LoMA: Lossless Compressed Memory Attention

Large Language Models (LLMs) face limitations due to the high demand on GPU memory and computational resources when handling long contexts. While sparsify the Key-Value (KV) cache of transformer model is a typical strategy to alleviate resource usage, it unavoidably results in the loss of information. We introduce Lossless Compressed Memory Attention (LoMA), a novel approach that enables lossless compression of the KV cache, thereby reducing the memory and computational demands during autoregressive generation. LoMA incorporates a specialized training or …

attention cache computational cs.cl cs.lg demand face gpu information key language language models large language large language models limitations llms loss memory novel resources strategy the key transformer transformer model usage value