Qwen2.5-7B-MLC/train_results.json

{
    "epoch": 3.0,
    "total_flos": 8.768589976326636e+17,
    "train_loss": 0.23950207320631442,
    "train_runtime": 868.6403,
    "train_samples_per_second": 9.836,
    "train_steps_per_second": 0.307
}