train_results.json

{
    "epoch": 2.9575757575757575,
    "num_input_tokens_seen": 6043392,
    "total_flos": 8.724962359229645e+16,
    "train_loss": 0.5365108480563138,
    "train_runtime": 4757.115,
    "train_samples_per_second": 1.249,
    "train_steps_per_second": 0.038
}