MoE-LLaVA: Mixture of Experts for Large Vision-Language Models

Benchmark Model Rank Results
visual-question-answering-on-mm-vetMoE-LLaVA-2.7B×4-Top2#104GPT-4 score: 35.9