Tree of Thoughts: Deliberate Problem Solving with Large Language Models

Abstract

We introduce Tree of Thoughts (ToT), a framework that generalizes over chain-of-thought prompting and enables exploration over coherent units of text (thoughts) that serve as intermediate steps toward problem solving. ToT allows language models to perform deliberate decision making by considering multiple different reasoning paths and self-evaluating choices to decide the next course of action.

Key Findings

• 1Generalized chain-of-thought to allow branching and backtracking reasoning
• 2Enabled LLMs to explore multiple reasoning paths and self-evaluate
• 3Achieved dramatic improvements on tasks requiring search and planning
• 4Solved the Game of 24 task with 74% success vs 4% for chain-of-thought
• 5Introduced a framework for deliberate, systematic problem solving with LLMs

Impact & Significance