| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.979856759176365, | |
| "eval_steps": 500, | |
| "global_step": 44500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.9998880931065354, | |
| "grad_norm": 1.9907540082931519, | |
| "learning_rate": 1.6001342882721576e-05, | |
| "loss": 0.5252, | |
| "step": 8935 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.3881196975708008, | |
| "eval_runtime": 304.5007, | |
| "eval_samples_per_second": 58.693, | |
| "eval_steps_per_second": 7.337, | |
| "step": 8936 | |
| }, | |
| { | |
| "epoch": 1.9997761862130707, | |
| "grad_norm": 1.3169994354248047, | |
| "learning_rate": 1.2004028648164728e-05, | |
| "loss": 0.3435, | |
| "step": 17870 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.32426688075065613, | |
| "eval_runtime": 734.2495, | |
| "eval_samples_per_second": 24.341, | |
| "eval_steps_per_second": 3.043, | |
| "step": 17872 | |
| }, | |
| { | |
| "epoch": 2.999664279319606, | |
| "grad_norm": 1.234189748764038, | |
| "learning_rate": 8.005819158460162e-06, | |
| "loss": 0.2993, | |
| "step": 26805 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.2952452301979065, | |
| "eval_runtime": 310.5637, | |
| "eval_samples_per_second": 57.547, | |
| "eval_steps_per_second": 7.193, | |
| "step": 26808 | |
| }, | |
| { | |
| "epoch": 3.9995523724261415, | |
| "grad_norm": 1.4175560474395752, | |
| "learning_rate": 4.007609668755595e-06, | |
| "loss": 0.2709, | |
| "step": 35740 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.2747509181499481, | |
| "eval_runtime": 785.3908, | |
| "eval_samples_per_second": 22.756, | |
| "eval_steps_per_second": 2.844, | |
| "step": 35744 | |
| } | |
| ], | |
| "logging_steps": 8935, | |
| "max_steps": 44680, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.619840752181084e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |