Flow-DPO: Improving LLM Mathematical Reasoning through Online Multi-Agent Learning

TL;DR

Flow-DPO enhances LLM mathematical reasoning by using online multi-agent collaboration and preference optimization to generate better reasoning traces for fine-tuning.

Contribution

This work introduces a novel online multi-agent learning framework with DPO for improving reasoning trace quality in LLMs.

Findings

Improved reasoning trace quality over direct inference methods

Effective online training with DPO in collaborative multi-agent setup

Enhanced LLM performance on mathematical reasoning tasks

Abstract

Mathematical reasoning is a crucial capability for Large Language Models (LLMs), yet generating detailed and accurate reasoning traces remains a significant challenge. This paper introduces a novel approach to produce high-quality reasoning traces for LLM fine-tuning using online learning \textbf{Flows}. Our method employs an incremental output production Flow, where component LLMs collaboratively construct solutions through iterative communication. We train the Flow using online Direct Preference Optimization (DPO) learning with rollouts, generating DPO pairs for each training example and updating models in real-time. We directly compare the quality of reasoning traces generated by our method with those produced through direct model inference, demonstrating the effectiveness of our approach in improving LLM performance in mathematical reasoning tasks.