| { | |
| "best_metric": 13.708574434508153, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 31.446540880503143, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.6799999999999996e-07, | |
| "loss": 1.2045, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.68e-07, | |
| "loss": 0.997, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.1679999999999999e-06, | |
| "loss": 0.6901, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.568e-06, | |
| "loss": 0.4398, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.968e-06, | |
| "loss": 0.3942, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.3679999999999996e-06, | |
| "loss": 0.3263, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.7679999999999996e-06, | |
| "loss": 0.3056, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.168e-06, | |
| "loss": 0.2628, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.568e-06, | |
| "loss": 0.2526, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.968e-06, | |
| "loss": 0.257, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.368e-06, | |
| "loss": 0.2296, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.768e-06, | |
| "loss": 0.231, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.168e-06, | |
| "loss": 0.194, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.567999999999999e-06, | |
| "loss": 0.1423, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 5.9679999999999994e-06, | |
| "loss": 0.1473, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.368e-06, | |
| "loss": 0.1454, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.767999999999999e-06, | |
| "loss": 0.143, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.168e-06, | |
| "loss": 0.141, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.567999999999999e-06, | |
| "loss": 0.1471, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 7.967999999999999e-06, | |
| "loss": 0.1336, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.980631578947368e-06, | |
| "loss": 0.1264, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.95957894736842e-06, | |
| "loss": 0.1324, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.938526315789473e-06, | |
| "loss": 0.1179, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.917473684210526e-06, | |
| "loss": 0.1243, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 7.896421052631578e-06, | |
| "loss": 0.1209, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.875368421052631e-06, | |
| "loss": 0.0874, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 7.854315789473684e-06, | |
| "loss": 0.0612, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.833263157894736e-06, | |
| "loss": 0.0569, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 7.812210526315789e-06, | |
| "loss": 0.0566, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.791157894736842e-06, | |
| "loss": 0.0522, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.770105263157894e-06, | |
| "loss": 0.0558, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.749052631578947e-06, | |
| "loss": 0.0547, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.728e-06, | |
| "loss": 0.0568, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.706947368421052e-06, | |
| "loss": 0.0569, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.685894736842105e-06, | |
| "loss": 0.0573, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 7.664842105263157e-06, | |
| "loss": 0.0538, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 7.64378947368421e-06, | |
| "loss": 0.054, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.622736842105263e-06, | |
| "loss": 0.0553, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.6016842105263155e-06, | |
| "loss": 0.0316, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 7.580631578947368e-06, | |
| "loss": 0.0265, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "eval_loss": 0.3689558207988739, | |
| "eval_runtime": 1285.0181, | |
| "eval_samples_per_second": 1.66, | |
| "eval_steps_per_second": 0.208, | |
| "eval_wer": 14.760652288269332, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 7.559578947368421e-06, | |
| "loss": 0.0247, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 7.5385263157894734e-06, | |
| "loss": 0.0285, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.517473684210526e-06, | |
| "loss": 0.0266, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 7.496421052631579e-06, | |
| "loss": 0.0242, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 7.475368421052631e-06, | |
| "loss": 0.0286, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.454315789473684e-06, | |
| "loss": 0.0254, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 7.433263157894736e-06, | |
| "loss": 0.0303, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 7.4122105263157885e-06, | |
| "loss": 0.0281, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.391157894736841e-06, | |
| "loss": 0.0267, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 7.370105263157895e-06, | |
| "loss": 0.0272, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 7.349052631578947e-06, | |
| "loss": 0.0233, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 7.328e-06, | |
| "loss": 0.0133, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 7.306947368421053e-06, | |
| "loss": 0.0148, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 7.285894736842105e-06, | |
| "loss": 0.0153, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 7.264842105263158e-06, | |
| "loss": 0.016, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 7.243789473684211e-06, | |
| "loss": 0.0146, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 7.222736842105262e-06, | |
| "loss": 0.0147, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 7.201684210526315e-06, | |
| "loss": 0.0151, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 7.180631578947368e-06, | |
| "loss": 0.0135, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 7.15957894736842e-06, | |
| "loss": 0.0159, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 7.138526315789473e-06, | |
| "loss": 0.0166, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 7.117473684210526e-06, | |
| "loss": 0.0118, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 7.096421052631578e-06, | |
| "loss": 0.015, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 7.075368421052632e-06, | |
| "loss": 0.0121, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 7.0543157894736845e-06, | |
| "loss": 0.0098, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 7.033263157894737e-06, | |
| "loss": 0.0083, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 7.012210526315789e-06, | |
| "loss": 0.0092, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 6.991157894736842e-06, | |
| "loss": 0.0076, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 6.970105263157894e-06, | |
| "loss": 0.009, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 6.949052631578947e-06, | |
| "loss": 0.0098, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 6.9279999999999996e-06, | |
| "loss": 0.0069, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 6.906947368421052e-06, | |
| "loss": 0.0095, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 6.885894736842105e-06, | |
| "loss": 0.009, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 6.8648421052631575e-06, | |
| "loss": 0.0103, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 6.84378947368421e-06, | |
| "loss": 0.0106, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 6.822736842105263e-06, | |
| "loss": 0.0114, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 6.8016842105263155e-06, | |
| "loss": 0.0089, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 6.780631578947368e-06, | |
| "loss": 0.0055, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "learning_rate": 6.759578947368421e-06, | |
| "loss": 0.0051, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 6.7385263157894735e-06, | |
| "loss": 0.0063, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "eval_loss": 0.4341892600059509, | |
| "eval_runtime": 1246.3959, | |
| "eval_samples_per_second": 1.711, | |
| "eval_steps_per_second": 0.214, | |
| "eval_wer": 14.092582851130983, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 6.717473684210526e-06, | |
| "loss": 0.006, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 6.696421052631579e-06, | |
| "loss": 0.005, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 6.6753684210526314e-06, | |
| "loss": 0.0047, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 6.654315789473684e-06, | |
| "loss": 0.0066, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 6.633263157894737e-06, | |
| "loss": 0.0057, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 6.612210526315789e-06, | |
| "loss": 0.0063, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 6.591157894736841e-06, | |
| "loss": 0.0067, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 6.570105263157894e-06, | |
| "loss": 0.006, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 6.5490526315789465e-06, | |
| "loss": 0.0066, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 6.527999999999999e-06, | |
| "loss": 0.0056, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 6.506947368421053e-06, | |
| "loss": 0.0036, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 6.485894736842105e-06, | |
| "loss": 0.0034, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 6.464842105263158e-06, | |
| "loss": 0.0027, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 6.443789473684211e-06, | |
| "loss": 0.0044, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 7.47, | |
| "learning_rate": 6.422736842105263e-06, | |
| "loss": 0.005, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 6.401684210526316e-06, | |
| "loss": 0.0034, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 6.380631578947368e-06, | |
| "loss": 0.0028, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 6.35957894736842e-06, | |
| "loss": 0.0037, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 6.338526315789473e-06, | |
| "loss": 0.0045, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 6.317473684210526e-06, | |
| "loss": 0.005, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 6.296421052631578e-06, | |
| "loss": 0.0063, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.275368421052631e-06, | |
| "loss": 0.0051, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 6.254315789473684e-06, | |
| "loss": 0.0038, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 6.233263157894737e-06, | |
| "loss": 0.0035, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 6.21221052631579e-06, | |
| "loss": 0.0036, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 6.1911578947368425e-06, | |
| "loss": 0.0034, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 6.170105263157894e-06, | |
| "loss": 0.0028, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 6.149052631578947e-06, | |
| "loss": 0.0028, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 6.128e-06, | |
| "loss": 0.0024, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 6.106947368421052e-06, | |
| "loss": 0.0029, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 6.085894736842105e-06, | |
| "loss": 0.0032, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 6.0648421052631576e-06, | |
| "loss": 0.0031, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 6.04378947368421e-06, | |
| "loss": 0.0023, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 6.022736842105263e-06, | |
| "loss": 0.0028, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.0016842105263155e-06, | |
| "loss": 0.0032, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 5.980631578947368e-06, | |
| "loss": 0.0037, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 5.95957894736842e-06, | |
| "loss": 0.0021, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 5.9385263157894735e-06, | |
| "loss": 0.0021, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 5.917473684210526e-06, | |
| "loss": 0.0022, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 5.896421052631579e-06, | |
| "loss": 0.0016, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "eval_loss": 0.48472946882247925, | |
| "eval_runtime": 1245.478, | |
| "eval_samples_per_second": 1.713, | |
| "eval_steps_per_second": 0.214, | |
| "eval_wer": 14.360862703840086, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 5.8753684210526315e-06, | |
| "loss": 0.002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 5.854315789473684e-06, | |
| "loss": 0.0027, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 5.833263157894737e-06, | |
| "loss": 0.0027, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 5.8122105263157894e-06, | |
| "loss": 0.0038, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 5.791157894736842e-06, | |
| "loss": 0.0027, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 5.770105263157895e-06, | |
| "loss": 0.0021, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 5.7490526315789465e-06, | |
| "loss": 0.0024, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 5.727999999999999e-06, | |
| "loss": 0.0017, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 5.706947368421052e-06, | |
| "loss": 0.0024, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 5.6858947368421045e-06, | |
| "loss": 0.0023, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 5.664842105263157e-06, | |
| "loss": 0.0018, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 5.643789473684211e-06, | |
| "loss": 0.0019, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 5.622736842105263e-06, | |
| "loss": 0.0027, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 5.601684210526316e-06, | |
| "loss": 0.0027, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 5.580631578947369e-06, | |
| "loss": 0.0022, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 5.559578947368421e-06, | |
| "loss": 0.0019, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 5.538526315789473e-06, | |
| "loss": 0.0018, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 5.517473684210526e-06, | |
| "loss": 0.002, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 5.496421052631578e-06, | |
| "loss": 0.0034, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.475368421052631e-06, | |
| "loss": 0.0017, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 5.454315789473684e-06, | |
| "loss": 0.002, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "learning_rate": 5.433263157894736e-06, | |
| "loss": 0.0019, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "learning_rate": 5.412210526315789e-06, | |
| "loss": 0.0022, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 5.391157894736842e-06, | |
| "loss": 0.0015, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 5.370105263157895e-06, | |
| "loss": 0.0014, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 5.349052631578948e-06, | |
| "loss": 0.0017, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 5.328e-06, | |
| "loss": 0.0015, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "learning_rate": 5.306947368421052e-06, | |
| "loss": 0.0016, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 11.71, | |
| "learning_rate": 5.285894736842105e-06, | |
| "loss": 0.0026, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 5.264842105263158e-06, | |
| "loss": 0.0026, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 11.87, | |
| "learning_rate": 5.24378947368421e-06, | |
| "loss": 0.0026, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 11.95, | |
| "learning_rate": 5.222736842105263e-06, | |
| "loss": 0.0019, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 5.2016842105263156e-06, | |
| "loss": 0.0019, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 12.11, | |
| "learning_rate": 5.180631578947368e-06, | |
| "loss": 0.0018, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 5.159578947368421e-06, | |
| "loss": 0.0021, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "learning_rate": 5.1385263157894735e-06, | |
| "loss": 0.0015, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 12.34, | |
| "learning_rate": 5.117473684210525e-06, | |
| "loss": 0.0019, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 12.42, | |
| "learning_rate": 5.096421052631578e-06, | |
| "loss": 0.0026, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 5.0753684210526315e-06, | |
| "loss": 0.0015, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 12.58, | |
| "learning_rate": 5.054315789473684e-06, | |
| "loss": 0.002, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 12.58, | |
| "eval_loss": 0.4919339120388031, | |
| "eval_runtime": 1243.533, | |
| "eval_samples_per_second": 1.715, | |
| "eval_steps_per_second": 0.215, | |
| "eval_wer": 14.171488690163073, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 5.033263157894737e-06, | |
| "loss": 0.0017, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 12.74, | |
| "learning_rate": 5.0122105263157895e-06, | |
| "loss": 0.0025, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 4.991157894736842e-06, | |
| "loss": 0.0025, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "learning_rate": 4.970105263157895e-06, | |
| "loss": 0.0016, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 4.9490526315789474e-06, | |
| "loss": 0.0027, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 4.928e-06, | |
| "loss": 0.0021, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 4.906947368421052e-06, | |
| "loss": 0.0012, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 4.8858947368421045e-06, | |
