| { | |
| "best_metric": 6.331942299477541, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 35.2112676056338, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4e-08, | |
| "loss": 0.9427, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.4e-08, | |
| "loss": 0.9466, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.44e-07, | |
| "loss": 0.9071, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.94e-07, | |
| "loss": 0.8369, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.4399999999999996e-07, | |
| "loss": 0.6547, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.9399999999999996e-07, | |
| "loss": 0.3677, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.4399999999999996e-07, | |
| "loss": 0.27, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.94e-07, | |
| "loss": 0.2291, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.44e-07, | |
| "loss": 0.2124, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.94e-07, | |
| "loss": 0.2101, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.44e-07, | |
| "loss": 0.1942, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.939999999999999e-07, | |
| "loss": 0.1848, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.44e-07, | |
| "loss": 0.1725, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.939999999999999e-07, | |
| "loss": 0.1731, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.44e-07, | |
| "loss": 0.1801, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.94e-07, | |
| "loss": 0.1664, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.439999999999999e-07, | |
| "loss": 0.1754, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.939999999999999e-07, | |
| "loss": 0.1595, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 9.439999999999999e-07, | |
| "loss": 0.157, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.94e-07, | |
| "loss": 0.1671, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.976842105263159e-07, | |
| "loss": 0.1665, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.950526315789475e-07, | |
| "loss": 0.1534, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.924210526315788e-07, | |
| "loss": 0.1694, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.897894736842104e-07, | |
| "loss": 0.1327, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9.87157894736842e-07, | |
| "loss": 0.1252, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.845263157894736e-07, | |
| "loss": 0.128, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.818947368421052e-07, | |
| "loss": 0.1287, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.792631578947367e-07, | |
| "loss": 0.12, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 9.766315789473683e-07, | |
| "loss": 0.1268, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 9.74e-07, | |
| "loss": 0.1218, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 9.713684210526315e-07, | |
| "loss": 0.1294, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 9.68736842105263e-07, | |
| "loss": 0.1373, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 9.661052631578947e-07, | |
| "loss": 0.1259, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 9.634736842105262e-07, | |
| "loss": 0.1338, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.608421052631578e-07, | |
| "loss": 0.1096, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 9.582105263157894e-07, | |
| "loss": 0.1041, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 9.55578947368421e-07, | |
| "loss": 0.1001, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 9.529473684210526e-07, | |
| "loss": 0.0935, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 9.503157894736841e-07, | |
| "loss": 0.0986, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 9.476842105263157e-07, | |
| "loss": 0.0999, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "eval_loss": 0.1753089725971222, | |
| "eval_runtime": 1240.803, | |
| "eval_samples_per_second": 7.311, | |
| "eval_steps_per_second": 0.229, | |
| "eval_wer": 6.331942299477541, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 9.450526315789473e-07, | |
| "loss": 0.1026, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 9.424210526315789e-07, | |
| "loss": 0.1011, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 9.397894736842105e-07, | |
| "loss": 0.0995, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 9.371578947368421e-07, | |
| "loss": 0.0967, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 9.345263157894736e-07, | |
| "loss": 0.1025, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 9.318947368421052e-07, | |
| "loss": 0.0857, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 9.292631578947368e-07, | |
| "loss": 0.0773, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 9.266315789473684e-07, | |
| "loss": 0.0814, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 9.24e-07, | |
| "loss": 0.0797, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 9.213684210526315e-07, | |
| "loss": 0.0736, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 9.187368421052631e-07, | |
| "loss": 0.0794, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 9.161052631578947e-07, | |
| "loss": 0.0784, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 9.134736842105263e-07, | |
| "loss": 0.0785, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 9.108421052631579e-07, | |
| "loss": 0.079, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 9.082105263157895e-07, | |
| "loss": 0.0783, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 9.05578947368421e-07, | |
| "loss": 0.0826, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.029473684210526e-07, | |
| "loss": 0.0773, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 9.003157894736841e-07, | |
| "loss": 0.0607, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 8.976842105263157e-07, | |
| "loss": 0.0628, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 8.950526315789473e-07, | |
| "loss": 0.0607, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 8.924210526315788e-07, | |
| "loss": 0.0655, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 8.897894736842105e-07, | |
| "loss": 0.0619, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 8.871578947368421e-07, | |
| "loss": 0.0601, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 8.845263157894737e-07, | |
| "loss": 0.0605, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 8.818947368421053e-07, | |
| "loss": 0.0641, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 8.792631578947369e-07, | |
| "loss": 0.065, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 8.766315789473684e-07, | |
| "loss": 0.0614, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 8.739999999999999e-07, | |
| "loss": 0.0578, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 8.713684210526315e-07, | |
| "loss": 0.0461, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 8.687368421052631e-07, | |
| "loss": 0.0451, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 8.661052631578947e-07, | |
| "loss": 0.0506, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 8.634736842105262e-07, | |
| "loss": 0.0467, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 8.608421052631578e-07, | |
| "loss": 0.0442, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 8.582105263157894e-07, | |
| "loss": 0.0496, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 8.555789473684211e-07, | |
| "loss": 0.0486, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 8.529473684210527e-07, | |
| "loss": 0.0466, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 8.503157894736843e-07, | |
| "loss": 0.0497, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 8.476842105263157e-07, | |
| "loss": 0.0475, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 8.450526315789473e-07, | |
| "loss": 0.0465, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 8.424210526315789e-07, | |
| "loss": 0.0436, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "eval_loss": 0.2026793211698532, | |
| "eval_runtime": 1172.7718, | |
| "eval_samples_per_second": 7.736, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 6.5520980514573, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 8.397894736842105e-07, | |
| "loss": 0.0322, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 8.371578947368421e-07, | |
| "loss": 0.0342, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 8.345263157894736e-07, | |
| "loss": 0.0334, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 8.318947368421052e-07, | |
| "loss": 0.0323, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 8.292631578947368e-07, | |
| "loss": 0.0319, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 8.266315789473684e-07, | |
| "loss": 0.0351, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 8.24e-07, | |
| "loss": 0.0365, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 8.213684210526314e-07, | |
| "loss": 0.0332, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 8.187368421052631e-07, | |
| "loss": 0.0347, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 8.161052631578947e-07, | |
| "loss": 0.0347, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.134736842105263e-07, | |
| "loss": 0.0315, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 8.108421052631579e-07, | |
| "loss": 0.0207, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 8.082105263157895e-07, | |
| "loss": 0.0246, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 8.05578947368421e-07, | |
| "loss": 0.023, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 8.029473684210526e-07, | |
| "loss": 0.0242, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 8.003157894736842e-07, | |
| "loss": 0.0233, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 7.976842105263158e-07, | |
| "loss": 0.0228, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 7.950526315789473e-07, | |
| "loss": 0.0234, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 7.924210526315788e-07, | |
| "loss": 0.0236, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 7.897894736842104e-07, | |
| "loss": 0.022, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 7.87157894736842e-07, | |
| "loss": 0.0244, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 7.845263157894737e-07, | |
| "loss": 0.0228, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 7.818947368421053e-07, | |
| "loss": 0.0192, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 7.792631578947369e-07, | |
| "loss": 0.0166, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 7.766315789473684e-07, | |
| "loss": 0.0145, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 7.74e-07, | |
| "loss": 0.0165, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 7.713684210526316e-07, | |
| "loss": 0.0149, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 7.687368421052631e-07, | |
| "loss": 0.0162, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 7.661052631578947e-07, | |
| "loss": 0.0155, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 7.634736842105262e-07, | |
| "loss": 0.0174, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 7.608421052631578e-07, | |
| "loss": 0.0173, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 7.582105263157894e-07, | |
| "loss": 0.0157, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 7.55578947368421e-07, | |
| "loss": 0.0164, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 7.529473684210526e-07, | |
| "loss": 0.0134, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 7.503157894736842e-07, | |
| "loss": 0.0117, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 7.476842105263158e-07, | |
| "loss": 0.011, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 7.450526315789474e-07, | |
| "loss": 0.0109, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 7.424210526315789e-07, | |
| "loss": 0.0114, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 7.397894736842105e-07, | |
| "loss": 0.0116, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 7.371578947368421e-07, | |
