| { | |
| "best_metric": 0.4411051273345947, | |
| "best_model_checkpoint": "DP16/checkpoint-750", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 750, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6825, | |
| "eval_loss": 0.6708120703697205, | |
| "eval_model_preparation_time": 0.0058, | |
| "eval_runtime": 79.0561, | |
| "eval_samples_per_second": 40.478, | |
| "eval_steps_per_second": 5.06, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9265625, | |
| "eval_loss": 0.5907767415046692, | |
| "eval_model_preparation_time": 0.0058, | |
| "eval_runtime": 79.1025, | |
| "eval_samples_per_second": 40.454, | |
| "eval_steps_per_second": 5.057, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.97125, | |
| "eval_loss": 0.5124214291572571, | |
| "eval_model_preparation_time": 0.0058, | |
| "eval_runtime": 78.7295, | |
| "eval_samples_per_second": 40.646, | |
| "eval_steps_per_second": 5.081, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 1.4777089357376099, | |
| "learning_rate": 3.5714285714285716e-07, | |
| "loss": 0.5571, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.98625, | |
| "eval_loss": 0.45859000086784363, | |
| "eval_model_preparation_time": 0.0058, | |
| "eval_runtime": 79.6112, | |
| "eval_samples_per_second": 40.195, | |
| "eval_steps_per_second": 5.024, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9884375, | |
| "eval_loss": 0.4411051273345947, | |
| "eval_model_preparation_time": 0.0058, | |
| "eval_runtime": 79.2925, | |
| "eval_samples_per_second": 40.357, | |
| "eval_steps_per_second": 5.045, | |
| "step": 750 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 750, | |
| "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": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.859807750750208e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |