| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.03031705623698218, | |
| "global_step": 262, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9997685720897943e-05, | |
| "loss": 2.316, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.993982874334645e-05, | |
| "loss": 3.0145, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9881971765794957e-05, | |
| "loss": 2.9775, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9821800509141405e-05, | |
| "loss": 2.9141, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.976162925248785e-05, | |
| "loss": 2.8213, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9701457995834298e-05, | |
| "loss": 2.8003, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9641286739180746e-05, | |
| "loss": 2.8823, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9581115482527195e-05, | |
| "loss": 2.8253, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9520944225873643e-05, | |
| "loss": 2.7823, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.946077296922009e-05, | |
| "loss": 2.7374, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9402915991668595e-05, | |
| "loss": 2.8325, | |
| "step": 260 | |
| } | |
| ], | |
| "max_steps": 8642, | |
| "num_train_epochs": 1, | |
| "total_flos": 487908402266112.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |