all_results.json

{
    "epoch": 0.9999343530493009,
    "eval_logits/chosen": -0.8756721019744873,
    "eval_logits/rejected": -0.9693624377250671,
    "eval_logps/chosen": -103.03213500976562,
    "eval_logps/rejected": -254.58119201660156,
    "eval_loss": 0.0005562438163906336,
    "eval_rewards/accuracies": 1.0,
    "eval_rewards/chosen": -3.5039849281311035,
    "eval_rewards/margins": 14.41648006439209,
    "eval_rewards/rejected": -17.92046546936035,
    "eval_runtime": 161.905,
    "eval_samples_per_second": 33.211,
    "eval_steps_per_second": 4.157
}