Mipha: A Comprehensive Overhaul of Multimodal Assistant with Small Language Models

Benchmark Model Rank Results
visual-question-answering-on-mm-vetMipha-3B#131GPT-4 score: 32.1Params: 3B