PageIndexMCTS

PageIndexMCTS is a Python library that implements Monte Carlo Tree Search (MCTS) for efficient page indexing and retrieval, integrated with Large Language Models (LLMs).

Features

• MCTS Implementation: Includes standard MCTS and Hybrid MCTS (PUCT) algorithms.
• LLM Integration: Supports using LLMs (like ChatGPT) for value estimation and policy priors.
• Parallel Search: Optimized for performance with parallel simulation steps.
• Document Indexing: Tools for indexing PDF and Markdown documents.

Installation

You can install the package directly from source:

pip install .

Usage

See the example directory for usage demonstrations.

Basic MCTS Search

from search.tree.mcts import PageIndexMCTS

# Initialize your structure and query
# ...
search = PageIndexMCTS(structure, query, use_llm=True, n_threads=4)
search.run_search(iterations=50)
results = search.get_top_results(n=3)

License

This project is licensed under the MIT License - see the LICENSE file for details.