deepseek-code-only/train_results.json

{
    "epoch": 4.0,
    "train_loss": 0.7115756135984702,
    "train_runtime": 15560.6826,
    "train_samples": 29976,
    "train_samples_per_second": 7.706,
    "train_steps_per_second": 1.926
}