From Code to Correctness: Closing the Last Mile of Code Generation with Hierarchical Debugging

While large language models have made significant strides in code generation, the pass rate of the generated code is bottlenecked on subtle errors, often requiring human intervention to pass tests, especially for complex problems. Existing LLM-based debugging systems treat generated programs as holistic units, failing to address bugs at multiple levels of granularity, from low-level syntax errors to high-level algorithmic flaws. In this paper, we introduce Multi-Granularity Debugger (MGDebugger), a hierarchical code debugger that isolates, identifies, and resolves bugs at various levels of granularity. MGDebugger decomposes problematic code into a hierarchical tree structure of subfunctions, with each level representing a particular granularity of error. During debugging, it analyzes each subfunction and iteratively resolves bugs in a bottom-up manner. To effectively test each subfunction, we propose an LLM-simulated Python executor, which traces code execution and tracks important variable states to pinpoint errors accurately. Extensive experiments with both open-source and commercial LLMs demonstrate that MGDebugger significantly outperforms existing debugging systems, achieving up to 18.9% improvement in accuracy over seed generations in HumanEval and a 97.6% repair success rate in Hu-manEvalFix. Furthermore, MGDebugger effectively generalizes to real-world software defects, fixing 129 bugs in Defects4J v1.2 and 64 bugs in Defects4J v2.0, outperforming state-of-the-art program repair approaches by 13.2% and 8.5% respectively. These results demonstrate MGDebugger's robustness and effectiveness across different scenarios, establishing it as a powerful tool for closing the last mile of code generation. 1