pytorch tutorial notes

Input Pipeline

1
2

# torchvision.datasets => torch.utils.data.DataLoader

Train && Val Diff

1
2
3
4
5
6
7
8

#train = true of false
train_dataset = torchvision.datasets.MNIST(root='../data/',
                                           train=True,
                                           transform=transforms.ToTensor(),
                                           download=True)
val_dataset = torchvision.datasets.MNIST(root='../data/',
                                         train=False,
                                         transform=transforms.ToTensor())

torch.max ‘s return : first is the max value,second is the index

计算准确率

1

correct += (predicted == labels).sum().item()

Output size of Conv

$$dfrac {W-F+2p}{S}+1$$

Summary.

 To summarize, the Conv Layer:

• Accepts a volume of size

  $$W_1 times H_1 times D_1$$

• Requires four hyperparameters:

  • Number of filters K,
  • their spatial extent F,
  • the stride S,
  • the amount of zero padding P.

• Produces a volume of sizeW₂×H₂×D₂

  where:

  • $$W_2 = (W_1 - F + 2P)/S + 1$$
  • $$H_2 = (H_1 - F + 2P)/S + 1$$

  • D₂=K

• With parameter sharing, it introduces

  $$F cdot F cdot D_1$$

  weights per filter, for a total of

  $$(F cdot F cdot D_1) cdot K$$

  weights and K biases.

• In the output volume, the d-th depth slice (of size W₂×H₂) is the result of performing a valid convolution of the d-th filter over the input volume with a stride of S, and then offset by d-th bias.