| "loss": 0.0113, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "eval_loss": 0.3134610652923584, | |
| "eval_runtime": 1171.9847, | |
| "eval_samples_per_second": 7.741, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 6.736108829231427, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 7.345263157894736e-07, | |
| "loss": 0.0103, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 7.318947368421052e-07, | |
| "loss": 0.0122, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 10.83, | |
| "learning_rate": 7.292631578947368e-07, | |
| "loss": 0.0115, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 7.266315789473684e-07, | |
| "loss": 0.0129, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 7.24e-07, | |
| "loss": 0.012, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "learning_rate": 7.213684210526316e-07, | |
| "loss": 0.008, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 11.18, | |
| "learning_rate": 7.187368421052631e-07, | |
| "loss": 0.008, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "learning_rate": 7.161052631578946e-07, | |
| "loss": 0.0081, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "learning_rate": 7.134736842105263e-07, | |
| "loss": 0.008, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 7.108421052631579e-07, | |
| "loss": 0.0095, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 11.53, | |
| "learning_rate": 7.082105263157895e-07, | |
| "loss": 0.0088, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 7.05578947368421e-07, | |
| "loss": 0.0078, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 11.71, | |
| "learning_rate": 7.029473684210526e-07, | |
| "loss": 0.0085, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 11.8, | |
| "learning_rate": 7.003157894736842e-07, | |
| "loss": 0.0072, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 6.976842105263158e-07, | |
| "loss": 0.0081, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 6.950526315789474e-07, | |
| "loss": 0.0091, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 6.92421052631579e-07, | |
| "loss": 0.0074, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 6.897894736842104e-07, | |
| "loss": 0.006, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 12.24, | |
| "learning_rate": 6.87157894736842e-07, | |
| "loss": 0.0066, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 12.32, | |
| "learning_rate": 6.845263157894736e-07, | |
| "loss": 0.0066, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 6.818947368421052e-07, | |
| "loss": 0.006, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 6.792631578947368e-07, | |
| "loss": 0.0063, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 6.766315789473684e-07, | |
| "loss": 0.0064, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 12.68, | |
| "learning_rate": 6.74e-07, | |
| "loss": 0.0061, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 12.76, | |
| "learning_rate": 6.713684210526316e-07, | |
| "loss": 0.0059, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 6.687368421052632e-07, | |
| "loss": 0.006, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 6.661052631578948e-07, | |
| "loss": 0.0065, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 6.634736842105262e-07, | |
| "loss": 0.0055, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 6.608421052631578e-07, | |
| "loss": 0.0048, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 6.582105263157894e-07, | |
| "loss": 0.0049, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "learning_rate": 6.55578947368421e-07, | |
| "loss": 0.0052, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 13.38, | |
| "learning_rate": 6.529473684210526e-07, | |
| "loss": 0.0054, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 13.47, | |
| "learning_rate": 6.503157894736842e-07, | |
| "loss": 0.0045, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 6.476842105263157e-07, | |
| "loss": 0.0047, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 13.64, | |
| "learning_rate": 6.450526315789473e-07, | |
| "loss": 0.0049, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 13.73, | |
| "learning_rate": 6.424210526315789e-07, | |
| "loss": 0.0048, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 6.397894736842106e-07, | |
| "loss": 0.0045, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 6.371578947368421e-07, | |
| "loss": 0.005, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 6.345263157894736e-07, | |
| "loss": 0.0048, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 14.08, | |
| "learning_rate": 6.318947368421052e-07, | |
| "loss": 0.0041, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 14.08, | |
| "eval_loss": 0.3616132140159607, | |
| "eval_runtime": 1172.6332, | |
| "eval_samples_per_second": 7.736, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 6.888903492918871, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 14.17, | |
| "learning_rate": 6.292631578947368e-07, | |
| "loss": 0.0038, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 6.266315789473684e-07, | |
| "loss": 0.004, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 6.24e-07, | |
| "loss": 0.0037, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 6.213684210526316e-07, | |
| "loss": 0.0042, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "learning_rate": 6.187368421052631e-07, | |
| "loss": 0.0039, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "learning_rate": 6.161052631578947e-07, | |
| "loss": 0.0042, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 6.134736842105263e-07, | |
| "loss": 0.0041, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 6.108421052631578e-07, | |
| "loss": 0.004, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 6.082105263157894e-07, | |
| "loss": 0.0038, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 14.96, | |
| "learning_rate": 6.05578947368421e-07, | |
| "loss": 0.0037, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 6.029473684210526e-07, | |
| "loss": 0.0036, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 15.14, | |
