Knowledge Fusion of Large Language Models Via Modular SkillPacks
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914351186706432 |
|---|---|
| author | Du, Guodong Li, Zhuo Zhou, Xuanning Li, Junlin Shi, Zesheng Lin, Wanyu Tang, Ho-Kin Li, Xiucheng Liu, Fangming Wang, Wenya Zhang, Min Li, Jing |
| author_facet | Du, Guodong Li, Zhuo Zhou, Xuanning Li, Junlin Shi, Zesheng Lin, Wanyu Tang, Ho-Kin Li, Xiucheng Liu, Fangming Wang, Wenya Zhang, Min Li, Jing |
| contents | Cross-capability transfer is a key challenge in large language model (LLM) research, with applications in multi-task integration, model compression, and continual learning. Recent works like FuseLLM and FuseChat have demonstrated the potential of transferring multiple model capabilities to lightweight models, enhancing adaptability and efficiency, which motivates our investigation into more efficient cross-capability transfer methods. However, existing approaches primarily focus on small, homogeneous models, limiting their applicability. For large, heterogeneous models, knowledge distillation with full-parameter fine-tuning often overlooks the student model's intrinsic capacity and risks catastrophic forgetting, while PEFT methods struggle to effectively absorb knowledge from source LLMs. To address these issues, we introduce GraftLLM, a novel method that stores source model capabilities in a target model with SkillPack format. This approach preserves general capabilities, reduces parameter conflicts, and supports forget-free continual learning and model fusion. We employ a module-aware adaptive compression strategy to compress parameter updates, ensuring efficient storage while maintaining task-specific knowledge. The resulting SkillPack serves as a compact and transferable knowledge carrier, ideal for heterogeneous model fusion and continual learning. Experiments across various scenarios demonstrate that GraftLLM outperforms existing techniques in knowledge transfer, knowledge fusion, and forget-free learning, providing a scalable and efficient solution for cross-capability transfer. The code is publicly available at: https://github.com/duguodong7/GraftLLM. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_18502 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Knowledge Fusion of Large Language Models Via Modular SkillPacks Du, Guodong Li, Zhuo Zhou, Xuanning Li, Junlin Shi, Zesheng Lin, Wanyu Tang, Ho-Kin Li, Xiucheng Liu, Fangming Wang, Wenya Zhang, Min Li, Jing Artificial Intelligence Computation and Language Machine Learning Cross-capability transfer is a key challenge in large language model (LLM) research, with applications in multi-task integration, model compression, and continual learning. Recent works like FuseLLM and FuseChat have demonstrated the potential of transferring multiple model capabilities to lightweight models, enhancing adaptability and efficiency, which motivates our investigation into more efficient cross-capability transfer methods. However, existing approaches primarily focus on small, homogeneous models, limiting their applicability. For large, heterogeneous models, knowledge distillation with full-parameter fine-tuning often overlooks the student model's intrinsic capacity and risks catastrophic forgetting, while PEFT methods struggle to effectively absorb knowledge from source LLMs. To address these issues, we introduce GraftLLM, a novel method that stores source model capabilities in a target model with SkillPack format. This approach preserves general capabilities, reduces parameter conflicts, and supports forget-free continual learning and model fusion. We employ a module-aware adaptive compression strategy to compress parameter updates, ensuring efficient storage while maintaining task-specific knowledge. The resulting SkillPack serves as a compact and transferable knowledge carrier, ideal for heterogeneous model fusion and continual learning. Experiments across various scenarios demonstrate that GraftLLM outperforms existing techniques in knowledge transfer, knowledge fusion, and forget-free learning, providing a scalable and efficient solution for cross-capability transfer. The code is publicly available at: https://github.com/duguodong7/GraftLLM. |
| title | Knowledge Fusion of Large Language Models Via Modular SkillPacks |
| topic | Artificial Intelligence Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2505.18502 |