ReHoPER: Receding-Horizon Planning for Enhanced Reasoning

2026-09-30

AI Summary

This study addresses the limited compositional reasoning capabilities of large language models in the absence of labeled data by proposing a zero-shot reasoning enhancement method. Grounded in receding horizon planning, the approach generates and resolves multi-path intermediate questions, progressively refining the final answer through an iterative replanning mechanism. Requiring only generic instructions without task-specific prompts or annotated examples, it achieves plug-and-play applicability across datasets and models. Experiments demonstrate that the proposed method significantly outperforms strong baselines on multiple benchmarks, with particularly notable gains in complex compositional reasoning scenarios. Furthermore, this work introduces a new benchmark, iLLC, and releases the associated open-source code to facilitate future research.

Abstract

We propose ReHoPER, an inference-only, zero-shot method that improves large language models' reasoning by generating and answering intermediate questions along multiple paths before the final answer. It iteratively plans a horizon of candidate intermediate questions, selects one to answer, and replans from the updated history. ReHoPER is task-agnostic, using the same generic instructions across datasets and models without labeled data or task-specific prompt design. Across multiple datasets, including iLLC, a new controlled benchmark for compositional reasoning, ReHoPER outperforms strong baselines, with the largest gains in the most compositional settings. Our implementation and the iLLC generator are publicly available to support future work.

Problem

• Large Language Models
• Reasoning
• Compositional Reasoning
• Zero-shot Inference

Innovation

• Receding-Horizon Planning
• Zero-shot Reasoning
• Intermediate Questions
• Compositional Reasoning
• Inference-only