Please Open Source Training Datasets and Evaluation Codes for MFTCoder

I feel excited to read papers focusing on multi-task learning for code-related tasks. Yet I noticed that the 5 training datasets and evaluation codes are not open-sourced, which makes the experimental results irreproducible. I believe this work would make an even bigger impact if it could be fully open-sourced.

No files, tests, or entry points are named in the issue. Start by inspecting the repository to locate the five training datasets and evaluation code, then determine how they can be published; done means all five datasets and the evaluation code are open-sourced and reproducible.

Scritto dal modello di indicizzazione a partire dal testo della issue.