.09,
"learning_rate": 8.79263157894737e-06,
"loss": 0.2327,
"step": 1650
},
{
"epoch": 1.11,
"learning_rate": 8.766315789473684e-06,
"loss": 0.229,
"step": 1675
},
{
"epoch": 1.12,
"learning_rate": 8.740000000000001e-06,
"loss": 0.2304,
"step": 1700
},
{
"epoch": 1.14,
"learning_rate": 8.713684210526316e-06,
"loss": 0.2201,
"step": 1725
},
{
"epoch": 1.16,
"learning_rate": 8.687368421052633e-06,
"loss": 0.2231,
"step": 1750
},
{
"epoch": 1.17,
"learning_rate": 8.661052631578948e-06,
"loss": 0.2259,
"step": 1775
},
{
"epoch": 1.19,
"learning_rate": 8.634736842105264e-06,
"loss": 0.2203,
"step": 1800
},
{
"epoch": 1.21,
"learning_rate": 8.60842105263158e-06,
"loss": 0.2188,
"step": 1825
},
{
"epoch": 1.22,
"learning_rate": 8.582105263157894e-06,
"loss": 0.2175,
"step": 1850
},
{
"epoch": 1.24,
"learning_rate": 8.555789473684213e-06,
"loss": 0.2167,
"step": 1875
},
{
"epoch": 1.25,
"learning_rate": 8.529473684210528e-06,
"loss": 0.2161,
"step": 1900
},
{
"epoch": 1.27,
"learning_rate": 8.503157894736843e-06,
"loss": 0.2113,
"step": 1925
},
{
"epoch": 1.29,
"learning_rate": 8.47684210526316e-06,
"loss": 0.2159,
"step": 1950
},
{
"epoch": 1.3,
"learning_rate": 8.450526315789474e-06,
"loss": 0.2156,
"step": 1975
},
{
"epoch": 1.32,
"learning_rate": 8.42421052631579e-06,
"loss": 0.2097,
"step": 2000
},
{
"epoch": 1.32,
"eval_loss": 0.3735288083553314,
"eval_runtime": 48.7039,
"eval_samples_per_second": 34.72,
"eval_steps_per_second": 0.287,
"eval_wer": 20.887973023604346,
"step": 2000
},
{
"epoch": 1.34,
"learning_rate": 8.397894736842106e-06,
"loss": 0.2144,
"step": 2025
},
{
"epoch": 1.35,
"learning_rate": 8.371578947368422e-06,
"loss": 0.2078,
"step": 2050
},
{
"epoch": 1.37,
"learning_rate": 8.345263157894737e-06,
"loss": 0.2036,
"step": 2075
},
{
"epoch": 1.39,
"learning_rate": 8.318947368421052e-06,
"loss": 0.2115,
"step": 2100
},
{
"epoch": 1.4,
"learning_rate": 8.292631578947369e-06,
"loss": 0.2059,
"step": 2125
},
{
"epoch": 1.42,
"learning_rate": 8.266315789473684e-06,
"loss": 0.2047,
"step": 2150
},
{
"epoch": 1.44,
"learning_rate": 8.24e-06,
"loss": 0.2052,
"step": 2175
},
{
"epoch": 1.45,
"learning_rate": 8.213684210526316e-06,
"loss": 0.2062,
"step": 2200
},
{
"epoch": 1.47,
"learning_rate": 8.187368421052632e-06,
"loss": 0.2041,
"step": 2225
},
{
"epoch": 1.49,
"learning_rate": 8.161052631578949e-06,
"loss": 0.1987,
"step": 2250
},
{
"epoch": 1.5,
"learning_rate": 8.134736842105264e-06,
"loss": 0.1976,
"step": 2275
},
{
"epoch": 1.52,
"learning_rate": 8.10842105263158e-06,
"loss": 0.1933,
"step": 2300
},
{
"epoch": 1.54,
"learning_rate": 8.082105263157896e-06,
"loss": 0.1935,
"step": 2325
},
{
"epoch": 1.55,
"learning_rate": 8.05578947368421e-06,
"loss": 0.2028,
"step": 2350
},
{
"epoch": 1.57,
"learning_rate": 8.029473684210527e-06,
"loss": 0.1969,
"step": 2375
},
{
"epoch": 1.59,
"learning_rate": 8.003157894736842e-06,
"loss": 0.1955,
"step": 2400
},
{
"epoch": 1.6,
"learning_rate": 7.976842105263159e-06,
"loss": 0.1916,
"step": 2425
},
{
"epoch": 1.62,
"learning_rate": 7.950526315789474e-06,
"loss": 0.1938,
"step": 2450
},
{
"epoch": 1.63,
"learning_rate": 7.92421052631579e-06,
"loss": 0.187,
"step": 2475
},
{
"epoch": 1.65,
"learning_rate": 7.897894736842106e-06,
"loss": 0.1875,
"step": 2500
},
{
"epoch": 1.67,
"learning_rate": 7.87157894736842e-06,
"loss": 0.1906,
"step": 2525
},
{
"epoch": 1.68,
"learning_rate": 7.845263157894739e-06,
"loss": 0.1889,
"step": 2550
},
{
"epoch": 1.7,
"learning_rate": 7.818947368421054e-06,
"loss": 0.1871,
"step": 2575
},
{
"epoch": 1.72,
"learning_rate": 7.792631578947369e-06,
"loss": 0.1877,
"step": 2600
},
{
"epoch": 1.73,
"learning_rate": 7.766315789473685e-06,
"loss": 0.1912,
"step": 2625
},
{
"epoch": 1.75,
"learning_rate": 7.74e-06,
"loss": 0.1873,
"step": 2650
},
{
"epoch": 1.77,
"learning_rate": 7.713684210526317e-06,
"loss": 0.1825,
"step": 2675
},
{
"epoch": 1.78,
"learning_rate": 7.687368421052632e-06,
"loss": 0.186,
"step": 2700
},
