Llama-3.1-8B-Instruct_SFT_mathfisher_v00.02/train_results.json

{
    "ewc_loss": 2.944469451904297e-05,
    "total_flos": 4.051147348618982e+19,
    "train_loss": 0.40901430871242717,
    "train_runtime": 39420.4574,
    "train_samples": 125770,
    "train_samples_per_second": 9.571,
    "train_steps_per_second": 0.598
}