Breaking the Lock-in: Diversifying Text-to-Image Generation via Representation ModulationDahee Kwon, Ha-Eun Lee, and Jaesik ChoiInternational Conference on Machine Learning (ICML), 2026.Generative models📄Paper
