03. PyTorch Computer Vision . Prepeare a datamodel problem

[genprokuror]

Hey i just reviewed one part of code.
Check out what's inside the training dataloader train_features_batch, train_labels_batch = next(iter(train_dataloader)) train_features_batch.shape, train_labels_batch.shape
I cant understand why it gives this when launching
(torch.Size([32, 1, 28, 28]), torch.Size([32]))
Why theyr sizes and values are not the same? Dont author assigned them together here? How it works?
train_features_batch, train_labels_batch = next(iter(train_dataloader))

[ZephyrusBlaze]

In PyTorch, the training dataloader typically yields batches of input features and their corresponding labels. In your case:

• train_features_batch has a shape of (32, 1, 28, 28), indicating it's a batch of 32 samples, each with 1 channel (grayscale image) and a size of 28x28 pixels.
• train_labels_batch has a shape of (32,), indicating it's a batch of 32 labels.

The sizes aren't the same because each element in train_features_batch corresponds to an image, while each element in train_labels_batch corresponds to the label of the respective image.

When you iterate over train_dataloader, it returns batches of features and labels, which you then unpack into train_features_batch and train_labels_batch. So, yes, they are assigned together, but they represent different aspects of your dataset: features and labels.

[LuluW8071]

To put it simply,
Train_dataloader contains tensor images (color_channels, height, width)
After,

 train_features_batch, train_labels_batch = next(iter(train_dataloader))

Train_features_batch contains images tensor arrays accompanied by batch size
(Batch_size, color_channels, height, width)

While, Train_labels_batch contain labels of those image data

• index of image labels accompanied by batch size
  (Batch_size)

The code snippet is just to assign batch_size (32) to load 32 images and respective 32 labels at a time in training device for model training.