1 paper
Shintaro Ozaki, Tomoyuki Jinno, Kazuki Hayashi +6
When generating images from prompts that include specific entities, the model must retain as much entity-specific knowledge as possible. However, the number of entities is almost c…