How does Pytorch’s “Fold” and “Unfold” work?

by

unfold imagines a tensor as a longer tensor with repeated columns/rows of values ‘folded’ on top of each other, which is then “unfolded”:

  • size determines how large the folds are
  • step determines how often it is folded

E.g. for a 2×5 tensor, unfolding it with step=1, and patch size=2 across dim=1:

x = torch.tensor([[1,2,3,4,5],
                  [6,7,8,9,10]])

>>> x.unfold(1,2,1)
tensor([[[ 1,  2], [ 2,  3], [ 3,  4], [ 4,  5]],
        [[ 6,  7], [ 7,  8], [ 8,  9], [ 9, 10]]])

enter image description here

fold is roughly the opposite of this operation, but “overlapping” values are summed in the output.

More Related Contents:

  1. How to initialize weights in PyTorch?
  2. How do I initialize weights in PyTorch?
  3. What does model.eval() do in pytorch?
  4. PyTorch NotImplementedError in forward
  5. What’s the difference between torch.stack() and torch.cat() functions?
  6. How to construct a network with two inputs in PyTorch
  7. RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same
  8. PyTorch Binary Classification – same network structure, ‘simpler’ data, but worse performance?
  9. Will it lead to Overfitting / Curse of Dimensionality
  10. How to get the location of all text present in an image using OpenCV?
  11. Best way to save a trained model in PyTorch? [closed]
  12. What is a `”Python”` layer in caffe?
  13. How to apply gradient clipping in TensorFlow?
  14. How do I save a trained model in PyTorch?
  15. Keras Sequential model input layer
  16. Why do we need to call zero_grad() in PyTorch?
  17. Does Any one got “AttributeError: ‘str’ object has no attribute ‘decode’ ” , while Loading a Keras Saved Model
  18. Keras custom loss function: Accessing current input pattern
  19. How can I use a pre-trained neural network with grayscale images?
  20. How to understand the term `tensor` in TensorFlow?
  21. Pytorch – RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed
  22. Removing then Inserting a New Middle Layer in a Keras Model
  23. error: (-2:Unspecified error) The function is not implemented. Rebuild the library with Windows, GTK+ 2.x or Cocoa support
  24. Unbalanced data and weighted cross entropy
  25. Custom loss function in Keras
  26. What is the role of TimeDistributed layer in Keras?
  27. How to handle non-determinism when training on a GPU?
  28. keras: how to save the training history attribute of the history object
  29. How do I split a custom dataset into training and test datasets?
  30. Pytorch ValueError: optimizer got an empty parameter list

Leave a Comment