Research
Read note
Wrong Answer, Different Wrong Answer
Vision-language models are being used for action tracking now. I built a test to check whether one was really watching, and the number it gave me turned out to mean nothing.