Enhancing Reasoning Capabilities of Small Language Models with Blueprints and Prompt Template Search

2025-06-10

AI Summary

Small language models (SLMs) suffer from weak reasoning capabilities and high sensitivity to prompt formulation. Method: This paper proposes a parameter-free, lightweight enhancement framework comprising two stages: (1) leveraging a large language model (LLM) as a teacher to autonomously generate structured, high-level reasoning blueprints; and (2) performing discrete optimization over prompt templates to achieve robust zero-shot reasoning orchestration. Contribution/Results: To our knowledge, this is the first work to synergistically integrate LLM-generated abstract reasoning plans with lightweight template search. Evaluated on GSM8K, MBPP, and BBH benchmarks, the method substantially improves SLM performance—approaching that of large models—while incurring minimal computational overhead. Its efficiency and zero-shot adaptability make it particularly suitable for edge deployment and resource-constrained environments.

Technology Category

Planning, Routing, and Scheduling: Planning with Language ModelsMachine Learning: Large Multimodal Models (LMMs)Search and Optimization: Learning to Search

Application Category

Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved informationUser Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendationEconomics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Small language models (SLMs) offer promising and efficient alternatives to large language models (LLMs). However, SLMs' limited capacity restricts their reasoning capabilities and makes them sensitive to prompt variations. To address these challenges, we propose a novel framework that enhances SLM reasoning capabilities through LLM generated blueprints. The blueprints provide structured, high-level reasoning guides that help SLMs systematically tackle related problems. Furthermore, our framework integrates a prompt template search mechanism to mitigate the SLMs' sensitivity to prompt variations. Our framework demonstrates improved SLM performance across various tasks, including math (GSM8K), coding (MBPP), and logic reasoning (BBH). Our approach improves the reasoning capabilities of SLMs without increasing model size or requiring additional training, offering a lightweight and deployment-friendly solution for on-device or resource-constrained environments.

Problem

• Enhancing reasoning capabilities of small language models (SLMs)
• Reducing SLMs' sensitivity to prompt variations
• Improving SLM performance without increasing model size

Innovation

• LLM generated blueprints enhance SLM reasoning
• Prompt template search reduces SLM sensitivity
• Lightweight solution without model size increase