Tweeted By @stanfordnlp

ConVIRT—Contrastive Learning of Visual Representations from Paired Images & Text—by @yuhaozhangx @hjian42 Miura @chrmanning @curtlanglotz pioneers effective use of human-written descriptive text to improve unsupervised contrastive learning from images https://t.co/HuKfDUTrxv pic.twitter.com/Z62R9FJ1rk

— Stanford NLP Group (@stanfordnlp) March 1, 2021