Llama-3.1-8B-Instruct_SFT_Math-220kv00.19/all_results.json

{
    "total_flos": 1.396241431747625e+18,
    "train_loss": 0.023126277875374347,
    "train_runtime": 17196.7264,
    "train_samples": 86158,
    "train_samples_per_second": 15.03,
    "train_steps_per_second": 0.015
}