Llama-3.1-8B-Instruct_SFT_sciencefisher_v00.01/train_results.json

{
    "ewc_loss": 5.841255187988281e-05,
    "total_flos": 3.1829322264618205e+19,
    "train_loss": 0.9239483515919018,
    "train_runtime": 37400.1397,
    "train_samples": 145693,
    "train_samples_per_second": 11.687,
    "train_steps_per_second": 0.73
}