Llama-3.2-1B-Instruct_SFT_sciencev00.02/train_results.json

{
    "total_flos": 1.377941890090926e+18,
    "train_loss": 1.2564991597543682,
    "train_runtime": 1307.2842,
    "train_samples": 145693,
    "train_samples_per_second": 111.621,
    "train_steps_per_second": 0.436
}