The development of multimodal foundation models relies heavily on OpenAI's CLIP vision encoder. It was so exciting to see the release of OpenVision in May, and now the even more powerful and scalable OpenVision2 ( by @cihangxie team) is here. That's so cooooooooool! 🚀(Young)