If it is possible to release the random seed or index of the pictures about these 6 datasets, thank you very much!

[xingyueye5]

Cause the choice of pictures may matters, providing the indexes in json file may help us better reproduct the result

[liuzhuang13]

Sorry we won't be releasing it, but in general these do not matter that much. The variance is very small. For a more complete instruction on datasets please refer to our follow-up paper's repo, if you are still interested :)

https://github.com/boyazeng/understand_bias