| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.998099762470309, | |
| "global_step": 2104, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.880703422053232e-05, | |
| "loss": 0.7156, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.5905963302752294, | |
| "eval_loss": 0.6729157567024231, | |
| "eval_runtime": 2.6255, | |
| "eval_samples_per_second": 332.129, | |
| "eval_steps_per_second": 10.665, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.692490494296578e-05, | |
| "loss": 0.6745, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6387614678899083, | |
| "eval_loss": 0.6506585478782654, | |
| "eval_runtime": 2.6291, | |
| "eval_samples_per_second": 331.667, | |
| "eval_steps_per_second": 10.65, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.5042775665399238e-05, | |
| "loss": 0.6574, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6456422018348624, | |
| "eval_loss": 0.6401416659355164, | |
| "eval_runtime": 2.6431, | |
| "eval_samples_per_second": 329.91, | |
| "eval_steps_per_second": 10.593, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 3.1606463878326996e-06, | |
| "loss": 0.6507, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6444954128440367, | |
| "eval_loss": 0.6368556618690491, | |
| "eval_runtime": 2.6019, | |
| "eval_samples_per_second": 335.136, | |
| "eval_steps_per_second": 10.761, | |
| "step": 2104 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 2104, | |
| "total_flos": 9.369725771789107e+16, | |
| "train_loss": 0.6729737561011949, | |
| "train_runtime": 2059.9591, | |
| "train_samples_per_second": 130.777, | |
| "train_steps_per_second": 1.021 | |
| } | |
| ], | |
| "max_steps": 2104, | |
| "num_train_epochs": 4, | |
| "total_flos": 9.369725771789107e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |