Llama-3.1-8B-Instruct_SFT_MoTv00.02/train_results.json

{
    "total_flos": 7.271307435875238e+19,
    "train_loss": 0.7722587241576269,
    "train_runtime": 61686.092,
    "train_samples": 269513,
    "train_samples_per_second": 8.738,
    "train_steps_per_second": 0.091
}