小模型在有限计算下实现高效多步推理
Dual-Track CoT: Budget-Aware Stepwise Guidance for Small LMs
- 双轨思维链：分步引导+预算控制，提升小模型推理效率
- 相同令牌数下，准确率显著优于传统采样方法
- 适合资源受限场景，如手机端或低延迟应用
大语言模型（LLMs）通过思维链（CoT）提示解决诸多推理任务，但参数量约7到80亿的小型语言模型（SLMs）在计算和令牌预算紧张时仍难以完成多步推理。现有测试时推理方法如自洽性（多次采样并投票）、思维树（搜索中间思考路径）和批判修正循环虽能提升性能，但通常代价高昂且缺乏细粒度的步骤级控制。本研究旨在填补这一空白：小型语言模型能否在相同或更少令牌消耗下实现可靠推理？该问题兼具科学与实际意义。科学上，它探讨过程监督与简单测试时控制（如令牌预算、冗余步骤剔除）是否可替代模型规模或大量采样。实际上，许多部署场景（如设备端、低延迟或成本受限环境）无法承担大模型或每查询数十次采样。因此，在固定成本下提升小模型推理能力具有直接实用性。
原文摘要 · Abstract (English)
Large Language Models (LLMs) solve many reasoning tasks via chain-of-thought (CoT) prompting, but smaller models (about 7 to 8B parameters) still struggle with multi-step reasoning under tight compute and token budgets. Existing test time reasoning methods such as self consistency (sampling multiple rationales and voting), Tree-of-Thoughts (search over intermediate thoughts), and critique revise loops improve performance, but often at high token cost and without fine-grained step-level control. This project1 aims to address that gap: can Small Language Models (SLMs) reason reliably using the same or fewer tokens? This question is both scientific and practical. Scientifically, it probes whether process supervision and simple test-time controls (such as token budgets and rejection of redundant steps) can substitute for model scale or large sampling counts. Practically, many deployments (on-device, low-latency, or cost-constrained settings) cannot afford huge models or dozens of sampled rationales per query. A method that improves SLM reasoning at fixed cost would therefore be directly useful.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。