| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 844, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6907926418393052, | |
| "eval_loss": 1.4545308351516724, | |
| "eval_runtime": 71.1157, | |
| "eval_samples_per_second": 15.791, | |
| "eval_steps_per_second": 3.951, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6881468917656793, | |
| "eval_loss": 1.4871423244476318, | |
| "eval_runtime": 71.2388, | |
| "eval_samples_per_second": 15.764, | |
| "eval_steps_per_second": 3.944, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.6221244244890336e-05, | |
| "loss": 1.4748, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6840786486403011, | |
| "eval_loss": 1.5352827310562134, | |
| "eval_runtime": 70.948, | |
| "eval_samples_per_second": 15.828, | |
| "eval_steps_per_second": 3.961, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6798846867100433, | |
| "eval_loss": 1.5906230211257935, | |
| "eval_runtime": 70.949, | |
| "eval_samples_per_second": 15.828, | |
| "eval_steps_per_second": 3.961, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 844, | |
| "total_flos": 2.3279302128697344e+16, | |
| "train_loss": 1.2069585357232115, | |
| "train_runtime": 779.6175, | |
| "train_samples_per_second": 4.32, | |
| "train_steps_per_second": 1.083 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6798846867100433, | |
| "eval_loss": 1.5906230211257935, | |
| "eval_runtime": 69.9412, | |
| "eval_samples_per_second": 16.056, | |
| "eval_steps_per_second": 4.018, | |
| "step": 844 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 844, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 2.3279302128697344e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |