A Method for Building Large Language Models with Predefined KV Cache Capacity (2411.15785v2)

Abstract: This paper introduces a novel approach, the Bounded-Cache Transformer (BCT), for building LLMs with a predefined Key-Value (KV) cache capacity. The BCT addresses the excessive memory consumption issue in traditional KV caches by implementing a bounded-length KV cache, which is particularly suitable for the attention layers in Transformer decode-only architectures. By dynamically updating the key-value vector sequences, the BCT achieves efficient inference within limited cache capacity, significantly reducing memory usage while maintaining model performance and system throughput. Experimental results demonstrate that the BCT significantly reduces memory usage while maintaining the model's inference quality, offering a new solution for efficient inference in LLMs.