| "loss": 0.0018, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "learning_rate": 4.864842105263157e-06, | |
| "loss": 0.0012, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 13.36, | |
| "learning_rate": 4.84378947368421e-06, | |
| "loss": 0.0012, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 4.8227368421052625e-06, | |
| "loss": 0.0008, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 13.52, | |
| "learning_rate": 4.801684210526316e-06, | |
| "loss": 0.001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 4.780631578947369e-06, | |
| "loss": 0.001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 4.759578947368421e-06, | |
| "loss": 0.001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 4.738526315789474e-06, | |
| "loss": 0.0011, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 13.84, | |
| "learning_rate": 4.717473684210527e-06, | |
| "loss": 0.0011, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 13.92, | |
| "learning_rate": 4.6964210526315784e-06, | |
| "loss": 0.0008, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "learning_rate": 4.675368421052631e-06, | |
| "loss": 0.0021, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 4.654315789473684e-06, | |
| "loss": 0.0018, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 4.633263157894736e-06, | |
| "loss": 0.0013, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 14.23, | |
| "learning_rate": 4.612210526315789e-06, | |
| "loss": 0.0009, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 14.31, | |
| "learning_rate": 4.591157894736842e-06, | |
| "loss": 0.0012, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 14.39, | |
| "learning_rate": 4.570105263157894e-06, | |
| "loss": 0.0009, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 4.549052631578947e-06, | |
| "loss": 0.002, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 4.528e-06, | |
| "loss": 0.0011, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 4.506947368421053e-06, | |
| "loss": 0.0006, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 4.485894736842105e-06, | |
| "loss": 0.0013, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 4.464842105263158e-06, | |
| "loss": 0.0011, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 4.44378947368421e-06, | |
| "loss": 0.0013, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 4.422736842105263e-06, | |
| "loss": 0.0012, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 4.401684210526316e-06, | |
| "loss": 0.0024, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 15.09, | |
| "learning_rate": 4.380631578947368e-06, | |
| "loss": 0.0013, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 4.359578947368421e-06, | |
| "loss": 0.0013, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 15.25, | |
| "learning_rate": 4.3385263157894736e-06, | |
| "loss": 0.0014, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 4.317473684210526e-06, | |
| "loss": 0.0015, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 4.296421052631579e-06, | |
| "loss": 0.0006, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 15.49, | |
| "learning_rate": 4.275368421052631e-06, | |
| "loss": 0.0008, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 4.254315789473683e-06, | |
| "loss": 0.0008, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 4.233263157894737e-06, | |
| "loss": 0.0007, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 15.72, | |
| "learning_rate": 4.2122105263157895e-06, | |
| "loss": 0.0013, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 15.72, | |
| "eval_loss": 0.5114014744758606, | |
| "eval_runtime": 1251.3958, | |
| "eval_samples_per_second": 1.704, | |
| "eval_steps_per_second": 0.213, | |
| "eval_wer": 14.229352972119939, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 15.8, | |
| "learning_rate": 4.191157894736842e-06, | |
| "loss": 0.0011, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 15.88, | |
| "learning_rate": 4.170105263157895e-06, | |
| "loss": 0.0011, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 15.96, | |
| "learning_rate": 4.1490526315789475e-06, | |
| "loss": 0.0007, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 4.128e-06, | |
| "loss": 0.0011, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 16.12, | |
| "learning_rate": 4.106947368421053e-06, | |
| "loss": 0.0012, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 16.19, | |
| "learning_rate": 4.0858947368421054e-06, | |
| "loss": 0.0005, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 16.27, | |
| "learning_rate": 4.064842105263157e-06, | |
| "loss": 0.0016, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 16.35, | |
| "learning_rate": 4.04378947368421e-06, | |
| "loss": 0.0009, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 16.43, | |
| "learning_rate": 4.0227368421052625e-06, | |
| "loss": 0.0005, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 16.51, | |
| "learning_rate": 4.001684210526315e-06, | |
| "loss": 0.0006, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 3.980631578947369e-06, | |
| "loss": 0.0005, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 3.9595789473684205e-06, | |
| "loss": 0.0005, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 16.75, | |
| "learning_rate": 3.939368421052631e-06, | |
| "loss": 0.0015, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 16.82, | |
| "learning_rate": 3.918315789473684e-06, | |
| "loss": 0.0006, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 16.9, | |
| "learning_rate": 3.897263157894737e-06, | |
| "loss": 0.0005, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 16.98, | |
| "learning_rate": 3.87621052631579e-06, | |
| "loss": 0.0014, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 3.855157894736842e-06, | |
| "loss": 0.0012, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 3.834105263157894e-06, | |
| "loss": 0.0024, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 17.22, | |
| "learning_rate": 3.813052631578947e-06, | |
| "loss": 0.0007, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 3.7919999999999994e-06, | |
| "loss": 0.001, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 3.7709473684210525e-06, | |
| "loss": 0.0012, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "learning_rate": 3.749894736842105e-06, | |
| "loss": 0.0008, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 17.53, | |
| "learning_rate": 3.728842105263158e-06, | |
| "loss": 0.0004, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 3.7077894736842105e-06, | |
| "loss": 0.0005, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 3.6867368421052627e-06, | |
| "loss": 0.0005, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 3.6656842105263154e-06, | |
| "loss": 0.0007, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 3.644631578947368e-06, | |
| "loss": 0.0011, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 17.92, | |
| "learning_rate": 3.623578947368421e-06, | |
| "loss": 0.0005, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 3.6025263157894738e-06, | |
| "loss": 0.0004, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 18.08, | |
| "learning_rate": 3.5814736842105264e-06, | |
| "loss": 0.0007, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 18.16, | |
| "learning_rate": 3.5604210526315786e-06, | |
| "loss": 0.0003, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 3.5393684210526313e-06, | |
| "loss": 0.0003, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 18.32, | |
| "learning_rate": 3.518315789473684e-06, | |
| "loss": 0.0003, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 3.4972631578947366e-06, | |
| "loss": 0.001, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 18.47, | |
| "learning_rate": 3.4762105263157897e-06, | |
| "loss": 0.0003, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 18.55, | |
| "learning_rate": 3.455157894736842e-06, | |
| "loss": 0.0004, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 18.63, | |
| "learning_rate": 3.4341052631578946e-06, | |
| "loss": 0.0003, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 18.71, | |
| "learning_rate": 3.4130526315789472e-06, | |
| "loss": 0.0004, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 18.79, | |
| "learning_rate": 3.392e-06, | |
| "loss": 0.0004, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 18.87, | |
| "learning_rate": 3.370947368421052e-06, | |
| "loss": 0.0014, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 18.87, | |
| "eval_loss": 0.5197107791900635, | |
| "eval_runtime": 1246.3616, | |
| "eval_samples_per_second": 1.711, | |
| "eval_steps_per_second": 0.214, | |
| "eval_wer": 13.913729615991583, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 18.95, | |
| "learning_rate": 3.349894736842105e-06, | |
| "loss": 0.0003, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 3.328842105263158e-06, | |
| "loss": 0.0003, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 19.1, | |
| "learning_rate": 3.3077894736842105e-06, | |
| "loss": 0.0004, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 19.18, | |
| "learning_rate": 3.286736842105263e-06, | |
| "loss": 0.0002, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 3.2656842105263154e-06, | |
| "loss": 0.0002, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 19.34, | |
| "learning_rate": 3.244631578947368e-06, | |
| "loss": 0.0004, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 19.42, | |
| "learning_rate": 3.2235789473684207e-06, | |
| "loss": 0.0005, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 19.5, | |
| "learning_rate": 3.2025263157894738e-06, | |
| "loss": 0.0003, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 19.58, | |
| "learning_rate": 3.1814736842105264e-06, | |
| "loss": 0.0007, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 19.65, | |
| "learning_rate": 3.160421052631579e-06, | |
| "loss": 0.0005, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 19.73, | |
| "learning_rate": 3.1393684210526313e-06, | |
| "loss": 0.0007, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 3.118315789473684e-06, | |
| "loss": 0.0002, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 3.0972631578947366e-06, | |
| "loss": 0.0008, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 19.97, | |
| "learning_rate": 3.0762105263157893e-06, | |
| "loss": 0.0013, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 20.05, | |
| "learning_rate": 3.0551578947368424e-06, | |
| "loss": 0.0004, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 20.13, | |
| "learning_rate": 3.0341052631578946e-06, | |
| "loss": 0.0002, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 20.2, | |
| "learning_rate": 3.0130526315789472e-06, | |
| "loss": 0.0004, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 20.28, | |
| "learning_rate": 2.992e-06, | |
| "loss": 0.0003, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 20.36, | |
| "learning_rate": 2.9709473684210526e-06, | |
| "loss": 0.0008, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 2.9498947368421048e-06, | |
| "loss": 0.0006, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 20.52, | |
| "learning_rate": 2.9288421052631574e-06, | |
| "loss": 0.0008, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 20.6, | |
| "learning_rate": 2.9077894736842105e-06, | |
| "loss": 0.0007, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 20.68, | |
| "learning_rate": 2.886736842105263e-06, | |
| "loss": 0.0007, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 20.75, | |
| "learning_rate": 2.865684210526316e-06, | |
| "loss": 0.0003, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 20.83, | |
| "learning_rate": 2.8446315789473685e-06, | |
| "loss": 0.0004, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 20.91, | |
| "learning_rate": 2.8235789473684207e-06, | |
| "loss": 0.0006, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 20.99, | |
| "learning_rate": 2.8025263157894734e-06, | |
| "loss": 0.0006, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 21.07, | |
| "learning_rate": 2.781473684210526e-06, | |
| "loss": 0.0008, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 21.15, | |
| "learning_rate": 2.760421052631579e-06, | |
| "loss": 0.0003, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 21.23, | |
| "learning_rate": 2.7393684210526318e-06, | |
| "loss": 0.0002, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 21.31, | |
| "learning_rate": 2.718315789473684e-06, | |
| "loss": 0.0002, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 21.38, | |
| "learning_rate": 2.6972631578947366e-06, | |
| "loss": 0.0003, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 21.46, | |
| "learning_rate": 2.6762105263157893e-06, | |
| "loss": 0.0005, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 21.54, | |
| "learning_rate": 2.655157894736842e-06, | |
| "loss": 0.0003, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 21.62, | |
| "learning_rate": 2.634105263157895e-06, | |
| "loss": 0.0004, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 2.6130526315789473e-06, | |
| "loss": 0.0003, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 21.78, | |
| "learning_rate": 2.592e-06, | |
| "loss": 0.0003, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 21.86, | |
| "learning_rate": 2.5709473684210526e-06, | |
| "loss": 0.0008, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 21.93, | |
| "learning_rate": 2.5498947368421052e-06, | |
| "loss": 0.0006, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 2.5288421052631575e-06, | |
| "loss": 0.0003, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "eval_loss": 0.5421546101570129, | |
| "eval_runtime": 1231.4206, | |
| "eval_samples_per_second": 1.732, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 14.1977906365071, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 22.09, | |
| "learning_rate": 2.50778947368421e-06, | |
| "loss": 0.0005, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 22.17, | |
| "learning_rate": 2.486736842105263e-06, | |
| "loss": 0.0002, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 22.25, | |
| "learning_rate": 2.465684210526316e-06, | |
| "loss": 0.0003, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 22.33, | |
| "learning_rate": 2.4446315789473685e-06, | |
| "loss": 0.0002, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 22.41, | |
| "learning_rate": 2.423578947368421e-06, | |
| "loss": 0.0003, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 22.48, | |
| "learning_rate": 2.4025263157894734e-06, | |
| "loss": 0.0002, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 22.56, | |
| "learning_rate": 2.381473684210526e-06, | |
| "loss": 0.0004, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 22.64, | |
| "learning_rate": 2.3604210526315787e-06, | |
| "loss": 0.0002, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "learning_rate": 2.3393684210526318e-06, | |
| "loss": 0.0004, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 22.8, | |
| "learning_rate": 2.3183157894736844e-06, | |
| "loss": 0.0002, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 22.88, | |
| "learning_rate": 2.2972631578947367e-06, | |
| "loss": 0.0002, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 22.96, | |
| "learning_rate": 2.2762105263157893e-06, | |
| "loss": 0.0002, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 2.255157894736842e-06, | |
| "loss": 0.0003, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 23.11, | |
| "learning_rate": 2.2341052631578946e-06, | |
| "loss": 0.0002, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 23.19, | |
| "learning_rate": 2.213052631578947e-06, | |
| "loss": 0.0002, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 23.27, | |
| "learning_rate": 2.192e-06, | |
| "loss": 0.0001, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 23.35, | |
| "learning_rate": 2.1709473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 23.43, | |
| "learning_rate": 2.1498947368421052e-06, | |
| "loss": 0.0001, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 23.51, | |
| "learning_rate": 2.128842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 23.58, | |
| "learning_rate": 2.10778947368421e-06, | |
| "loss": 0.0001, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 23.66, | |
| "learning_rate": 2.0867368421052628e-06, | |
| "loss": 0.0001, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 23.74, | |
| "learning_rate": 2.065684210526316e-06, | |
| "loss": 0.0001, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 23.82, | |
| "learning_rate": 2.0446315789473685e-06, | |
| "loss": 0.0001, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 23.9, | |
| "learning_rate": 2.023578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 23.98, | |
| "learning_rate": 2.002526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 24.06, | |
| "learning_rate": 1.981473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 24.14, | |
| "learning_rate": 1.9604210526315787e-06, | |
| "loss": 0.0001, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 24.21, | |
| "learning_rate": 1.9393684210526314e-06, | |
| "loss": 0.0001, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 24.29, | |
| "learning_rate": 1.918315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 24.37, | |
| "learning_rate": 1.8972631578947367e-06, | |
| "loss": 0.0001, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 24.45, | |
| "learning_rate": 1.8762105263157895e-06, | |
| "loss": 0.0001, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 24.53, | |