{
"epoch": 1.8,
"learning_rate": 7.661052631578949e-06,
"loss": 0.1883,
"step": 2725
},
{
"epoch": 1.82,
"learning_rate": 7.634736842105264e-06,
"loss": 0.18,
"step": 2750
},
{
"epoch": 1.83,
"learning_rate": 7.6084210526315795e-06,
"loss": 0.1897,
"step": 2775
},
{
"epoch": 1.85,
"learning_rate": 7.582105263157895e-06,
"loss": 0.1779,
"step": 2800
},
{
"epoch": 1.87,
"learning_rate": 7.555789473684211e-06,
"loss": 0.1766,
"step": 2825
},
{
"epoch": 1.88,
"learning_rate": 7.529473684210526e-06,
"loss": 0.1813,
"step": 2850
},
{
"epoch": 1.9,
"learning_rate": 7.503157894736842e-06,
"loss": 0.1873,
"step": 2875
},
{
"epoch": 1.92,
"learning_rate": 7.476842105263159e-06,
"loss": 0.1816,
"step": 2900
},
{
"epoch": 1.93,
"learning_rate": 7.4505263157894744e-06,
"loss": 0.1833,
"step": 2925
},
{
"epoch": 1.95,
"learning_rate": 7.42421052631579e-06,
"loss": 0.1783,
"step": 2950
},
{
"epoch": 1.96,
"learning_rate": 7.397894736842106e-06,
"loss": 0.1785,
"step": 2975
},
{
"epoch": 1.98,
"learning_rate": 7.371578947368422e-06,
"loss": 0.179,
"step": 3000
},
{
"epoch": 1.98,
"eval_loss": 0.34564170241355896,
"eval_runtime": 48.6517,
"eval_samples_per_second": 34.757,
"eval_steps_per_second": 0.288,
"eval_wer": 18.789808917197455,
"step": 3000
},
{
"epoch": 2.0,
"learning_rate": 7.345263157894738e-06,
"loss": 0.1768,
"step": 3025
},
{
"epoch": 2.01,
"learning_rate": 7.3189473684210535e-06,
"loss": 0.1647,
"step": 3050
},
{
"epoch": 2.03,
"learning_rate": 7.292631578947369e-06,
"loss": 0.1619,
"step": 3075
},
{
"epoch": 2.05,
"learning_rate": 7.266315789473684e-06,
"loss": 0.1625,
"step": 3100
},
{
"epoch": 2.06,
"learning_rate": 7.24e-06,
"loss": 0.1644,
"step": 3125
},
{
"epoch": 2.08,
"learning_rate": 7.213684210526316e-06,
"loss": 0.1714,
"step": 3150
},
{
"epoch": 2.1,
"learning_rate": 7.187368421052632e-06,
"loss": 0.1657,
"step": 3175
},
{
"epoch": 2.11,
"learning_rate": 7.161052631578948e-06,
"loss": 0.1624,
"step": 3200
},
{
"epoch": 2.13,
"learning_rate": 7.134736842105264e-06,
"loss": 0.1634,
"step": 3225
},
{
"epoch": 2.15,
"learning_rate": 7.10842105263158e-06,
"loss": 0.1603,
"step": 3250
},
{
"epoch": 2.16,
"learning_rate": 7.082105263157896e-06,
"loss": 0.1594,
"step": 3275
},
{
"epoch": 2.18,
"learning_rate": 7.055789473684212e-06,
"loss": 0.1613,
"step": 3300
},
{
"epoch": 2.2,
"learning_rate": 7.0294736842105275e-06,
"loss": 0.159,
"step": 3325
},
{
"epoch": 2.21,
"learning_rate": 7.0031578947368425e-06,
"loss": 0.1606,
"step": 3350
},
{
"epoch": 2.23,
"learning_rate": 6.976842105263158e-06,
"loss": 0.16,
"step": 3375
},
{
"epoch": 2.25,
"learning_rate": 6.950526315789474e-06,
"loss": 0.1567,
"step": 3400
},
{
"epoch": 2.26,
"learning_rate": 6.92421052631579e-06,
"loss": 0.1599,
"step": 3425
},
{
"epoch": 2.28,
"learning_rate": 6.897894736842106e-06,
"loss": 0.1604,
"step": 3450
},
{
"epoch": 2.3,
"learning_rate": 6.871578947368422e-06,
"loss": 0.1598,
"step": 3475
},
{
"epoch": 2.31,
"learning_rate": 6.845263157894737e-06,
"loss": 0.1593,
"step": 3500
},
{
"epoch": 2.33,
"learning_rate": 6.818947368421052e-06,
"loss": 0.1596,
"step": 3525
},
{
"epoch": 2.34,
"learning_rate": 6.792631578947368e-06,
"loss": 0.1596,
"step": 3550
},
{
"epoch": 2.36,
"learning_rate": 6.766315789473685e-06,
"loss": 0.1566,
"step": 3575
},
{
"epoch": 2.38,
"learning_rate": 6.740000000000001e-06,
"loss": 0.1554,
"step": 3600
},
{
"epoch": 2.39,
"learning_rate": 6.7136842105263165e-06,
"loss": 0.1594,
"step": 3625
},
{
"epoch": 2.41,
"learning_rate": 6.687368421052632e-06,
"loss": 0.1589,
"step": 3650
},
{
"epoch": 2.43,
"learning_rate": 6.661052631578948e-06,
"loss": 0.1565,
"step": 3675
},
{
"epoch": 2.44,
"learning_rate": 6.634736842105264e-06,
"loss": 0.1536,
"step": 3700
},
{
"epoch": 2.46,
"learning_rate": 6.60842105263158e-06,
"loss": 0.1591,
"step": 3725
},
{
"epoch": 2.48,
"learning_rate": 6.582105263157896e-06,
"loss": 0.1571,
"step": 3750
},
{
"epoch": 2.49,
"learning_rate": 6.5557894736842106e-06,
