| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.8181818181818183, | |
| "eval_steps": 500, | |
| "global_step": 264, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00012592592592592592, | |
| "loss": 1.7865, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00019409282700421943, | |
| "loss": 1.5839, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.000179746835443038, | |
| "loss": 1.3442, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00016540084388185655, | |
| "loss": 1.2967, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0001510548523206751, | |
| "loss": 1.2723, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00013670886075949366, | |
| "loss": 1.2075, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00012236286919831225, | |
| "loss": 1.1807, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00010801687763713079, | |
| "loss": 1.0787, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.367088607594936e-05, | |
| "loss": 1.0961, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.932489451476794e-05, | |
| "loss": 1.0747, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.49789029535865e-05, | |
| "loss": 1.0936, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.0632911392405066e-05, | |
| "loss": 1.0467, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.628691983122363e-05, | |
| "loss": 0.9777, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.1940928270042196e-05, | |
| "loss": 1.0002, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 7.5949367088607605e-06, | |
| "loss": 0.948, | |
| "step": 255 | |
| } | |
| ], | |
| "logging_steps": 17, | |
| "max_steps": 264, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 2.1873419863719936e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |