Multi-LLM Collaborative Search for Complex Problem Solving
This addresses the limitation of single-model approaches in AI reasoning for complex tasks like mathematics and commonsense, though it appears incremental as it builds on existing multi-agent and MCTS methods.
The paper tackles the problem of LLMs struggling with complex reasoning tasks by proposing the Mixture-of-Search-Agents (MoSA) paradigm, which leverages multiple LLMs to enhance search-based reasoning, resulting in improved accuracy across four reasoning benchmarks.
Large language models (LLMs) often struggle with complex reasoning tasks due to their limitations in addressing the vast reasoning space and inherent ambiguities of natural language. We propose the Mixture-of-Search-Agents (MoSA) paradigm, a novel approach leveraging the collective expertise of multiple LLMs to enhance search-based reasoning. MoSA integrates diverse reasoning pathways by combining independent exploration with iterative refinement among LLMs, mitigating the limitations of single-model approaches. Using Monte Carlo Tree Search (MCTS) as a backbone, MoSA enables multiple agents to propose and aggregate reasoning steps, resulting in improved accuracy. Our comprehensive evaluation across four reasoning benchmarks demonstrates MoSA's consistent performance improvements over single-agent and other multi-agent baselines, particularly in complex mathematical and commonsense reasoning tasks.