"loss": 0.1537,
"step": 3775
},
{
"epoch": 2.51,
"learning_rate": 6.529473684210526e-06,
"loss": 0.1566,
"step": 3800
},
{
"epoch": 2.53,
"learning_rate": 6.503157894736842e-06,
"loss": 0.1503,
"step": 3825
},
{
"epoch": 2.54,
"learning_rate": 6.476842105263158e-06,
"loss": 0.1525,
"step": 3850
},
{
"epoch": 2.56,
"learning_rate": 6.450526315789474e-06,
"loss": 0.1535,
"step": 3875
},
{
"epoch": 2.58,
"learning_rate": 6.42421052631579e-06,
"loss": 0.1521,
"step": 3900
},
{
"epoch": 2.59,
"learning_rate": 6.397894736842106e-06,
"loss": 0.154,
"step": 3925
},
{
"epoch": 2.61,
"learning_rate": 6.371578947368422e-06,
"loss": 0.1517,
"step": 3950
},
{
"epoch": 2.63,
"learning_rate": 6.345263157894738e-06,
"loss": 0.154,
"step": 3975
},
{
"epoch": 2.64,
"learning_rate": 6.318947368421054e-06,
"loss": 0.1488,
"step": 4000
},
{
"epoch": 2.64,
"eval_loss": 0.3319928050041199,
"eval_runtime": 48.7458,
"eval_samples_per_second": 34.69,
"eval_steps_per_second": 0.287,
"eval_wer": 17.316098413887847,
"step": 4000
},
{
"epoch": 2.66,
"learning_rate": 6.292631578947369e-06,
"loss": 0.1486,
"step": 4025
},
{
"epoch": 2.68,
"learning_rate": 6.266315789473685e-06,
"loss": 0.1537,
"step": 4050
},
{
"epoch": 2.69,
"learning_rate": 6.24e-06,
"loss": 0.1517,
"step": 4075
},
{
"epoch": 2.71,
"learning_rate": 6.213684210526316e-06,
"loss": 0.1528,
"step": 4100
},
{
"epoch": 2.72,
"learning_rate": 6.187368421052632e-06,
"loss": 0.1497,
"step": 4125
},
{
"epoch": 2.74,
"learning_rate": 6.161052631578948e-06,
"loss": 0.1498,
"step": 4150
},
{
"epoch": 2.76,
"learning_rate": 6.134736842105264e-06,
"loss": 0.1504,
"step": 4175
},
{
"epoch": 2.77,
"learning_rate": 6.108421052631579e-06,
"loss": 0.1517,
"step": 4200
},
{
"epoch": 2.79,
"learning_rate": 6.0821052631578945e-06,
"loss": 0.1458,
"step": 4225
},
{
"epoch": 2.81,
"learning_rate": 6.055789473684212e-06,
"loss": 0.1459,
"step": 4250
},
{
"epoch": 2.82,
"learning_rate": 6.029473684210527e-06,
"loss": 0.1459,
"step": 4275
},
{
"epoch": 2.84,
"learning_rate": 6.003157894736843e-06,
"loss": 0.1467,
"step": 4300
},
{
"epoch": 2.86,
"learning_rate": 5.976842105263159e-06,
"loss": 0.1457,
"step": 4325
},
{
"epoch": 2.87,
"learning_rate": 5.950526315789474e-06,
"loss": 0.1481,
"step": 4350
},
{
"epoch": 2.89,
"learning_rate": 5.92421052631579e-06,
"loss": 0.1466,
"step": 4375
},
{
"epoch": 2.91,
"learning_rate": 5.897894736842106e-06,
"loss": 0.145,
"step": 4400
},
{
"epoch": 2.92,
"learning_rate": 5.871578947368422e-06,
"loss": 0.1499,
"step": 4425
},
{
"epoch": 2.94,
"learning_rate": 5.845263157894737e-06,
"loss": 0.1455,
"step": 4450
},
{
"epoch": 2.96,
"learning_rate": 5.818947368421053e-06,
"loss": 0.1462,
"step": 4475
},
{
"epoch": 2.97,
"learning_rate": 5.7926315789473685e-06,
"loss": 0.1461,
"step": 4500
},
{
"epoch": 2.99,
"learning_rate": 5.766315789473684e-06,
"loss": 0.1457,
"step": 4525
},
{
"epoch": 3.01,
"learning_rate": 5.74e-06,
"loss": 0.1402,
"step": 4550
},
{
"epoch": 3.02,
"learning_rate": 5.713684210526316e-06,
"loss": 0.1351,
"step": 4575
},
{
"epoch": 3.04,
"learning_rate": 5.687368421052633e-06,
"loss": 0.1316,
"step": 4600
},
{
"epoch": 3.05,
"learning_rate": 5.661052631578948e-06,
"loss": 0.1344,
"step": 4625
},
{
"epoch": 3.07,
"learning_rate": 5.634736842105264e-06,
"loss": 0.136,
"step": 4650
},
{
"epoch": 3.09,
"learning_rate": 5.60842105263158e-06,
"loss": 0.1323,
"step": 4675
},
{
"epoch": 3.1,
"learning_rate": 5.582105263157895e-06,
"loss": 0.1365,
"step": 4700
},
{
"epoch": 3.12,
"learning_rate": 5.555789473684211e-06,
"loss": 0.134,
"step": 4725
},
{
"epoch": 3.14,
"learning_rate": 5.529473684210527e-06,
"loss": 0.1322,
"step": 4750
},
{
"epoch": 3.15,
"learning_rate": 5.5031578947368425e-06,
"loss": 0.1365,
"step": 4775
},
{
"epoch": 3.17,
"learning_rate": 5.476842105263158e-06,
"loss": 0.1297,
"step": 4800
},
{
"epoch": 3.19,
"learning_rate": 5.450526315789474e-06,
"loss": 0.136,
"step": 4825
},
{
"epoch": 3.2,
