| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9941379571543597e-05, | |
| "loss": 1.0685, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9476531711828027e-05, | |
| "loss": 0.9694, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.8568571761675893e-05, | |
| "loss": 0.9469, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.725995491923131e-05, | |
| "loss": 0.9373, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5611870653623826e-05, | |
| "loss": 0.9309, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.3701381553399147e-05, | |
| "loss": 0.9124, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.161781996552765e-05, | |
| "loss": 0.9115, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.458610914145826e-06, | |
| "loss": 0.906, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.324716614707794e-06, | |
| "loss": 0.9042, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.3159155930021e-06, | |
| "loss": 0.9025, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.5261371521817247e-06, | |
| "loss": 0.8989, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.0390693429435626e-06, | |
| "loss": 0.893, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 9.242458032904311e-07, | |
| "loss": 0.8912, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 2.3379444289913344e-07, | |
| "loss": 0.8924, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.8895, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 150, | |
| "total_flos": 4.584413079177003e+18, | |
| "train_loss": 0.9236494954427084, | |
| "train_runtime": 2513.3699, | |
| "train_samples_per_second": 89.997, | |
| "train_steps_per_second": 0.06 | |
| } | |
| ], | |
| "max_steps": 150, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.584413079177003e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |