DevEval: Evaluating Code Generation in Practical Software Projects

TL;DR

DevEval is a new benchmark designed to evaluate large language models' code generation capabilities in realistic software project scenarios, addressing limitations of previous benchmarks.

Contribution

The paper introduces DevEval, a practical, large-scale benchmark aligned with real-world projects, and evaluates popular LLMs' performance on it.

Findings

GPT-3.5-turbo achieves a Pass@1 of 42

DevEval reveals current LLMs' limitations in practical code generation

Open-sourced benchmark facilitates future research

Abstract

How to evaluate Large Language Models (LLMs) in code generation is an open question. Many benchmarks have been proposed but are inconsistent with practical software projects, e.g., unreal program distributions, insufficient dependencies, and small-scale project contexts. Thus, the capabilities of LLMs in practical projects are still unclear. In this paper, we propose a new benchmark named DevEval, aligned with Developers' experiences in practical projects. DevEval is collected through a rigorous pipeline, containing 2,690 samples from 119 practical projects and covering 10 domains. Compared to previous benchmarks, DevEval aligns to practical projects in multiple dimensions, e.g., real program distributions, sufficient dependencies, and enough-scale project contexts. We assess five popular LLMs on DevEval (e.g., gpt-4, gpt-3.5-turbo, CodeLLaMa, and StarCoder) and reveal their actual abilities in code generation. For instance, the highest Pass@1 of gpt-3.5-turbo only is 42 in our experiments. We also discuss the challenges and future directions of code generation in practical projects. We open-source DevEval and hope it can facilitate the development of code generation in practical projects.