"learning_rate": 5.42421052631579e-06,
"loss": 0.1345,
"step": 4850
},
{
"epoch": 3.22,
"learning_rate": 5.397894736842105e-06,
"loss": 0.1321,
"step": 4875
},
{
"epoch": 3.24,
"learning_rate": 5.371578947368421e-06,
"loss": 0.133,
"step": 4900
},
{
"epoch": 3.25,
"learning_rate": 5.345263157894738e-06,
"loss": 0.1319,
"step": 4925
},
{
"epoch": 3.27,
"learning_rate": 5.318947368421053e-06,
"loss": 0.1307,
"step": 4950
},
{
"epoch": 3.29,
"learning_rate": 5.292631578947369e-06,
"loss": 0.1328,
"step": 4975
},
{
"epoch": 3.3,
"learning_rate": 5.266315789473685e-06,
"loss": 0.1348,
"step": 5000
},
{
"epoch": 3.3,
"eval_loss": 0.3174799382686615,
"eval_runtime": 48.3109,
"eval_samples_per_second": 35.002,
"eval_steps_per_second": 0.29,
"eval_wer": 16.772823779193207,
"step": 5000
},
{
"epoch": 3.32,
"learning_rate": 5.240000000000001e-06,
"loss": 0.1317,
"step": 5025
},
{
"epoch": 3.34,
"learning_rate": 5.2136842105263165e-06,
"loss": 0.1362,
"step": 5050
},
{
"epoch": 3.35,
"learning_rate": 5.187368421052632e-06,
"loss": 0.1323,
"step": 5075
},
{
"epoch": 3.37,
"learning_rate": 5.161052631578948e-06,
"loss": 0.128,
"step": 5100
},
{
"epoch": 3.39,
"learning_rate": 5.134736842105263e-06,
"loss": 0.1308,
"step": 5125
},
{
"epoch": 3.4,
"learning_rate": 5.108421052631579e-06,
"loss": 0.1336,
"step": 5150
},
{
"epoch": 3.42,
"learning_rate": 5.082105263157895e-06,
"loss": 0.1332,
"step": 5175
},
{
"epoch": 3.43,
"learning_rate": 5.0557894736842105e-06,
"loss": 0.1317,
"step": 5200
},
{
"epoch": 3.45,
"learning_rate": 5.029473684210526e-06,
"loss": 0.1308,
"step": 5225
},
{
"epoch": 3.47,
"learning_rate": 5.003157894736842e-06,
"loss": 0.1335,
"step": 5250
},
{
"epoch": 3.48,
"learning_rate": 4.976842105263158e-06,
"loss": 0.1298,
"step": 5275
},
{
"epoch": 3.5,
"learning_rate": 4.950526315789474e-06,
"loss": 0.1326,
"step": 5300
},
{
"epoch": 3.52,
"learning_rate": 4.92421052631579e-06,
"loss": 0.1295,
"step": 5325
},
{
"epoch": 3.53,
"learning_rate": 4.897894736842106e-06,
"loss": 0.1306,
"step": 5350
},
{
"epoch": 3.55,
"learning_rate": 4.871578947368421e-06,
"loss": 0.1323,
"step": 5375
},
{
"epoch": 3.57,
"learning_rate": 4.845263157894737e-06,
"loss": 0.132,
"step": 5400
},
{
"epoch": 3.58,
"learning_rate": 4.818947368421053e-06,
"loss": 0.1305,
"step": 5425
},
{
"epoch": 3.6,
"learning_rate": 4.792631578947369e-06,
"loss": 0.1341,
"step": 5450
},
{
"epoch": 3.62,
"learning_rate": 4.7663157894736845e-06,
"loss": 0.1303,
"step": 5475
},
{
"epoch": 3.63,
"learning_rate": 4.74e-06,
"loss": 0.1306,
"step": 5500
},
{
"epoch": 3.65,
"learning_rate": 4.713684210526316e-06,
"loss": 0.131,
"step": 5525
},
{
"epoch": 3.67,
"learning_rate": 4.687368421052632e-06,
"loss": 0.1256,
"step": 5550
},
{
"epoch": 3.68,
"learning_rate": 4.661052631578948e-06,
"loss": 0.1296,
"step": 5575
},
{
"epoch": 3.7,
"learning_rate": 4.634736842105264e-06,
"loss": 0.1322,
"step": 5600
},
{
"epoch": 3.72,
"learning_rate": 4.6084210526315794e-06,
"loss": 0.1263,
"step": 5625
},
{
"epoch": 3.73,
"learning_rate": 4.582105263157895e-06,
"loss": 0.1317,
"step": 5650
},
{
"epoch": 3.75,
"learning_rate": 4.55578947368421e-06,
"loss": 0.1299,
"step": 5675
},
{
"epoch": 3.76,
"learning_rate": 4.529473684210527e-06,
"loss": 0.1332,
"step": 5700
},
{
"epoch": 3.78,
"learning_rate": 4.503157894736843e-06,
"loss": 0.1287,
"step": 5725
},
{
"epoch": 3.8,
"learning_rate": 4.4768421052631585e-06,
"loss": 0.1288,
"step": 5750
},
{
"epoch": 3.81,
"learning_rate": 4.450526315789474e-06,
"loss": 0.1323,
"step": 5775
},
{
"epoch": 3.83,
"learning_rate": 4.424210526315789e-06,
"loss": 0.1293,
"step": 5800
},
{
"epoch": 3.85,
"learning_rate": 4.397894736842105e-06,
"loss": 0.1269,
"step": 5825
},
{
"epoch": 3.86,
"learning_rate": 4.371578947368421e-06,
"loss": 0.1259,
"step": 5850
},
{
"epoch": 3.88,
"learning_rate": 4.345263157894738e-06,
"loss": 0.1261,
"step": 5875
},
{
"epoch": 3.9,
"learning_rate": 4.3189473684210535e-06,
"loss": 0.1265,
"step": 5900
},
{
"epoch": 3.91,
"learning_rate": 4.2926315789473684e-06,
"loss": 0.1282,
"step": 5925
},
{
"epoch": 3.93,
"learning_rate": 4.266315789473684e-06,
"loss": 0.1274,
"step": 5950
},
{
"epoch": 3.95,
"learning_rate": 4.24e-06,
"loss": 0.1279,
"step": 5975
},
{
"epoch": 3.96,
"learning_rate": 4.213684210526316e-06,
"loss": 0.1271,
"step": 6000
},
{
"epoch": 3.96,
"eval_loss": 0.3109683692455292,
"eval_runtime": 47.2397,
"eval_samples_per_second": 35.796,
"eval_steps_per_second": 0.296,
"eval_wer": 16.142125640064943,
"step": 6000
},
{
"epoch": 3.98,
"learning_rate": 4.1873684210526325e-06,
"loss": 0.13,
"step": 6025
},
{
"epoch": 4.0,
"learning_rate": 4.1610526315789475e-06,
"loss": 0.1302,
"step": 6050
},
{
"epoch": 4.01,
"learning_rate": 4.134736842105263e-06,
"loss": 0.1208,
"step": 6075
},
{
"epoch": 4.03,
"learning_rate": 4.108421052631579e-06,
"loss": 0.1188,
"step": 6100
},
{
"epoch": 4.05,
"learning_rate": 4.082105263157895e-06,
"loss": 0.12,
"step": 6125
},
{
"epoch": 4.06,
"learning_rate": 4.055789473684211e-06,
"loss": 0.1225,
"step": 6150
},
{
"epoch": 4.08,
"learning_rate": 4.029473684210527e-06,
"loss": 0.1171,
"step": 6175
},
{
"epoch": 4.1,
"learning_rate": 4.0031578947368424e-06,
"loss": 0.1183,
"step": 6200
},
{
"epoch": 4.11,
"learning_rate": 3.976842105263158e-06,
"loss": 0.1203,
"step": 6225
},
{
"epoch": 4.13,
"learning_rate": 3.950526315789474e-06,
"loss": 0.12,
"step": 6250
},
{
"epoch": 4.14,
"learning_rate": 3.92421052631579e-06,
"loss": 0.1189,
"step": 6275
},
{
"epoch": 4.16,
"learning_rate": 3.897894736842106e-06,
"loss": 0.1196,
"step": 6300
},
{
"epoch": 4.18,
"learning_rate": 3.8715789473684215e-06,
"loss": 0.1195,
"step": 6325
},
{
"epoch": 4.19,
"learning_rate": 3.845263157894737e-06,
"loss": 0.1176,
"step": 6350
},
{
"epoch": 4.21,
"learning_rate": 3.818947368421053e-06,
"loss": 0.115,
"step": 6375
},
{
"epoch": 4.23,
"learning_rate": 3.792631578947369e-06,
"loss": 0.1188,
"step": 6400
},
{
"epoch": 4.24,
"learning_rate": 3.766315789473685e-06,
"loss": 0.1212,
"step": 6425
},
{
"epoch": 4.26,
"learning_rate": 3.74e-06,
"loss": 0.1154,
"step": 6450
},
{
"epoch": 4.28,
"learning_rate": 3.713684210526316e-06,
"loss": 0.1193,
"step": 6475
},
{
"epoch": 4.29,
"learning_rate": 3.687368421052632e-06,
"loss": 0.118,
"step": 6500
},
{
"epoch": 4.31,
"learning_rate": 3.6610526315789472e-06,
"loss": 0.1182,
"step": 6525
},
{
"epoch": 4.33,
"learning_rate": 3.6347368421052635e-06,
"loss": 0.12,
"step": 6550
},
{
"epoch": 4.34,
"learning_rate": 3.6084210526315793e-06,
"loss": 0.1181,
"step": 6575
},
{
"epoch": 4.36,
"learning_rate": 3.582105263157895e-06,
"loss": 0.1181,
"step": 6600
},
{
"epoch": 4.38,
"learning_rate": 3.555789473684211e-06,
"loss": 0.1152,
"step": 6625
},
{
"epoch": 4.39,
"learning_rate": 3.5294736842105263e-06,
"loss": 0.1185,
"step": 6650
},
{
"epoch": 4.41,
"learning_rate": 3.503157894736842e-06,
"loss": 0.1185,
"step": 6675
},
{
"epoch": 4.43,
"learning_rate": 3.476842105263158e-06,
"loss": 0.1142,
"step": 6700
},
{
"epoch": 4.44,
"learning_rate": 3.450526315789474e-06,
"loss": 0.1187,
"step": 6725
},
{
"epoch": 4.46,
"learning_rate": 3.42421052631579e-06,
"loss": 0.1152,
"step": 6750
},
{
"epoch": 4.47,
"learning_rate": 3.3978947368421054e-06,
"loss": 0.118,
"step": 6775
},
{
"epoch": 4.49,
"learning_rate": 3.3715789473684212e-06,
"loss": 0.1158,
"step": 6800
},
{
"epoch": 4.51,
"learning_rate": 3.345263157894737e-06,
"loss": 0.1155,
"step": 6825
},
{
"epoch": 4.52,
"learning_rate": 3.318947368421053e-06,
"loss": 0.1175,
"step": 6850
},
{
"epoch": 4.54,
"learning_rate": 3.292631578947369e-06,
"loss": 0.1196,
"step": 6875
},
{
"epoch": 4.56,
"learning_rate": 3.2663157894736845e-06,
"loss": 0.1151,
"step": 6900
},
{
"epoch": 4.57,
"learning_rate": 3.2400000000000003e-06,
"loss": 0.1164,
"step": 6925
},
{
"epoch": 4.59,
"learning_rate": 3.213684210526316e-06,
"loss": 0.1175,
"step": 6950
},
{
"epoch": 4.61,
"learning_rate": 3.187368421052632e-06,
"loss": 0.1156,
"step": 6975
},
{
"epoch": 4.62,
"learning_rate": 3.1610526315789474e-06,
"loss": 0.1176,
"step": 7000
},
{
"epoch": 4.62,
"eval_loss": 0.3098698854446411,
"eval_runtime": 46.3105,
"eval_samples_per_second": 36.514,
"eval_steps_per_second": 0.302,
"eval_wer": 15.70500811789684,
"step": 7000
},
{
"epoch": 4.64,
"learning_rate": 3.134736842105263e-06,
"loss": 0.1186,
"step": 7025
},
{
"epoch": 4.66,
"learning_rate": 3.1084210526315794e-06,
"loss": 0.1161,
"step": 7050
},
{
"epoch": 4.67,
"learning_rate": 3.0821052631578952e-06,
"loss": 0.1153,
"step": 7075
},
{
"epoch": 4.69,
"learning_rate": 3.055789473684211e-06,
"loss": 0.1202,
"step": 7100
},
{
"epoch": 4.71,
"learning_rate": 3.0294736842105264e-06,
"loss": 0.114,
"step": 7125
},
{
"epoch": 4.72,
"learning_rate": 3.0031578947368423e-06,
"loss": 0.1145,
"step": 7150
},
{
"epoch": 4.74,
"learning_rate": 2.976842105263158e-06,
"loss": 0.1173,
"step": 7175
},
{
"epoch": 4.76,
"learning_rate": 2.9505263157894735e-06,
"loss": 0.1164,
"step": 7200
},
{
"epoch": 4.77,
"learning_rate": 2.92421052631579e-06,
"loss": 0.117,
"step": 7225
},
{
"epoch": 4.79,
"learning_rate": 2.8978947368421055e-06,
"loss": 0.1191,
"step": 7250
},
{
"epoch": 4.81,
"learning_rate": 2.8715789473684214e-06,
"loss": 0.1144,
"step": 7275
},
{
"epoch": 4.82,
"learning_rate": 2.845263157894737e-06,
"loss": 0.1166,
"step": 7300
},
{
"epoch": 4.84,
"learning_rate": 2.8189473684210526e-06,
"loss": 0.1151,
"step": 7325
},
{
"epoch": 4.85,
"learning_rate": 2.7926315789473684e-06,
"loss": 0.1156,
"step": 7350
},
{
"epoch": 4.87,
"learning_rate": 2.766315789473684e-06,
"loss": 0.1152,
"step": 7375
},
{
"epoch": 4.89,
"learning_rate": 2.7400000000000004e-06,
"loss": 0.1135,
"step": 7400
},
{
"epoch": 4.9,
"learning_rate": 2.7136842105263163e-06,
"loss": 0.115,
"step": 7425
},
{
"epoch": 4.92,
"learning_rate": 2.6873684210526317e-06,
"loss": 0.117,
"step": 7450
},
{
"epoch": 4.94,
"learning_rate": 2.6610526315789475e-06,
"loss": 0.118,
"step": 7475
},
{
"epoch": 4.95,
"learning_rate": 2.6347368421052633e-06,
"loss": 0.1162,
"step": 7500
},
{
"epoch": 4.97,
"learning_rate": 2.608421052631579e-06,
"loss": 0.1178,
"step": 7525
},
{
"epoch": 4.99,
"learning_rate": 2.5821052631578954e-06,
"loss": 0.1168,
"step": 7550
},
{
"epoch": 5.0,
"learning_rate": 2.5557894736842108e-06,
"loss": 0.1144,
"step": 7575
},
{
"epoch": 5.02,
"learning_rate": 2.5294736842105266e-06,
"loss": 0.1069,
"step": 7600
},
{
"epoch": 5.04,
"learning_rate": 2.5031578947368424e-06,
"loss": 0.1087,
"step": 7625
},
{
"epoch": 5.05,
"learning_rate": 2.476842105263158e-06,
"loss": 0.1092,
"step": 7650
},
{
"epoch": 5.07,
"learning_rate": 2.4505263157894736e-06,
"loss": 0.1063,
"step": 7675
},
{
"epoch": 5.09,
"learning_rate": 2.42421052631579e-06,
"loss": 0.11,
"step": 7700
},
{
"epoch": 5.1,
"learning_rate": 2.3978947368421052e-06,
"loss": 0.1084,
"step": 7725
},
{
"epoch": 5.12,
"learning_rate": 2.371578947368421e-06,
"loss": 0.111,
"step": 7750
},
{
"epoch": 5.14,
"learning_rate": 2.3452631578947373e-06,
"loss": 0.1119,
"step": 7775
},
{
"epoch": 5.15,
"learning_rate": 2.3189473684210527e-06,
"loss": 0.1076,
"step": 7800
},
{
"epoch": 5.17,
"learning_rate": 2.2926315789473685e-06,
"loss": 0.1064,
"step": 7825
},
{
"epoch": 5.18,
"learning_rate": 2.2663157894736843e-06,
"loss": 0.1125,
"step": 7850
},
{
"epoch": 5.2,
"learning_rate": 2.24e-06,
"loss": 0.112,
"step": 7875
},
{
"epoch": 5.22,
"learning_rate": 2.213684210526316e-06,
"loss": 0.1093,
"step": 7900
},
{
"epoch": 5.23,
"learning_rate": 2.187368421052632e-06,
"loss": 0.1105,
"step": 7925
},
{
"epoch": 5.25,
"learning_rate": 2.1610526315789476e-06,
"loss": 0.1122,
"step": 7950
},
{
"epoch": 5.27,
"learning_rate": 2.1347368421052634e-06,
"loss": 0.1114,
"step": 7975
},
{
"epoch": 5.28,
"learning_rate": 2.1084210526315792e-06,
"loss": 0.1095,
"step": 8000
},
{
"epoch": 5.28,
"eval_loss": 0.3078354001045227,
"eval_runtime": 46.3045,
"eval_samples_per_second": 36.519,
"eval_steps_per_second": 0.302,
"eval_wer": 15.286624203821656,
"step": 8000
},
{
"epoch": 5.3,
"learning_rate": 2.082105263157895e-06,
"loss": 0.1085,
"step": 8025
},
{
"epoch": 5.32,
"learning_rate": 2.055789473684211e-06,
"loss": 0.1085,
"step": 8050
},
{
"epoch": 5.33,
"learning_rate": 2.0294736842105263e-06,
"loss": 0.1079,
"step": 8075
},
{
"epoch": 5.35,
"learning_rate": 2.003157894736842e-06,
"loss": 0.1083,
"step": 8100
},
{
"epoch": 5.37,
"learning_rate": 1.976842105263158e-06,
"loss": 0.1117,
"step": 8125
},
{
"epoch": 5.38,
"learning_rate": 1.9505263157894737e-06,
"loss": 0.1118,
"step": 8150
},
{
"epoch": 5.4,
"learning_rate": 1.9242105263157896e-06,
"loss": 0.1094,
"step": 8175
},
{
"epoch": 5.42,
"learning_rate": 1.8978947368421056e-06,
"loss": 0.1079,
"step": 8200
},
{
"epoch": 5.43,
"learning_rate": 1.8715789473684212e-06,
"loss": 0.1081,
"step": 8225
},
{
"epoch": 5.45,
"learning_rate": 1.845263157894737e-06,
"loss": 0.108,
"step": 8250
},
{
"epoch": 5.47,
"learning_rate": 1.8189473684210528e-06,
"loss": 0.1097,
"step": 8275
},
{
"epoch": 5.48,
"learning_rate": 1.7926315789473686e-06,
"loss": 0.1085,
"step": 8300
},
{
"epoch": 5.5,
"learning_rate": 1.7663157894736843e-06,
"loss": 0.1087,
"step": 8325
},
{
"epoch": 5.52,
"learning_rate": 1.74e-06,
"loss": 0.1092,
"step": 8350
},
{
"epoch": 5.53,
"learning_rate": 1.713684210526316e-06,
"loss": 0.1102,
"step": 8375
},
{
"epoch": 5.55,
"learning_rate": 1.6873684210526317e-06,
"loss": 0.1068,
"step": 8400
},
{
"epoch": 5.56,
"learning_rate": 1.6610526315789473e-06,
"loss": 0.1076,
"step": 8425
},
{
"epoch": 5.58,
"learning_rate": 1.6347368421052633e-06,
"loss": 0.1119,
"step": 8450
},
{
"epoch": 5.6,
"learning_rate": 1.6084210526315792e-06,
"loss": 0.1088,
"step": 8475
},
{
"epoch": 5.61,
"learning_rate": 1.5821052631578948e-06,
"loss": 0.11,
"step": 8500
},
{
"epoch": 5.63,
"learning_rate": 1.5557894736842106e-06,
"loss": 0.1089,
"step": 8525
},
{
"epoch": 5.65,
"learning_rate": 1.5294736842105264e-06,
"loss": 0.1077,
"step": 8550
},
{
"epoch": 5.66,
"learning_rate": 1.5031578947368422e-06,
"loss": 0.11,
"step": 8575
},
{
"epoch": 5.68,
"learning_rate": 1.4768421052631578e-06,
"loss": 0.1106,
"step": 8600
},
{
"epoch": 5.7,
"learning_rate": 1.4505263157894739e-06,
"loss": 0.1089,
"step": 8625
},
{
"epoch": 5.71,
"learning_rate": 1.4242105263157897e-06,
"loss": 0.1067,
"step": 8650
},
{
"epoch": 5.73,
"learning_rate": 1.3978947368421053e-06,
"loss": 0.1075,
"step": 8675
},
{
"epoch": 5.75,
"learning_rate": 1.3715789473684213e-06,
"loss": 0.1062,
"step": 8700
},
{
"epoch": 5.76,
"learning_rate": 1.345263157894737e-06,
"loss": 0.1087,
"step": 8725
},
{
"epoch": 5.78,
"learning_rate": 1.3189473684210527e-06,
"loss": 0.1102,
"step": 8750
},
{
"epoch": 5.8,
"learning_rate": 1.2926315789473683e-06,
"loss": 0.1068,
"step": 8775
},
{
"epoch": 5.81,
"learning_rate": 1.2663157894736844e-06,
"loss": 0.1117,
"step": 8800
},
{
"epoch": 5.83,
"learning_rate": 1.2400000000000002e-06,
"loss": 0.1079,
"step": 8825
},
{
"epoch": 5.85,
"learning_rate": 1.213684210526316e-06,
"loss": 0.1073,
"step": 8850
},
{
"epoch": 5.86,
"learning_rate": 1.1873684210526316e-06,
"loss": 0.1038,
"step": 8875
},
{
"epoch": 5.88,
"learning_rate": 1.1610526315789474e-06,
"loss": 0.1056,
"step": 8900
},
{
"epoch": 5.89,
"learning_rate": 1.1347368421052633e-06,
"loss": 0.1094,
"step": 8925
},
{
"epoch": 5.91,
"learning_rate": 1.108421052631579e-06,
"loss": 0.1089,
"step": 8950
},
{
"epoch": 5.93,
"learning_rate": 1.082105263157895e-06,
"loss": 0.109,
"step": 8975
},
{
"epoch": 5.94,
"learning_rate": 1.0557894736842105e-06,
"loss": 0.11,
"step": 9000
},
{
"epoch": 5.94,
"eval_loss": 0.3059212267398834,
"eval_runtime": 46.4619,
"eval_samples_per_second": 36.395,
"eval_steps_per_second": 0.301,
"eval_wer": 15.280379667790683,
"step": 9000
},
{
"epoch": 5.96,
"learning_rate": 1.0294736842105265e-06,
"loss": 0.107,
"step": 9025
},
{
"epoch": 5.98,
