1 paper · 2 filters
Sujie Hu, Huaisong Zhang, Pujing Yang +7
Recent reinforcement learning approaches have improved visual text generation, yet generated images still suffer from incorrect wording and malformed glyphs. We identify two couple…