TRAPSBench: Vision-Language Models Encode but Fail to Express Epistemic Restraint

Add code
Aug 13, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: