Reproduce “OptiTree: Hierarchical Thoughts Generation with Tree Search for LLM Optimization Modeling” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haoyang Liu, Jie Wang, Yuyang Cai, Xiongwei Han, Yufei Kuang, Jianye Hao
OptiTree introduces a hierarchical tree search approach to enhance large language models' ability to generate optimization models for complex operations research problems. The framework achieves over 10% higher modeling accuracy on challenging datasets compared to existing methods, demonstrating improved generalization and a higher code pass rate by adaptively decomposing problems.
Key insights
- Complex operations research problems can often be effectively modeled by decomposing them into a series of simpler, standard subproblems.
- A hierarchical structure for organizing problem types and associated "modeling thoughts" allows LLMs to leverage expert-like reasoning, reducing common errors like incorrect variable definitions.
- Guiding LLMs with distilled "statement thoughts" for subproblem identification and flexible "modeling thoughts" for model construction is essential for robust and accurate generation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.