Fetching the paper…

Align before Fuse: Vision and Language Representation Learning with Momentum Distillation · Around