Why these tensor have all same values?

vision
1

I am trying to debug and watch the values of a code for studying machine learning. Here’s the code snippet:

train_data = DataGenerator(lines[:num_train], input_shape, True)  # num_train is 1066
......
gen_train = DataLoader(train_data, batch_size=4) 
......
epochs = 20 
for epoch in range(epochs):
    total_train = 0  
    for data in gen_train:
        img, label = data    # breakpoint here
        with torch.no_grad():
            img = img.to(device)
            label = label.to(device)
        optim.zero_grad()
        output = net(img)
        train_loss = nn.CrossEntropyLoss()(output, label).to(device)
        train_loss.backward()  
        optim.step()  
        total_train += train_loss  
......

here in this code, img are 4 images loaded, and label are 4 integers that corresponds to the images’ class. when I put a breakpoint, I saw the images has 4 “blocks”, each block has all same values:

image
image593×540 20.6 KB

(as I am new user, I can only upload one image, but all other 3 blocks has the same value)

these blocks should be the images, but why they all have the same numbers?

2

Check if these values represent borders with a static color by visualizing the images via e.g. matplotlib.