Questions for the seminar Paper "Adding Conditional Control to Text-to-Image Diffusion Models"
-----------------------------------------------------------------------------------------------
Please send your answers to: sommerl@cs.uni-freiburg.de

1) What do the three different tables in the main paper measure? (~3 sentences)
2) What limitations of text-to-image diffusion models is the proposed approach addressing? (~1 sentence)
3) What is the main argument for zero initialized convolutions? (~1 sentences)