| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.347593582887701, | |
| "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": 0.7019778490066528, | |
| "eval_loss": 0.5544637441635132, | |
| "eval_runtime": 9.2773, | |
| "eval_samples_per_second": 321.539, | |
| "eval_steps_per_second": 20.157, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.744552493095398, | |
| "eval_loss": 0.49814581871032715, | |
| "eval_runtime": 9.3471, | |
| "eval_samples_per_second": 319.136, | |
| "eval_steps_per_second": 20.006, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 2.6737967914438503, | |
| "grad_norm": 11.23240852355957, | |
| "learning_rate": 1.7634576095057786e-06, | |
| "loss": 0.5749, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.763660728931427, | |
| "eval_loss": 0.4700300395488739, | |
| "eval_runtime": 9.4418, | |
| "eval_samples_per_second": 315.936, | |
| "eval_steps_per_second": 19.806, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7747234106063843, | |
| "eval_loss": 0.4598151743412018, | |
| "eval_runtime": 9.504, | |
| "eval_samples_per_second": 313.868, | |
| "eval_steps_per_second": 19.676, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7864565849304199, | |
| "eval_loss": 0.4558647572994232, | |
| "eval_runtime": 9.4827, | |
| "eval_samples_per_second": 314.573, | |
| "eval_steps_per_second": 19.72, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 5.347593582887701, | |
| "grad_norm": 16.31344985961914, | |
| "learning_rate": 1.1198599418029397e-06, | |
| "loss": 0.4329, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1870, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.1927468911210596e+16, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 2.4070552772086175e-06, | |
| "per_device_train_batch_size": 64 | |
| } | |
| } | |