| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 15, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0003, | |
| "loss": 2.6389, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00027857142857142854, | |
| "loss": 2.6419, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002571428571428571, | |
| "loss": 2.6501, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00023571428571428569, | |
| "loss": 2.6429, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00021428571428571427, | |
| "loss": 2.6434, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00019285714285714286, | |
| "loss": 2.6262, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0001714285714285714, | |
| "loss": 2.6372, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00015, | |
| "loss": 2.6339, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00012857142857142855, | |
| "loss": 2.6276, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00010714285714285714, | |
| "loss": 2.5977, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.57142857142857e-05, | |
| "loss": 2.6049, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 6.428571428571427e-05, | |
| "loss": 2.6189, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.285714285714285e-05, | |
| "loss": 2.6324, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 2.1428571428571425e-05, | |
| "loss": 2.564, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.5782, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 15, | |
| "total_flos": 5.963598898365727e+17, | |
| "train_loss": 2.6225359439849854, | |
| "train_runtime": 96.9726, | |
| "train_samples_per_second": 9.838, | |
| "train_steps_per_second": 0.155 | |
| } | |
| ], | |
| "max_steps": 15, | |
| "num_train_epochs": 3, | |
| "total_flos": 5.963598898365727e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |