Are vision language models robust to uncertain inputs?

AAML
VLM
Re-assign community