| { | |
| "best_metric": 5.48054433340421, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 2.046, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.973833272194737e-06, | |
| "loss": 0.5401, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.195318418690893e-06, | |
| "loss": 0.1681, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.881634451095711e-06, | |
| "loss": 0.1456, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.361221988663844e-06, | |
| "loss": 0.1779, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.730207550743121e-06, | |
| "loss": 0.1771, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.03016458599496e-06, | |
| "loss": 0.1744, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.282894746203441e-06, | |
| "loss": 0.2568, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.501266121799902e-06, | |
| "loss": 0.2241, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.693512601774437e-06, | |
| "loss": 0.1958, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.865222471593567e-06, | |
| "loss": 0.1475, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.020362953730323e-06, | |
| "loss": 0.1671, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.161852281961698e-06, | |
| "loss": 0.1486, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.29189975311636e-06, | |
| "loss": 0.1477, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.412218256259678e-06, | |
| "loss": 0.1357, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.524162683365145e-06, | |
| "loss": 0.1299, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.62882322733502e-06, | |
| "loss": 0.1261, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.727090137141168e-06, | |
| "loss": 0.128, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.819699807237934e-06, | |
| "loss": 0.1318, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.907268307310855e-06, | |
| "loss": 0.1528, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.990316248055788e-06, | |
| "loss": 0.1296, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.1347, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.1501, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.1415, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.1686, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.1445, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.1379, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.15, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.1481, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.1234, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.1332, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.1321, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.156, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.1204, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.1182, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.1191, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.1345, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.1332, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.1638, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.2081, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.1951, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.18505859375, | |
| "eval_runtime": 3130.1973, | |
| "eval_samples_per_second": 4.793, | |
| "eval_steps_per_second": 0.15, | |
| "eval_wer": 7.626780778226664, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.1445, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.133, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.1212, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.1564, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.1511, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.1618, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.1673, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.139, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.1688, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.1454, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.1539, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.1483, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.1384, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.1554, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.147, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.1412, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.1444, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.1527, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.1323, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.1649, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.1539, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.1463, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.1496, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.1437, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.1369, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.1412, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.1238, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.1439, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.1226, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.1386, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.1584, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.1749, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.1539, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.1423, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.1402, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.1539, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.1361, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.1136, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.1096, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.1311, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.156494140625, | |
| "eval_runtime": 3130.1703, | |
| "eval_samples_per_second": 4.793, | |
| "eval_steps_per_second": 0.15, | |
| "eval_wer": 6.2885392302785466, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.6244444444444445e-06, | |
| "loss": 0.1408, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.568888888888889e-06, | |
| "loss": 0.1441, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.513333333333333e-06, | |
| "loss": 0.1216, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.457777777777778e-06, | |
| "loss": 0.1377, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.402222222222223e-06, | |
| "loss": 0.1104, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.346666666666668e-06, | |
| "loss": 0.1133, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.291111111111111e-06, | |
| "loss": 0.1381, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.235555555555556e-06, | |
| "loss": 0.1237, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.18e-06, | |
| "loss": 0.0971, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.124444444444445e-06, | |
| "loss": 0.1464, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.06888888888889e-06, | |
| "loss": 0.1185, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 6.013333333333335e-06, | |
| "loss": 0.1667, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.957777777777778e-06, | |
| "loss": 0.1366, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.902222222222223e-06, | |
| "loss": 0.1271, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.846666666666667e-06, | |
| "loss": 0.1199, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.791111111111112e-06, | |
| "loss": 0.0914, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.735555555555557e-06, | |
| "loss": 0.0658, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.68e-06, | |
| "loss": 0.0611, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.624444444444445e-06, | |
| "loss": 0.0752, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.56888888888889e-06, | |
| "loss": 0.0673, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5.513333333333334e-06, | |
| "loss": 0.063, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5.4577777777777785e-06, | |
| "loss": 0.0669, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.402222222222223e-06, | |
| "loss": 0.0678, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.346666666666667e-06, | |
| "loss": 0.0482, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 5.2911111111111115e-06, | |
| "loss": 0.0531, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 5.235555555555556e-06, | |
| "loss": 0.0586, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.18e-06, | |
| "loss": 0.0609, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.124444444444445e-06, | |
| "loss": 0.0536, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 5.06888888888889e-06, | |
| "loss": 0.0484, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 5.013333333333333e-06, | |
| "loss": 0.0532, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.957777777777778e-06, | |
| "loss": 0.0573, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.902222222222222e-06, | |
| "loss": 0.0593, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.846666666666667e-06, | |
| "loss": 0.0672, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.791111111111111e-06, | |
| "loss": 0.0884, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.735555555555556e-06, | |
| "loss": 0.0847, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.680000000000001e-06, | |
| "loss": 0.0695, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.624444444444445e-06, | |
| "loss": 0.0622, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.568888888888889e-06, | |
| "loss": 0.0541, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.513333333333333e-06, | |
| "loss": 0.0715, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.457777777777778e-06, | |
| "loss": 0.0681, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "eval_loss": 0.155517578125, | |
| "eval_runtime": 3146.5974, | |
| "eval_samples_per_second": 4.768, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 6.039363172443121, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.406666666666667e-06, | |
| "loss": 0.0763, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.351111111111111e-06, | |
| "loss": 0.0846, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.295555555555556e-06, | |
| "loss": 0.0799, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.0726, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.184444444444445e-06, | |
| "loss": 0.0662, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.12888888888889e-06, | |
| "loss": 0.0679, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.073333333333334e-06, | |
| "loss": 0.0698, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.017777777777778e-06, | |
| "loss": 0.0605, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.962222222222222e-06, | |
| "loss": 0.0505, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.906666666666667e-06, | |
| "loss": 0.0677, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.851111111111112e-06, | |
| "loss": 0.074, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.7955555555555557e-06, | |
| "loss": 0.07, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.74e-06, | |
| "loss": 0.0635, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.684444444444445e-06, | |
| "loss": 0.0642, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.628888888888889e-06, | |
| "loss": 0.0509, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.5733333333333336e-06, | |
| "loss": 0.0574, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.5177777777777784e-06, | |
| "loss": 0.069, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.4622222222222225e-06, | |
| "loss": 0.0575, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.406666666666667e-06, | |
| "loss": 0.0486, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.351111111111112e-06, | |
| "loss": 0.0691, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.295555555555556e-06, | |
| "loss": 0.0612, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 0.08, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.1844444444444444e-06, | |
| "loss": 0.079, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.1288888888888892e-06, | |
| "loss": 0.0718, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.0733333333333337e-06, | |
| "loss": 0.0624, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.0177777777777777e-06, | |
| "loss": 0.069, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9622222222222226e-06, | |
| "loss": 0.0738, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.906666666666667e-06, | |
| "loss": 0.0616, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.851111111111111e-06, | |
| "loss": 0.0752, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.795555555555556e-06, | |
| "loss": 0.065, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.7400000000000004e-06, | |
| "loss": 0.0656, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.6844444444444445e-06, | |
| "loss": 0.0652, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.6288888888888894e-06, | |
| "loss": 0.0572, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.573333333333334e-06, | |
| "loss": 0.0677, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.517777777777778e-06, | |
| "loss": 0.0644, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.4622222222222223e-06, | |
| "loss": 0.0679, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.4066666666666668e-06, | |
| "loss": 0.0621, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.3511111111111112e-06, | |
| "loss": 0.0667, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.2955555555555557e-06, | |
| "loss": 0.0615, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.067, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_loss": 0.14697265625, | |
| "eval_runtime": 3238.9845, | |
| "eval_samples_per_second": 4.632, | |
| "eval_steps_per_second": 0.145, | |
| "eval_wer": 5.614767169891559, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.188888888888889e-06, | |
| "loss": 0.0698, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.133333333333334e-06, | |
| "loss": 0.0682, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.077777777777778e-06, | |
| "loss": 0.0634, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.0222222222222223e-06, | |
| "loss": 0.0615, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.9666666666666668e-06, | |
| "loss": 0.0622, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.9111111111111112e-06, | |
| "loss": 0.0615, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.8555555555555557e-06, | |
| "loss": 0.0572, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 0.0643, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.7444444444444448e-06, | |
| "loss": 0.0562, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.688888888888889e-06, | |
| "loss": 0.0595, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.6333333333333335e-06, | |
| "loss": 0.0897, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5777777777777778e-06, | |
| "loss": 0.0797, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5222222222222224e-06, | |
| "loss": 0.0627, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.4666666666666669e-06, | |
| "loss": 0.064, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.4111111111111111e-06, | |
| "loss": 0.0818, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.3555555555555558e-06, | |
| "loss": 0.0752, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.3e-06, | |
| "loss": 0.1079, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.2444444444444445e-06, | |
| "loss": 0.1028, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.188888888888889e-06, | |
| "loss": 0.1115, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.1333333333333334e-06, | |
| "loss": 0.0666, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.0777777777777779e-06, | |
| "loss": 0.0711, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.0222222222222223e-06, | |
| "loss": 0.0609, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.666666666666668e-07, | |
| "loss": 0.0629, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.111111111111113e-07, | |
| "loss": 0.0591, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.555555555555556e-07, | |
| "loss": 0.0581, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.0587, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.444444444444444e-07, | |
| "loss": 0.0551, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.88888888888889e-07, | |
| "loss": 0.0549, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.333333333333334e-07, | |
| "loss": 0.0618, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.777777777777778e-07, | |
| "loss": 0.0647, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.222222222222223e-07, | |
| "loss": 0.0521, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.666666666666667e-07, | |
| "loss": 0.0483, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.111111111111112e-07, | |
| "loss": 0.0467, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.555555555555556e-07, | |
| "loss": 0.0459, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 0.0364, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.444444444444445e-07, | |
| "loss": 0.0346, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.888888888888889e-07, | |
| "loss": 0.0403, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.3333333333333336e-07, | |
| "loss": 0.038, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.777777777777778e-08, | |
| "loss": 0.0308, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.2222222222222224e-08, | |
| "loss": 0.0336, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "eval_loss": 0.1456298828125, | |
| "eval_runtime": 3116.6143, | |
| "eval_samples_per_second": 4.814, | |
| "eval_steps_per_second": 0.15, | |
| "eval_wer": 5.48054433340421, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "step": 5000, | |
| "total_flos": 3.265302963514633e+20, | |
| "train_loss": 0.10500919036865235, | |
| "train_runtime": 64843.4319, | |
| "train_samples_per_second": 4.935, | |
| "train_steps_per_second": 0.077 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 3.265302963514633e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |