Alignment-Lab-AI/sophia

1{2  "best_metric": null,3  "best_model_checkpoint": null,4  "epoch": 2.742857142857143,5  "global_step": 24,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.11,12      "learning_rate": 4.9786121534345265e-05,13      "loss": 2.3656,14      "step": 115    },16    {17      "epoch": 0.23,18      "learning_rate": 4.914814565722671e-05,19      "loss": 2.3106,20      "step": 221    },22    {23      "epoch": 0.34,24      "learning_rate": 4.8096988312782174e-05,25      "loss": 2.2921,26      "step": 327    },28    {29      "epoch": 0.46,30      "learning_rate": 4.665063509461097e-05,31      "loss": 2.1831,32      "step": 433    },34    {35      "epoch": 0.57,36      "learning_rate": 4.4833833507280884e-05,37      "loss": 2.2563,38      "step": 539    },40    {41      "epoch": 0.69,42      "learning_rate": 4.267766952966369e-05,43      "loss": 2.2585,44      "step": 645    },46    {47      "epoch": 0.8,48      "learning_rate": 4.021903572521802e-05,49      "loss": 2.3297,50      "step": 751    },52    {53      "epoch": 0.91,54      "learning_rate": 3.7500000000000003e-05,55      "loss": 2.2337,56      "step": 857    },58    {59      "epoch": 1.03,60      "learning_rate": 3.456708580912725e-05,61      "loss": 2.2176,62      "step": 963    },64    {65      "epoch": 1.14,66      "learning_rate": 3.147047612756302e-05,67      "loss": 2.2138,68      "step": 1069    },70    {71      "epoch": 1.26,72      "learning_rate": 2.8263154805501297e-05,73      "loss": 2.23,74      "step": 1175    },76    {77      "epoch": 1.37,78      "learning_rate": 2.5e-05,79      "loss": 2.2902,80      "step": 1281    },82    {83      "epoch": 1.49,84      "learning_rate": 2.173684519449872e-05,85      "loss": 2.2732,86      "step": 1387    },88    {89      "epoch": 1.6,90      "learning_rate": 1.852952387243698e-05,91      "loss": 2.2587,92      "step": 1493    },94    {95      "epoch": 1.71,96      "learning_rate": 1.5432914190872757e-05,97      "loss": 2.1618,98      "step": 1599    },100    {101      "epoch": 1.83,102      "learning_rate": 1.2500000000000006e-05,103      "loss": 2.2363,104      "step": 16105    },106    {107      "epoch": 1.94,108      "learning_rate": 9.780964274781984e-06,109      "loss": 2.2207,110      "step": 17111    },112    {113      "epoch": 2.06,114      "learning_rate": 7.3223304703363135e-06,115      "loss": 2.2272,116      "step": 18117    },118    {119      "epoch": 2.17,120      "learning_rate": 5.166166492719124e-06,121      "loss": 2.2384,122      "step": 19123    },124    {125      "epoch": 2.29,126      "learning_rate": 3.3493649053890326e-06,127      "loss": 2.1726,128      "step": 20129    },130    {131      "epoch": 2.4,132      "learning_rate": 1.9030116872178316e-06,133      "loss": 2.1893,134      "step": 21135    },136    {137      "epoch": 2.51,138      "learning_rate": 8.51854342773295e-07,139      "loss": 2.2168,140      "step": 22141    },142    {143      "epoch": 2.63,144      "learning_rate": 2.1387846565474045e-07,145      "loss": 2.2572,146      "step": 23147    },148    {149      "epoch": 2.74,150      "learning_rate": 0.0,151      "loss": 2.2699,152      "step": 24153    },154    {155      "epoch": 2.74,156      "step": 24,157      "total_flos": 7.741968380801843e+16,158      "train_loss": 2.2459669013818107,159      "train_runtime": 103.0683,160      "train_samples_per_second": 32.425,161      "train_steps_per_second": 0.233162    }163  ],164  "max_steps": 24,165  "num_train_epochs": 3,166  "total_flos": 7.741968380801843e+16,167  "trial_name": null,168  "trial_params": null169}170