Subsampling rate for Human3.6m

[WeiyuDu]

Hi,

Thanks for sharing your code! I have a question about the subsampling rates for Human3.6m dataset. It seems that the training data is subsampled 6 times in your code while test data is not subsampled. But the preprocessing codebase you linked to subsampled test data 64 times in their code. I was wondering if you changed that parameter? What is the correct subsampling rates for training and testing to recreate your results?

Thank you so much!