From Parameters to Data: A Task-Parameter-Guided Fine-Tuning Pipeline for Efficient LLM Alignment
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917518066581504 |
|---|---|
| author | Chen, Hao Zhang, Qi Li, Liyao Shen, Zhanming Ye, Wentao Gao, Lirong Wang, Ningtao Fu, Xing Shen, Xiaoyu Zhao, Junbo |
| author_facet | Chen, Hao Zhang, Qi Li, Liyao Shen, Zhanming Ye, Wentao Gao, Lirong Wang, Ningtao Fu, Xing Shen, Xiaoyu Zhao, Junbo |
| contents | Adapting Large Language Models (LLMs) to specialized domains typically incurs high data and computational overhead. While prior efficiency efforts have largely treated data selection and parameter-efficient fine-tuning as isolated processes, our empirical analysis suggests they may be intrinsically coupled. We posit the Strong Map Hypothesis: a sparse subset of attention heads plays a dominant role in task-specific adaptation, acting as keys that unlock specific data patterns. Building on this observation, we propose From Parameters to Data (P2D), a unified framework that leverages these task-sensitive attention heads as a dual compass for both sample mining and structural pruning. To rigorously quantify the total pipeline cost, we introduce the Alignment Efficiency Ratio (AER) metric for both selection latency and training time. Mechanistically, P2D identifies critical heads via a lightweight proxy and uses them as a functional filter to curate high-affinity data, establishing a synergistic pipeline. Empirically, by updating merely 10% of attention heads on 10% of the data, P2D achieves an 8.3 pp performance gain over strong baselines and delivers a 7.0x end-to-end time speedup. These results validate that precise parameter-data synchronization eliminates redundancy, offering a new paradigm for efficient alignment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_21558 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | From Parameters to Data: A Task-Parameter-Guided Fine-Tuning Pipeline for Efficient LLM Alignment Chen, Hao Zhang, Qi Li, Liyao Shen, Zhanming Ye, Wentao Gao, Lirong Wang, Ningtao Fu, Xing Shen, Xiaoyu Zhao, Junbo Machine Learning Computation and Language Adapting Large Language Models (LLMs) to specialized domains typically incurs high data and computational overhead. While prior efficiency efforts have largely treated data selection and parameter-efficient fine-tuning as isolated processes, our empirical analysis suggests they may be intrinsically coupled. We posit the Strong Map Hypothesis: a sparse subset of attention heads plays a dominant role in task-specific adaptation, acting as keys that unlock specific data patterns. Building on this observation, we propose From Parameters to Data (P2D), a unified framework that leverages these task-sensitive attention heads as a dual compass for both sample mining and structural pruning. To rigorously quantify the total pipeline cost, we introduce the Alignment Efficiency Ratio (AER) metric for both selection latency and training time. Mechanistically, P2D identifies critical heads via a lightweight proxy and uses them as a functional filter to curate high-affinity data, establishing a synergistic pipeline. Empirically, by updating merely 10% of attention heads on 10% of the data, P2D achieves an 8.3 pp performance gain over strong baselines and delivers a 7.0x end-to-end time speedup. These results validate that precise parameter-data synchronization eliminates redundancy, offering a new paradigm for efficient alignment. |
| title | From Parameters to Data: A Task-Parameter-Guided Fine-Tuning Pipeline for Efficient LLM Alignment |
| topic | Machine Learning Computation and Language |
| url | https://arxiv.org/abs/2605.21558 |