| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9648784253184098, | |
| "global_step": 12500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9742699086581757e-05, | |
| "loss": 2.8717, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9485398173163515e-05, | |
| "loss": 2.2456, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9228097259745274e-05, | |
| "loss": 2.0513, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8970796346327032e-05, | |
| "loss": 1.9335, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8713495432908788e-05, | |
| "loss": 1.8139, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8456194519490546e-05, | |
| "loss": 1.7416, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8198893606072305e-05, | |
| "loss": 1.6725, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.794159269265406e-05, | |
| "loss": 1.6281, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.768429177923582e-05, | |
| "loss": 1.5546, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7426990865817574e-05, | |
| "loss": 1.5645, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.7169689952399332e-05, | |
| "loss": 1.5436, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.691238903898109e-05, | |
| "loss": 1.504, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.6655088125562846e-05, | |
| "loss": 1.4839, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.6397787212144604e-05, | |
| "loss": 1.4755, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.614048629872636e-05, | |
| "loss": 1.4362, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5883185385308118e-05, | |
| "loss": 1.4063, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5625884471889877e-05, | |
| "loss": 1.4216, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5368583558471635e-05, | |
| "loss": 1.3668, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.5111282645053392e-05, | |
| "loss": 1.3787, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4853981731635147e-05, | |
| "loss": 1.3967, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.4596680818216906e-05, | |
| "loss": 1.3867, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.4339379904798663e-05, | |
| "loss": 1.3715, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.4082078991380421e-05, | |
| "loss": 1.3416, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3824778077962178e-05, | |
| "loss": 1.3113, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.3567477164543936e-05, | |
| "loss": 1.323, | |
| "step": 12500 | |
| } | |
| ], | |
| "max_steps": 38865, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.95972567552e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |