Llama-3.1-8B-Instruct_SFT_sciencefisher_v00.09/train_results.json

{
    "ewc_loss": 0.00012063980102539062,
    "total_flos": 3.1829322264618205e+19,
    "train_loss": 0.9450448354238018,
    "train_runtime": 37078.6153,
    "train_samples": 145693,
    "train_samples_per_second": 11.788,
    "train_steps_per_second": 0.737
}