| { |
| "best_metric": 0.4507129896372791, |
| "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-0/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": 1.803922773535424e-05, |
| "loss": 0.5874, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.6919057110953206e-05, |
| "loss": 0.5524, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 1.5798886486552175e-05, |
| "loss": 0.5657, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.4678715862151142e-05, |
| "loss": 0.5372, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_loss": 0.5069713592529297, |
| "eval_matthews_correlation": 0.4467807407096838, |
| "eval_runtime": 0.7552, |
| "eval_samples_per_second": 1381.121, |
| "eval_steps_per_second": 87.396, |
| "step": 2138 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 1.3558545237750109e-05, |
| "loss": 0.4356, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 1.2438374613349074e-05, |
| "loss": 0.483, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 1.131820398894804e-05, |
| "loss": 0.4814, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.0198033364547007e-05, |
| "loss": 0.4779, |
| "step": 4000 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 0.9718611240386963, |
| "eval_matthews_correlation": 0.4507129896372791, |
| "eval_runtime": 0.7395, |
| "eval_samples_per_second": 1410.379, |
| "eval_steps_per_second": 89.247, |
| "step": 4276 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 8552, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 4, |
| "save_steps": 500, |
| "total_flos": 75899679823680.0, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": { |
| "learning_rate": 1.9159398359755275e-05, |
| "num_train_epochs": 4, |
| "per_device_train_batch_size": 4, |
| "seed": 20 |
| } |
| } |
|
|
