Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912665384779776 |
|---|---|
| author | Fu, Yu Cai, Zefan Asi, Abedelkadir Xiong, Wayne Dong, Yue Xiao, Wen |
| author_facet | Fu, Yu Cai, Zefan Asi, Abedelkadir Xiong, Wayne Dong, Yue Xiao, Wen |
| contents | Key-Value (KV) caching is a common technique to enhance the computational efficiency of Large Language Models (LLMs), but its memory overhead grows rapidly with input length. Prior work has shown that not all tokens are equally important for text generation, proposing layer-level KV cache compression to selectively retain key information. Recognizing the distinct roles of attention heads in generation, we propose HeadKV, a head-level KV cache compression method, and HeadKV-R2, which leverages a novel contextual reasoning ability estimation for compression. Our approach operates at the level of individual heads, estimating their importance for contextual QA tasks that require both retrieval and reasoning capabilities. Extensive experiments across diverse benchmarks (LongBench, LooGLE), model architectures (e.g., Llama-3-8B-Instruct, Mistral-7B-Instruct), and long-context abilities tests demonstrate that our head-level KV cache compression significantly outperforms strong baselines, particularly in low-resource settings (KV size = 64 & 128). Notably, our method retains just 1.5% of the KV cache while achieving 97% of the performance of the full KV cache on the contextual question answering benchmark. Codes are available at https://github.com/FYYFU/HeadKV |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_19258 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning Fu, Yu Cai, Zefan Asi, Abedelkadir Xiong, Wayne Dong, Yue Xiao, Wen Computation and Language Artificial Intelligence Key-Value (KV) caching is a common technique to enhance the computational efficiency of Large Language Models (LLMs), but its memory overhead grows rapidly with input length. Prior work has shown that not all tokens are equally important for text generation, proposing layer-level KV cache compression to selectively retain key information. Recognizing the distinct roles of attention heads in generation, we propose HeadKV, a head-level KV cache compression method, and HeadKV-R2, which leverages a novel contextual reasoning ability estimation for compression. Our approach operates at the level of individual heads, estimating their importance for contextual QA tasks that require both retrieval and reasoning capabilities. Extensive experiments across diverse benchmarks (LongBench, LooGLE), model architectures (e.g., Llama-3-8B-Instruct, Mistral-7B-Instruct), and long-context abilities tests demonstrate that our head-level KV cache compression significantly outperforms strong baselines, particularly in low-resource settings (KV size = 64 & 128). Notably, our method retains just 1.5% of the KV cache while achieving 97% of the performance of the full KV cache on the contextual question answering benchmark. Codes are available at https://github.com/FYYFU/HeadKV |
| title | Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2410.19258 |