ORGANISATION

CodeSearchNet

Dataset of paired code snippets and natural-language documentation across six programming languages, used for code search and understanding.

About Dataset

CodeSearchNet is a dataset of function-level code snippets paired with their natural-language documentation, sourced from open-source repositories across six programming languages including Python, Java, and JavaScript. Released by GitHub in collaboration with academic researchers, the dataset was designed to advance research in semantic code search, where natural-language queries are matched to relevant code. Each entry links a code snippet with an associated docstring or comment describing its function.

Purpose of Dataset

Codesearchnet Is Used To Train And Evaluate Models On Code Search, Code Summarization, And Code-to-text And Text-to-code Tasks. It Supports Research In Bridging Natural Language Understanding With Programming Language Semantics, Enabling Tools Such As Semantic Code Search Engines And Documentation Generators. The Dataset Remains A Widely Cited Benchmark For Code Understanding Research.