two questions about the use of the Hateful Memes dataset

What do you base that on? It’s not accurate according to this post:

2 Likes