RepoQA

coding official site →

RepoQA is a benchmark for evaluating long-context code understanding capabilities of Large Language Models through the Searching Needle Function (SNF) task, where LLMs must locate specific functions in code repositories using natural language descriptions. The benchmark contains 500 code search tasks spanning 50 repositories across 5 modern programming languages (Python, Java, TypeScript, C++, and Rust), tested on 26 general and code-specific LLMs to assess their ability to comprehend and navigate code repositories.

Methodology

Imported from llm-stats public benchmark metadata. Modality: text. Max score: 1. Categories: code, long_context, reasoning. Language: en. Verified by llm-stats: no.

Leaderboard

  1. Phi-3.5-MoE-instruct self-reported llm-stats
    85.0%
  2. Phi-3.5-mini-instruct self-reported llm-stats
    77.0%