llama-1B-ddm/train_results.json

{
    "epoch": 1.0,
    "total_flos": 2.295939729457152e+19,
    "train_loss": 8.168445116853714,
    "train_runtime": 153668.5765,
    "train_samples_per_second": 49.978,
    "train_steps_per_second": 3.124
}