Dear all,
I have a 4D tensor [batch_size, temporal_dimension, data[0], data[1]], the 3d tensor of [temporal_dimension, data[0], data[1]] is actually my input data to the network. I would shuffle the tensor along the second dimension, which is my temporal dimension to check if the network is learning something from the temporal dimension or not. Will be glad if this shuffling is kind of reproducible .
Is there a certain way to do that?
Thanks.