FutureMind: Equipping Small Language Models with Strategic Thinking-Pattern Priors via Adaptive Knowledge Distillation

TL;DR

FutureMind enhances small language models with strategic reasoning abilities through adaptive knowledge distillation, enabling efficient handling of complex, knowledge-intensive tasks with improved accuracy and reasoning skills.

Contribution

The paper introduces a modular reasoning framework that equips small language models with structured thinking-pattern priors via adaptive knowledge distillation from large models.

Findings

Outperforms strong baselines on multi-hop QA benchmarks

Achieves state-of-the-art results across diverse SLM architectures

Reveals cognitive bias bottleneck in reasoning skill transfer

Abstract

Small Language Models (SLMs) are attractive for cost-sensitive and resource-limited settings due to their efficient, low-latency inference. However, they often struggle with complex, knowledge-intensive tasks that require structured reasoning and effective retrieval. To address these limitations, we propose FutureMind, a modular reasoning framework that equips SLMs with strategic thinking-pattern priors via adaptive knowledge distillation from large language models (LLMs). FutureMind introduces a dynamic reasoning pipeline composed of four key modules: Problem Analysis, Logical Reasoning, Strategy Planning, and Retrieval Guidance. This pipeline is augmented by three distinct retrieval paradigms that decompose complex queries into tractable subproblems, ensuring efficient and accurate retrieval execution. Extensive experiments on multi-hop QA benchmarks, including 2WikiMultihopQA, MuSiQue, Bamboogle, and Frames, demonstrate the superiority of FutureMind. It consistently outperforms strong baselines such as Search-o1, achieving state-of-the-art results under free training conditions across diverse SLM architectures and scales. Beyond empirical gains, our analysis reveals that the process of thinking-pattern distillation is restricted by the cognitive bias bottleneck between the teacher (LLMs) and student (SLMs) models. This provides new perspectives on the transferability of reasoning skills, paving the way for the development of SLMs that combine efficiency with genuine cognitive capability.