DI-BENCH: Benchmarking Large Language Models on Dependency Inference with Testable Repositories at Scale

TL;DR

DI-BENCH is a large-scale benchmark designed to evaluate how well Large Language Models can infer software dependencies across multiple programming languages, revealing significant gaps in current model capabilities.

Contribution

Introduces DI-BENCH, a comprehensive benchmark with 581 repositories for assessing LLMs' dependency inference, highlighting the need for improved models in software synthesis.

Findings

Current best model achieves only 42.9% execution pass rate

Dependency issues cause over 40% of runtime errors in generated repositories

DI-BENCH provides a new evaluation perspective for LLM-based software development

Abstract

Large Language Models have advanced automated software development, however, it remains a challenge to correctly infer dependencies, namely, identifying the internal components and external packages required for a repository to successfully run. Existing studies highlight that dependency-related issues cause over 40\% of observed runtime errors on the generated repository. To address this, we introduce DI-BENCH, a large-scale benchmark and evaluation framework specifically designed to assess LLMs' capability on dependency inference. The benchmark features 581 repositories with testing environments across Python, C#, Rust, and JavaScript. Extensive experiments with textual and execution-based metrics reveal that the current best-performing model achieves only a 42.9% execution pass rate, indicating significant room for improvement. DI-BENCH establishes a new viewpoint for evaluating LLM performance on repositories, paving the way for more robust end-to-end software synthesis.