| { | |
| "best_metric": 0.27657164760495423, | |
| "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-3/checkpoint-4276", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 4276, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.3504098486610872e-06, | |
| "loss": 0.6235, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1715949852260916e-06, | |
| "loss": 0.6084, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.92780121791096e-07, | |
| "loss": 0.5687, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.139652583561002e-07, | |
| "loss": 0.5599, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5971149802207947, | |
| "eval_matthews_correlation": 0.0463559874942472, | |
| "eval_runtime": 0.7351, | |
| "eval_samples_per_second": 1418.884, | |
| "eval_steps_per_second": 89.786, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.351503949211046e-07, | |
| "loss": 0.5449, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.563355314861089e-07, | |
| "loss": 0.5175, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.7752066805111325e-07, | |
| "loss": 0.535, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.870580461611762e-08, | |
| "loss": 0.5162, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.571711540222168, | |
| "eval_matthews_correlation": 0.27657164760495423, | |
| "eval_runtime": 0.7318, | |
| "eval_samples_per_second": 1425.327, | |
| "eval_steps_per_second": 90.193, | |
| "step": 4276 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 4276, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "total_flos": 65111866045632.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.529224712096083e-06, | |
| "num_train_epochs": 2, | |
| "per_device_train_batch_size": 4, | |
| "seed": 5 | |
| } | |
| } | |