qwen4B_models/qwen_4B_d0_iter1_model/train_results.json

{
    "epoch": 2.0,
    "total_flos": 22867310280704.0,
    "train_loss": 1.0144876452053295,
    "train_runtime": 349.3742,
    "train_samples_per_second": 4.631,
    "train_steps_per_second": 0.292
}