Search papers, labs, and topics across Lattice.
This paper introduces DyRetriever, a novel context retrieval method that utilizes partial dependency graphs to enhance repository-level code generation with LLMs. By mimicking human developers' iterative inspection of code dependencies, DyRetriever eliminates the need for static global graphs and manual rules, allowing for dynamic graph construction that is both efficient and flexible. Experimental results demonstrate that DyCoder, which integrates DyRetriever with a similarity-based retriever, significantly outperforms existing methods, achieving up to 59.73% improvement in Pass@1 metrics while being 7.4 times faster.
DyRetriever achieves a remarkable 25.63% improvement in code generation accuracy while slashing retrieval times by over 7x compared to traditional methods.
LLM-based repository-level code generation aims to generate code using the context available in a software repository, requiring LLMs to reason over complex code dependencies. Due to limited context windows and insufficient repository-specific understanding, LLMs typically rely on retrieval-augmented generation (RAG) to incorporate relevant code. Early RAG approaches primarily employ similarity-based retrieval, which often fails to retrieve code snippets that the target function depends on. Recent work introduces graph-based retrieval to model such dependencies, but typically relies on manually designed rules and static global graphs, leading to limited flexibility and high construction and maintenance costs. In contrast, human developers collect helpful context by implicitly constructing a partial dependency graph and iteratively inspecting along it. Inspired by this behavior, we propose DyRetriever, an efficient context retrieval method via partial dependency graphs. DyRetriever uses an LLM to first select a set of entry-point functions and then perform multi-hop reasoning along the code dependency graph. During multi-hop reasoning, it uses the LLM's semantic understanding to validate whether a function can help generate the target function, eliminating manually designed rules and enabling flexibility across scenarios. Instead of statically constructing a global dependency graph, DyRetriever builds a partial graph on demand and discards it after use, reducing construction and maintenance costs. We integrate DyRetriever with a similarity-based code retriever to build DyCoder and evaluate it on CoderEval and DevEval. Experimental results show that DyCoder achieves relative Pass@1 improvements of 25.63% and 59.73% on CoderEval and DevEval, respectively, compared with existing RAG-based methods, while being 7.4x faster than baselines based on static dependency graph construction.
