How to get around in place operation error if index leaf variable for gradient update?

I just find out: to update the Variable, should use ht.data[idx]. use data to access tensor directly.