| { | |
| "best_metric": 9.891409525857435, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 1.0, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 0.4442, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.1027, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.0877, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.0788, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.0815, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.0817, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.0614, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.049, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.0471, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.0456, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.951578947368423e-06, | |
| "loss": 0.0483, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.898947368421054e-06, | |
| "loss": 0.051, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.846315789473684e-06, | |
| "loss": 0.0526, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.793684210526316e-06, | |
| "loss": 0.0609, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.741052631578947e-06, | |
| "loss": 0.052, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.68842105263158e-06, | |
| "loss": 0.0476, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.635789473684212e-06, | |
| "loss": 0.0433, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.583157894736842e-06, | |
| "loss": 0.0508, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.530526315789474e-06, | |
| "loss": 0.0551, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.477894736842106e-06, | |
| "loss": 0.0413, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.425263157894737e-06, | |
| "loss": 0.0313, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.372631578947369e-06, | |
| "loss": 0.0411, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.32e-06, | |
| "loss": 0.0496, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.267368421052632e-06, | |
| "loss": 0.0442, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.214736842105264e-06, | |
| "loss": 0.0445, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.162105263157895e-06, | |
| "loss": 0.0536, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.109473684210527e-06, | |
| "loss": 0.036, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.056842105263159e-06, | |
| "loss": 0.034, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.00421052631579e-06, | |
| "loss": 0.0305, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.951578947368422e-06, | |
| "loss": 0.0301, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.898947368421054e-06, | |
| "loss": 0.0362, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.846315789473685e-06, | |
| "loss": 0.0323, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.793684210526317e-06, | |
| "loss": 0.0416, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.741052631578949e-06, | |
| "loss": 0.033, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.688421052631579e-06, | |
| "loss": 0.0372, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.63578947368421e-06, | |
| "loss": 0.0332, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.583157894736843e-06, | |
| "loss": 0.0319, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.530526315789475e-06, | |
| "loss": 0.037, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.477894736842107e-06, | |
| "loss": 0.0437, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.425263157894737e-06, | |
| "loss": 0.0383, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.22878967225551605, | |
| "eval_runtime": 2943.9063, | |
| "eval_samples_per_second": 1.722, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 12.202988925921945, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.372631578947368e-06, | |
| "loss": 0.0308, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.32e-06, | |
| "loss": 0.0329, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.267368421052632e-06, | |
| "loss": 0.0351, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.214736842105265e-06, | |
| "loss": 0.0333, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.162105263157895e-06, | |
| "loss": 0.0349, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.109473684210527e-06, | |
| "loss": 0.0339, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.056842105263158e-06, | |
| "loss": 0.0276, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.00421052631579e-06, | |
| "loss": 0.0344, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.951578947368421e-06, | |
| "loss": 0.0368, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.898947368421053e-06, | |
| "loss": 0.0331, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.846315789473685e-06, | |
| "loss": 0.028, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.793684210526316e-06, | |
| "loss": 0.0333, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.741052631578948e-06, | |
| "loss": 0.0299, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.68842105263158e-06, | |
| "loss": 0.0252, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.635789473684211e-06, | |
| "loss": 0.0288, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.583157894736842e-06, | |
| "loss": 0.0294, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.5305263157894745e-06, | |
| "loss": 0.0265, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.477894736842106e-06, | |
| "loss": 0.0239, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.425263157894738e-06, | |
| "loss": 0.0248, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.3726315789473694e-06, | |
| "loss": 0.0306, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.32e-06, | |
| "loss": 0.0262, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.267368421052632e-06, | |
| "loss": 0.0257, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.2147368421052635e-06, | |
| "loss": 0.0229, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.162105263157896e-06, | |
| "loss": 0.0238, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.109473684210528e-06, | |
| "loss": 0.0187, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.056842105263158e-06, | |
| "loss": 0.0232, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.00421052631579e-06, | |
| "loss": 0.0273, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.951578947368422e-06, | |
| "loss": 0.0326, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.8989473684210525e-06, | |
| "loss": 0.0266, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.846315789473684e-06, | |
