PyTorch MCQ Questions and Answers

Q1. What is a tensor in PyTorch?

A. A multi-dimensional array with additional features for GPU computations
B. A simple Python list
C. A database management system
D. A type of neural network
Check Answer

Q2. Which function is used to convert a numpy array to a PyTorch tensor?

A. torch.array()
B. torch.from_numpy()
C. torch.to_tensor()
D. numpy.to_torch()
Check Answer

Q3. What is autograd in PyTorch?

A. Automatic GPU acceleration
B. Automatic gradient computation
C. Automatic model training
D. Automatic data loading
Check Answer

Q4. Which PyTorch module is used for creating neural networks?

A. torch.nn
B. torch.network
C. torch.neural
D. torch.layers
Check Answer

Q5. What is the purpose of torch.cuda.is_available()?

A. To check if CUDA is installed
B. To enable GPU computing
C. To check if a GPU is available for computation
D. To install CUDA drivers
Check Answer

Q6. Which loss function is commonly used for binary classification?

A. MSELoss
B. CrossEntropyLoss
C. BCELoss
D. L1Loss
Check Answer

Q7. What does torch.no_grad() do?

A. Disables GPU usage
B. Disables gradient computation
C. Disables model training
D. Disables data loading
Check Answer

Q8. Which optimizer is NOT available in PyTorch?

A. SGD
B. Adam
C. RMSprop
D. QuickProp
Check Answer

Q9. What is the purpose of DataLoader in PyTorch?

A. To load pretrained models
B. To load data in batches with various options
C. To load model weights
D. To load GPU drivers
Check Answer

Q10. What does requires_grad=True do in PyTorch?

A. Enables GPU acceleration
B. Enables gradient computation for the tensor
C. Requires gradual training
D. Requires gradient descent optimizer
Check Answer

Q11. Which method is used to move a tensor to GPU?

A. tensor.gpu()
B. tensor.cuda()
C. tensor.to_gpu()
D. tensor.move_gpu()
Check Answer

Q12. What is the purpose of torch.save()?

A. To save training data
B. To save model parameters and state
C. To save GPU memory
D. To save computation time
Check Answer

Q13. What is a Module in PyTorch?

A. A Python package
B. A base class for all neural network layers
C. A type of tensor
D. A data loading utility
Check Answer

Q14. Which function is used to compute gradients in PyTorch?

A. forward()
B. compute_grad()
C. backward()
D. gradient()
Check Answer

Q15. What is the purpose of torch.nn.Sequential?

A. To create sequential data
B. To sequence model training
C. To create a sequential container of modules
D. To load data sequentially
Check Answer

Q16. What is the difference between model.train() and model.eval()?

A. They are the same thing
B. train() enables gradient computation, eval() disables it
C. train() is for training, eval() is for testing only
D. train() enables dropout and batch norm updates, eval() disables them
Check Answer

Q17. What is the purpose of torch.randn()?

A. Creates random integers
B. Creates random normal distributed numbers
C. Creates random binary numbers
D. Creates random uniform distributed numbers
Check Answer

Q18. What is torch.nn.Module?

A. A data structure for tensors
B. Base class for all neural network modules
C. A mathematical operation
D. A type of optimizer
Check Answer

Q19. What does torch.cuda.is_available() check?

A. If CUDA version is latest
B. If GPU memory is sufficient
C. If CUDA-enabled GPU is available
D. If model is CUDA compatible
Check Answer

Q20. What is the purpose of torch.no_grad()?

A. Disables gradient calculation
B. Removes all gradients
C. Stops model training
D. Prevents model updates
Check Answer

Q21. What is the default value of requires_grad for a tensor?

A. True
B. False
C. None
D. Depends on tensor type
Check Answer

Q22. Which function is used to concatenate tensors?

A. torch.stack()
B. torch.cat()
C. torch.join()
D. torch.combine()
Check Answer

Q23. What is the purpose of torch.squeeze()?

A. Reduces tensor dimensions
B. Compresses tensor data
C. Removes dimensions of size 1
D. Reduces tensor size
Check Answer

Q24. What is the main purpose of DataLoader in PyTorch?

A. To load data from files
B. To wrap dataset for batch processing
C. To preprocess data
D. To save data to disk
Check Answer

Q25. What does criterion typically refer to in PyTorch?

A. Model architecture
B. Loss function
C. Optimization algorithm
D. Evaluation metric
Check Answer

Q26. What is the purpose of torch.autograd?

A. Automatic model creation
B. Automatic data loading
C. Automatic differentiation
D. Automatic optimization
Check Answer

Q27. What does .to(device) do in PyTorch?

A. Converts tensor type
B. Moves tensor to specified device
C. Saves tensor to device
D. Copies tensor to memory
Check Answer

Q28. What is the purpose of torch.save()?

A. Saves model architecture
B. Saves training data
C. Saves model parameters
D. Saves Python objects
Check Answer

Q29. What is the purpose of detach() in PyTorch?

A. Removes tensor from computation graph
B. Deletes tensor from memory
C. Detaches GPU from computation
D. Removes model parameters
Check Answer

Q30. What is torch.nn.Sequential?

A. A type of RNN
B. A container for modules
C. A sequential optimizer
D. A data loading method
Check Answer

Q31. What does require_grad_() do?

A. Changes tensor type
B. Enables gradient computation
C. Requires model gradients
D. Evaluates gradients
Check Answer

Q32. What is the purpose of backward() in PyTorch?

A. Moves backward in training data
B. Computes gradients
C. Reverses tensor operations
D. Decrements model parameters
Check Answer