Consistency LLM: converting LLMs to parallel decoders accelerates inference 3.5x (hao-ai-lab.github.io)
Artificial Intelligence, Inference Optimization, Parallel Computing
461 points by zhisbug 197 days ago | 98 comments