train_results.json

{
    "epoch": 1.9985964912280703,
    "num_input_tokens_seen": 5878272,
    "total_flos": 8.486575409523917e+16,
    "train_loss": 0.5569543002220394,
    "train_runtime": 4198.4776,
    "train_samples_per_second": 1.358,
    "train_steps_per_second": 0.085
}