all_results.json

{
    "epoch": 0.9982631930527722,
    "eval_logits/chosen": 0.10372808575630188,
    "eval_logits/rejected": 0.08583810925483704,
    "eval_logps/chosen": -0.3924722671508789,
    "eval_logps/rejected": -0.5309580564498901,
    "eval_loss": 1.2116273641586304,
    "eval_rewards/accuracies": 0.7113820910453796,
    "eval_rewards/chosen": -0.3924722671508789,
    "eval_rewards/margins": 0.13848578929901123,
    "eval_rewards/rejected": -0.5309580564498901,
    "eval_runtime": 427.6213,
    "eval_samples": 1961,
    "eval_samples_per_second": 4.586,
    "eval_steps_per_second": 0.288,
    "total_flos": 0.0,
    "train_loss": 1.2289682925385632,
    "train_runtime": 21112.5317,
    "train_samples": 59876,
    "train_samples_per_second": 2.836,
    "train_steps_per_second": 0.022
}