whisper-large-v2-es / trainer_state.json
arpagon's picture
End of training
0fa3e08
raw
history blame contribute delete
No virus
51.1 kB
{
"best_metric": 5.288186684683748,
"best_model_checkpoint": "./checkpoint-10000",
"epoch": 5.03,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.800000000000001e-07,
"loss": 0.8424,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.800000000000001e-07,
"loss": 0.5831,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 0.3011,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.9200000000000003e-06,
"loss": 0.2563,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.42e-06,
"loss": 0.2321,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.92e-06,
"loss": 0.1945,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.4200000000000007e-06,
"loss": 0.1634,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.920000000000001e-06,
"loss": 0.1808,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.42e-06,
"loss": 0.1676,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 4.92e-06,
"loss": 0.1741,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 5.420000000000001e-06,
"loss": 0.1598,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.92e-06,
"loss": 0.1417,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.42e-06,
"loss": 0.1426,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 6.92e-06,
"loss": 0.152,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 7.420000000000001e-06,
"loss": 0.1561,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 7.92e-06,
"loss": 0.179,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.42e-06,
"loss": 0.1732,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.920000000000001e-06,
"loss": 0.1928,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 9.42e-06,
"loss": 0.189,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 9.920000000000002e-06,
"loss": 0.1814,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 9.977894736842106e-06,
"loss": 0.1767,
"step": 525
},
{
"epoch": 0.06,
"learning_rate": 9.951578947368423e-06,
"loss": 0.1584,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 9.925263157894738e-06,
"loss": 0.1637,
"step": 575
},
{
"epoch": 0.06,
"learning_rate": 9.898947368421054e-06,
"loss": 0.1689,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 9.87263157894737e-06,
"loss": 0.1696,
"step": 625
},
{
"epoch": 0.07,
"learning_rate": 9.846315789473684e-06,
"loss": 0.1647,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 9.820000000000001e-06,
"loss": 0.1868,
"step": 675
},
{
"epoch": 0.07,
"learning_rate": 9.793684210526316e-06,
"loss": 0.1532,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 9.767368421052632e-06,
"loss": 0.155,
"step": 725
},
{
"epoch": 0.07,
"learning_rate": 9.741052631578947e-06,
"loss": 0.1521,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 9.714736842105264e-06,
"loss": 0.1613,
"step": 775
},
{
"epoch": 0.08,
"learning_rate": 9.68842105263158e-06,
"loss": 0.1665,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 9.662105263157896e-06,
"loss": 0.1795,
"step": 825
},
{
"epoch": 0.09,
"learning_rate": 9.635789473684212e-06,
"loss": 0.1635,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 9.609473684210527e-06,
"loss": 0.1762,
"step": 875
},
{
"epoch": 0.09,
"learning_rate": 9.583157894736842e-06,
"loss": 0.1709,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 9.556842105263159e-06,
"loss": 0.1837,
"step": 925
},
{
"epoch": 0.1,
"learning_rate": 9.530526315789474e-06,
"loss": 0.1917,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 9.50421052631579e-06,
"loss": 0.159,
"step": 975
},
{
"epoch": 0.1,
"learning_rate": 9.477894736842106e-06,
"loss": 0.1738,
"step": 1000
},
{
"epoch": 0.1,
"eval_loss": 0.20310187339782715,
"eval_runtime": 16231.5599,
"eval_samples_per_second": 0.956,
"eval_steps_per_second": 0.239,
"eval_wer": 7.038355594823553,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 9.451578947368422e-06,
"loss": 0.1611,
"step": 1025
},
{
"epoch": 0.1,
"learning_rate": 9.425263157894737e-06,
"loss": 0.1583,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 9.398947368421052e-06,
"loss": 0.1584,
"step": 1075
},
{
"epoch": 0.11,
"learning_rate": 9.372631578947369e-06,
"loss": 0.1689,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 9.346315789473684e-06,
"loss": 0.1584,
"step": 1125
},
{
"epoch": 0.12,
"learning_rate": 9.32e-06,
"loss": 0.1616,
"step": 1150
},
{
"epoch": 0.12,
"learning_rate": 9.293684210526317e-06,
"loss": 0.1608,
"step": 1175
},
{
"epoch": 0.12,
"learning_rate": 9.267368421052632e-06,
"loss": 0.1697,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 9.241052631578949e-06,
"loss": 0.1714,
"step": 1225
},
{
"epoch": 0.12,
"learning_rate": 9.214736842105264e-06,
"loss": 0.1694,
"step": 1250
},
{
"epoch": 0.13,
"learning_rate": 9.18842105263158e-06,
"loss": 0.1822,
"step": 1275
},
{
"epoch": 0.13,
"learning_rate": 9.162105263157895e-06,
"loss": 0.1608,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 9.13578947368421e-06,
"loss": 0.1611,
"step": 1325
},
{
"epoch": 0.14,
"learning_rate": 9.109473684210527e-06,
"loss": 0.1595,
"step": 1350
},
{
"epoch": 0.14,
"learning_rate": 9.083157894736842e-06,
"loss": 0.1437,
"step": 1375
},
{
"epoch": 0.14,
"learning_rate": 9.056842105263159e-06,
"loss": 0.1564,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 9.030526315789474e-06,
"loss": 0.1457,
"step": 1425
},
{
"epoch": 0.14,
"learning_rate": 9.00421052631579e-06,
"loss": 0.1289,
"step": 1450
},
{
"epoch": 0.15,
"learning_rate": 8.978947368421055e-06,
"loss": 0.1647,
"step": 1475
},
{
"epoch": 0.15,
"learning_rate": 8.95263157894737e-06,
"loss": 0.1748,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 8.926315789473685e-06,
"loss": 0.1646,
"step": 1525
},
{
"epoch": 0.15,
"learning_rate": 8.900000000000001e-06,
"loss": 0.1314,
"step": 1550
},
{
"epoch": 0.16,
"learning_rate": 8.873684210526316e-06,
"loss": 0.1776,
"step": 1575
},
{
"epoch": 0.16,
"learning_rate": 8.847368421052633e-06,
"loss": 0.1917,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 8.821052631578948e-06,
"loss": 0.1529,
"step": 1625
},
{
"epoch": 0.17,
"learning_rate": 8.794736842105264e-06,
"loss": 0.1609,
"step": 1650
},
{
"epoch": 0.17,
"learning_rate": 8.76842105263158e-06,
"loss": 0.1655,
"step": 1675
},
{
"epoch": 0.17,
"learning_rate": 8.742105263157894e-06,
"loss": 0.1585,
"step": 1700
},
{
"epoch": 0.17,
"learning_rate": 8.715789473684211e-06,
"loss": 0.1561,
"step": 1725
},
{
"epoch": 0.17,
"learning_rate": 8.689473684210526e-06,
"loss": 0.171,
"step": 1750
},
{
"epoch": 0.18,
"learning_rate": 8.663157894736843e-06,
"loss": 0.1908,
"step": 1775
},
{
"epoch": 0.18,
"learning_rate": 8.63684210526316e-06,
"loss": 0.1721,
"step": 1800
},
{
"epoch": 0.18,
"learning_rate": 8.610526315789474e-06,
"loss": 0.1744,
"step": 1825
},
{
"epoch": 0.18,
"learning_rate": 8.584210526315791e-06,
"loss": 0.1535,
"step": 1850
},
{
"epoch": 0.19,
"learning_rate": 8.557894736842106e-06,
"loss": 0.1684,
"step": 1875
},
{
"epoch": 0.19,
"learning_rate": 8.531578947368423e-06,
"loss": 0.1642,
"step": 1900
},
{
"epoch": 0.19,
"learning_rate": 8.505263157894738e-06,
"loss": 0.1786,
"step": 1925
},
{
"epoch": 1.0,
"learning_rate": 8.478947368421053e-06,
"loss": 0.2292,
"step": 1950
},
{
"epoch": 1.0,
"learning_rate": 8.45263157894737e-06,
"loss": 0.2358,
"step": 1975
},
{
"epoch": 1.01,
"learning_rate": 8.426315789473684e-06,
"loss": 0.2108,
"step": 2000
},
{
"epoch": 1.01,
"eval_loss": 0.18848001956939697,
"eval_runtime": 16376.339,
"eval_samples_per_second": 0.948,
"eval_steps_per_second": 0.237,
"eval_wer": 6.666753287251183,
"step": 2000
},
{
"epoch": 1.01,
"learning_rate": 8.400000000000001e-06,
"loss": 0.1974,
"step": 2025
},
{
"epoch": 1.01,
"learning_rate": 8.373684210526316e-06,
"loss": 0.1971,
"step": 2050
},
{
"epoch": 1.01,
"learning_rate": 8.347368421052633e-06,
"loss": 0.1891,
"step": 2075
},
{
"epoch": 1.02,
"learning_rate": 8.32105263157895e-06,
"loss": 0.1976,
"step": 2100
},
{
"epoch": 1.02,
"learning_rate": 8.294736842105264e-06,
"loss": 0.1978,
"step": 2125
},
{
"epoch": 1.02,
"learning_rate": 8.26842105263158e-06,
"loss": 0.1976,
"step": 2150
},
{
"epoch": 1.02,
"learning_rate": 8.242105263157896e-06,
"loss": 0.2031,
"step": 2175
},
{
"epoch": 1.03,
"learning_rate": 8.21578947368421e-06,
"loss": 0.2168,
"step": 2200
},
{
"epoch": 1.03,
"learning_rate": 8.189473684210527e-06,
"loss": 0.1731,
"step": 2225
},
{
"epoch": 1.03,
"learning_rate": 8.163157894736842e-06,
"loss": 0.1829,
"step": 2250
},
{
"epoch": 1.03,
"learning_rate": 8.136842105263159e-06,
"loss": 0.1714,
"step": 2275
},
{
"epoch": 1.04,
"learning_rate": 8.110526315789474e-06,
"loss": 0.1494,
"step": 2300
},
{
"epoch": 1.04,
"learning_rate": 8.08421052631579e-06,
"loss": 0.1627,
"step": 2325
},
{
"epoch": 1.04,
"learning_rate": 8.057894736842106e-06,
"loss": 0.1622,
"step": 2350
},
{
"epoch": 1.04,
"learning_rate": 8.03157894736842e-06,
"loss": 0.1926,
"step": 2375
},
{
"epoch": 1.05,
"learning_rate": 8.005263157894737e-06,
"loss": 0.1586,
"step": 2400
},
{
"epoch": 1.05,
"learning_rate": 7.978947368421052e-06,
"loss": 0.1773,
"step": 2425
},
{
"epoch": 1.05,
"learning_rate": 7.952631578947369e-06,
"loss": 0.1831,
"step": 2450
},
{
"epoch": 1.05,
"learning_rate": 7.926315789473686e-06,
"loss": 0.1753,
"step": 2475
},
{
"epoch": 1.06,
"learning_rate": 7.9e-06,
"loss": 0.163,
"step": 2500
},
{
"epoch": 1.06,
"learning_rate": 7.874736842105263e-06,
"loss": 0.1701,
"step": 2525
},
{
"epoch": 1.06,
"learning_rate": 7.84842105263158e-06,
"loss": 0.1937,
"step": 2550
},
{
"epoch": 1.06,
"learning_rate": 7.822105263157895e-06,
"loss": 0.2363,
"step": 2575
},
{
"epoch": 1.07,
"learning_rate": 7.795789473684211e-06,
"loss": 0.2011,
"step": 2600
},
{
"epoch": 1.07,
"learning_rate": 7.769473684210526e-06,
"loss": 0.1569,
"step": 2625
},
{
"epoch": 1.07,
"learning_rate": 7.743157894736843e-06,
"loss": 0.135,
"step": 2650
},
{
"epoch": 1.07,
"learning_rate": 7.716842105263158e-06,
"loss": 0.1419,
"step": 2675
},
{
"epoch": 1.08,
"learning_rate": 7.690526315789475e-06,
"loss": 0.1272,
"step": 2700
},
{
"epoch": 1.08,
"learning_rate": 7.664210526315791e-06,
"loss": 0.15,
"step": 2725
},
{
"epoch": 1.08,
"learning_rate": 7.637894736842106e-06,
"loss": 0.1363,
"step": 2750
},
{
"epoch": 1.08,
"learning_rate": 7.611578947368422e-06,
"loss": 0.1448,
"step": 2775
},
{
"epoch": 1.09,
"learning_rate": 7.585263157894738e-06,
"loss": 0.1358,
"step": 2800
},
{
"epoch": 1.09,
"learning_rate": 7.558947368421054e-06,
"loss": 0.1378,
"step": 2825
},
{
"epoch": 1.09,
"learning_rate": 7.53263157894737e-06,
"loss": 0.1534,
"step": 2850
},
{
"epoch": 1.09,
"learning_rate": 7.5063157894736846e-06,
"loss": 0.1557,
"step": 2875
},
{
"epoch": 1.1,
"learning_rate": 7.48e-06,
"loss": 0.1477,
"step": 2900
},
{
"epoch": 1.1,
"learning_rate": 7.453684210526316e-06,
"loss": 0.1466,
"step": 2925
},
{
"epoch": 1.1,
"learning_rate": 7.427368421052632e-06,
"loss": 0.1505,
"step": 2950
},
{
"epoch": 1.1,
"learning_rate": 7.401052631578948e-06,
"loss": 0.1431,
"step": 2975
},
{
"epoch": 1.11,
"learning_rate": 7.374736842105264e-06,
"loss": 0.1599,
"step": 3000
},
{
"epoch": 1.11,
"eval_loss": 0.1814190298318863,
"eval_runtime": 16373.1622,
"eval_samples_per_second": 0.948,
"eval_steps_per_second": 0.237,
"eval_wer": 6.534223792942155,
"step": 3000
},
{
"epoch": 1.11,
"learning_rate": 7.3484210526315795e-06,
"loss": 0.1431,
"step": 3025
},
{
"epoch": 1.11,
"learning_rate": 7.3221052631578944e-06,
"loss": 0.1301,
"step": 3050
},
{
"epoch": 1.11,
"learning_rate": 7.295789473684212e-06,
"loss": 0.1275,
"step": 3075
},
{
"epoch": 1.12,
"learning_rate": 7.269473684210527e-06,
"loss": 0.1422,
"step": 3100
},
{
"epoch": 1.12,
"learning_rate": 7.243157894736843e-06,
"loss": 0.1383,
"step": 3125
},
{
"epoch": 1.12,
"learning_rate": 7.2168421052631586e-06,
"loss": 0.153,
"step": 3150
},
{
"epoch": 1.12,
"learning_rate": 7.190526315789474e-06,
"loss": 0.1536,
"step": 3175
},
{
"epoch": 1.13,
"learning_rate": 7.16421052631579e-06,
"loss": 0.156,
"step": 3200
},
{
"epoch": 1.13,
"learning_rate": 7.137894736842106e-06,
"loss": 0.1622,
"step": 3225
},
{
"epoch": 1.13,
"learning_rate": 7.111578947368422e-06,
"loss": 0.1419,
"step": 3250
},
{
"epoch": 1.13,
"learning_rate": 7.085263157894738e-06,
"loss": 0.1364,
"step": 3275
},
{
"epoch": 1.14,
"learning_rate": 7.058947368421053e-06,
"loss": 0.1639,
"step": 3300
},
{
"epoch": 1.14,
"learning_rate": 7.0326315789473684e-06,
"loss": 0.1512,
"step": 3325
},
{
"epoch": 1.14,
"learning_rate": 7.006315789473684e-06,
"loss": 0.1535,
"step": 3350
},
{
"epoch": 1.14,
"learning_rate": 6.98e-06,
"loss": 0.1381,
"step": 3375
},
{
"epoch": 1.15,
"learning_rate": 6.953684210526316e-06,
"loss": 0.1253,
"step": 3400
},
{
"epoch": 1.15,
"learning_rate": 6.9273684210526326e-06,
"loss": 0.1474,
"step": 3425
},
{
"epoch": 1.15,
"learning_rate": 6.901052631578948e-06,
"loss": 0.1394,
"step": 3450
},
{
"epoch": 1.15,
"learning_rate": 6.874736842105264e-06,
"loss": 0.1434,
"step": 3475
},
{
"epoch": 1.16,
"learning_rate": 6.84842105263158e-06,
"loss": 0.1503,
"step": 3500
},
{
"epoch": 1.16,
"learning_rate": 6.822105263157896e-06,
"loss": 0.1528,
"step": 3525
},
{
"epoch": 1.16,
"learning_rate": 6.795789473684211e-06,
"loss": 0.1496,
"step": 3550
},
{
"epoch": 1.16,
"learning_rate": 6.769473684210527e-06,
"loss": 0.1237,
"step": 3575
},
{
"epoch": 1.17,
"learning_rate": 6.7431578947368425e-06,
"loss": 0.1376,
"step": 3600
},
{
"epoch": 1.17,
"learning_rate": 6.716842105263158e-06,
"loss": 0.128,
"step": 3625
},
{
"epoch": 1.17,
"learning_rate": 6.690526315789474e-06,
"loss": 0.1152,
"step": 3650
},
{
"epoch": 1.17,
"learning_rate": 6.66421052631579e-06,
"loss": 0.1319,
"step": 3675
},
{
"epoch": 1.18,
"learning_rate": 6.637894736842106e-06,
"loss": 0.1348,
"step": 3700
},
{
"epoch": 1.18,
"learning_rate": 6.611578947368421e-06,
"loss": 0.1307,
"step": 3725
},
{
"epoch": 1.18,
"learning_rate": 6.585263157894738e-06,
"loss": 0.149,
"step": 3750
},
{
"epoch": 1.18,
"learning_rate": 6.558947368421054e-06,
"loss": 0.1372,
"step": 3775
},
{
"epoch": 1.19,
"learning_rate": 6.532631578947369e-06,
"loss": 0.154,
"step": 3800
},
{
"epoch": 1.19,
"learning_rate": 6.506315789473685e-06,
"loss": 0.1456,
"step": 3825
},
{
"epoch": 1.19,
"learning_rate": 6.480000000000001e-06,
"loss": 0.1654,
"step": 3850
},
{
"epoch": 1.19,
"learning_rate": 6.4536842105263165e-06,
"loss": 0.1377,
"step": 3875
},
{
"epoch": 2.0,
"learning_rate": 6.427368421052632e-06,
"loss": 0.1147,
"step": 3900
},
{
"epoch": 2.0,
"learning_rate": 6.401052631578948e-06,
"loss": 0.0916,
"step": 3925
},
{
"epoch": 2.01,
"learning_rate": 6.374736842105264e-06,
"loss": 0.0903,
"step": 3950
},
{
"epoch": 2.01,
"learning_rate": 6.348421052631579e-06,
"loss": 0.092,
"step": 3975
},
{
"epoch": 2.01,
"learning_rate": 6.322105263157895e-06,
"loss": 0.0794,
"step": 4000
},
{
"epoch": 2.01,
"eval_loss": 0.179224893450737,
"eval_runtime": 16334.157,
"eval_samples_per_second": 0.95,
"eval_steps_per_second": 0.238,
"eval_wer": 6.031391299828491,
"step": 4000
},
{
"epoch": 2.01,
"learning_rate": 6.2957894736842105e-06,
"loss": 0.0904,
"step": 4025
},
{
"epoch": 2.02,
"learning_rate": 6.269473684210526e-06,
"loss": 0.0903,
"step": 4050
},
{
"epoch": 2.02,
"learning_rate": 6.243157894736842e-06,
"loss": 0.0861,
"step": 4075
},
{
"epoch": 2.02,
"learning_rate": 6.216842105263159e-06,
"loss": 0.0975,
"step": 4100
},
{
"epoch": 2.02,
"learning_rate": 6.190526315789475e-06,
"loss": 0.0879,
"step": 4125
},
{
"epoch": 2.03,
"learning_rate": 6.1642105263157905e-06,
"loss": 0.0756,
"step": 4150
},
{
"epoch": 2.03,
"learning_rate": 6.137894736842106e-06,
"loss": 0.0714,
"step": 4175
},
{
"epoch": 2.03,
"learning_rate": 6.111578947368422e-06,
"loss": 0.0704,
"step": 4200
},
{
"epoch": 2.03,
"learning_rate": 6.085263157894737e-06,
"loss": 0.07,
"step": 4225
},
{
"epoch": 2.04,
"learning_rate": 6.058947368421053e-06,
"loss": 0.0713,
"step": 4250
},
{
"epoch": 2.04,
"learning_rate": 6.032631578947369e-06,
"loss": 0.0705,
"step": 4275
},
{
"epoch": 2.04,
"learning_rate": 6.0063157894736845e-06,
"loss": 0.0703,
"step": 4300
},
{
"epoch": 2.04,
"learning_rate": 5.98e-06,
"loss": 0.0661,
"step": 4325
},
{
"epoch": 2.05,
"learning_rate": 5.953684210526316e-06,
"loss": 0.0726,
"step": 4350
},
{
"epoch": 2.05,
"learning_rate": 5.927368421052632e-06,
"loss": 0.0723,
"step": 4375
},
{
"epoch": 2.05,
"learning_rate": 5.901052631578947e-06,
"loss": 0.0691,
"step": 4400
},
{
"epoch": 2.05,
"learning_rate": 5.8747368421052645e-06,
"loss": 0.0776,
"step": 4425
},
{
"epoch": 2.06,
"learning_rate": 5.84842105263158e-06,
"loss": 0.0676,
"step": 4450
},
{
"epoch": 2.06,
"learning_rate": 5.822105263157895e-06,
"loss": 0.0772,
"step": 4475
},
{
"epoch": 2.06,
"learning_rate": 5.795789473684211e-06,
"loss": 0.0895,
"step": 4500
},
{
"epoch": 2.06,
"learning_rate": 5.769473684210527e-06,
"loss": 0.0812,
"step": 4525
},
{
"epoch": 2.07,
"learning_rate": 5.743157894736843e-06,
"loss": 0.0771,
"step": 4550
},
{
"epoch": 2.07,
"learning_rate": 5.7168421052631585e-06,
"loss": 0.0655,
"step": 4575
},
{
"epoch": 2.07,
"learning_rate": 5.690526315789474e-06,
"loss": 0.0533,
"step": 4600
},
{
"epoch": 2.07,
"learning_rate": 5.66421052631579e-06,
"loss": 0.0543,
"step": 4625
},
{
"epoch": 2.08,
"learning_rate": 5.637894736842105e-06,
"loss": 0.0637,
"step": 4650
},
{
"epoch": 2.08,
"learning_rate": 5.611578947368421e-06,
"loss": 0.0545,
"step": 4675
},
{
"epoch": 2.08,
"learning_rate": 5.585263157894737e-06,
"loss": 0.0536,
"step": 4700
},
{
"epoch": 2.08,
"learning_rate": 5.558947368421053e-06,
"loss": 0.0502,
"step": 4725
},
{
"epoch": 2.09,
"learning_rate": 5.532631578947368e-06,
"loss": 0.0453,
"step": 4750
},
{
"epoch": 2.09,
"learning_rate": 5.506315789473685e-06,
"loss": 0.0585,
"step": 4775
},
{
"epoch": 2.09,
"learning_rate": 5.480000000000001e-06,
"loss": 0.0683,
"step": 4800
},
{
"epoch": 2.09,
"learning_rate": 5.453684210526317e-06,
"loss": 0.0637,
"step": 4825
},
{
"epoch": 2.1,
"learning_rate": 5.4273684210526325e-06,
"loss": 0.0517,
"step": 4850
},
{
"epoch": 2.1,
"learning_rate": 5.401052631578948e-06,
"loss": 0.0731,
"step": 4875
},
{
"epoch": 2.1,
"learning_rate": 5.374736842105263e-06,
"loss": 0.0613,
"step": 4900
},
{
"epoch": 2.1,
"learning_rate": 5.348421052631579e-06,
"loss": 0.0576,
"step": 4925
},
{
"epoch": 2.11,
"learning_rate": 5.322105263157895e-06,
"loss": 0.0591,
"step": 4950
},
{
"epoch": 2.11,
"learning_rate": 5.295789473684211e-06,
"loss": 0.0629,
"step": 4975
},
{
"epoch": 2.11,
"learning_rate": 5.269473684210527e-06,
"loss": 0.0477,
"step": 5000
},
{
"epoch": 2.11,
"eval_loss": 0.1935829371213913,
"eval_runtime": 16412.3819,
"eval_samples_per_second": 0.946,
"eval_steps_per_second": 0.236,
"eval_wer": 6.179512499350345,
"step": 5000
},
{
"epoch": 2.11,
"learning_rate": 5.243157894736842e-06,
"loss": 0.0562,
"step": 5025
},
{
"epoch": 2.12,
"learning_rate": 5.216842105263158e-06,
"loss": 0.0559,
"step": 5050
},
{
"epoch": 2.12,
"learning_rate": 5.190526315789474e-06,
"loss": 0.0678,
"step": 5075
},
{
"epoch": 2.12,
"learning_rate": 5.164210526315791e-06,
"loss": 0.0588,
"step": 5100
},
{
"epoch": 2.12,
"learning_rate": 5.1378947368421065e-06,
"loss": 0.0671,
"step": 5125
},
{
"epoch": 2.13,
"learning_rate": 5.1115789473684215e-06,
"loss": 0.0744,
"step": 5150
},
{
"epoch": 2.13,
"learning_rate": 5.085263157894737e-06,
"loss": 0.0659,
"step": 5175
},
{
"epoch": 2.13,
"learning_rate": 5.058947368421053e-06,
"loss": 0.0731,
"step": 5200
},
{
"epoch": 2.13,
"learning_rate": 5.032631578947369e-06,
"loss": 0.0531,
"step": 5225
},
{
"epoch": 2.14,
"learning_rate": 5.006315789473685e-06,
"loss": 0.0618,
"step": 5250
},
{
"epoch": 2.14,
"learning_rate": 4.980000000000001e-06,
"loss": 0.0642,
"step": 5275
},
{
"epoch": 2.14,
"learning_rate": 4.953684210526316e-06,
"loss": 0.0577,
"step": 5300
},
{
"epoch": 2.14,
"learning_rate": 4.927368421052631e-06,
"loss": 0.0654,
"step": 5325
},
{
"epoch": 2.15,
"learning_rate": 4.901052631578947e-06,
"loss": 0.058,
"step": 5350
},
{
"epoch": 2.15,
"learning_rate": 4.874736842105264e-06,
"loss": 0.0637,
"step": 5375
},
{
"epoch": 2.15,
"learning_rate": 4.84842105263158e-06,
"loss": 0.0531,
"step": 5400
},
{
"epoch": 2.15,
"learning_rate": 4.8221052631578955e-06,
"loss": 0.0774,
"step": 5425
},
{
"epoch": 2.16,
"learning_rate": 4.7957894736842105e-06,
"loss": 0.068,
"step": 5450
},
{
"epoch": 2.16,
"learning_rate": 4.769473684210526e-06,
"loss": 0.059,
"step": 5475
},
{
"epoch": 2.16,
"learning_rate": 4.743157894736842e-06,
"loss": 0.0611,
"step": 5500
},
{
"epoch": 2.16,
"learning_rate": 4.716842105263159e-06,
"loss": 0.0546,
"step": 5525
},
{
"epoch": 2.17,
"learning_rate": 4.690526315789475e-06,
"loss": 0.0602,
"step": 5550
},
{
"epoch": 2.17,
"learning_rate": 4.6642105263157896e-06,
"loss": 0.0434,
"step": 5575
},
{
"epoch": 2.17,
"learning_rate": 4.637894736842105e-06,
"loss": 0.0513,
"step": 5600
},
{
"epoch": 2.17,
"learning_rate": 4.611578947368421e-06,
"loss": 0.0548,
"step": 5625
},
{
"epoch": 2.18,
"learning_rate": 4.585263157894737e-06,
"loss": 0.0659,
"step": 5650
},
{
"epoch": 2.18,
"learning_rate": 4.558947368421053e-06,
"loss": 0.058,
"step": 5675
},
{
"epoch": 2.18,
"learning_rate": 4.532631578947369e-06,
"loss": 0.0656,
"step": 5700
},
{
"epoch": 2.18,
"learning_rate": 4.5063157894736845e-06,
"loss": 0.0638,
"step": 5725
},
{
"epoch": 2.19,
"learning_rate": 4.48e-06,
"loss": 0.0617,
"step": 5750
},
{
"epoch": 2.19,
"learning_rate": 4.453684210526316e-06,
"loss": 0.0663,
"step": 5775
},
{
"epoch": 2.19,
"learning_rate": 4.427368421052632e-06,
"loss": 0.0639,
"step": 5800
},
{
"epoch": 3.0,
"learning_rate": 4.401052631578948e-06,
"loss": 0.0642,
"step": 5825
},
{
"epoch": 3.0,
"learning_rate": 4.374736842105264e-06,
"loss": 0.0452,
"step": 5850
},
{
"epoch": 3.01,
"learning_rate": 4.348421052631579e-06,
"loss": 0.0359,
"step": 5875
},
{
"epoch": 3.01,
"learning_rate": 4.322105263157895e-06,
"loss": 0.0384,
"step": 5900
},
{
"epoch": 3.01,
"learning_rate": 4.295789473684211e-06,
"loss": 0.0337,
"step": 5925
},
{
"epoch": 3.01,
"learning_rate": 4.269473684210527e-06,
"loss": 0.0313,
"step": 5950
},
{
"epoch": 3.02,
"learning_rate": 4.243157894736843e-06,
"loss": 0.0364,
"step": 5975
},
{
"epoch": 3.02,
"learning_rate": 4.2168421052631585e-06,
"loss": 0.0341,
"step": 6000
},
{
"epoch": 3.02,
"eval_loss": 0.2037559300661087,
"eval_runtime": 16334.9327,
"eval_samples_per_second": 0.95,
"eval_steps_per_second": 0.238,
"eval_wer": 6.01125201392859,
"step": 6000
},
{
"epoch": 3.02,
"learning_rate": 4.1905263157894735e-06,
"loss": 0.0345,
"step": 6025
},
{
"epoch": 3.02,
"learning_rate": 4.16421052631579e-06,
"loss": 0.0412,
"step": 6050
},
{
"epoch": 3.03,
"learning_rate": 4.137894736842106e-06,
"loss": 0.0374,
"step": 6075
},
{
"epoch": 3.03,
"learning_rate": 4.111578947368422e-06,
"loss": 0.03,
"step": 6100
},
{
"epoch": 3.03,
"learning_rate": 4.085263157894737e-06,
"loss": 0.0341,
"step": 6125
},
{
"epoch": 3.03,
"learning_rate": 4.0589473684210526e-06,
"loss": 0.0329,
"step": 6150
},
{
"epoch": 3.04,
"learning_rate": 4.032631578947368e-06,
"loss": 0.0273,
"step": 6175
},
{
"epoch": 3.04,
"learning_rate": 4.006315789473684e-06,
"loss": 0.0342,
"step": 6200
},
{
"epoch": 3.04,
"learning_rate": 3.980000000000001e-06,
"loss": 0.0277,
"step": 6225
},
{
"epoch": 3.04,
"learning_rate": 3.953684210526316e-06,
"loss": 0.0334,
"step": 6250
},
{
"epoch": 3.05,
"learning_rate": 3.927368421052632e-06,
"loss": 0.0312,
"step": 6275
},
{
"epoch": 3.05,
"learning_rate": 3.9010526315789475e-06,
"loss": 0.0271,
"step": 6300
},
{
"epoch": 3.05,
"learning_rate": 3.874736842105263e-06,
"loss": 0.0295,
"step": 6325
},
{
"epoch": 3.05,
"learning_rate": 3.848421052631579e-06,
"loss": 0.0295,
"step": 6350
},
{
"epoch": 3.06,
"learning_rate": 3.822105263157895e-06,
"loss": 0.0303,
"step": 6375
},
{
"epoch": 3.06,
"learning_rate": 3.795789473684211e-06,
"loss": 0.0257,
"step": 6400
},
{
"epoch": 3.06,
"learning_rate": 3.7694736842105266e-06,
"loss": 0.0283,
"step": 6425
},
{
"epoch": 3.06,
"learning_rate": 3.7431578947368424e-06,
"loss": 0.0405,
"step": 6450
},
{
"epoch": 3.07,
"learning_rate": 3.716842105263158e-06,
"loss": 0.031,
"step": 6475
},
{
"epoch": 3.07,
"learning_rate": 3.690526315789474e-06,
"loss": 0.0262,
"step": 6500
},
{
"epoch": 3.07,
"learning_rate": 3.6642105263157894e-06,
"loss": 0.0257,
"step": 6525
},
{
"epoch": 3.07,
"learning_rate": 3.6378947368421057e-06,
"loss": 0.0216,
"step": 6550
},
{
"epoch": 3.08,
"learning_rate": 3.6115789473684215e-06,
"loss": 0.0231,
"step": 6575
},
{
"epoch": 3.08,
"learning_rate": 3.5852631578947373e-06,
"loss": 0.023,
"step": 6600
},
{
"epoch": 3.08,
"learning_rate": 3.558947368421053e-06,
"loss": 0.0215,
"step": 6625
},
{
"epoch": 3.08,
"learning_rate": 3.5326315789473685e-06,
"loss": 0.0221,
"step": 6650
},
{
"epoch": 3.09,
"learning_rate": 3.5063157894736843e-06,
"loss": 0.0201,
"step": 6675
},
{
"epoch": 3.09,
"learning_rate": 3.48e-06,
"loss": 0.0195,
"step": 6700
},
{
"epoch": 3.09,
"learning_rate": 3.4536842105263164e-06,
"loss": 0.0203,
"step": 6725
},
{
"epoch": 3.09,
"learning_rate": 3.427368421052632e-06,
"loss": 0.0226,
"step": 6750
},
{
"epoch": 3.1,
"learning_rate": 3.4010526315789476e-06,
"loss": 0.02,
"step": 6775
},
{
"epoch": 3.1,
"learning_rate": 3.3747368421052634e-06,
"loss": 0.0237,
"step": 6800
},
{
"epoch": 3.1,
"learning_rate": 3.3484210526315792e-06,
"loss": 0.0229,
"step": 6825
},
{
"epoch": 3.1,
"learning_rate": 3.3221052631578946e-06,
"loss": 0.0274,
"step": 6850
},
{
"epoch": 3.11,
"learning_rate": 3.2957894736842104e-06,
"loss": 0.027,
"step": 6875
},
{
"epoch": 3.11,
"learning_rate": 3.2694736842105267e-06,
"loss": 0.0271,
"step": 6900
},
{
"epoch": 3.11,
"learning_rate": 3.2431578947368425e-06,
"loss": 0.022,
"step": 6925
},
{
"epoch": 3.11,
"learning_rate": 3.2168421052631583e-06,
"loss": 0.023,
"step": 6950
},
{
"epoch": 3.12,
"learning_rate": 3.1905263157894737e-06,
"loss": 0.02,
"step": 6975
},
{
"epoch": 3.12,
"learning_rate": 3.1642105263157895e-06,
"loss": 0.0264,
"step": 7000
},
{
"epoch": 3.12,
"eval_loss": 0.2110663503408432,
"eval_runtime": 16366.3852,
"eval_samples_per_second": 0.948,
"eval_steps_per_second": 0.237,
"eval_wer": 5.841042565355231,
"step": 7000
},
{
"epoch": 3.12,
"learning_rate": 3.1378947368421054e-06,
"loss": 0.0282,
"step": 7025
},
{
"epoch": 3.12,
"learning_rate": 3.111578947368421e-06,
"loss": 0.0249,
"step": 7050
},
{
"epoch": 3.13,
"learning_rate": 3.0852631578947374e-06,
"loss": 0.0286,
"step": 7075
},
{
"epoch": 3.13,
"learning_rate": 3.058947368421053e-06,
"loss": 0.0275,
"step": 7100
},
{
"epoch": 3.13,
"learning_rate": 3.0326315789473686e-06,
"loss": 0.0215,
"step": 7125
},
{
"epoch": 3.13,
"learning_rate": 3.0063157894736844e-06,
"loss": 0.0229,
"step": 7150
},
{
"epoch": 3.14,
"learning_rate": 2.9800000000000003e-06,
"loss": 0.0263,
"step": 7175
},
{
"epoch": 3.14,
"learning_rate": 2.9536842105263157e-06,
"loss": 0.0265,
"step": 7200
},
{
"epoch": 3.14,
"learning_rate": 2.927368421052632e-06,
"loss": 0.0264,
"step": 7225
},
{
"epoch": 3.14,
"learning_rate": 2.9010526315789477e-06,
"loss": 0.0227,
"step": 7250
},
{
"epoch": 3.15,
"learning_rate": 2.8747368421052635e-06,
"loss": 0.0221,
"step": 7275
},
{
"epoch": 3.15,
"learning_rate": 2.8484210526315794e-06,
"loss": 0.0203,
"step": 7300
},
{
"epoch": 3.15,
"learning_rate": 2.8221052631578948e-06,
"loss": 0.025,
"step": 7325
},
{
"epoch": 3.15,
"learning_rate": 2.7957894736842106e-06,
"loss": 0.0294,
"step": 7350
},
{
"epoch": 3.16,
"learning_rate": 2.7694736842105264e-06,
"loss": 0.0308,
"step": 7375
},
{
"epoch": 3.16,
"learning_rate": 2.7431578947368426e-06,
"loss": 0.0242,
"step": 7400
},
{
"epoch": 3.16,
"learning_rate": 2.7168421052631585e-06,
"loss": 0.0223,
"step": 7425
},
{
"epoch": 3.16,
"learning_rate": 2.690526315789474e-06,
"loss": 0.0217,
"step": 7450
},
{
"epoch": 3.17,
"learning_rate": 2.6642105263157897e-06,
"loss": 0.0229,
"step": 7475
},
{
"epoch": 3.17,
"learning_rate": 2.6378947368421055e-06,
"loss": 0.025,
"step": 7500
},
{
"epoch": 3.17,
"learning_rate": 2.6115789473684213e-06,
"loss": 0.0205,
"step": 7525
},
{
"epoch": 3.17,
"learning_rate": 2.5852631578947367e-06,
"loss": 0.0207,
"step": 7550
},
{
"epoch": 3.18,
"learning_rate": 2.558947368421053e-06,
"loss": 0.0248,
"step": 7575
},
{
"epoch": 3.18,
"learning_rate": 2.5326315789473688e-06,
"loss": 0.0226,
"step": 7600
},
{
"epoch": 3.18,
"learning_rate": 2.5063157894736846e-06,
"loss": 0.0258,
"step": 7625
},
{
"epoch": 3.18,
"learning_rate": 2.4800000000000004e-06,
"loss": 0.0212,
"step": 7650
},
{
"epoch": 3.19,
"learning_rate": 2.453684210526316e-06,
"loss": 0.0288,
"step": 7675
},
{
"epoch": 3.19,
"learning_rate": 2.427368421052632e-06,
"loss": 0.0299,
"step": 7700
},
{
"epoch": 3.19,
"learning_rate": 2.4010526315789474e-06,
"loss": 0.0247,
"step": 7725
},
{
"epoch": 3.19,
"learning_rate": 2.3747368421052632e-06,
"loss": 0.0255,
"step": 7750
},
{
"epoch": 4.0,
"learning_rate": 2.348421052631579e-06,
"loss": 0.0654,
"step": 7775
},
{
"epoch": 4.0,
"learning_rate": 2.322105263157895e-06,
"loss": 0.0913,
"step": 7800
},
{
"epoch": 4.01,
"learning_rate": 2.2957894736842107e-06,
"loss": 0.1062,
"step": 7825
},
{
"epoch": 4.01,
"learning_rate": 2.2694736842105265e-06,
"loss": 0.0783,
"step": 7850
},
{
"epoch": 4.01,
"learning_rate": 2.2431578947368423e-06,
"loss": 0.0794,
"step": 7875
},
{
"epoch": 4.01,
"learning_rate": 2.216842105263158e-06,
"loss": 0.0683,
"step": 7900
},
{
"epoch": 4.02,
"learning_rate": 2.190526315789474e-06,
"loss": 0.0555,
"step": 7925
},
{
"epoch": 4.02,
"learning_rate": 2.16421052631579e-06,
"loss": 0.0654,
"step": 7950
},
{
"epoch": 4.02,
"learning_rate": 2.1378947368421056e-06,
"loss": 0.0584,
"step": 7975
},
{
"epoch": 4.02,
"learning_rate": 2.111578947368421e-06,
"loss": 0.0608,
"step": 8000
},
{
"epoch": 4.02,
"eval_loss": 0.18235287070274353,
"eval_runtime": 16380.7046,
"eval_samples_per_second": 0.947,
"eval_steps_per_second": 0.237,
"eval_wer": 5.906657658125877,
"step": 8000
},
{
"epoch": 4.03,
"learning_rate": 2.085263157894737e-06,
"loss": 0.0518,
"step": 8025
},
{
"epoch": 4.03,
"learning_rate": 2.058947368421053e-06,
"loss": 0.0416,
"step": 8050
},
{
"epoch": 4.03,
"learning_rate": 2.0326315789473685e-06,
"loss": 0.039,
"step": 8075
},
{
"epoch": 4.03,
"learning_rate": 2.0063157894736843e-06,
"loss": 0.0411,
"step": 8100
},
{
"epoch": 4.04,
"learning_rate": 1.98e-06,
"loss": 0.0412,
"step": 8125
},
{
"epoch": 4.04,
"learning_rate": 1.953684210526316e-06,
"loss": 0.043,
"step": 8150
},
{
"epoch": 4.04,
"learning_rate": 1.9273684210526317e-06,
"loss": 0.0503,
"step": 8175
},
{
"epoch": 4.04,
"learning_rate": 1.9010526315789476e-06,
"loss": 0.0517,
"step": 8200
},
{
"epoch": 4.05,
"learning_rate": 1.8747368421052634e-06,
"loss": 0.0499,
"step": 8225
},
{
"epoch": 4.05,
"learning_rate": 1.848421052631579e-06,
"loss": 0.0393,
"step": 8250
},
{
"epoch": 4.05,
"learning_rate": 1.8221052631578948e-06,
"loss": 0.0444,
"step": 8275
},
{
"epoch": 4.05,
"learning_rate": 1.7957894736842108e-06,
"loss": 0.04,
"step": 8300
},
{
"epoch": 4.06,
"learning_rate": 1.7694736842105264e-06,
"loss": 0.0439,
"step": 8325
},
{
"epoch": 4.06,
"learning_rate": 1.7431578947368423e-06,
"loss": 0.0346,
"step": 8350
},
{
"epoch": 4.06,
"learning_rate": 1.716842105263158e-06,
"loss": 0.0415,
"step": 8375
},
{
"epoch": 4.06,
"learning_rate": 1.6905263157894739e-06,
"loss": 0.0363,
"step": 8400
},
{
"epoch": 4.07,
"learning_rate": 1.6642105263157895e-06,
"loss": 0.0393,
"step": 8425
},
{
"epoch": 4.07,
"learning_rate": 1.6378947368421053e-06,
"loss": 0.0362,
"step": 8450
},
{
"epoch": 4.07,
"learning_rate": 1.6115789473684211e-06,
"loss": 0.0357,
"step": 8475
},
{
"epoch": 4.07,
"learning_rate": 1.585263157894737e-06,
"loss": 0.0311,
"step": 8500
},
{
"epoch": 4.08,
"learning_rate": 1.5589473684210526e-06,
"loss": 0.0311,
"step": 8525
},
{
"epoch": 4.08,
"learning_rate": 1.5326315789473686e-06,
"loss": 0.0307,
"step": 8550
},
{
"epoch": 4.08,
"learning_rate": 1.5063157894736844e-06,
"loss": 0.032,
"step": 8575
},
{
"epoch": 4.08,
"learning_rate": 1.48e-06,
"loss": 0.0368,
"step": 8600
},
{
"epoch": 4.09,
"learning_rate": 1.453684210526316e-06,
"loss": 0.0387,
"step": 8625
},
{
"epoch": 4.09,
"learning_rate": 1.4273684210526317e-06,
"loss": 0.046,
"step": 8650
},
{
"epoch": 4.09,
"learning_rate": 1.402105263157895e-06,
"loss": 0.0422,
"step": 8675
},
{
"epoch": 4.09,
"learning_rate": 1.3757894736842106e-06,
"loss": 0.0385,
"step": 8700
},
{
"epoch": 4.1,
"learning_rate": 1.3494736842105262e-06,
"loss": 0.0448,
"step": 8725
},
{
"epoch": 4.1,
"learning_rate": 1.3231578947368423e-06,
"loss": 0.0405,
"step": 8750
},
{
"epoch": 4.1,
"learning_rate": 1.296842105263158e-06,
"loss": 0.0373,
"step": 8775
},
{
"epoch": 4.1,
"learning_rate": 1.2705263157894737e-06,
"loss": 0.0441,
"step": 8800
},
{
"epoch": 4.11,
"learning_rate": 1.2442105263157895e-06,
"loss": 0.0404,
"step": 8825
},
{
"epoch": 4.11,
"learning_rate": 1.2178947368421053e-06,
"loss": 0.0496,
"step": 8850
},
{
"epoch": 4.11,
"learning_rate": 1.1915789473684211e-06,
"loss": 0.0303,
"step": 8875
},
{
"epoch": 4.11,
"learning_rate": 1.165263157894737e-06,
"loss": 0.0315,
"step": 8900
},
{
"epoch": 4.12,
"learning_rate": 1.1389473684210526e-06,
"loss": 0.0411,
"step": 8925
},
{
"epoch": 4.12,
"learning_rate": 1.1126315789473686e-06,
"loss": 0.0435,
"step": 8950
},
{
"epoch": 4.12,
"learning_rate": 1.0863157894736844e-06,
"loss": 0.0415,
"step": 8975
},
{
"epoch": 4.12,
"learning_rate": 1.06e-06,
"loss": 0.0523,
"step": 9000
},
{
"epoch": 4.12,
"eval_loss": 0.1768079549074173,
"eval_runtime": 16390.3907,
"eval_samples_per_second": 0.947,
"eval_steps_per_second": 0.237,
"eval_wer": 5.394080349254197,
"step": 9000
},
{
"epoch": 4.13,
"learning_rate": 1.0336842105263158e-06,
"loss": 0.0368,
"step": 9025
},
{
"epoch": 4.13,
"learning_rate": 1.0073684210526317e-06,
"loss": 0.0424,
"step": 9050
},
{
"epoch": 4.13,
"learning_rate": 9.810526315789475e-07,
"loss": 0.0365,
"step": 9075
},
{
"epoch": 4.13,
"learning_rate": 9.547368421052633e-07,
"loss": 0.03,
"step": 9100
},
{
"epoch": 4.14,
"learning_rate": 9.28421052631579e-07,
"loss": 0.035,
"step": 9125
},
{
"epoch": 4.14,
"learning_rate": 9.021052631578948e-07,
"loss": 0.0342,
"step": 9150
},
{
"epoch": 4.14,
"learning_rate": 8.757894736842105e-07,
"loss": 0.0245,
"step": 9175
},
{
"epoch": 4.14,
"learning_rate": 8.494736842105264e-07,
"loss": 0.0266,
"step": 9200
},
{
"epoch": 4.15,
"learning_rate": 8.231578947368423e-07,
"loss": 0.0243,
"step": 9225
},
{
"epoch": 4.15,
"learning_rate": 7.96842105263158e-07,
"loss": 0.0361,
"step": 9250
},
{
"epoch": 4.15,
"learning_rate": 7.705263157894738e-07,
"loss": 0.0282,
"step": 9275
},
{
"epoch": 4.15,
"learning_rate": 7.442105263157895e-07,
"loss": 0.0436,
"step": 9300
},
{
"epoch": 4.16,
"learning_rate": 7.178947368421053e-07,
"loss": 0.0309,
"step": 9325
},
{
"epoch": 4.16,
"learning_rate": 6.91578947368421e-07,
"loss": 0.0366,
"step": 9350
},
{
"epoch": 4.16,
"learning_rate": 6.652631578947369e-07,
"loss": 0.0405,
"step": 9375
},
{
"epoch": 4.16,
"learning_rate": 6.389473684210527e-07,
"loss": 0.0289,
"step": 9400
},
{
"epoch": 4.17,
"learning_rate": 6.126315789473684e-07,
"loss": 0.0333,
"step": 9425
},
{
"epoch": 4.17,
"learning_rate": 5.863157894736843e-07,
"loss": 0.0366,
"step": 9450
},
{
"epoch": 4.17,
"learning_rate": 5.6e-07,
"loss": 0.0292,
"step": 9475
},
{
"epoch": 4.17,
"learning_rate": 5.336842105263159e-07,
"loss": 0.0401,
"step": 9500
},
{
"epoch": 4.18,
"learning_rate": 5.073684210526316e-07,
"loss": 0.0409,
"step": 9525
},
{
"epoch": 4.18,
"learning_rate": 4.810526315789474e-07,
"loss": 0.0445,
"step": 9550
},
{
"epoch": 4.18,
"learning_rate": 4.547368421052632e-07,
"loss": 0.0375,
"step": 9575
},
{
"epoch": 4.18,
"learning_rate": 4.28421052631579e-07,
"loss": 0.0361,
"step": 9600
},
{
"epoch": 4.19,
"learning_rate": 4.021052631578948e-07,
"loss": 0.0372,
"step": 9625
},
{
"epoch": 4.19,
"learning_rate": 3.7578947368421056e-07,
"loss": 0.0422,
"step": 9650
},
{
"epoch": 4.19,
"learning_rate": 3.494736842105263e-07,
"loss": 0.0359,
"step": 9675
},
{
"epoch": 4.19,
"learning_rate": 3.231578947368421e-07,
"loss": 0.0421,
"step": 9700
},
{
"epoch": 5.0,
"learning_rate": 2.968421052631579e-07,
"loss": 0.138,
"step": 9725
},
{
"epoch": 5.0,
"learning_rate": 2.705263157894737e-07,
"loss": 0.0981,
"step": 9750
},
{
"epoch": 5.01,
"learning_rate": 2.442105263157895e-07,
"loss": 0.0728,
"step": 9775
},
{
"epoch": 5.01,
"learning_rate": 2.1789473684210528e-07,
"loss": 0.0806,
"step": 9800
},
{
"epoch": 5.01,
"learning_rate": 1.9157894736842107e-07,
"loss": 0.0617,
"step": 9825
},
{
"epoch": 5.01,
"learning_rate": 1.6526315789473687e-07,
"loss": 0.085,
"step": 9850
},
{
"epoch": 5.02,
"learning_rate": 1.3894736842105266e-07,
"loss": 0.0835,
"step": 9875
},
{
"epoch": 5.02,
"learning_rate": 1.1263157894736842e-07,
"loss": 0.0805,
"step": 9900
},
{
"epoch": 5.02,
"learning_rate": 8.631578947368422e-08,
"loss": 0.0931,
"step": 9925
},
{
"epoch": 5.03,
"learning_rate": 6.000000000000001e-08,
"loss": 0.0704,
"step": 9950
},
{
"epoch": 5.03,
"learning_rate": 3.3684210526315793e-08,
"loss": 0.0719,
"step": 9975
},
{
"epoch": 5.03,
"learning_rate": 7.3684210526315796e-09,
"loss": 0.0984,
"step": 10000
},
{
"epoch": 5.03,
"eval_loss": 0.17024469375610352,
"eval_runtime": 16316.0705,
"eval_samples_per_second": 0.951,
"eval_steps_per_second": 0.238,
"eval_wer": 5.288186684683748,
"step": 10000
},
{
"epoch": 5.03,
"step": 10000,
"total_flos": 3.39707363328e+20,
"train_loss": 0.09585554905235767,
"train_runtime": 213640.7293,
"train_samples_per_second": 0.749,
"train_steps_per_second": 0.047
}
],
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"total_flos": 3.39707363328e+20,
"trial_name": null,
"trial_params": null
}