让小模型学会自我纠错，提升复杂问题求解能力
Project Aletheia: Verifier-Guided Distillation of Backtracking for Small Language Models
- 用验证器引导训练，教小模型如何发现并回溯错误
- 70亿参数模型在包含错误修正的推理轨迹上训练后，可识别矛盾并修正假设
- 适合需要高可靠性、低资源消耗的设备端应用
小型语言模型（SLMs，参数量低于100亿）因其适合私密、本地部署而受到关注，但常因线性且过度自信的推理过程，在需严格满足约束的问题上失败。我们提出验证器引导的蒸馏方法，将错误修复过程——即显式的冲突检测与回溯——作为知识迁移内容，而非仅传递最终正确答案。通过在包含错误与自我修正的验证推理轨迹上训练一个70亿参数模型，我们证明小模型中可涌现出潜在的验证行为，使其能偶尔停止、识别矛盾，并修正先前假设。
原文摘要 · Abstract (English)
Small Language Models (SLMs, under 10B parameters) are attractive for private, on-device deployment, yet they frequently fail on strict constraint-satisfaction problems due to linear, overconfident reasoning traces that do not recover from early mistakes. We introduce Verifier-Guided Distillation, a training protocol that transfers the process of error repair - explicit conflict detection and backtracking - rather than only correct final answers. By training a 7B model on verified reasoning traces that include mistakes and self-corrections, we show that latent verification behavior can emerge in small models, enabling them to occasionally stop, detect contradictions, and revise earlier assumptions.
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。