Fetching the paper…

Break the Sequential Dependency of LLM Inference Using Lookahead Decoding · Around