Home
This webpage is created by Reliable AI Lab at the University of Maryland, advised by Prof. Soheil Feizi. Contributors include Keivan Rezaei, Samyadeep Basu, Sriram Balasubramanian, and Arman Zarei.
We showcase research on the interpretability of text-to-image and multi-modal generative models, with applications in compositionality, reducing spurious correlations, knowledge updating in MLLMs, and editing text-to-image models to prevent specific artistic or copyrighted content generation.
Posts
subscribe via RSS