llama3.2-1B-pt/train_results.json

{
    "epoch": 1.0,
    "total_flos": 2.295939729457152e+19,
    "train_loss": 3.214656148060163,
    "train_runtime": 141094.738,
    "train_samples_per_second": 54.432,
    "train_steps_per_second": 3.402
}