UniDDT: Unifying Multimodal Understanding and Generation with Decoupled Diffusion Transformer
Open paper
Benchmark
Model
Rank
Results
multimodal-reasoning-on-mmmu
VLM-UniDDT
–
Accuracy: 52.6
text-to-image-generation-on-geneval
VLM-UniDDT
–
Overall: 0.87
vision-language-understanding-on-mmstar
VLM-UniDDT
–
Accuracy: 57.7
Rank counts only results with a code link.