| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 5352, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.53288490284006e-05, | |
| "loss": 4.6706, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.06576980568012e-05, | |
| "loss": 3.2966, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.598654708520179e-05, | |
| "loss": 3.1131, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.131539611360239e-05, | |
| "loss": 3.023, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6644245142002988e-05, | |
| "loss": 2.8461, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.197309417040359e-05, | |
| "loss": 2.793, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7301943198804185e-05, | |
| "loss": 2.7901, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2630792227204785e-05, | |
| "loss": 2.6794, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.959641255605381e-06, | |
| "loss": 2.6457, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.2884902840059795e-06, | |
| "loss": 2.6186, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 5352, | |
| "total_flos": 1.0118640236829696e+16, | |
| "train_loss": 3.0209487093938305, | |
| "train_runtime": 15633.4435, | |
| "train_samples_per_second": 0.684, | |
| "train_steps_per_second": 0.342 | |
| } | |
| ], | |
| "max_steps": 5352, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.0118640236829696e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |