| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9961977186311786, | |
| "eval_steps": 150, | |
| "global_step": 393, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 3.4180147647857666, | |
| "learning_rate": 0.00026183206106870227, | |
| "loss": 0.9544, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.5740360617637634, | |
| "learning_rate": 0.00022366412213740454, | |
| "loss": 0.7865, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 2.0023133754730225, | |
| "learning_rate": 0.00018549618320610686, | |
| "loss": 0.6667, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "eval_loss": 0.670988142490387, | |
| "eval_runtime": 68.8887, | |
| "eval_samples_per_second": 15.126, | |
| "eval_steps_per_second": 1.902, | |
| "eval_wer": 0.36541157620668935, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.8345832824707031, | |
| "learning_rate": 0.00014732824427480916, | |
| "loss": 0.5926, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 0.9624394178390503, | |
| "learning_rate": 0.00010916030534351144, | |
| "loss": 0.569, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "grad_norm": 1.4039088487625122, | |
| "learning_rate": 7.099236641221374e-05, | |
| "loss": 0.5023, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "eval_loss": 0.6015787720680237, | |
| "eval_runtime": 68.721, | |
| "eval_samples_per_second": 15.163, | |
| "eval_steps_per_second": 1.906, | |
| "eval_wer": 0.34132785900260365, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 0.5447213053703308, | |
| "learning_rate": 3.282442748091603e-05, | |
| "loss": 0.4813, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 393, | |
| "total_flos": 6.537821345695587e+18, | |
| "train_loss": 0.10885493445942421, | |
| "train_runtime": 326.0744, | |
| "train_samples_per_second": 38.577, | |
| "train_steps_per_second": 1.205 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 393, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 150, | |
| "total_flos": 6.537821345695587e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |