{"ID":2855289,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.13935","arxiv_id":"2510.13935","title":"Big Reasoning with Small Models: Instruction Retrieval at Inference Time","abstract":"Small language models (SLMs) enable low-cost, private, on-device inference, but they often fail on problems that require specialized domain knowledge or multi-step reasoning. Existing approaches for improving reasoning either rely on scale (e.g., chain-of-thought prompting), require task-specific training that limits reuse and generality (e.g., distillation), or retrieve unstructured information that still leaves the SLM to determine an appropriate reasoning strategy. We propose instruction retrieval, an inference-time intervention that augments an SLM with structured, reusable reasoning procedures rather than raw passages. We construct an Instruction Corpus by clustering similar training questions and using a teacher model to generate generalizable guides that pair domain background with explicit step-by-step procedures. At inference, the SLM retrieves the instructions most relevant to a given query and executes the associated procedures without any additional fine-tuning. Across three challenging domains: medicine, law, and mathematics, instruction retrieval yields consistent gains for models with at least 3B parameters, improving accuracy by 9.4%, 7.9%, and 5.1%, respectively, with the strongest 14B model surpassing GPT-4o's zero-shot performance on knowledge-intensive tasks.","short_abstract":"Small language models (SLMs) enable low-cost, private, on-device inference, but they often fail on problems that require specialized domain knowledge or multi-step reasoning. Existing approaches for improving reasoning either rely on scale (e.g., chain-of-thought prompting), require task-specific training that limits r...","url_abs":"https://arxiv.org/abs/2510.13935","url_pdf":"https://arxiv.org/pdf/2510.13935v2","authors":"[\"Kenan Alkiek\",\"David Jurgens\",\"Vinod Vydiswaran\"]","published":"2025-10-15T15:51:13Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.AI\"]","methods":"[\"Language Model\"]","has_code":false}