| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.966292134831461, | |
| "global_step": 132, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.242424242424242e-05, | |
| "loss": 2.0363, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.484848484848486e-05, | |
| "loss": 1.5648, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.727272727272727e-05, | |
| "loss": 1.4116, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.96969696969697e-05, | |
| "loss": 1.374, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.212121212121213e-05, | |
| "loss": 1.1637, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.4545454545454546e-05, | |
| "loss": 0.9509, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.696969696969697e-05, | |
| "loss": 0.9321, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.939393939393939e-05, | |
| "loss": 0.828, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.181818181818182e-05, | |
| "loss": 0.9276, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.4242424242424244e-05, | |
| "loss": 0.6383, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.6272, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.5016, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.5151515151515152e-06, | |
| "loss": 0.4818, | |
| "step": 130 | |
| } | |
| ], | |
| "max_steps": 132, | |
| "num_train_epochs": 3, | |
| "total_flos": 1791640197021696.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |