all_results.json

{
    "epoch": 2.981366459627329,
    "eval_logits/chosen": -2.2030229568481445,
    "eval_logits/rejected": -2.1740832328796387,
    "eval_logps/chosen": -268.22247314453125,
    "eval_logps/rejected": -258.3641662597656,
    "eval_loss": 0.6179147958755493,
    "eval_rewards/accuracies": 0.7777777910232544,
    "eval_rewards/chosen": 2.2275123596191406,
    "eval_rewards/margins": 2.3490304946899414,
    "eval_rewards/rejected": -0.1215180829167366,
    "eval_runtime": 74.2544,
    "eval_samples_per_second": 15.353,
    "eval_steps_per_second": 0.242,
    "total_flos": 1414680891359232.0,
    "train_loss": 0.3555192892750104,
    "train_runtime": 4293.0666,
    "train_samples_per_second": 7.168,
    "train_steps_per_second": 0.028
}