PipeLLM: Fast and Confidential Large Language Model Services with Speculative Pipelined Encryption.

Yifan Tan, Cheng Tan, Zeyu Mi, Haibo Chen

Browse the full ASPLOS paper archive.