Deconstructing Long Chain-of-Thought: A Structured Reasoning Optimization Framework for Long CoT Distillation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909545862791168 |
|---|---|
| author | Luo, Yijia Song, Yulin Zhang, Xingyao Liu, Jiaheng Wang, Weixun Chen, GengRu Su, Wenbo Zheng, Bo |
| author_facet | Luo, Yijia Song, Yulin Zhang, Xingyao Liu, Jiaheng Wang, Weixun Chen, GengRu Su, Wenbo Zheng, Bo |
| contents | Recent advancements in large language models (LLMs) have demonstrated remarkable reasoning capabilities through long chain-of-thought (CoT) reasoning. The R1 distillation scheme has emerged as a promising approach for training cost-effective models with enhanced reasoning abilities. However, the underlying mechanisms driving its effectiveness remain unclear. This study examines the universality of distillation data and identifies key components that enable the efficient transfer of long-chain reasoning capabilities in LLM distillation. Our findings reveal that the effectiveness of long CoT reasoning distillation from teacher models like Qwen-QwQ degrades significantly on nonhomologous models, challenging the assumed universality of current distillation methods. To gain deeper insights into the structure and patterns of long CoT reasoning, we propose DLCoT (Deconstructing Long Chain-of-Thought), a distillation data enhancement framework. DLCoT consists of three key steps: (1) data segmentation to decompose complex long CoT structures, (2) simplification by eliminating unsolvable and redundant solutions, and (3) optimization of intermediate error states. Our approach significantly improves model performance and token efficiency, facilitating the development of high-performance LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_16385 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Deconstructing Long Chain-of-Thought: A Structured Reasoning Optimization Framework for Long CoT Distillation Luo, Yijia Song, Yulin Zhang, Xingyao Liu, Jiaheng Wang, Weixun Chen, GengRu Su, Wenbo Zheng, Bo Artificial Intelligence Recent advancements in large language models (LLMs) have demonstrated remarkable reasoning capabilities through long chain-of-thought (CoT) reasoning. The R1 distillation scheme has emerged as a promising approach for training cost-effective models with enhanced reasoning abilities. However, the underlying mechanisms driving its effectiveness remain unclear. This study examines the universality of distillation data and identifies key components that enable the efficient transfer of long-chain reasoning capabilities in LLM distillation. Our findings reveal that the effectiveness of long CoT reasoning distillation from teacher models like Qwen-QwQ degrades significantly on nonhomologous models, challenging the assumed universality of current distillation methods. To gain deeper insights into the structure and patterns of long CoT reasoning, we propose DLCoT (Deconstructing Long Chain-of-Thought), a distillation data enhancement framework. DLCoT consists of three key steps: (1) data segmentation to decompose complex long CoT structures, (2) simplification by eliminating unsolvable and redundant solutions, and (3) optimization of intermediate error states. Our approach significantly improves model performance and token efficiency, facilitating the development of high-performance LLMs. |
| title | Deconstructing Long Chain-of-Thought: A Structured Reasoning Optimization Framework for Long CoT Distillation |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2503.16385 |