| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 10, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 2.571428571428571, |
| "grad_norm": 1.0920410795291435, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.7242, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15573155879974365, |
| "step": 5, |
| "valid_targets_mean": 3221.0, |
| "valid_targets_min": 805 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.6903794426972766, |
| "learning_rate": 1.2061475842818337e-06, |
| "loss": 0.6204, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2023322582244873, |
| "step": 10, |
| "valid_targets_mean": 3909.8, |
| "valid_targets_min": 569 |
| }, |
| { |
| "epoch": 5.0, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.2023322582244873, |
| "step": 10, |
| "total_flos": 1.3906329399721984e+16, |
| "train_loss": 0.6722828388214112, |
| "train_runtime": 195.3945, |
| "train_samples_per_second": 2.559, |
| "train_steps_per_second": 0.051, |
| "valid_targets_mean": 3909.8, |
| "valid_targets_min": 569 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 10, |
| "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": false, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.3906329399721984e+16, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
