llama-3-8b-base-margin-dpo-hh-helpful-batch-64/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.572698849610295,
    "train_runtime": 1998.3785,
    "train_samples": 43598,
    "train_samples_per_second": 21.817,
    "train_steps_per_second": 0.341
}