Hai, Nam LeLe, Anh DucVu, TuVan, Linh Ngo2026-10-162026-10-160950-7051WOS:001824873800001https://hdl.handle.net/1885/733816306Large 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 MultiLevelOT, Universal Logit Distillation (ULD) and Dual-Space Knowledge Distillation (DSKD) address vocabulary mismatches, they overlook the critical reasoning-aware distillation aspect. To bridge this gap, we propose DISCORD, 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 DISCORD outperforms existing KD methods across different vocabulary settings, improving reasoning capabilities and robustness in both domain-specific and cross-task scenarios.This work is funded by the Ministry of Education and Training of Vietnam under Contract Number B2026–BKA–01 .12enPublisher Copyright: © 2026 Elsevier B.V.Chain-of-thoughtDifferent tokenizersKnowledge distillationOptimal transport2026-10-0910.1016/j.knosys.2026.116606105046476316