train_results.json

{
    "epoch": 5.971724787935909,
    "num_input_tokens_seen": 8063376,
    "total_flos": 9.844904305885184e+16,
    "train_loss": 2.431757736085641,
    "train_runtime": 2652.6299,
    "train_samples_per_second": 9.6,
    "train_steps_per_second": 0.299
}