Skip-Thinking：分块思维链蒸馏使小型语言模型推理更优更快
Skip-Thinking: Chunk-wise Chain-of-Thought Distillation Enable Smaller Language Models to Reason Better and Faster
AI总结：
提出分块训练（CWT）与跳跃思维训练（STT），通过将长思维链划分为语义块并仅学习推理块，使小型语言模型推理更快更准。
摘要：
思维链（CoT）蒸馏允许大型语言模型（LLM）在推理任务中指导小型语言模型（SLM）。现有方法训练SLM在一次迭代中学习完整的长篇推理过程，导致两个问题：1）长推理过程导致训练时token级批次大小过大，使得核心推理token（即直接影响后续推理正确性的token）的梯度被过度平滑，因为它们在推理过程中占比极小。结果，SLM收敛到尖锐最小值，无法掌握推理逻辑。2）响应速度慢，因为SLM必须先生成长篇推理过程才能到达答案。因此，我们提出分块训练（CWT），它使用启发式搜索将推理过程划分为内部语义连贯的块，并让SLM每次迭代仅学习一个块。通过这种方式，CWT自然地将不涉及核心推理token的非推理块（例如总结块和过渡块）从SLM的推理块学习中隔离出来，使得核心推理token在相应迭代中的占比增加。基于CWT，我们提出了跳跃思维训练（STT）。STT使SLM自动跳过非推理的中间块以到达答案，在保持准确性的同时提高推理速度。我们在多种SLM和多个推理任务上验证了我们的方法。
英文摘要：
Chain-of-thought (CoT) distillation allows a large language model (LLM) to guide a small language model (SLM) in reasoning tasks. Existing methods train the SLM to learn the long rationale in one iteration, resulting in two issues: 1) Long rationales lead to a large token-level batch size during training, making gradients of core reasoning tokens (i.e., the token will directly affect the correctness of subsequent reasoning) over-smoothed as they contribute a tiny fraction of the rationale. As a result, the SLM converges to sharp minima where it fails to grasp the reasoning logic. 2) The response is slow, as the SLM must generate a long rationale before reaching the answer. Therefore, we propose chunk-wise training (CWT), which uses a heuristic search to divide the rationale into internal semantically coherent chunks and focuses SLM on learning from only one chunk per iteration. In this way, CWT naturally isolates non-reasoning chunks that do not involve the core reasoning token (e.g., summary and transitional chunks) from the SLM learning for reasoning chunks, making the fraction of the core reasoning token increase in the corresponding iteration. Based on CWT, skip-thinking training (STT) is proposed. STT makes the SLM automatically skip non-reasoning medium chunks to reach the answer, improving reasoning speed while maintaining accuracy. We validate our approach on a variety of SLMs and multiple reasoning tasks.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
