| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 5490, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00010911475409836066, | |
| "loss": 1.7923, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.820765027322405e-05, | |
| "loss": 1.3442, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.727868852459017e-05, | |
| "loss": 1.218, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.63497267759563e-05, | |
| "loss": 1.1314, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.54207650273224e-05, | |
| "loss": 1.0894, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.4513661202185795e-05, | |
| "loss": 0.881, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.3584699453551917e-05, | |
| "loss": 0.7084, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.265573770491803e-05, | |
| "loss": 0.6909, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.1726775956284153e-05, | |
| "loss": 0.682, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.0797814207650273e-05, | |
| "loss": 0.6723, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 5490, | |
| "total_flos": 5.164033933049242e+16, | |
| "train_loss": 0.9859708636619137, | |
| "train_runtime": 2796.8161, | |
| "train_samples_per_second": 94.217, | |
| "train_steps_per_second": 1.963 | |
| } | |
| ], | |
| "max_steps": 5490, | |
| "num_train_epochs": 2, | |
| "total_flos": 5.164033933049242e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |