{
    "epoch": 1.0,
    "eval_logits/chosen": -2.550788640975952,
    "eval_logits/rejected": -2.5080249309539795,
    "eval_logps/chosen": -362.1445007324219,
    "eval_logps/rejected": -394.6997375488281,
    "eval_loss": 0.5302429795265198,
    "eval_rewards/accuracies": 0.7182539701461792,
    "eval_rewards/chosen": -0.7890641093254089,
    "eval_rewards/margins": 0.6776372790336609,
    "eval_rewards/rejected": -1.4667012691497803,
    "eval_runtime": 166.4681,
    "eval_samples": 500,
    "eval_samples_per_second": 3.004,
    "eval_steps_per_second": 0.378,
    "total_flos": 0.0,
    "train_loss": 0.5873338260650635,
    "train_runtime": 15803.1996,
    "train_samples": 20000,
    "train_samples_per_second": 1.266,
    "train_steps_per_second": 0.079
}