Our work on trustworthy reasoning in LLMs was accepted as an oral presentation.
A new open dataset for evaluating robust visual question answering.