FEA-Bench: A Benchmark for Evaluating Repository-Level Code Generation for Feature Implementation
Wei Li, Xin Zhang, Zhongxin Guo, Shaoguang Mao, Wen Luo, Guangyue Peng, Yangyu Huang, Houfeng Wang, Scarlett Li
2025-03-11
Summary
This paper talks about FEA-Bench, a test that checks if AI can add new features to existing code projects by making changes across multiple files, like a programmer working on a big team project.
What's the problem?
Current AI tests only check small code snippets, but real coding involves updating many parts of a project at once, which AI struggles with because it can’t track all the connections between files.
What's the solution?
FEA-Bench uses real coding project updates from GitHub, pairs them with test files, and checks if AI can correctly add features while keeping everything working, like a teacher grading homework with real-world problems.
Why it matters?
This helps improve AI tools for programmers by identifying weaknesses in handling complex projects, making future AI assistants better at collaborating on large codebases.
Abstract
Implementing new features in repository-level codebases is a crucial application of code generation models. However, current benchmarks lack a dedicated evaluation framework for this capability. To fill this gap, we introduce FEA-Bench, a benchmark designed to assess the ability of large language models (LLMs) to perform incremental development within code repositories. We collect pull requests from 83 GitHub repositories and use rule-based and intent-based filtering to construct task instances focused on new feature development. Each task instance containing code changes is paired with relevant unit test files to ensure that the solution can be verified. The feature implementation requires LLMs to simultaneously possess code completion capabilities for new components and code editing abilities for other relevant parts in the code repository, providing a more comprehensive evaluation method of LLMs' automated software engineering capabilities. Experimental results show that LLMs perform significantly worse in the FEA-Bench, highlighting considerable challenges in such repository-level incremental code development.