| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.742857142857143, | |
| "global_step": 24, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9786121534345265e-05, | |
| "loss": 2.3656, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.914814565722671e-05, | |
| "loss": 2.3106, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8096988312782174e-05, | |
| "loss": 2.2921, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.665063509461097e-05, | |
| "loss": 2.1831, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.4833833507280884e-05, | |
| "loss": 2.2563, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.267766952966369e-05, | |
| "loss": 2.2585, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.021903572521802e-05, | |
| "loss": 2.3297, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 2.2337, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.456708580912725e-05, | |
| "loss": 2.2176, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.147047612756302e-05, | |
| "loss": 2.2138, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8263154805501297e-05, | |
| "loss": 2.23, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.5e-05, | |
| "loss": 2.2902, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.173684519449872e-05, | |
| "loss": 2.2732, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.852952387243698e-05, | |
| "loss": 2.2587, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.5432914190872757e-05, | |
| "loss": 2.1618, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.2500000000000006e-05, | |
| "loss": 2.2363, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.780964274781984e-06, | |
| "loss": 2.2207, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.3223304703363135e-06, | |
| "loss": 2.2272, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.166166492719124e-06, | |
| "loss": 2.2384, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.3493649053890326e-06, | |
| "loss": 2.1726, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.9030116872178316e-06, | |
| "loss": 2.1893, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.51854342773295e-07, | |
| "loss": 2.2168, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.1387846565474045e-07, | |
| "loss": 2.2572, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0, | |
| "loss": 2.2699, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "step": 24, | |
| "total_flos": 7.741968380801843e+16, | |
| "train_loss": 2.2459669013818107, | |
| "train_runtime": 103.0683, | |
| "train_samples_per_second": 32.425, | |
| "train_steps_per_second": 0.233 | |
| } | |
| ], | |
| "max_steps": 24, | |
| "num_train_epochs": 3, | |
| "total_flos": 7.741968380801843e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |