English

Boosting Logical Reasoning in Large Language Models through a New Framework: The Graph of Thought

Machine Learning 2023-08-21 v1 Artificial Intelligence Computation and Language

Abstract

Recent advancements in large-scale models, such as GPT-4, have showcased remarkable capabilities in addressing standard queries. However, when facing complex problems that require multi-step logical reasoning, their accuracy dramatically decreases. Current research has explored the realm of \textit{prompting engineering} to bolster the inferential capacities of these models. Our paper unveils a pioneering prompting technique, dubbed \textit{Graph of Thoughts (GoT)}. Through testing on a trio of escalating challenges: the 24-point game, resolution of high-degree polynomial equations, and derivation of formulas for recursive sequences, our method outperformed GPT-4, achieving accuracy improvements of 89.7%89.7\%, 86%86\%, and 56%56\% for each respective task. Moreover, when juxtaposed with the state-of-the-art (SOTA) prompting method, \textit{Tree of Thought (ToT)}, our approach registered an average accuracy boost of 23%23\%, 24%24\%, and 15%15\%.

Keywords

Cite

@article{arxiv.2308.08614,
  title  = {Boosting Logical Reasoning in Large Language Models through a New Framework: The Graph of Thought},
  author = {Bin Lei and pei-Hung Lin and Chunhua Liao and Caiwen Ding},
  journal= {arXiv preprint arXiv:2308.08614},
  year   = {2023}
}
R2 v1 2026-06-28T11:57:24.749Z