ZSMerge: Zero-Shot KV Cache Compression for Memory-Efficient Long-Context LLMs
This addresses memory efficiency and inference speed issues for users of large language models in long-context processing, representing a strong specific gain rather than a foundational breakthrough.
The paper tackles the memory and computational bottlenecks in long-context LLMs by proposing ZSMerge, a zero-shot KV cache compression framework, which achieves a 20:1 compression ratio and triple throughput gains with negligible performance degradation.
The linear growth of key-value (KV) cache memory and quadratic computational in attention mechanisms complexity pose significant bottlenecks for large language models (LLMs) in long-context processing. While existing KV cache optimization methods address these challenges through token pruning or feature merging, they often incur irreversible information loss or require costly parameter retraining. To this end, we propose ZSMerge, a dynamic KV cache compression framework designed for efficient cache management, featuring three key operations: (1) fine-grained memory allocation guided by multi-dimensional token importance metrics at head-level granularity, (2) a residual merging mechanism that preserves critical context through compensated attention scoring, and (3) a zero-shot adaptation mechanism compatible with diverse LLM architectures without requiring retraining. ZSMerge significantly enhances memory efficiency and inference speed with negligible performance degradation across LLMs. When applied to LLaMA2-7B, it demonstrates a 20:1 compression ratio for key-value cache retention (reducing memory footprint to 5\% of baseline) while sustaining comparable generation quality, coupled with triple throughput gains at extreme 54k-token contexts that eliminate out-of-memory failures. The code is available at https://github.com/SusCom-Lab/ZSMerge.