arXiv 2502.168062025-03-01cs.CL

CoT2Align：通过最优传输对齐为具有不同分词器的语言模型进行跨思维链蒸馏

CoT2Align: Cross-Chain of Thought Distillation via Optimal Transport Alignment for Language Models with Different Tokenizers

Hanoi University of Science and Technology（河内理工大学）；Oraichain Labs Inc.（Oraichain实验室公司）；ByteDance Inc（字节跳动公司）；VNU University of Engineering and Technology（越南国家大学工程技术大学）；Monash University（莫纳什大学）；University of Oregon（俄勒冈大学）

Anh Duc Le, Tu Vu, Nam Le Hai, Nguyen Thi Ngoc Diep, Linh Ngo Van, Trung Le, Thien Huu Nguyen

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出CoT2Align通用知识蒸馏框架，通过集成思维链增强与Cross-CoT Alignment，结合序列级和层级最优传输对齐，解决了不同分词器模型间的推理感知蒸馏问题，显著提升了推理能力与鲁棒性。

摘要：

大型语言模型（LLMs）在各种NLP任务中取得了最先进的性能，但由于高昂的计算成本和内存限制，面临部署挑战。知识蒸馏（KD）是一种有前景的解决方案，将知识从大型教师模型转移到较小的学生模型。然而，现有的KD方法通常假设共享词汇表和分词器，限制了其灵活性。虽然Universal Logit Distillation（ULD）和Dual-Space Knowledge Distillation（DSKD）等方法解决了词汇不匹配问题，但它们忽略了关键的推理感知蒸馏方面。为了弥合这一差距，我们提出了CoT2Align，这是一个通用KD框架，集成了思维链（CoT）增强，并引入了Cross-CoT Alignment以增强推理迁移。此外，我们将Optimal Transport扩展到超越逐词对齐，成为一种序列级和层级的对齐方法，该方法能够适应不同的序列长度，同时保持上下文完整性。综合实验表明，CoT2Align在不同的词汇设置下优于现有的KD方法，提高了推理能力和领域特定任务的鲁棒性。

英文摘要：

Large Language Models (LLMs) achieve state-of-the-art performance across various NLP tasks but face deployment challenges due to high computational costs and memory constraints. Knowledge distillation (KD) is a promising solution, transferring knowledge from large teacher models to smaller student models. However, existing KD methods often assume shared vocabularies and tokenizers, limiting their flexibility. While approaches like Universal Logit Distillation (ULD) and Dual-Space Knowledge Distillation (DSKD) address vocabulary mismatches, they overlook the critical \textbf{reasoning-aware distillation} aspect. To bridge this gap, we propose CoT2Align a universal KD framework that integrates Chain-of-Thought (CoT) augmentation and introduces Cross-CoT Alignment to enhance reasoning transfer. Additionally, we extend Optimal Transport beyond token-wise alignment to a sequence-level and layer-wise alignment approach that adapts to varying sequence lengths while preserving contextual integrity. Comprehensive experiments demonstrate that CoT2Align outperforms existing KD methods across different vocabulary settings, improving reasoning capabilities and robustness in domain-specific 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。