Why deployment serves predictions in PyTorch - Test Your Understanding

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a PyTorch model for deployment.

PyTorch

model = torch.load([1])

Drag options to blanks, or click blank then click option'

Atorch.model('model.pth')

Bmodel.pth

Cload_model('model.pth')

D'model.pth'

Attempts:

3 left

💡 Hint

Common Mistakes

Forgetting quotes around the file path string.

Using incorrect function names to load the model.

The model file path must be a string, so it needs quotes around it.

2fill in blank

medium

Complete the code to set the model to evaluation mode before serving predictions.

PyTorch

model.[1]()

Drag options to blanks, or click blank then click option'

Atrain

Bpredict

Ceval

Dserve

Attempts:

3 left

💡 Hint

Common Mistakes

Using model.train() which enables training mode.

Trying to call non-existent methods like predict or serve.

Calling model.eval() sets the model to evaluation mode, disabling training behaviors like dropout.

3fill in blank

hard

Fix the error in the code to get predictions from the model without tracking gradients.

PyTorch

with torch.[1]():
    output = model(input_tensor)

Drag options to blanks, or click blank then click option'

Ano_grad

Btrack_grad

Cenable_grad

Dgrad

Attempts:

3 left

💡 Hint

Common Mistakes

Using torch.grad() which does not exist.

Not disabling gradients during prediction causing unnecessary computation.

torch.no_grad() disables gradient tracking, which is important during prediction to save memory and computation.

4fill in blank

hard

Fill both blanks to convert model output logits to predicted class indices.

PyTorch

predictions = torch.[1](output, dim=[2])

Drag options to blanks, or click blank then click option'

Aargmax

Bsoftmax

C1

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Using softmax instead of argmax returns probabilities, not class indices.

Using wrong dimension for argmax.

torch.argmax with dim=1 returns the index of the highest logit for each sample, which is the predicted class.

5fill in blank

hard

Fill all three blanks to prepare input tensor, run prediction, and convert output to numpy array for deployment.

PyTorch

input_tensor = torch.tensor([1], dtype=[2])
with torch.no_grad():
    output = model(input_tensor)
predictions = output.[3]()

Drag options to blanks, or click blank then click option'

A[1.0, 2.0, 3.0]

Btorch.float32

Cnumpy

Dnumpy()

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong data type for tensor.

Forgetting to disable gradients during prediction.

Trying to convert tensor to numpy without calling the method.

Input data is converted to a float tensor, model runs without gradients, and output is converted to a numpy array for easy use outside PyTorch.