BRAVE: Broadening the visual encoding of vision-language models

Open in new window