chanmuzi
<Multi-modal> ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision