{
    "epoch": 0.9947089947089947,
    "eval_logits/chosen": -2.934617280960083,
    "eval_logits/rejected": -3.0282351970672607,
    "eval_logps/chosen": -524.4794921875,
    "eval_logps/rejected": -741.4998779296875,
    "eval_loss": 0.1684425175189972,
    "eval_rewards/accuracies": 0.9375,
    "eval_rewards/chosen": -0.7677375078201294,
    "eval_rewards/margins": 1.8244943618774414,
    "eval_rewards/rejected": -2.5922317504882812,
    "eval_runtime": 1.3947,
    "eval_samples": 100,
    "eval_samples_per_second": 71.701,
    "eval_steps_per_second": 0.717,
    "total_flos": 0.0,
    "train_loss": 0.5087684798747936,
    "train_runtime": 2521.2346,
    "train_samples": 96652,
    "train_samples_per_second": 38.335,
    "train_steps_per_second": 0.037
}