patrickloeber / pytorchtutorial Goto Github PK

Where is the dataset data/hymenoptera_data in transfer learning tutorial from? It seems that in previous tutorials it was not mentioned how this dataset is derived.

When executing 09_dataloader.py got this error.

Currently running on Pytorch 1.5.

• I'm using a Windows 10 Pro OS and in the data loader, dataiter.next() isn't working.
  Exception has occurred: AttributeError 'DataLoader' object has no attribute 'next' File "C:\Users\xxx\Desktop\xxxx\xxxxx\code\pytorchbasics\datasetanddataloaderclass.py", line 40, in <module> datatiter = dataloader.next()

Hi @python-engineer ,
on line 131, shouldn't we assign z again on CPU?
z = z.to("cpu") before then calling z.numpy()
I got following error without doing so:

TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

Thanks for all the great material!
Cheers,
Gautam

Please change the name of the file transformers to transforms.

Hello Patrick, Can you please update the repo with the data used in tutorial 15?

Hi @python-engineer ,
Your codes and videos are really helpful thanks a lot!
Just a small doubt: in 14_cnn.py line 117 shouldn't be min(labels.size(0),batch_size) instead of batch_size . Otherwise if batch size = 3 you get an error.

Hi, Excellent tutorials! But I have a question. Form tutorial 13 and on you change the place where the zero_grad method is called and I do not get why?
Before 13 was:

loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
optimizer.zero_grad()

After 13:

loss = criterion(outputs, labels)
optimizer.zero_grad() # Here is the change
loss.backward()
optimizer.step()

Now I am wondering if you set to zero the gradients, then, how the optimizer could update the parameters without any information about the gradient?

Where is the dataset data/hymenoptera_data in transfer learning tutorial from? It seems that in previous tutorials it was not mentioned how this dataset was obtained.

Where is the dataset data/hymenoptera_data in transfer learning tutorial from? It seems that in previous tutorials it was not mentioned how this dataset is obtained.

Hi, i have an question for the .py

line 141 :

class_labels.append(predicted)   

why not

class_labels.append(labels)

the add_pr_curve() function should use the ground truth label?

torch.version => '2.0.0+cu118'
L100: weights.grad.zero_()
show error "AttributeError: 'Tensor' object has no attribute 'grad_zero_'"

Hello @python-engineer
thanks a lot for your videos. they are very helpful.
I have a cuda support and when I tried to run the code I got the error:
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_addmm)

I solved that by adding:
.to(device) to the lines (50, 51, 52)

        self.l1= nn.Linear(input_size, hidden_size).to(device)
        self.relu= nn.ReLU().to(device)
        self.l2 = nn.Linear(hidden_size, num_classes).to(device)

and it runs now well

In 09_dataloader.py line 69, I guess you want to access data (from DataLoader) and not first_data (from DataSet, line 55). Best wishes!

Awesome tuts! Keep it up!
I think you mixed up names for 06_1_loss_and_optimizer.py (contains model) and 06_2_model_loss_and_ optimizer.py (does not).