Search papers, labs, and topics across Lattice.
Shanghai AI Laboratory
3
0
5
A 4B-parameter model, InternVL-U, punches above its weight, outperforming 14B-parameter models in multimodal generation and editing by using a novel data synthesis pipeline and architecture.
You can now get 4x faster text-to-image generation from masked image models like Lumina-DiMOO, without sacrificing quality, by predicting feature evolution.
A 5B model just crushed the image generation and editing performance of models 5-16x larger, thanks to smarter feature fusion and a novel RL training strategy.
