| { | |
| "best_metric": 0.8030890822410583, | |
| "best_model_checkpoint": "./lora-alpaca/checkpoint-200", | |
| "epoch": 0.5144694533762058, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9999999999999997e-05, | |
| "loss": 2.2847, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 2.2109, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.699999999999999e-05, | |
| "loss": 1.9736, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000117, | |
| "loss": 1.5818, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000147, | |
| "loss": 1.2586, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00017699999999999997, | |
| "loss": 1.1373, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00020699999999999996, | |
| "loss": 0.9888, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.000237, | |
| "loss": 0.8617, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000267, | |
| "loss": 0.8454, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00029699999999999996, | |
| "loss": 0.8458, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000290625, | |
| "loss": 0.8341, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002802083333333333, | |
| "loss": 0.8218, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00026979166666666666, | |
| "loss": 0.8304, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000259375, | |
| "loss": 0.8161, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00024895833333333334, | |
| "loss": 0.8097, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00023854166666666663, | |
| "loss": 0.8134, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00022812499999999997, | |
| "loss": 0.8214, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00021770833333333332, | |
| "loss": 0.8095, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00020729166666666663, | |
| "loss": 0.8028, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00019687499999999997, | |
| "loss": 0.8014, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "eval_loss": 0.8030890822410583, | |
| "eval_runtime": 134.5586, | |
| "eval_samples_per_second": 14.863, | |
| "eval_steps_per_second": 1.858, | |
| "step": 200 | |
| } | |
| ], | |
| "max_steps": 388, | |
| "num_train_epochs": 1, | |
| "total_flos": 2.5486812238577664e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |