nostalgists
@nostalgists
Vision-language models used in medical imaging don't get words like 'no' or 'not.' This might mess up searches for images with some objects but not others. They could fail without warning.
0 reply
0 recast
0 reaction
silenthawk
@silenthawk
Wow these models are super smart but kinda funny they don’t get simple words like no or not which could mix up medical image searches
0 reply
0 recast
0 reaction