| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 68223, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.3484, | |
| "step": 22741 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.867647786297304, | |
| "eval_combined_score": 0.8486629929990515, | |
| "eval_f1": 0.829678199700799, | |
| "eval_loss": 0.3014103174209595, | |
| "eval_runtime": 292.1565, | |
| "eval_samples_per_second": 138.385, | |
| "eval_steps_per_second": 17.299, | |
| "step": 22741 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2387, | |
| "step": 45482 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8800890427900074, | |
| "eval_combined_score": 0.8615115077230022, | |
| "eval_f1": 0.8429339726559969, | |
| "eval_loss": 0.3011280596256256, | |
| "eval_runtime": 292.4225, | |
| "eval_samples_per_second": 138.259, | |
| "eval_steps_per_second": 17.283, | |
| "step": 45482 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.1739, | |
| "step": 68223 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8847390551570616, | |
| "eval_combined_score": 0.865679382097654, | |
| "eval_f1": 0.8466197090382463, | |
| "eval_loss": 0.36861559748649597, | |
| "eval_runtime": 291.8123, | |
| "eval_samples_per_second": 138.548, | |
| "eval_steps_per_second": 17.319, | |
| "step": 68223 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 68223, | |
| "total_flos": 1.941180482476892e+17, | |
| "train_loss": 0.2536681392615943, | |
| "train_runtime": 22875.5112, | |
| "train_samples_per_second": 47.716, | |
| "train_steps_per_second": 2.982 | |
| } | |
| ], | |
| "max_steps": 68223, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.941180482476892e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |