ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision.

Wonjae Kim, Bokyung Son, Ildoo Kim

Browse the full ICML paper archive.