| "learning_rate": 1.855157894736842e-06, | |
| "loss": 0.0001, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 24.61, | |
| "learning_rate": 1.8341052631578946e-06, | |
| "loss": 0.0001, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 24.69, | |
| "learning_rate": 1.8130526315789473e-06, | |
| "loss": 0.0001, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 24.76, | |
| "learning_rate": 1.792e-06, | |
| "loss": 0.0001, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 24.84, | |
| "learning_rate": 1.7709473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 24.92, | |
| "learning_rate": 1.749894736842105e-06, | |
| "loss": 0.0001, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 1.728842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 25.08, | |
| "learning_rate": 1.7077894736842104e-06, | |
| "loss": 0.0001, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 25.16, | |
| "learning_rate": 1.686736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 25.16, | |
| "eval_loss": 0.5658935904502869, | |
| "eval_runtime": 1242.8998, | |
| "eval_samples_per_second": 1.716, | |
| "eval_steps_per_second": 0.215, | |
| "eval_wer": 13.871646501841136, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 25.24, | |
| "learning_rate": 1.6656842105263159e-06, | |
| "loss": 0.0001, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 25.31, | |
| "learning_rate": 1.6446315789473683e-06, | |
| "loss": 0.0001, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 25.39, | |
| "learning_rate": 1.623578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 25.47, | |
| "learning_rate": 1.6025263157894734e-06, | |
| "loss": 0.0001, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 25.55, | |
| "learning_rate": 1.5814736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 25.63, | |
| "learning_rate": 1.560421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 25.71, | |
| "learning_rate": 1.5393684210526314e-06, | |
| "loss": 0.0001, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 25.79, | |
| "learning_rate": 1.5183157894736843e-06, | |
| "loss": 0.0001, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 25.86, | |
| "learning_rate": 1.4972631578947367e-06, | |
| "loss": 0.0001, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 25.94, | |
| "learning_rate": 1.4762105263157894e-06, | |
| "loss": 0.0001, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.4551578947368422e-06, | |
| "loss": 0.0001, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 26.1, | |
| "learning_rate": 1.4341052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 26.18, | |
| "learning_rate": 1.4130526315789473e-06, | |
| "loss": 0.0001, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 26.26, | |
| "learning_rate": 1.3919999999999998e-06, | |
| "loss": 0.0001, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 26.34, | |
| "learning_rate": 1.3709473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 26.42, | |
| "learning_rate": 1.3498947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 26.49, | |
| "learning_rate": 1.3288421052631577e-06, | |
| "loss": 0.0001, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 26.57, | |
| "learning_rate": 1.3077894736842106e-06, | |
| "loss": 0.0001, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 26.65, | |
| "learning_rate": 1.286736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 26.73, | |
| "learning_rate": 1.2656842105263157e-06, | |
| "loss": 0.0001, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 26.81, | |
| "learning_rate": 1.2446315789473683e-06, | |
| "loss": 0.0001, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 26.89, | |
| "learning_rate": 1.223578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 26.97, | |
| "learning_rate": 1.2025263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 27.04, | |
| "learning_rate": 1.181473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 27.12, | |
| "learning_rate": 1.160421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 27.2, | |
| "learning_rate": 1.1393684210526316e-06, | |
| "loss": 0.0001, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 27.28, | |
| "learning_rate": 1.118315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 27.36, | |
| "learning_rate": 1.097263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 27.44, | |
| "learning_rate": 1.0762105263157894e-06, | |
| "loss": 0.0001, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 27.52, | |
| "learning_rate": 1.055157894736842e-06, | |
| "loss": 0.0001, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "learning_rate": 1.0341052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 27.67, | |
| "learning_rate": 1.0130526315789473e-06, | |
| "loss": 0.0001, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 27.75, | |
| "learning_rate": 9.92e-07, | |
| "loss": 0.0001, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 27.83, | |
| "learning_rate": 9.709473684210526e-07, | |
| "loss": 0.0001, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 27.91, | |
| "learning_rate": 9.498947368421052e-07, | |
| "loss": 0.0001, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 27.99, | |
| "learning_rate": 9.288421052631578e-07, | |
| "loss": 0.0001, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "learning_rate": 9.077894736842104e-07, | |
| "loss": 0.0001, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 28.14, | |
| "learning_rate": 8.867368421052632e-07, | |
| "loss": 0.0001, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 28.22, | |
| "learning_rate": 8.656842105263158e-07, | |
| "loss": 0.0001, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 28.3, | |
| "learning_rate": 8.446315789473684e-07, | |
| "loss": 0.0001, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 28.3, | |
| "eval_loss": 0.5772180557250977, | |
| "eval_runtime": 1231.7677, | |
| "eval_samples_per_second": 1.732, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 13.729615991583376, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 28.38, | |
| "learning_rate": 8.23578947368421e-07, | |
| "loss": 0.0001, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 28.46, | |
| "learning_rate": 8.025263157894736e-07, | |
| "loss": 0.0001, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 28.54, | |
| "learning_rate": 7.814736842105263e-07, | |
| "loss": 0.0001, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 28.62, | |
| "learning_rate": 7.60421052631579e-07, | |
| "loss": 0.0001, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 28.69, | |
| "learning_rate": 7.393684210526315e-07, | |
| "loss": 0.0001, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 28.77, | |
| "learning_rate": 7.183157894736842e-07, | |
| "loss": 0.0001, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 28.85, | |
| "learning_rate": 6.972631578947367e-07, | |
| "loss": 0.0001, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 28.93, | |
| "learning_rate": 6.762105263157895e-07, | |
| "loss": 0.0001, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 6.55157894736842e-07, | |
| "loss": 0.0001, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 29.09, | |
| "learning_rate": 6.341052631578947e-07, | |
| "loss": 0.0001, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 29.17, | |
| "learning_rate": 6.130526315789474e-07, | |
| "loss": 0.0001, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 29.25, | |
| "learning_rate": 5.919999999999999e-07, | |
| "loss": 0.0001, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 29.32, | |
| "learning_rate": 5.709473684210527e-07, | |
| "loss": 0.0001, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 29.4, | |
| "learning_rate": 5.498947368421052e-07, | |
| "loss": 0.0001, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 29.48, | |
| "learning_rate": 5.288421052631579e-07, | |
| "loss": 0.0001, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 29.56, | |
| "learning_rate": 5.077894736842105e-07, | |
| "loss": 0.0001, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 29.64, | |
| "learning_rate": 4.867368421052631e-07, | |
| "loss": 0.0001, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 29.72, | |
| "learning_rate": 4.656842105263158e-07, | |
| "loss": 0.0001, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 29.8, | |
| "learning_rate": 4.4463157894736843e-07, | |
| "loss": 0.0001, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 29.87, | |
| "learning_rate": 4.2357894736842103e-07, | |
| "loss": 0.0001, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 29.95, | |
| "learning_rate": 4.0252631578947364e-07, | |
| "loss": 0.0001, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 30.03, | |
| "learning_rate": 3.814736842105263e-07, | |
| "loss": 0.0001, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 30.11, | |
| "learning_rate": 3.604210526315789e-07, | |
| "loss": 0.0001, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 30.19, | |
| "learning_rate": 3.393684210526316e-07, | |
| "loss": 0.0001, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 30.27, | |
| "learning_rate": 3.183157894736842e-07, | |
| "loss": 0.0001, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 30.35, | |
| "learning_rate": 2.972631578947368e-07, | |
| "loss": 0.0001, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 30.42, | |
| "learning_rate": 2.7621052631578946e-07, | |
| "loss": 0.0001, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 30.5, | |
| "learning_rate": 2.5515789473684206e-07, | |
| "loss": 0.0001, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 30.58, | |
| "learning_rate": 2.3410526315789472e-07, | |
| "loss": 0.0001, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 30.66, | |
| "learning_rate": 2.1305263157894734e-07, | |
| "loss": 0.0001, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 30.74, | |
| "learning_rate": 1.92e-07, | |
| "loss": 0.0001, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 30.82, | |
| "learning_rate": 1.7094736842105263e-07, | |
| "loss": 0.0001, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 30.9, | |
| "learning_rate": 1.4989473684210526e-07, | |
| "loss": 0.0001, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 30.97, | |
| "learning_rate": 1.2884210526315788e-07, | |
| "loss": 0.0001, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 31.05, | |
| "learning_rate": 1.0778947368421053e-07, | |
| "loss": 0.0001, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 31.13, | |
| "learning_rate": 8.673684210526315e-08, | |
| "loss": 0.0001, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 31.21, | |
| "learning_rate": 6.568421052631578e-08, | |
| "loss": 0.0001, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 31.29, | |
| "learning_rate": 4.463157894736842e-08, | |
| "loss": 0.0001, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 31.37, | |
| "learning_rate": 2.3578947368421052e-08, | |
| "loss": 0.0001, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 31.45, | |
| "learning_rate": 2.526315789473684e-09, | |
| "loss": 0.0001, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 31.45, | |
| "eval_loss": 0.5813759565353394, | |
| "eval_runtime": 1234.2331, | |
| "eval_samples_per_second": 1.728, | |
| "eval_steps_per_second": 0.216, | |
| "eval_wer": 13.708574434508153, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 31.45, | |
| "step": 10000, | |
| "total_flos": 3.263088216372019e+20, | |
| "train_loss": 0.0230206538159051, | |
| "train_runtime": 56008.0991, | |
| "train_samples_per_second": 5.713, | |
| "train_steps_per_second": 0.179 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 32, | |
| "total_flos": 3.263088216372019e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |