AIA: Rethinking Architecture Decoupling Strategy In Unified Multimodal Model

Open paper
Benchmark Model Rank Results
visual-question-answering-on-mm-vetJanus-Pro + AIA#68GPT-4 score: 49.8
text-to-image-generation-on-genevalJanus-Pro + AIA#95Overall: 0.81
multimodal-reasoning-on-mmmuJanus-Pro + AIA#119Accuracy: 42.1

Rank counts only results with a code link.