| { |
| "best_metric": 2.138493299484253, |
| "best_model_checkpoint": "gen_test/checkpoint-471", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 471, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.0, |
| "eval_loss": 2.1953625679016113, |
| "eval_runtime": 56.1633, |
| "eval_samples_per_second": 178.052, |
| "eval_steps_per_second": 2.795, |
| "step": 157 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 2.150932550430298, |
| "eval_runtime": 55.9253, |
| "eval_samples_per_second": 178.81, |
| "eval_steps_per_second": 2.807, |
| "step": 314 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_loss": 2.138493299484253, |
| "eval_runtime": 56.1775, |
| "eval_samples_per_second": 178.007, |
| "eval_steps_per_second": 2.795, |
| "step": 471 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 628, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 4, |
| "save_steps": 500, |
| "total_flos": 1.567752192e+16, |
| "train_batch_size": 64, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
