用合成连续思维提升大模型推理效率与准确率
SynAdapt: Learning Adaptive Reasoning in Large Language Models via Synthetic Continuous Chain-of-Thought
- 生成合成连续思维作为精准对齐目标，指导模型直接学习连续推理
- 在多个难度级别的基准上实现最优准确率与效率平衡
- 自适应识别难题并触发重思考，适合追求高效推理的场景
尽管思维链（CoT）推理能提升模型性能，但因其生成离散思维链（DCoT） token 而带来显著时间开销。连续思维链（CCoT）是更高效的替代方案，但现有方法受限于间接微调、对齐不足或目标不一致。为此，我们提出 SynAdapt——一种创新的高效推理框架。SynAdapt 生成合成连续思维（synthetic CCoT）作为大语言模型（LLM）的精确对齐目标，明确引导模型学习连续推理并直接得出准确答案。此外，仅依赖 CCoT 无法有效解决难题。为此，SynAdapt 引入难度分类器，结合问题上下文与 CCoT 判定难题。经过简短推理后，该分类器可有效识别难题，并自适应地提示模型重新思考，从而提升性能。大量实验结果表明，该方法在多个不同难度级别的基准上均表现优异，实现了最佳的准确率-效率权衡。
原文摘要 · Abstract (English)
While Chain-of-Thought (CoT) reasoning improves model performance, it incurs significant time costs due to the generation of discrete CoT tokens (DCoT). Continuous CoT (CCoT) offers a more efficient alternative, but existing CCoT methods are hampered by indirect fine-tuning, limited alignment, or inconsistent targets. To overcome these limitations, we propose \textit{SynAdapt}, an innovative efficient reasoning framework. Specifically, \textit{SynAdapt} generates the synthetic CCoT to serve as a precise and effective alignment target for LLMs. This synthetic CCoT explicitly guides the LLM to learn CCoT and derive accurate answers directly. Furthermore, relying solely on CCoT is insufficient for solving hard questions. To address this, \textit{SynAdapt} integrates a difficulty classifier that leverages both question context and CCoT to identify hard questions. CCoT can effectively help identify hard questions after some brief reasoning. We then adaptively prompt the LLM to re-think these hard questions for improved performance. Extensive experimental results across various benchmarks from different difficulty levels strongly demonstrate the effectiveness of our method, achieving the best accuracy-efficiency trade-off.
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。