DAPO-8B/train_results.json

{
    "total_flos": 0.0,
    "train_loss": 0.009447227440541611,
    "train_runtime": 101500.2967,
    "train_samples": 7000,
    "train_samples_per_second": 0.095,
    "train_steps_per_second": 0.002
}