| { | |
| "best_metric": 0.8988341093063354, | |
| "best_model_checkpoint": "./results/run-d045ac8a/checkpoint-1764", | |
| "epoch": 3.0, | |
| "global_step": 2646, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.970769826170644e-08, | |
| "loss": 2.104, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1": 0.5132287012657224, | |
| "eval_loss": 0.9356955289840698, | |
| "eval_runtime": 3.9376, | |
| "eval_samples_per_second": 597.314, | |
| "eval_steps_per_second": 74.664, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.823832044239721e-05, | |
| "loss": 1.1539, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_f1": 0.6100360403521365, | |
| "eval_loss": 0.8988341093063354, | |
| "eval_runtime": 3.9403, | |
| "eval_samples_per_second": 596.902, | |
| "eval_steps_per_second": 74.613, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.5007263065485174e-05, | |
| "loss": 0.6207, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_f1": 0.6201819619499414, | |
| "eval_loss": 1.0120480060577393, | |
| "eval_runtime": 3.9533, | |
| "eval_samples_per_second": 594.945, | |
| "eval_steps_per_second": 74.368, | |
| "step": 2646 | |
| } | |
| ], | |
| "max_steps": 2646, | |
| "num_train_epochs": 3, | |
| "total_flos": 1840940131235520.0, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 4.985384913085322e-05, | |
| "num_train_epochs": 3, | |
| "per_device_train_batch_size": 8, | |
| "warmup_steps": 500, | |
| "weight_decay": 0.20061124234729208 | |
| } | |
| } | |