| "learning_rate": 6.003157894736842e-07, | |
| "loss": 0.0032, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 5.976842105263158e-07, | |
| "loss": 0.0032, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 5.950526315789474e-07, | |
| "loss": 0.0035, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 15.4, | |
| "learning_rate": 5.92421052631579e-07, | |
| "loss": 0.0032, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 15.49, | |
| "learning_rate": 5.897894736842105e-07, | |
| "loss": 0.0035, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "learning_rate": 5.871578947368421e-07, | |
| "loss": 0.0034, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 5.845263157894736e-07, | |
| "loss": 0.0033, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 15.76, | |
| "learning_rate": 5.818947368421052e-07, | |
| "loss": 0.0032, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 5.792631578947368e-07, | |
| "loss": 0.0032, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "learning_rate": 5.766315789473683e-07, | |
| "loss": 0.0032, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.739999999999999e-07, | |
| "loss": 0.0032, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 5.713684210526315e-07, | |
| "loss": 0.003, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "learning_rate": 5.687368421052632e-07, | |
| "loss": 0.0028, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 5.661052631578948e-07, | |
| "loss": 0.0028, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 16.37, | |
| "learning_rate": 5.634736842105264e-07, | |
| "loss": 0.0028, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "learning_rate": 5.608421052631579e-07, | |
| "loss": 0.0029, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 16.55, | |
| "learning_rate": 5.582105263157894e-07, | |
| "loss": 0.0027, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 16.64, | |
| "learning_rate": 5.55578947368421e-07, | |
| "loss": 0.0028, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 5.529473684210526e-07, | |
| "loss": 0.0028, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 5.503157894736842e-07, | |
| "loss": 0.0029, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 16.9, | |
| "learning_rate": 5.476842105263157e-07, | |
| "loss": 0.0029, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 16.99, | |
| "learning_rate": 5.450526315789473e-07, | |
| "loss": 0.0029, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 17.08, | |
| "learning_rate": 5.424210526315789e-07, | |
| "loss": 0.0026, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 17.17, | |
| "learning_rate": 5.397894736842105e-07, | |
| "loss": 0.0025, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 17.25, | |
| "learning_rate": 5.371578947368421e-07, | |
| "loss": 0.0025, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 5.345263157894738e-07, | |
| "loss": 0.0026, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 17.43, | |
| "learning_rate": 5.318947368421052e-07, | |
| "loss": 0.0024, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 17.52, | |
| "learning_rate": 5.292631578947368e-07, | |
| "loss": 0.0025, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 5.266315789473684e-07, | |
| "loss": 0.0026, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "eval_loss": 0.3908345401287079, | |
| "eval_runtime": 1175.3684, | |
| "eval_samples_per_second": 7.718, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 7.0564847369631645, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 5.24e-07, | |
| "loss": 0.0025, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 17.78, | |
| "learning_rate": 5.213684210526316e-07, | |
| "loss": 0.0024, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 5.187368421052631e-07, | |
| "loss": 0.0025, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 17.96, | |
| "learning_rate": 5.161052631578947e-07, | |
| "loss": 0.0025, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 5.134736842105263e-07, | |
| "loss": 0.0024, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 18.13, | |
| "learning_rate": 5.108421052631579e-07, | |
| "loss": 0.0022, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 5.082105263157895e-07, | |
| "loss": 0.0022, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 18.31, | |
| "learning_rate": 5.05578947368421e-07, | |
| "loss": 0.0022, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 5.029473684210525e-07, | |
| "loss": 0.0023, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 18.49, | |
| "learning_rate": 5.003157894736841e-07, | |
| "loss": 0.0023, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 18.57, | |
| "learning_rate": 4.976842105263158e-07, | |
| "loss": 0.0022, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 18.66, | |
| "learning_rate": 4.950526315789474e-07, | |
| "loss": 0.0023, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 4.924210526315789e-07, | |
| "loss": 0.0021, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 18.84, | |
| "learning_rate": 4.897894736842105e-07, | |
| "loss": 0.0023, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 4.871578947368421e-07, | |
| "loss": 0.0021, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.845263157894737e-07, | |
| "loss": 0.0021, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 19.1, | |
| "learning_rate": 4.818947368421053e-07, | |
| "loss": 0.0021, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 4.792631578947368e-07, | |
| "loss": 0.0021, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 19.28, | |
| "learning_rate": 4.766315789473684e-07, | |
| "loss": 0.002, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 19.37, | |
| "learning_rate": 4.7399999999999993e-07, | |
| "loss": 0.002, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 19.45, | |
| "learning_rate": 4.7136842105263156e-07, | |
| "loss": 0.0019, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 19.54, | |
| "learning_rate": 4.6873684210526315e-07, | |
| "loss": 0.002, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 19.63, | |
| "learning_rate": 4.6610526315789473e-07, | |
| "loss": 0.002, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 4.634736842105263e-07, | |
| "loss": 0.0019, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 4.6084210526315784e-07, | |
| "loss": 0.002, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 4.582105263157894e-07, | |
| "loss": 0.002, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 19.98, | |
| "learning_rate": 4.55578947368421e-07, | |
| "loss": 0.0019, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 20.07, | |
| "learning_rate": 4.5294736842105264e-07, | |
| "loss": 0.0018, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 20.16, | |
| "learning_rate": 4.503157894736842e-07, | |
| "loss": 0.0018, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 20.25, | |
| "learning_rate": 4.476842105263158e-07, | |
| "loss": 0.0018, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 20.33, | |
| "learning_rate": 4.4505263157894733e-07, | |
| "loss": 0.0018, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 20.42, | |
| "learning_rate": 4.424210526315789e-07, | |
| "loss": 0.0019, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 20.51, | |
| "learning_rate": 4.397894736842105e-07, | |
| "loss": 0.0018, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 20.6, | |
| "learning_rate": 4.371578947368421e-07, | |
| "loss": 0.0018, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 20.69, | |
| "learning_rate": 4.345263157894737e-07, | |
| "loss": 0.0018, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "learning_rate": 4.3189473684210524e-07, | |
| "loss": 0.0018, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 20.86, | |
| "learning_rate": 4.292631578947368e-07, | |
| "loss": 0.0018, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 20.95, | |
| "learning_rate": 4.266315789473684e-07, | |
| "loss": 0.0017, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 21.04, | |
| "learning_rate": 4.24e-07, | |
| "loss": 0.0017, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 21.13, | |
| "learning_rate": 4.213684210526315e-07, | |
| "loss": 0.0016, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 21.13, | |
| "eval_loss": 0.40775156021118164, | |
| "eval_runtime": 1175.538, | |
| "eval_samples_per_second": 7.717, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 7.141918312358294, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 21.21, | |
| "learning_rate": 4.1873684210526315e-07, | |
| "loss": 0.0017, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 21.3, | |
| "learning_rate": 4.1610526315789473e-07, | |
| "loss": 0.0016, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 21.39, | |
| "learning_rate": 4.134736842105263e-07, | |
| "loss": 0.0016, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 21.48, | |
| "learning_rate": 4.108421052631579e-07, | |
| "loss": 0.0016, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 21.57, | |
| "learning_rate": 4.082105263157894e-07, | |
| "loss": 0.0016, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 21.65, | |
| "learning_rate": 4.05578947368421e-07, | |
| "loss": 0.0016, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 21.74, | |
| "learning_rate": 4.029473684210526e-07, | |
| "loss": 0.0017, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 21.83, | |
| "learning_rate": 4.003157894736842e-07, | |
| "loss": 0.0016, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 21.92, | |
| "learning_rate": 3.976842105263158e-07, | |
| "loss": 0.0016, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.950526315789474e-07, | |
| "loss": 0.0017, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 22.1, | |
| "learning_rate": 3.924210526315789e-07, | |
| "loss": 0.0015, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 22.18, | |
| "learning_rate": 3.897894736842105e-07, | |
| "loss": 0.0015, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 22.27, | |
| "learning_rate": 3.871578947368421e-07, | |
| "loss": 0.0015, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 22.36, | |
| "learning_rate": 3.8452631578947366e-07, | |
| "loss": 0.0014, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 22.45, | |
| "learning_rate": 3.818947368421053e-07, | |
| "loss": 0.0015, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 22.54, | |
| "learning_rate": 3.792631578947368e-07, | |
| "loss": 0.0015, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 22.62, | |
| "learning_rate": 3.766315789473684e-07, | |
| "loss": 0.0015, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 22.71, | |
| "learning_rate": 3.74e-07, | |
| "loss": 0.0015, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 22.8, | |
| "learning_rate": 3.7136842105263157e-07, | |
| "loss": 0.0015, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 22.89, | |
| "learning_rate": 3.687368421052631e-07, | |
| "loss": 0.0015, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 22.98, | |
| "learning_rate": 3.661052631578947e-07, | |
| "loss": 0.0014, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 23.06, | |
| "learning_rate": 3.634736842105263e-07, | |
| "loss": 0.0014, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 23.15, | |
| "learning_rate": 3.608421052631579e-07, | |
| "loss": 0.0014, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 23.24, | |
| "learning_rate": 3.582105263157895e-07, | |
| "loss": 0.0014, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 23.33, | |
| "learning_rate": 3.5557894736842106e-07, | |
| "loss": 0.0014, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 23.42, | |
| "learning_rate": 3.529473684210526e-07, | |
| "loss": 0.0014, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 23.5, | |
| "learning_rate": 3.5031578947368417e-07, | |
| "loss": 0.0014, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 23.59, | |
| "learning_rate": 3.4768421052631575e-07, | |
| "loss": 0.0013, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 23.68, | |
| "learning_rate": 3.450526315789474e-07, | |
| "loss": 0.0014, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 23.77, | |
| "learning_rate": 3.4242105263157897e-07, | |
| "loss": 0.0013, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 23.86, | |
| "learning_rate": 3.397894736842105e-07, | |
| "loss": 0.0014, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 23.94, | |
| "learning_rate": 3.371578947368421e-07, | |
| "loss": 0.0014, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 3.3452631578947366e-07, | |
| "loss": 0.0013, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 24.12, | |
| "learning_rate": 3.3189473684210524e-07, | |
| "loss": 0.0013, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 24.21, | |
| "learning_rate": 3.292631578947369e-07, | |
| "loss": 0.0012, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 24.3, | |
| "learning_rate": 3.266315789473684e-07, | |
| "loss": 0.0013, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 24.38, | |
| "learning_rate": 3.24e-07, | |
| "loss": 0.0013, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 24.47, | |
| "learning_rate": 3.2136842105263157e-07, | |
| "loss": 0.0013, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 24.56, | |
| "learning_rate": 3.1873684210526315e-07, | |
| "loss": 0.0013, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 24.65, | |
| "learning_rate": 3.161052631578947e-07, | |
| "loss": 0.0013, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 24.65, | |
| "eval_loss": 0.42271819710731506, | |
| "eval_runtime": 1176.399, | |
| "eval_samples_per_second": 7.712, | |
| "eval_steps_per_second": 0.241, | |
| "eval_wer": 7.153418985969179, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 24.74, | |
| "learning_rate": 3.1347368421052626e-07, | |
| "loss": 0.0012, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 24.82, | |
| "learning_rate": 3.108421052631579e-07, | |
| "loss": 0.0013, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 24.91, | |
| "learning_rate": 3.082105263157895e-07, | |
| "loss": 0.0012, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 3.0557894736842106e-07, | |
| "loss": 0.0013, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 25.09, | |
| "learning_rate": 3.0294736842105264e-07, | |
| "loss": 0.0012, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 25.18, | |
| "learning_rate": 3.0031578947368417e-07, | |
| "loss": 0.0012, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 25.26, | |
| "learning_rate": 2.9768421052631576e-07, | |
| "loss": 0.0012, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 25.35, | |
| "learning_rate": 2.9505263157894734e-07, | |
| "loss": 0.0012, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 25.44, | |
| "learning_rate": 2.9242105263157897e-07, | |
| "loss": 0.0012, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 25.53, | |
| "learning_rate": 2.8978947368421055e-07, | |
| "loss": 0.0012, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 25.62, | |
| "learning_rate": 2.871578947368421e-07, | |
| "loss": 0.0012, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 25.7, | |
| "learning_rate": 2.8452631578947366e-07, | |
| "loss": 0.0011, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 25.79, | |
| "learning_rate": 2.8189473684210525e-07, | |
| "loss": 0.0012, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 25.88, | |
| "learning_rate": 2.7926315789473683e-07, | |
| "loss": 0.0012, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 25.97, | |
| "learning_rate": 2.7663157894736836e-07, | |
| "loss": 0.0012, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 26.06, | |
| "learning_rate": 2.74e-07, | |
| "loss": 0.0011, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 26.14, | |
| "learning_rate": 2.713684210526316e-07, | |
| "loss": 0.0011, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 26.23, | |
| "learning_rate": 2.6873684210526316e-07, | |
| "loss": 0.0011, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 26.32, | |
| "learning_rate": 2.6610526315789474e-07, | |
| "loss": 0.0011, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 26.41, | |
| "learning_rate": 2.6347368421052627e-07, | |
| "loss": 0.0011, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 26.5, | |
| "learning_rate": 2.6084210526315785e-07, | |
| "loss": 0.0011, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 26.58, | |
| "learning_rate": 2.582105263157895e-07, | |
| "loss": 0.0011, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 26.67, | |
| "learning_rate": 2.5557894736842106e-07, | |
| "loss": 0.0011, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 26.76, | |
| "learning_rate": 2.5294736842105265e-07, | |
| "loss": 0.0011, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 26.85, | |
| "learning_rate": 2.5031578947368423e-07, | |
| "loss": 0.0011, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 26.94, | |
| "learning_rate": 2.4768421052631576e-07, | |
| "loss": 0.0011, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 2.4505263157894734e-07, | |
| "loss": 0.0011, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 27.11, | |
| "learning_rate": 2.424210526315789e-07, | |
| "loss": 0.001, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 27.2, | |
| "learning_rate": 2.397894736842105e-07, | |
| "loss": 0.001, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 27.29, | |
| "learning_rate": 2.3715789473684208e-07, | |
| "loss": 0.001, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 27.38, | |
| "learning_rate": 2.345263157894737e-07, | |
| "loss": 0.001, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 27.46, | |
| "learning_rate": 2.3189473684210525e-07, | |
| "loss": 0.0011, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 27.55, | |
| "learning_rate": 2.2926315789473683e-07, | |
| "loss": 0.001, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 27.64, | |
| "learning_rate": 2.266315789473684e-07, | |
| "loss": 0.001, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 27.73, | |
| "learning_rate": 2.24e-07, | |
| "loss": 0.001, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 27.82, | |
| "learning_rate": 2.2136842105263158e-07, | |
| "loss": 0.0011, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 27.9, | |
| "learning_rate": 2.1873684210526313e-07, | |
| "loss": 0.001, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 27.99, | |
| "learning_rate": 2.1610526315789474e-07, | |
| "loss": 0.001, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 28.08, | |
| "learning_rate": 2.134736842105263e-07, | |
| "loss": 0.001, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 28.17, | |
| "learning_rate": 2.1084210526315788e-07, | |
| "loss": 0.001, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 28.17, | |
| "eval_loss": 0.43434110283851624, | |
| "eval_runtime": 1175.8728, | |
| "eval_samples_per_second": 7.715, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 7.176420333190944, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 28.26, | |
| "learning_rate": 2.0821052631578949e-07, | |
| "loss": 0.001, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 28.35, | |
| "learning_rate": 2.0557894736842104e-07, | |
| "loss": 0.001, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 28.43, | |
| "learning_rate": 2.0294736842105262e-07, | |
| "loss": 0.001, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 28.52, | |
| "learning_rate": 2.0031578947368418e-07, | |
| "loss": 0.001, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 28.61, | |
| "learning_rate": 1.9768421052631579e-07, | |
| "loss": 0.001, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 28.7, | |
| "learning_rate": 1.9505263157894737e-07, | |
| "loss": 0.001, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 28.79, | |
| "learning_rate": 1.9242105263157892e-07, | |
| "loss": 0.001, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 28.87, | |
| "learning_rate": 1.8978947368421053e-07, | |
| "loss": 0.001, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 28.96, | |
| "learning_rate": 1.871578947368421e-07, | |
| "loss": 0.001, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 29.05, | |
| "learning_rate": 1.8452631578947367e-07, | |
| "loss": 0.0009, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 29.14, | |
| "learning_rate": 1.8189473684210528e-07, | |
| "loss": 0.0009, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 29.23, | |
| "learning_rate": 1.7926315789473683e-07, | |
| "loss": 0.0009, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 29.31, | |
| "learning_rate": 1.7663157894736841e-07, | |
| "loss": 0.0009, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 29.4, | |
| "learning_rate": 1.7399999999999997e-07, | |
| "loss": 0.0009, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 29.49, | |
| "learning_rate": 1.7136842105263158e-07, | |
| "loss": 0.0009, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 29.58, | |
| "learning_rate": 1.6873684210526316e-07, | |
| "loss": 0.0009, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 29.67, | |
| "learning_rate": 1.6610526315789472e-07, | |
| "loss": 0.0009, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 29.75, | |
| "learning_rate": 1.6347368421052632e-07, | |
| "loss": 0.0009, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 29.84, | |
| "learning_rate": 1.6084210526315788e-07, | |
| "loss": 0.0009, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 29.93, | |
| "learning_rate": 1.5821052631578946e-07, | |
| "loss": 0.0009, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 1.5557894736842104e-07, | |
| "loss": 0.0009, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 30.11, | |
| "learning_rate": 1.5294736842105262e-07, | |
| "loss": 0.0009, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 30.19, | |
| "learning_rate": 1.503157894736842e-07, | |
| "loss": 0.0009, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 30.28, | |
| "learning_rate": 1.4768421052631576e-07, | |
| "loss": 0.0009, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 30.37, | |
| "learning_rate": 1.4505263157894737e-07, | |
| "loss": 0.0009, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 30.46, | |
| "learning_rate": 1.4242105263157895e-07, | |
| "loss": 0.0009, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 30.55, | |
| "learning_rate": 1.397894736842105e-07, | |
| "loss": 0.0009, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 30.63, | |
| "learning_rate": 1.3715789473684212e-07, | |
| "loss": 0.0009, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 30.72, | |
| "learning_rate": 1.3452631578947367e-07, | |
| "loss": 0.0009, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 30.81, | |
| "learning_rate": 1.3189473684210525e-07, | |
| "loss": 0.0009, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 30.9, | |
| "learning_rate": 1.2926315789473683e-07, | |
| "loss": 0.0009, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 30.99, | |
| "learning_rate": 1.2663157894736842e-07, | |
| "loss": 0.0009, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 31.07, | |
| "learning_rate": 1.24e-07, | |
| "loss": 0.0008, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 31.16, | |
| "learning_rate": 1.2136842105263158e-07, | |
| "loss": 0.0008, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 1.1873684210526315e-07, | |
| "loss": 0.0008, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 31.34, | |
| "learning_rate": 1.1610526315789473e-07, | |
| "loss": 0.0008, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 31.43, | |
| "learning_rate": 1.134736842105263e-07, | |
| "loss": 0.0009, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 31.51, | |
| "learning_rate": 1.108421052631579e-07, | |
| "loss": 0.0008, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 31.6, | |
| "learning_rate": 1.0821052631578948e-07, | |
| "loss": 0.0009, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 31.69, | |
| "learning_rate": 1.0557894736842105e-07, | |
| "loss": 0.0008, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 31.69, | |
