Skip to content
← All tracks

CNNs from Scratch

PyTorch

Build convolutional networks one block at a time — from raw conv to skip connections and squeeze-excitation.

0 / 15 solved
Continue →
  1. 1. Not solved yet. Implement 1D Convolution
  2. 2. Not solved yet. Implement Conv2d
  3. 3. Not solved yet. Implement Max Pooling 1D
  4. 4. Not solved yet. Implement Average Pooling 1D
  5. 5. Not solved yet. Depthwise Separable Convolution
  6. 6. Not solved yet. Transposed Conv2d
  7. 7. Not solved yet. Skip Connection Block
  8. 8. Not solved yet. Squeeze-and-Excitation Block
  9. 9. Not solved yet. Simple CNN
  10. 10. Not solved yet. Train Image Classifier End-to-End
  11. 11. Not solved yet. Patch Embedding with CLS Token
  12. 12. Not solved yet. ViT Encoder Block
  13. 13. Not solved yet. ViT Classification Forward
  14. 14. Not solved yet. Train ViT Classifier End-to-End
  15. 15. Not solved yet. ViT with Mixup Augmentation

Check yourself

4 questions · one attempt each

These do not count toward finishing the track. They are here to catch the things that are easy to read past.

0 / 4

A 32x32 input through a 3x3 conv. What is the output size with (stride 1, pad 0) and with (stride 2, pad 1)? 

nn.Conv2d(1, 1, kernel_size=3, stride=1, padding=0)
nn.Conv2d(1, 1, kernel_size=3, stride=2, padding=1)
Question 1 of 4