In plain words: Instead of keeping a memory of every past word, this design squeezes earlier text into one anchor token carrying the context forward. It kept nearly the same accuracy on question answering while cutting stored memory up to 99% and running up to 3.5 times faster.
Abstract · Anchor-based Large Language Models
Large language models (LLMs) predominantly employ decoder-only transformer architectures, necessitating the retention of keys/values information for historical tokens to provide contextual information and avoid redundant computation. However, the substantial size and parameter volume of these LLMs require massive GPU memory. This memory demand increases with the length of the input text, leading to an urgent need for more efficient methods of information storage and processing. This study introduces Anchor-based LLMs (AnLLMs), which utilize an innovative anchor-based self-attention network (AnSAN) and also an anchor-based inference strategy. This approach enables LLMs to compress sequence information into an anchor token, reducing the keys/values cache and enhancing inference efficiency. Experiments on question-answering benchmarks reveal that AnLLMs maintain similar accuracy levels while achieving up to 99% keys/values cache reduction and up to 3.5 times faster inference. Despite a minor compromise in accuracy, the substantial enhancements of AnLLMs employing the AnSAN technique in resource utilization and computational efficiency underscore their potential for practical LLM applications.
Jianhui Pang, Fanghua Ye, Derek Fai Wong, Xin He, Wanshun Chen, Longyue Wang
arXiv:2402.07616 · cs.CL, cs.AI · submitted Feb 12, 2024 · updated Jun 1, 2024
abstract · pdf · html · The paper has been accepted by the ACL2024 conference. Work was done when Jianhui Pang and Fanghua Ye were interning at Tencent AI Lab
