Tools · Hugging Face Blog ·
Native-speed vLLM transformers modeling backend
Hugging Face announced a native-speed Transformers modeling backend built with vLLM. The update aims to improve inference performance for Transformers models by integrating vLLM more directly into the stack.