A mixture-of-experts visual question-answering server that enables visual grounding, compositional image question answering, and external knowledge-dependent image question answering through code generation and execution. Built on the ViperGPT framework with support for multiple computer vision mode
View & run on Rokha →