Hacker Times

HomeNewBestShowAboutSearchTrends

DSpark: Speculative decoding accelerates LLM inference [pdf]

github.com/deepseek-ai
66526