Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts Paper • 2102.08981 • Published Feb 17, 2021 • 1