Aggregating source code snippets from the POJ-104 collection, pre-processed via CodeXGLUE scripts for semantic clone detection tasks. It provides a structured environment for retrieving functionally equivalent code candidates from a repository based on a query snippet.
Use Cases
- Train a machine learning model to retrieve Top K codes with the same semantic meaning as an input query.
- Evaluate the accuracy of code embedding models using the MAP@R score.
- Develop semantic clone detection systems that identify functionally identical implementations within the collection of candidates.
Strengths
- Includes source code snippets and a collection of candidates for code-to-code retrieval.
- Pre-processed using official CodeXGLUE scripts for consistency with established benchmarks.
- Uses the MAP@R (Mean Average Precision at R) metric for evaluating retrieval performance.