all_results.json

{
    "epoch": 2.963855421686747,
    "eval_logits/chosen": 0.5211664438247681,
    "eval_logits/rejected": 0.5150557160377502,
    "eval_logps/chosen": -92.4592514038086,
    "eval_logps/rejected": -109.63432312011719,
    "eval_loss": 1.5016677379608154,
    "eval_rewards/accuracies": 0.6418919563293457,
    "eval_rewards/chosen": -3.8774402141571045,
    "eval_rewards/margins": 1.2819865942001343,
    "eval_rewards/rejected": -5.159426689147949,
    "eval_runtime": 16.1667,
    "eval_samples_per_second": 9.155,
    "eval_steps_per_second": 2.289,
    "total_flos": 8544080363520.0,
    "train_loss": 0.8105673431380978,
    "train_runtime": 814.398,
    "train_samples_per_second": 4.892,
    "train_steps_per_second": 0.151
}