Skip to content

ReMCTS: Reflection-Enhanced Monte Carlo Tree Search for Code Generation

Sep 2026 · arXiv.org · Vol abs/2609.34717 · 0 citations · 26 references
Computer Science

TL;DR

ReMCTS, an execution-grounded, memory-augmented, LLM-guided MCTS-style search framework that organizes program candidates as tree states, retains branch-local debugging context, retrieves failure experience across branches, and distinguishes failed checks from unavailable evidence is presented.

Abstract

Open-weight large language models (LLMs) can generate function-level programs from natural-language prompts, but plausible candidates still fail on hidden semantics and repeat mistakes across repair attempts. We present ReMCTS, an execution-grounded, memory-augmented, LLM-guided MCTS-style search framework. It organizes program candidates as tree states, retains branch-local debugging context, retrieves failure experience across branches, and distinguishes failed checks from unavailable evidence. On HumanEval and MBPP-Sanitized, visible-test ReMCTS improves over direct generation in 8 of 10 model-dataset pairs under held-out evaluation, whereas proxy-only search is less stable. Controlled tree-search, sampling, repair, and memory ablations characterize the source and limits of these gains. A 30-task HumanEval-X C++ pilot further demonstrates compatibility with compiler-backed execution, but does not constitute a broad multilingual evaluation.

View source

Similar papers

Book Open access Oct 2026

Evaluating Symbolic Feedback Granularity for LLM-Based Program Repair

Large Language Models (LLMs) are increasingly incorporated into verifier-in-the-loop software engineering systems, yet it remains unclear how the representation of verifier feedback affects their ability to repair incorrect programs. We study this question by isolating feedback representation from model, task, and prom...

Tao-Er Guo · 0 citations
Preprint Aug 2026

CHISEL-ing Back Source Code with AI-enabled Iterative Recovery

This work presents CHISEL, a test suite-free framework to iteratively recover source code from Ghidra-derived pseudo-C, and systematically evaluates CHISEL for compilation and semantic recovery, feedback oracle soundness, and iteration overhead on 120 ExeBench functions compiled for the x86-64 architecture.

Varun Kohli, N. Raghava, B. Sikdar et al. · 2 citations
#artificial intelligence Preprint Oct 2026

Characterizing Overconfident Failure in LLM-Based Code Generation

Large language models (LLMs) are increasingly used for automated code generation, but generated programs can appear syntactically plausible while still failing execution-based correctness checks. Existing validation methods, such as testing and program analysis, remain essential but are often incomplete, costly, or app...

Ravishka Rathnasuriya, Wei Yang · 3 citations
#small language model Book Open access Sep 2026

Code Transformation Rule Synthesis using LLMs: Potential and Limits

A novel empirical study targeting three domain- specific languages for transformation rules, namely Comby, GritQL, and Ast-Grep, which shows great potential for LLMs to generate sound, correct, generalizable, and reusable rules.

Axel Allain, Aymeric Blot, D. Khelladi et al. · 1 citation
#small language model Book Open access Oct 2026

LPR+: Diverse Transformations for LLM-Aided Program Reduction

Program reduction helps compiler and language-tool developers turn large failure-inducing programs into small, shareable bug reports. LPR (Large Language Models-Aided Program Reduction) demonstrated that large language models (LLMs) can complement syntax-guided reducers by proposing language-specific transformations on...

Ze-Hua Zhang, Jia-Tong Liu, Xue-Song Yao et al. · 0 citations
Open access 2026

Diagnosing Candidate Quality Bottlenecks in RL-Guided LLM Code Repair

An RL-guided repair loop is used as an instrumented diagnostic environment to separate correct-candidate availability from candidate selection and establish general superiority of RL, a selector, or an output representation to single-file Python algorithmic repair on QuixBugs.

Jun-Jun Zhang, Giseop Noh · 0 citations

Related blog posts

MIT News · Artificial Intelligence Nov 17, 2026

Learning the language of machine learning

By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.