Fix Initial Codes and Iteratively Refine Textual Directions Toward Safe Multi-Turn Code Correction

TL;DR

This paper introduces IRTD, a simple iterative method for multi-turn code correction that refines textual directions to improve inference performance, matching complex state-of-the-art methods.

Contribution

The paper proposes IRTD, a straightforward alternative to complex search methods, with theoretical safety guarantees and comparable empirical performance.

Findings

IRTD achieves inference performance comparable to state-of-the-art methods.

Refining initial codes with textual directions alone can significantly improve inference.

IRTD's simplicity allows for theoretical safety analysis using OGIS.

Abstract

Recent work on large language models (LLMs) has emphasized the importance of scaling inference compute. From this perspective, the state-of-the-art method Scattered Forest Search (SFS) has been proposed, employing Monte Carlo Tree Search with carefully crafted initial seeds and textual optimization for multi-turn code correction. However, its complexity makes it unclear what factors contribute to improvements in inference performance. To address this problem, we analyze SFS and propose a simpler method, Iterative Refinement of Textual Directions (IRTD), which fixes initial codes and iteratively refines textual directions. Because of the simplicity of IRTD, we theoretically establish the safety of IRTD using Oracle-Guided Inductive Synthesis (OGIS). Experiments on several code generation benchmarks suggest that IRTD achieves inference performance comparable to state-of-the-art methods. These results indicate that, even without complex search structures, refining initial codes with high-quality textual directions alone can effectively improve inference performance.