Submitted by Andres Marafioti 202 SmolVLM: Redefining small and efficient multimodal models Hugging Face 9