| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 12111, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.793576087854017e-05, | |
| "loss": 2.3279, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5871521757080346e-05, | |
| "loss": 1.6773, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.380728263562051e-05, | |
| "loss": 1.4905, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.174304351416068e-05, | |
| "loss": 1.3552, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9678804392700854e-05, | |
| "loss": 1.2864, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.761456527124102e-05, | |
| "loss": 1.2008, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.555032614978119e-05, | |
| "loss": 1.1758, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.348608702832136e-05, | |
| "loss": 1.122, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.1421847906861534e-05, | |
| "loss": 1.0693, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9357608785401702e-05, | |
| "loss": 1.037, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.729336966394187e-05, | |
| "loss": 1.0164, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5229130542482042e-05, | |
| "loss": 0.99, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.316489142102221e-05, | |
| "loss": 0.9796, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1100652299562382e-05, | |
| "loss": 0.9223, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9036413178102553e-05, | |
| "loss": 0.9296, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.697217405664272e-05, | |
| "loss": 0.8937, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4907934935182893e-05, | |
| "loss": 0.876, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2843695813723062e-05, | |
| "loss": 0.8717, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0779456692263231e-05, | |
| "loss": 0.8513, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.715217570803403e-06, | |
| "loss": 0.8391, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.650978449343573e-06, | |
| "loss": 0.8317, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.586739327883742e-06, | |
| "loss": 0.8231, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.522500206423912e-06, | |
| "loss": 0.8137, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.5826108496408225e-07, | |
| "loss": 0.8145, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 12111, | |
| "total_flos": 1.1295642461038182e+17, | |
| "train_loss": 1.0889082555776386, | |
| "train_runtime": 7447.3904, | |
| "train_samples_per_second": 16.262, | |
| "train_steps_per_second": 1.626 | |
| } | |
| ], | |
| "max_steps": 12111, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.1295642461038182e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |