| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.0964360587002098, |
| "eval_steps": 500, |
| "global_step": 1000, |
| "is_hyper_param_search": true, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.9119354838709678, |
| "eval_loss": 1.895792007446289, |
| "eval_runtime": 17.833, |
| "eval_samples_per_second": 173.835, |
| "eval_steps_per_second": 5.439, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.0482180293501049, |
| "grad_norm": 20.671356201171875, |
| "learning_rate": 1.4769392033542977e-05, |
| "loss": 7.6866, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_accuracy": 0.9461290322580646, |
| "eval_loss": 1.1595720052719116, |
| "eval_runtime": 18.9353, |
| "eval_samples_per_second": 163.715, |
| "eval_steps_per_second": 5.123, |
| "step": 954 |
| }, |
| { |
| "epoch": 2.0964360587002098, |
| "grad_norm": 7.095048904418945, |
| "learning_rate": 9.528301886792455e-06, |
| "loss": 1.2357, |
| "step": 1000 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 1908, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 4, |
| "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": 448725995623620.0, |
| "train_batch_size": 32, |
| "trial_name": null, |
| "trial_params": { |
| "alpha": 0.8316624474522443, |
| "num_train_epochs": 4, |
| "temperature": 2.959983034910542 |
| } |
| } |
|
|
