Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released a return torch.max_pool2d(input, kernel_size, stride, padding, dilation, ceil_mode)

1

I use google colab. colab was crash after getting this exception.
I pass the single image to model with for loop it’s working fine. but if i use dataloader then it’s returning this exception and colab was crash.

class Dataset(torch.utils.data.Dataset):
  def __init__(self,inputImage):
    outLabel=pd.read_csv('/content/drive/MyDrive/image_labels.csv',index_col="image")
    self.outLabel=outLabel
    self.imagePath=imagePath
  
  def __len__(self):
    return (len(inputImage))
    
  def __getitem__(self,index):
    target_encoded=[]
    input_data=inputImage[index]    
    # path='/content/drive/MyDrive//images/',input_data
    # path=path[0]+path[1]
    # img=Image.open(path)
    #     # # images = images.to(device)
    #     # # labels = labels.to(device)
    #     # convert_tensor = transforms.ToTensor()
    # convert_tensor = transforms.Compose([transforms.Resize(size=(720,720)),transforms.ToTensor()])
    # img=convert_tensor(img)
    rows = self.outLabel.loc[[input_data]]
    dums=np.zeros((80,4))
    lst=[]
    for i in range(len(rows)):
      dums[i][0],dums[i][1],dums[i][2],dums[i][3]=rows['left'][i],rows['width'][i],rows['top'][i],rows['height'][i]
      # dums[i][1]=rows['width'][i]
      # dums[i][2]=rows['top'][i]
      # dums[i][3]=rows['height'][i]
    target_encoded=torch.tensor(dums)
    return input_data,target_encoded
2

The crash is most likely not coming from this warning, which should already be fixed in this PR, so you might need to isolate the issue a bit more.