| { | |
| "best_metric": 0.9376567006111145, | |
| "best_model_checkpoint": "finetune_xlm-roberta-base/checkpoint-5476", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 5476, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8789870952033116e-05, | |
| "loss": 2.1799, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.7572437302167033e-05, | |
| "loss": 1.2786, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.635500365230095e-05, | |
| "loss": 1.1682, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.513757000243487e-05, | |
| "loss": 1.1019, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3920136352568787e-05, | |
| "loss": 1.0745, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.9917843341827393, | |
| "eval_runtime": 32.4148, | |
| "eval_samples_per_second": 326.086, | |
| "eval_steps_per_second": 2.561, | |
| "step": 2738 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.2702702702702702e-05, | |
| "loss": 0.9744, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.1487703920136353e-05, | |
| "loss": 0.9039, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.027027027027027e-05, | |
| "loss": 0.8688, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.052836620404189e-06, | |
| "loss": 0.8802, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.835402970538106e-06, | |
| "loss": 0.8816, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.9376567006111145, | |
| "eval_runtime": 32.177, | |
| "eval_samples_per_second": 328.495, | |
| "eval_steps_per_second": 2.579, | |
| "step": 5476 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 8214, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 3.4334001889975296e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |