| { | |
| "best_metric": 3.2501280307769775, | |
| "best_model_checkpoint": "./results-bart-base/checkpoint-38400", | |
| "epoch": 4.0, | |
| "global_step": 38400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5003125e-05, | |
| "loss": 3.4014, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 3.2687737941741943, | |
| "eval_runtime": 36.2388, | |
| "eval_samples_per_second": 40.951, | |
| "eval_steps_per_second": 40.951, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0005208333333334e-05, | |
| "loss": 3.1216, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.2523913383483887, | |
| "eval_runtime": 36.0255, | |
| "eval_samples_per_second": 41.193, | |
| "eval_steps_per_second": 41.193, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.006770833333334e-06, | |
| "loss": 2.984, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 3.2509608268737793, | |
| "eval_runtime": 36.1722, | |
| "eval_samples_per_second": 41.026, | |
| "eval_steps_per_second": 41.026, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 8.333333333333335e-09, | |
| "loss": 2.9004, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 3.2501280307769775, | |
| "eval_runtime": 36.2569, | |
| "eval_samples_per_second": 40.93, | |
| "eval_steps_per_second": 40.93, | |
| "step": 38400 | |
| } | |
| ], | |
| "max_steps": 38400, | |
| "num_train_epochs": 4, | |
| "total_flos": 3.277740271583232e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |