Chain-of-Visual-Thought: Teaching VLMs to See and Think Better with Continuous Visual Tokens Paper • 2511.19418 • Published Nov 24 • 27