why dose the pretrain corpus have 100w coco_qa pairs?

microsoft / Oscar

Oscar and VinVL

MIT License

1.04k stars 251 forks source link

Open logicwong opened 3 years ago

logicwong commented 3 years ago

This is inconsistent with the number reported in the paper

ckmstydy commented 3 years ago

VQA2.0 dataset use COCO iamges

logicwong commented 3 years ago

VQA2.0 dataset use COCO iamges

I know, but VQA2.0 dataset only have 66w qa pairs (train+val). Where did the extra 40w qa pairs come from?