Published April 1, 2026 | Version v1.0.0
Software Open

MCTS-Reasoning: Monte Carlo Tree Search for LLM Reasoning

Authors/Creators

  • 1. Independent Researcher

Description

A canonical Monte Carlo Tree Search implementation for LLM-based step-by-step reasoning. Explores multiple reasoning paths to find high-quality solutions using UCB1 exploration-exploitation balance, tree-building rollouts, and optional MCP tool integration.

Files

queelius/mcts-reasoning-v1.0.0.zip

Files (246.1 kB)

Name Size Download all
md5:a9759de03aab7f50b88f0d625e5945fc
246.1 kB Preview Download

Additional details

Related works

Is supplement to
https://github.com/queelius/mcts-reasoning (URL)

Software

Repository URL
https://github.com/queelius/mcts-reasoning

References

  • Kocsis, L., & Szepesvári, C. (2006). Bandit based monte-carlo planning. ECML.
  • Wei, J., et al. (2022). Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. NeurIPS.
  • Yao, S., et al. (2023). Tree of Thoughts: Deliberate Problem Solving with Large Language Models.