"learning_rate": 1.0031578947368421e-06,
"loss": 0.1065,
"step": 9050
},
{
"epoch": 5.99,
"learning_rate": 9.76842105263158e-07,
"loss": 0.1058,
"step": 9075
},
{
"epoch": 6.01,
"learning_rate": 9.505263157894738e-07,
"loss": 0.1062,
"step": 9100
},
{
"epoch": 6.03,
"learning_rate": 9.242105263157895e-07,
"loss": 0.1022,
"step": 9125
},
{
"epoch": 6.04,
"learning_rate": 8.978947368421054e-07,
"loss": 0.1079,
"step": 9150
},
{
"epoch": 6.06,
"learning_rate": 8.715789473684211e-07,
"loss": 0.1034,
"step": 9175
},
{
"epoch": 6.08,
"learning_rate": 8.452631578947369e-07,
"loss": 0.1041,
"step": 9200
},
{
"epoch": 6.09,
"learning_rate": 8.189473684210527e-07,
"loss": 0.1047,
"step": 9225
},
{
"epoch": 6.11,
"learning_rate": 7.926315789473685e-07,
"loss": 0.1052,
"step": 9250
},
{
"epoch": 6.13,
"learning_rate": 7.663157894736843e-07,
"loss": 0.1039,
"step": 9275
},
{
"epoch": 6.14,
"learning_rate": 7.4e-07,
"loss": 0.1032,
"step": 9300
},
{
"epoch": 6.16,
"learning_rate": 7.136842105263158e-07,
"loss": 0.1016,
"step": 9325
},
{
"epoch": 6.18,
"learning_rate": 6.873684210526315e-07,
"loss": 0.1054,
"step": 9350
},
{
"epoch": 6.19,
"learning_rate": 6.610526315789475e-07,
"loss": 0.106,
"step": 9375
},
{
"epoch": 6.21,
"learning_rate": 6.347368421052633e-07,
"loss": 0.1049,
"step": 9400
},
{
"epoch": 6.23,
"learning_rate": 6.08421052631579e-07,
"loss": 0.1031,
"step": 9425
},
{
"epoch": 6.24,
"learning_rate": 5.821052631578948e-07,
"loss": 0.1032,
"step": 9450
},
{
"epoch": 6.26,
"learning_rate": 5.557894736842106e-07,
"loss": 0.1051,
"step": 9475
},
{
"epoch": 6.27,
"learning_rate": 5.294736842105263e-07,
"loss": 0.1056,
"step": 9500
},
{
"epoch": 6.29,
"learning_rate": 5.031578947368422e-07,
"loss": 0.1055,
"step": 9525
},
{
"epoch": 6.31,
"learning_rate": 4.768421052631579e-07,
"loss": 0.1046,
"step": 9550
},
{
"epoch": 6.32,
"learning_rate": 4.505263157894737e-07,
"loss": 0.1052,
"step": 9575
},
{
"epoch": 6.34,
"learning_rate": 4.242105263157895e-07,
"loss": 0.1072,
"step": 9600
},
{
"epoch": 6.36,
"learning_rate": 3.9789473684210533e-07,
"loss": 0.1069,
"step": 9625
},
{
"epoch": 6.37,
"learning_rate": 3.715789473684211e-07,
"loss": 0.1066,
"step": 9650
},
{
"epoch": 6.39,
"learning_rate": 3.4526315789473686e-07,
"loss": 0.1032,
"step": 9675
},
{
"epoch": 6.41,
"learning_rate": 3.1894736842105263e-07,
"loss": 0.1034,
"step": 9700
},
{
"epoch": 6.42,
"learning_rate": 2.9263157894736845e-07,
"loss": 0.1046,
"step": 9725
},
{
"epoch": 6.44,
"learning_rate": 2.663157894736842e-07,
"loss": 0.1037,
"step": 9750
},
{
"epoch": 6.46,
"learning_rate": 2.4000000000000003e-07,
"loss": 0.1051,
"step": 9775
},
{
"epoch": 6.47,
"learning_rate": 2.136842105263158e-07,
"loss": 0.1004,
"step": 9800
},
{
"epoch": 6.49,
"learning_rate": 1.8736842105263162e-07,
"loss": 0.1016,
"step": 9825
},
{
"epoch": 6.51,
"learning_rate": 1.6105263157894738e-07,
"loss": 0.1081,
"step": 9850
},
{
"epoch": 6.52,
"learning_rate": 1.3473684210526317e-07,
"loss": 0.107,
"step": 9875
},
{
"epoch": 6.54,
"learning_rate": 1.0842105263157895e-07,
"loss": 0.1063,
"step": 9900
},
{
"epoch": 6.56,
"learning_rate": 8.210526315789474e-08,
"loss": 0.1016,
"step": 9925
},
{
"epoch": 6.57,
"learning_rate": 5.578947368421053e-08,
"loss": 0.1046,
"step": 9950
},
{
"epoch": 6.59,
"learning_rate": 2.9473684210526318e-08,
"loss": 0.1072,
"step": 9975
},
{
"epoch": 6.61,
"learning_rate": 3.1578947368421054e-09,
"loss": 0.1048,
"step": 10000
},
{
"epoch": 6.61,
"eval_loss": 0.30679601430892944,
"eval_runtime": 46.3547,
"eval_samples_per_second": 36.48,
"eval_steps_per_second": 0.302,
"eval_wer": 15.274135131759712,
"step": 10000
},
{
"epoch": 6.61,
"step": 10000,
"total_flos": 6.300100720410624e+19,
"train_loss": 0.23215757968425751,
"train_runtime": 16345.8402,
"train_samples_per_second": 156.615,
"train_steps_per_second": 0.612
}
],
"max_steps": 10000,
"num_train_epochs": 7,
"total_flos": 6.300100720410624e+19,
"trial_name": null,
"trial_params": null
}