Llama-3.1-8B-Instruct_SDFT_mathv00.06/train_results.json

{
    "epoch": 3.0,
    "total_flos": 0.0,
    "train_loss": 0.07083441573637131,
    "train_runtime": 38065.4229,
    "train_samples": 225129,
    "train_samples_per_second": 17.743,
    "train_steps_per_second": 0.035
}