Enhancing and Accelerating Large Language Models via Instruction-Aware Contextual Compression

Haowen Hou, Fei Ma, Binwen Bai +2

Large Language Models (LLMs) have garnered widespread attention due to their remarkable performance across various tasks. However, to mitigate the issue of hallucinations, LLMs oft…