ChineseVLBert
ChineseVLBert copied to clipboard
How about the performance of the Chinese pretrained mode.
Translating COCO dataset to Chinese sounds good, is there any further progross of this idea?