Privacy-preserved LLM Cascade via CoT-enhanced Policy Learning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913712052371456 |
|---|---|
| author | Zhang, Kai Wang, Congchao Peng, Liqian Go, Alec Liu, Xiaozhong |
| author_facet | Zhang, Kai Wang, Congchao Peng, Liqian Go, Alec Liu, Xiaozhong |
| contents | Large Language Models (LLMs) have gained significant attention in on-device applications due to their remarkable performance across real-world tasks. However, on-device LLMs often suffer from suboptimal performance due to hardware limitations. A promising solution to this challenge is cascading a weaker local (on-device) LLM with a more powerful server LLM. While existing research on LLM cascade primarily optimizes the performance-cost trade-off, real-world applications impose additional requirements, such as privacy preservation, which remain largely unaddressed. In this work, we move beyond existing confidence- and logit-based LLM cascade methods and propose $\mathbf{P^{3}Defer}$, a novel Chain-of-Thought (CoT)-enhanced \textbf{p}olicy learning framework for \textbf{p}rivacy-\textbf{p}reserved \textbf{defer}ral decision-making. Our approach effectively improves cascade efficiency while mitigating privacy risks. Extensive experiments on three benchmark datasets demonstrate the effectiveness and superiority of $\mathbf{P^{3}Defer}$ over existing methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_08014 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Privacy-preserved LLM Cascade via CoT-enhanced Policy Learning Zhang, Kai Wang, Congchao Peng, Liqian Go, Alec Liu, Xiaozhong Computation and Language Large Language Models (LLMs) have gained significant attention in on-device applications due to their remarkable performance across real-world tasks. However, on-device LLMs often suffer from suboptimal performance due to hardware limitations. A promising solution to this challenge is cascading a weaker local (on-device) LLM with a more powerful server LLM. While existing research on LLM cascade primarily optimizes the performance-cost trade-off, real-world applications impose additional requirements, such as privacy preservation, which remain largely unaddressed. In this work, we move beyond existing confidence- and logit-based LLM cascade methods and propose $\mathbf{P^{3}Defer}$, a novel Chain-of-Thought (CoT)-enhanced \textbf{p}olicy learning framework for \textbf{p}rivacy-\textbf{p}reserved \textbf{defer}ral decision-making. Our approach effectively improves cascade efficiency while mitigating privacy risks. Extensive experiments on three benchmark datasets demonstrate the effectiveness and superiority of $\mathbf{P^{3}Defer}$ over existing methods. |
| title | Privacy-preserved LLM Cascade via CoT-enhanced Policy Learning |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2410.08014 |