Target size (torch.Size([800])) must be the same as input size (torch.Size([800, 4]))

[Mubassir1820]

• Course video/timestamp: (Learn PyTorch for deep learning in a day(02. PyTorch neural network classification), timestamp 13:32)

Target size (torch.Size([800])) must be the same as input size (torch.Size([800, 4]))

TODO:I have tried the exact steps in the training loop(also tried copy paste from the pdf) but for some reason this error keeps showing up

My code

TODO: Add your code here, best to format with backticks as well, for example:

[
# Fit the multi-class model to the data
torch.manual_seed(42)
torch.cuda.manual_seed(42)

# Set number of epochs
epochs = 100

# Put the data to the target device
X_blob_train, y_blob_train = X_blob_train.to(device), y_blob_train.to(device)
X_blob_test, y_blob_test = X_blob_test.to(device), y_blob_test.to(device)

# Loop through data
for epoch in range(epochs):
  ### Training 
  model_4.train()

  y_logits = model_4(X_blob_train)
  y_pred = torch.softmax(y_logits, dim = 1).argmax(dim = 1)

  print(y_logits.shape, y_blob_train.shape)
  loss = loss_fn(y_logits, y_blob_train)
  acc = accuracy_fn(y_true = y_blob_train,
                    y_pred = y_pred)
  
  optimizer.zero_grad()
  loss.backward()
  optimizer.step()

  ###  Testing
  model_4.eval()
  with torch.inference_mode():
    test_logits = model_4(X_blob_test)
    test_preds = torch.softmax(test_logits, dim = 1).argmax(dim = 1)

    test_loss = loss_fn(test_logits, y_blob_test)
    test_acc = accuracy_fn(y_true=y_blob_test,
                           y_pred = test_preds)
    
  # Print out whats happening
  if epoch % 10 == 0:
    print(f'Epoch: {epochs} | Loss: {loss:.4f}, Acc: {acc:.2f}% | Test Loss: {test_loss:.4f}, Test acc: {test_acc:.2f}%')]
``` <- be sure to delete the slashes, only keep the backticks

## What I've tried so far

TODO: I have tried using squeeze() like binary classification problem, but it still has the same issue 

[Mubassir1820]

So, I finally figured it out. I wrote loss_dn instead of loss_fn and later referred to loss_fn. As the loss function helps to convert the shape of y_logits to match with y_blob_train, the error kept popping up