| "eval_loss": 0.44237083196640015, | |
| "eval_runtime": 1174.8847, | |
| "eval_samples_per_second": 7.722, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 7.207636447277626, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 31.78, | |
| "learning_rate": 1.0294736842105263e-07, | |
| "loss": 0.0008, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 31.87, | |
| "learning_rate": 1.003157894736842e-07, | |
| "loss": 0.0009, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 31.95, | |
| "learning_rate": 9.768421052631579e-08, | |
| "loss": 0.0008, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 32.04, | |
| "learning_rate": 9.505263157894737e-08, | |
| "loss": 0.0009, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 32.13, | |
| "learning_rate": 9.242105263157894e-08, | |
| "loss": 0.0008, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 32.22, | |
| "learning_rate": 8.978947368421052e-08, | |
| "loss": 0.0008, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 32.31, | |
| "learning_rate": 8.715789473684209e-08, | |
| "loss": 0.0008, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 32.39, | |
| "learning_rate": 8.452631578947369e-08, | |
| "loss": 0.0008, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 32.48, | |
| "learning_rate": 8.189473684210526e-08, | |
| "loss": 0.0008, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 32.57, | |
| "learning_rate": 7.926315789473684e-08, | |
| "loss": 0.0008, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 32.66, | |
| "learning_rate": 7.663157894736842e-08, | |
| "loss": 0.0008, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 32.75, | |
| "learning_rate": 7.399999999999999e-08, | |
| "loss": 0.0008, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 32.83, | |
| "learning_rate": 7.136842105263158e-08, | |
| "loss": 0.0008, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 32.92, | |
| "learning_rate": 6.873684210526315e-08, | |
| "loss": 0.0008, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 33.01, | |
| "learning_rate": 6.610526315789473e-08, | |
| "loss": 0.0008, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 33.1, | |
| "learning_rate": 6.347368421052632e-08, | |
| "loss": 0.0008, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 33.19, | |
| "learning_rate": 6.084210526315788e-08, | |
| "loss": 0.0008, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 33.27, | |
| "learning_rate": 5.821052631578947e-08, | |
| "loss": 0.0008, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 33.36, | |
| "learning_rate": 5.5578947368421054e-08, | |
| "loss": 0.0008, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 33.45, | |
| "learning_rate": 5.294736842105263e-08, | |
| "loss": 0.0008, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 33.54, | |
| "learning_rate": 5.0315789473684204e-08, | |
| "loss": 0.0008, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 33.63, | |
| "learning_rate": 4.7684210526315786e-08, | |
| "loss": 0.0008, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 33.71, | |
| "learning_rate": 4.505263157894736e-08, | |
| "loss": 0.0008, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 33.8, | |
| "learning_rate": 4.242105263157895e-08, | |
| "loss": 0.0008, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 33.89, | |
| "learning_rate": 3.9789473684210525e-08, | |
| "loss": 0.0008, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 33.98, | |
| "learning_rate": 3.71578947368421e-08, | |
| "loss": 0.0008, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 34.07, | |
| "learning_rate": 3.452631578947368e-08, | |
| "loss": 0.0008, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 34.15, | |
| "learning_rate": 3.189473684210526e-08, | |
| "loss": 0.0008, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 34.24, | |
| "learning_rate": 2.926315789473684e-08, | |
| "loss": 0.0008, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 34.33, | |
| "learning_rate": 2.6631578947368418e-08, | |
| "loss": 0.0008, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 34.42, | |
| "learning_rate": 2.4e-08, | |
| "loss": 0.0008, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 34.51, | |
| "learning_rate": 2.136842105263158e-08, | |
| "loss": 0.0008, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 34.6, | |
| "learning_rate": 1.8736842105263157e-08, | |
| "loss": 0.0008, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 34.68, | |
| "learning_rate": 1.6105263157894735e-08, | |
| "loss": 0.0008, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 34.77, | |
| "learning_rate": 1.3473684210526316e-08, | |
| "loss": 0.0008, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 34.86, | |
| "learning_rate": 1.0842105263157894e-08, | |
| "loss": 0.0008, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 34.95, | |
| "learning_rate": 8.210526315789473e-09, | |
| "loss": 0.0008, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 35.04, | |
| "learning_rate": 5.578947368421052e-09, | |
| "loss": 0.0007, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 35.12, | |
| "learning_rate": 2.9473684210526315e-09, | |
| "loss": 0.0008, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 35.21, | |
| "learning_rate": 3.157894736842105e-10, | |
| "loss": 0.0008, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 35.21, | |
| "eval_loss": 0.4463559687137604, | |
| "eval_runtime": 1177.1302, | |
| "eval_samples_per_second": 7.707, | |
| "eval_steps_per_second": 0.241, | |
| "eval_wer": 7.222423027634475, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 35.21, | |
| "step": 10000, | |
| "total_flos": 6.520440633163776e+20, | |
| "train_loss": 0.03515132477898151, | |
| "train_runtime": 76775.2561, | |
| "train_samples_per_second": 8.336, | |
| "train_steps_per_second": 0.13 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 36, | |
| "save_steps": 1000, | |
| "total_flos": 6.520440633163776e+20, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |