How to reproduce the results in the paper？

[Yuzuriha-Inori-x]

Hi,

Thank you for the good work, but when I tried to reproduce it, I discovered a problem.

In each.sh file, the file paths for input and output need to be filled in. However, I'm not sure what the corresponding files are like in each.sh file within the dataset you provided on HuggingFace. Could you provide a complete file? For example, the process of training and validation on Qwen2.5-7B.