Semantic Exploration with Adaptive Gating for Efficient Problem Solving with Language Models

TL;DR

SEAG introduces an adaptive gating mechanism for large language models that intelligently decides when to perform tree searches, reducing redundancy and computational costs while improving accuracy in complex reasoning tasks.

Contribution

The paper presents SEAG, a novel method that dynamically controls tree search based on answer confidence, enhancing efficiency and semantic exploration in reasoning with LLMs.

Findings

SEAG improves accuracy by 4.3% on average.

Reduces computational costs to 31% of existing methods.

Effective across diverse models and reasoning benchmarks.

Abstract

Recent advancements in large language models (LLMs) have shown remarkable potential in various complex tasks requiring multi-step reasoning methods like tree search to explore diverse reasoning paths. However, existing methods often suffer from computational inefficiency and redundancy. First, they overlook the diversity of task difficulties, leading to unnecessarily extensive searches even for easy tasks. Second, they neglect the semantics of reasoning paths, resulting in redundant exploration of semantically identical paths. To address these limitations, we propose Semantic Exploration with Adaptive Gating (SEAG), a computationally efficient method. SEAG employs an adaptive gating mechanism that dynamically decides whether to conduct a tree search, based on the confidence level of answers from a preceding simple reasoning method. Furthermore, its tree-based exploration consolidates semantically identical reasoning steps, reducing redundant explorations while maintaining or even improving accuracy. Our extensive experiments demonstrate that SEAG significantly improves accuracy by 4.3% on average while requiring only 31% of computational costs compared to existing tree search-based methods on complex reasoning benchmarks including GSM8K and ARC with diverse language models such as Llama2, Llama3, and Mistral. Our code is available at https://github.com/ml-postech/SEAG-semantic-exploration-with-adaptive-gating .