| "loss": 0.0245, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.793684210526317e-06, | |
| "loss": 0.0345, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.741052631578948e-06, | |
| "loss": 0.033, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.68842105263158e-06, | |
| "loss": 0.0293, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.635789473684211e-06, | |
| "loss": 0.0187, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.583157894736842e-06, | |
| "loss": 0.0181, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.530526315789474e-06, | |
| "loss": 0.0242, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.477894736842106e-06, | |
| "loss": 0.0214, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.425263157894738e-06, | |
| "loss": 0.0238, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.372631578947369e-06, | |
| "loss": 0.0168, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.3200000000000005e-06, | |
| "loss": 0.0239, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.2214207649230957, | |
| "eval_runtime": 2933.0533, | |
| "eval_samples_per_second": 1.728, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 11.38049672078271, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.267368421052632e-06, | |
| "loss": 0.0201, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.214736842105264e-06, | |
| "loss": 0.0149, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.1621052631578945e-06, | |
| "loss": 0.0167, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.109473684210527e-06, | |
| "loss": 0.0169, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.056842105263159e-06, | |
| "loss": 0.0157, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.00421052631579e-06, | |
| "loss": 0.0168, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.951578947368422e-06, | |
| "loss": 0.0173, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.898947368421053e-06, | |
| "loss": 0.0167, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.846315789473684e-06, | |
| "loss": 0.015, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.793684210526316e-06, | |
| "loss": 0.0212, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.7410526315789485e-06, | |
| "loss": 0.0208, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.68842105263158e-06, | |
| "loss": 0.0132, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.635789473684211e-06, | |
| "loss": 0.0133, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.5831578947368425e-06, | |
| "loss": 0.0146, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.530526315789474e-06, | |
| "loss": 0.0163, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.477894736842105e-06, | |
| "loss": 0.0141, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.425263157894737e-06, | |
| "loss": 0.0168, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.372631578947369e-06, | |
| "loss": 0.0189, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.320000000000001e-06, | |
| "loss": 0.0204, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.267368421052632e-06, | |
| "loss": 0.0165, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.214736842105263e-06, | |
| "loss": 0.0192, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.162105263157895e-06, | |
| "loss": 0.0164, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.1094736842105264e-06, | |
| "loss": 0.0123, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.056842105263158e-06, | |
| "loss": 0.0174, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.0042105263157906e-06, | |
| "loss": 0.0152, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.951578947368421e-06, | |
| "loss": 0.0156, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.898947368421053e-06, | |
| "loss": 0.0138, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.846315789473685e-06, | |
| "loss": 0.0138, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.793684210526316e-06, | |
| "loss": 0.0142, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.741052631578948e-06, | |
| "loss": 0.0119, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.6884210526315795e-06, | |
| "loss": 0.0154, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.63578947368421e-06, | |
| "loss": 0.0137, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.583157894736843e-06, | |
| "loss": 0.0167, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.5305263157894744e-06, | |
| "loss": 0.0172, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.477894736842105e-06, | |
| "loss": 0.013, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.425263157894737e-06, | |
| "loss": 0.0141, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.3726315789473685e-06, | |
| "loss": 0.0118, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.32e-06, | |
| "loss": 0.0179, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.267368421052632e-06, | |
| "loss": 0.0143, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.214736842105263e-06, | |
| "loss": 0.0125, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.22999617457389832, | |
| "eval_runtime": 2921.4658, | |
| "eval_samples_per_second": 1.735, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 10.98269003332975, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.162105263157895e-06, | |
| "loss": 0.0089, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.109473684210527e-06, | |
| "loss": 0.0113, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.056842105263158e-06, | |
| "loss": 0.0097, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.00421052631579e-06, | |
| "loss": 0.0135, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.951578947368422e-06, | |
| "loss": 0.0142, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.898947368421052e-06, | |
| "loss": 0.0132, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.846315789473685e-06, | |
| "loss": 0.0161, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.793684210526316e-06, | |
| "loss": 0.0141, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7410526315789473e-06, | |
| "loss": 0.0189, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6884210526315794e-06, | |
| "loss": 0.0143, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.635789473684211e-06, | |
| "loss": 0.0165, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.5831578947368422e-06, | |
| "loss": 0.0104, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.5305263157894743e-06, | |
| "loss": 0.011, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.4778947368421055e-06, | |
| "loss": 0.0146, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.0196, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.0139, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.0127, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.0124, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.0112, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.0149, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.0155, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.0215, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.0171, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.0205, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.0156, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.0151, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.0155, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.0109, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.688421052631579e-06, | |
| "loss": 0.0183, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.635789473684211e-06, | |
| "loss": 0.0125, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5831578947368425e-06, | |
| "loss": 0.0108, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5305263157894737e-06, | |
| "loss": 0.0141, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4778947368421053e-06, | |
| "loss": 0.012, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.425263157894737e-06, | |
| "loss": 0.0104, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3726315789473686e-06, | |
| "loss": 0.0099, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 0.0196, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.267368421052632e-06, | |
| "loss": 0.0201, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2147368421052635e-06, | |
| "loss": 0.0127, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1621052631578947e-06, | |
| "loss": 0.0166, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1094736842105264e-06, | |
| "loss": 0.0168, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.24368737637996674, | |
| "eval_runtime": 2933.3968, | |
| "eval_samples_per_second": 1.728, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 10.471992258896892, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.056842105263158e-06, | |
| "loss": 0.0111, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0042105263157896e-06, | |
| "loss": 0.0105, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9515789473684213e-06, | |
| "loss": 0.0109, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8989473684210527e-06, | |
| "loss": 0.0167, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8463157894736843e-06, | |
| "loss": 0.0102, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7936842105263158e-06, | |
| "loss": 0.0099, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7410526315789474e-06, | |
| "loss": 0.0157, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6884210526315792e-06, | |
| "loss": 0.0203, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6357894736842107e-06, | |
| "loss": 0.0222, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5831578947368423e-06, | |
| "loss": 0.0159, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5305263157894737e-06, | |
| "loss": 0.0172, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4778947368421054e-06, | |
| "loss": 0.0106, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.425263157894737e-06, | |
| "loss": 0.0077, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3726315789473684e-06, | |
| "loss": 0.0095, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.32e-06, | |
| "loss": 0.0073, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2673684210526315e-06, | |
| "loss": 0.0064, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2147368421052633e-06, | |
| "loss": 0.0088, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1621052631578948e-06, | |
| "loss": 0.0085, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1094736842105264e-06, | |
| "loss": 0.0111, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0568421052631578e-06, | |
| "loss": 0.0072, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0042105263157897e-06, | |
| "loss": 0.0175, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.515789473684212e-07, | |
| "loss": 0.0087, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.989473684210527e-07, | |
| "loss": 0.009, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.463157894736843e-07, | |
| "loss": 0.0103, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.936842105263158e-07, | |
| "loss": 0.0091, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.410526315789475e-07, | |
| "loss": 0.012, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.884210526315791e-07, | |
| "loss": 0.0113, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.357894736842106e-07, | |
| "loss": 0.0108, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.831578947368421e-07, | |
| "loss": 0.0094, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.305263157894737e-07, | |
| "loss": 0.0072, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.778947368421053e-07, | |
| "loss": 0.0078, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.2526315789473684e-07, | |
| "loss": 0.0065, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.726315789473685e-07, | |
| "loss": 0.0068, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.2e-07, | |
| "loss": 0.0105, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.6736842105263164e-07, | |
| "loss": 0.0054, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.1473684210526317e-07, | |
| "loss": 0.0077, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6210526315789476e-07, | |
| "loss": 0.0085, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0947368421052632e-07, | |
| "loss": 0.0092, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.68421052631579e-08, | |
| "loss": 0.0077, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.210526315789474e-09, | |
| "loss": 0.0071, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.24834245443344116, | |
| "eval_runtime": 2917.338, | |
| "eval_samples_per_second": 1.738, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 9.891409525857435, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 5000, | |
| "total_flos": 1.632967852032e+20, | |
| "train_loss": 0.025400285175442697, | |
| "train_runtime": 51804.3597, | |
| "train_samples_per_second": 3.089, | |
| "train_steps_per_second": 0.097 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.632967852032e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |