Enhancing Test-Time Scaling of Large Language Models with Hierarchical Retrieval-Augmented MCTS

TL;DR

This paper introduces R2-LLMs, a hierarchical retrieval-augmented reasoning framework that improves test-time scaling of large language models by enhancing in-context learning and step-wise reasoning without needing distillation.

Contribution

The paper presents a novel hierarchical retrieval-augmented reasoning framework, R2-LLMs, that enhances test-time inference in LLMs through dual-level retrieval and MCTS, without requiring CoT training data.

Findings

Up to 16% performance improvement on reasoning benchmarks.

Effective integration of hierarchical retrieval with MCTS.

Robust reasoning enhancement without model distillation.

Abstract

Test-time scaling has emerged as a promising paradigm in language modeling, leveraging additional computational resources at inference time to enhance model performance. In this work, we introduce R2-LLMs, a novel and versatile hierarchical retrieval-augmented reasoning framework designed to improve test-time scaling in large language models (LLMs) without requiring distillation from more advanced models to obtain chain-of-thought (CoT) training data. R2-LLMs enhances inference-time generalization by integrating dual-level retrieval-based in-context learning: (1) At the coarse level, our approach extracts abstract templates from complex reasoning problems and retrieves similar problem-answer pairs to facilitate high-level in-context learning; (2) At the fine level, during Monte Carlo Tree Search (MCTS), R2-LLMs efficiently retrieves analogous intermediate solution steps from reference mathematical problem datasets, refining step-wise reasoning with the aid of a process reward model (PRM) for scoring. R2-LLMs is a robust hierarchical reasoning-augmentation method that enhances in-context-level reasoning while seamlessly integrating with step-level tree search methods. Utilizing PRM, it refines both candidate generation and decision-making for improved reasoning accuracy. Empirical evaluations on the MATH500, GSM8K, and OlympiadBench-TO datasets achieve substantial relative improvement with an increase of up to 16% using LLaMA-3.1-8B compared to the baselines, showcasing the effectiveness of our approach in complex reasoning tasks.