DAFT/train_results.json

{
    "epoch": 1.0,
    "total_flos": 1.3702381181159342e+18,
    "train_loss": 0.11128336125916809,
    "train_runtime": 25797.0773,
    "train_samples_per_second": 1.359,
    "train_steps_per_second": 0.011
}