visual-generation · vllm-omniA framework for efficient model inference with omni-modality models 原始来源：https://github.com/vllm-project/vllm-omni