llama-3-8b-base-beta-dpo-hh-helpful-8xh200/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.5267414394546958,
    "train_runtime": 1440.2657,
    "train_samples": 43598,
    "train_samples_per_second": 30.271,
    "train_steps_per_second": 0.236
}