llama-3-8b-base-epsilon-dpo-hh-helpful-4xh200-batch-64-20260418-001920/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.8920053139482126,
    "train_runtime": 3235.1211,
    "train_samples": 43598,
    "train_samples_per_second": 13.476,
    "train_steps_per_second": 0.211
}