The size of tensor a (32) must match the size of tensor b (8) at non-singleton dimension 0 error

vision
1

I was training my model when this error surfaced up

<ipython-input-14-b4c5dff1539c> in <module>()
      1 model_conv = train_model(model_conv,train_dl,val_dl, criterion, optimizer_conv,
----> 2                          exp_lr_scheduler, num_epochs=25)

<ipython-input-13-f484d720bf26> in train_model(model, train_dl, val_dl, criterion, optimizer, scheduler, num_epochs)
     57            #stats
     58            running_validation_loss += loss_val.item()* image.size(0)
---> 59            running_corrects_validation += torch.sum(preds_val == labels)
     60 
     61 


The size of tensor a (32) must match the size of tensor b (8) at non-singleton dimension 0

The error was directed at this code snippet

for image,label in val_dl:
           model.eval()
           with torch.set_grad_enabled(False):
            image,label = image.to(device), label.to(device)

          
            outputs = model(image)
            _, preds_val = torch.max(outputs, 1)
            loss_val = criterion(outputs, label)


           #stats
           running_validation_loss += loss_val.item()* image.size(0)
   ----> running_corrects_validation += torch.sum(preds_val == labels)

So I figured it may be something wrong with the dimensions of my tensor, hence I decided to test out this theory

Input:

for images,labels in train_dl:
  output = model_conv(images)
  loss = F.cross_entropy(output,labels)
  _,preds_vals = torch.max(output,1)
  accuracy = torch.sum(preds_vals==labels)

  print(f"Output: {output.size()}" )
  print(f"labels: {labels.size()}")
  print(f"preds_vals: {preds_vals.size()}" )
  print(f"accuracy: {accuracy}")
  break

Output:

Output: torch.Size([32, 10])
labels: torch.Size([32])
preds_vals: torch.Size([32])
accuracy: 1

Nothing seems to be off, however after researching quite a bit, I think it may be the loss function which may be causing all of this problems. Does anybody have an idea why this is happening. My full code will be listed below

2

Full code

data = download_url("https://s3.amazonaws.com/fast-ai-imageclas/cifar10.tgz",".")

with tarfile.open('./cifar10.tgz', 'r:gz') as tar:
    tar.extractall(path='./data')




if os.path.exists("/content/data/cifar10/validate") is False:

  os.makedirs("/content/data/cifar10/validate")

  os.makedirs("/content/data/cifar10/validate/airplane")

  os.makedirs("/content/data/cifar10/validate/automobile")

  os.makedirs("/content/data/cifar10/validate/bird")

  os.makedirs("/content/data/cifar10/validate/cat")

  os.makedirs("/content/data/cifar10/validate/deer")

  os.makedirs("/content/data/cifar10/validate/dog")

  os.makedirs("/content/data/cifar10/validate/frog")

  os.makedirs("/content/data/cifar10/validate/horse")

  os.makedirs("/content/data/cifar10/validate/ship")

  os.makedirs("/content/data/cifar10/validate/truck") 

for i in sample(glob.glob("/content/data/cifar10/train/airplane/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/airplane")

for i in sample(glob.glob("/content/data/cifar10/train/automobile/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/automobile")

for i in sample(glob.glob("/content/data/cifar10/train/bird/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/bird")

for i in sample(glob.glob("/content/data/cifar10/train/cat/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/cat")

for i in sample(glob.glob("/content/data/cifar10/train/deer/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/deer")

for i in sample(glob.glob("/content/data/cifar10/train/dog/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/dog")

for i in sample(glob.glob("/content/data/cifar10/train/frog/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/frog")

for i in sample(glob.glob("/content/data/cifar10/train/horse/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/horse")

for i in sample(glob.glob("/content/data/cifar10/train/ship/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/ship")

for i in sample(glob.glob("/content/data/cifar10/train/truck/*.png"),500):
  shutil.move(i,"/content/data/cifar10/validate/truck")

   

from torch.utils.data.dataset import random_split
mean = np.array([0.5, 0.5, 0.5])
std = np.array([0.25, 0.25, 0.25])


transform =  transforms.Compose([
        transforms.RandomResizedCrop(224),
        transforms.RandomHorizontalFlip(),
        transforms.ToTensor(),
        transforms.Normalize(mean, std)
    ])


train_ds = ImageFolder("/content/data/cifar10/train",transform)
val_ds = ImageFolder("/content/data/cifar10/validate", transform)

train_dl = DataLoader(train_ds, batch_size = 32, shuffle=True, num_workers=2, pin_memory=True)
val_dl  =  DataLoader(val_ds,batch_size = 32,shuffle = False, num_workers=2, pin_memory=True)

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

def train_model(model, train_dl , val_dl, criterion, optimizer, scheduler, num_epochs=25):
    since = time.time()

    best_model_wts = copy.deepcopy(model.state_dict())
    best_acc = 0.0
    running_training_loss = 0.0
    running_validation_loss = 0.0
    running_corrects_training = 0
    running_corrects_validation = 0


    for epoch in range(num_epochs):
        print('Epoch {}/{}'.format(epoch, num_epochs - 1))
        print('-' * 10)

        #training loop
        for images,labels in train_dl:
          with torch.set_grad_enabled(True):
           
           images,labels = images.to(device), labels.to(device)

              model.train(True):
              outputs = model(images)
              _, preds_train = torch.max(outputs, 1)
              loss_train = criterion(outputs, labels)

              optimizer.zero_grad()
              loss_train.backward()
              optimizer.step()


          #stats
          running_training_loss += loss_train.item()
          running_corrects_training += torch.sum(preds_train == labels.data)


        #scheduler step      
        scheduler.step()

        #stats
        epoch_loss_training = running_training_loss / len(train_dl)
        epoch_acc_training = running_corrects_training.double() / len(train_dl)
        
        #validation loop
        for image,label in val_dl:
           with torch.set_grad_enabled(False):
            image,label = image.to(device), label.to(device)

              model.eval():
              outputs = model(image)
              _, preds_val = torch.max(outputs, 1)
              loss_val = criterion(outputs, label)


           #stats
           running_validation_loss += loss_val.item()
           running_corrects_validation += torch.sum(preds_val == labels.data)
          

        epoch_loss_validation = running_validation_loss / len(val_dl)
        epoch_acc_validation = running_corrects_validation.double() / len(val_dl)

        print("epoch {}, epoch training loss {}, epoch training acc {}, epoch_loss_validation{}, epoch_acc_validation" .format(epoch, epoch_loss_training, epoch_acc_training,
                                                                                                                           epoch_loss_validation, epoch_acc_validation))
        

        #finding best accuracy
        if running_corrects_validation > best_acc:
          best_acc = running_corrects_validation
          best_model_wts = copy.deepcopy(model.state_dict())
        print()


        # total time
        time_elapsed = time.time() - since
        print('Training complete in {:.0f}m {:.0f}s'.format(
           time_elapsed // 60, time_elapsed % 60))
        print('Best val Acc: {:4f}'.format(best_acc))

        

        # load best model weights
        model.load_state_dict(best_model_wts)
        return model


#downloading pretrained model/finetune
model_conv = torchvision.models.resnet18(pretrained=True)
for param in model_conv.parameters():
    param.requires_grad = False

#defining parameters

num_ftrs = model_conv.fc.in_features
model_conv.fc = nn.Linear(num_ftrs, 10)

model_conv = model_conv.to(device)

criterion = nn.CrossEntropyLoss()

optimizer_conv = optim.SGD(model_conv.fc.parameters(), lr=0.001, momentum=0.9)

exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)


model_conv = train_model(model_conv,train_dl, val_dl, criterion, optimizer_conv,exp_lr_scheduler, num_epochs=25)
3

In your training loop you define “image, labelS” and in your validation loop you define “image, label” (no “S” at the end of label).

Yet, when you do

running_corrects_validation += torch.sum(preds_val == labels.data)

You are using “labels” with an “s”.

1 Like