A unified library for building, evaluating, and storing speculative decoding algorithms for LLM inference in vLLM
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).