Llama-3.1-8B-Instruct_SDFT_mathv00.09/train_results.json

{
    "epoch": 3.0,
    "total_flos": 0.0,
    "train_loss": 0.06867336523107444,
    "train_runtime": 41172.7067,
    "train_samples": 225129,
    "train_samples_per_second": 16.404,
    "train_steps_per_second": 0.032
}