{
    "epoch": 0.9993060374739764,
    "eval_logits/chosen": -0.44369885325431824,
    "eval_logits/rejected": -0.31763577461242676,
    "eval_logps/chosen": -729.1251220703125,
    "eval_logps/rejected": -989.8275146484375,
    "eval_loss": 0.3316161632537842,
    "eval_rewards/accuracies": 0.8383458852767944,
    "eval_rewards/chosen": -2.6774144172668457,
    "eval_rewards/margins": 2.368194341659546,
    "eval_rewards/rejected": -5.0456085205078125,
    "eval_runtime": 378.3021,
    "eval_samples": 8491,
    "eval_samples_per_second": 22.445,
    "eval_steps_per_second": 0.352,
    "total_flos": 0.0,
    "train_loss": 0.4493948830498589,
    "train_runtime": 20175.981,
    "train_samples": 184443,
    "train_samples_per_second": 9.142,
    "train_steps_per_second": 0.036
}