I've created a pipeline using tf.Data (or more accurately a mix of Pandas and then tf.Data). My specific requirement is to create sequences using the window command (which applies a moving window with a specified window size, step and stride) and then batch these windows in order to train time series RNN's. The network is actually implemented in JAX/FLAX so I use as_numpy_iterator() to cast from Tensor to np.array

I'm considering moving to PyTorch and would like to migrate my pipeline. What is the equivalent in PyTorch? I've found torch.utils.data but I don't see any mention of windowing etc.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.