| { | |
| "best_metric": 0.5613207547169812, | |
| "best_model_checkpoint": "final_models/glue_models/structroberta_s2_50ep//finetune/multirc/checkpoint-400", | |
| "epoch": 10.0, | |
| "global_step": 730, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.74, | |
| "eval_accuracy": 0.6155531406402588, | |
| "eval_f1": 0.5517241379310345, | |
| "eval_loss": 0.6651450991630554, | |
| "eval_mcc": 0.22280007426012766, | |
| "eval_runtime": 1.9443, | |
| "eval_samples_per_second": 469.573, | |
| "eval_steps_per_second": 59.147, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "eval_accuracy": 0.5925520062446594, | |
| "eval_f1": 0.5613207547169812, | |
| "eval_loss": 0.7663992643356323, | |
| "eval_mcc": 0.181004001401683, | |
| "eval_runtime": 1.954, | |
| "eval_samples_per_second": 467.248, | |
| "eval_steps_per_second": 58.854, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.5753424657534248e-05, | |
| "loss": 0.6095, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "eval_accuracy": 0.5980284810066223, | |
| "eval_f1": 0.5312899106002553, | |
| "eval_loss": 0.9963139295578003, | |
| "eval_mcc": 0.1867939732247269, | |
| "eval_runtime": 1.9532, | |
| "eval_samples_per_second": 467.432, | |
| "eval_steps_per_second": 58.877, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 730, | |
| "total_flos": 4247678489011200.0, | |
| "train_loss": 0.5290867113087275, | |
| "train_runtime": 256.3743, | |
| "train_samples_per_second": 180.868, | |
| "train_steps_per_second": 2.847 | |
| } | |
| ], | |
| "max_steps": 730, | |
| "num_train_epochs": 10, | |
| "total_flos": 4247678489011200.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |