Fetching the paper…

Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision · Around