PairCoder Dataset Release

This release contains the complete dataset required for the PairCoder project. The dataset is organized as follows:

dataset/
├─codecontest    
│  ├─test        
│  └─valid       
├─humaneval
│  ├─plus
│  └─raw
└─mbpp
    ├─plus
    └─test

Instructions

1. Download the dataset.zip file from this release.
2. Unzip the file.
3. Replace the existing dataset directory in the project root with the extracted dataset folder.

Notes

• The dataset directory includes all .arrow files and metadata necessary for the project.
• Ensure that the structure matches the project requirements as shown above.

For any issues, please open a GitHub issue.