all_results.json

{
    "epoch": 1.971563981042654,
    "eval_logits/chosen": 99.97323608398438,
    "eval_logits/rejected": 94.49807739257812,
    "eval_logps/chosen": -365.3877868652344,
    "eval_logps/rejected": -382.0710754394531,
    "eval_loss": 0.7875852584838867,
    "eval_rewards/accuracies": 0.78125,
    "eval_rewards/chosen": -0.08597712963819504,
    "eval_rewards/margins": 0.9140273928642273,
    "eval_rewards/rejected": -1.0000044107437134,
    "eval_runtime": 122.4417,
    "eval_samples": 750,
    "eval_samples_per_second": 6.125,
    "eval_steps_per_second": 0.196,
    "total_flos": 0.0,
    "train_loss": 3302.790195810107,
    "train_runtime": 2170.5017,
    "train_samples": 6750,
    "train_samples_per_second": 6.22,
    "train_steps_per_second": 0.048
}