- What is PyTorch used for?
a) Natural Language Processing
b) Object Detection
c) Neural Network
d) All of the above
Answer: d) All of the above
- What is the difference between PyTorch and TensorFlow?
a) Tensorflow is easier to use than PyTorch
b) PyTorch has dynamic computation graph while TensorFlow has static computation graph
c) TensorFlow has more community support than PyTorch
d) None of the above
Answer: b) PyTorch has dynamic computation graph while TensorFlow has static computation graph
- What is dynamic Computation graph?
a) Graph generated at compile time
b) Graph generated at runtime
c) There is no concept of dynamic computation graph in PyTorch
d) None of the above
Answer: b) Graph generated at runtime
- How is data loaded in PyTorch?
a) By using Dataset and Dataloader
b) By using CSV files
c) By using numpy array
d) None of the above
Answer: a) By using Dataset and Dataloader
- How to add a new dimension to a tensor?
a) Tensor.ndim = new_dim
b) Tensor.reshape(new_dim)
c) Tensor.unsqueeze(new_dim)
d) None of the above
Answer: c) Tensor.unsqueeze(new_dim)
- What is Autograd in PyTorch?
a) A module for fast operations on images
b) A module for building neural network architectures
c) A module for gradient calculation and differentiation
d) None of the above
Answer: c) A module for gradient calculation and differentiation
- What is CUDA in PyTorch?
a) A high-level neural network library
b) A parallel computing platform and programming model
c) A module for data loading and transformation
d) None of the above
Answer: b) A parallel computing platform and programming model
- What is the purpose of .to() method in PyTorch?
a) To convert tensor to a specific device (CPU, GPU)
b) To convert tensor to numpy array
c) To transpose a tensor
d) None of the above
Answer: a) To convert tensor to a specific device (CPU, GPU)
- What is the purpose of Flatten layer in PyTorch?
a) To convert multi-dimensional tensor to one-dimensional tensor
b) To add extra dimensions to a tensor
c) To remove dimensions from a tensor
d) None of the above
Answer: a) To convert multi-dimensional tensor to one-dimensional tensor
- What is Stochastic Gradient Descent (SGD)?
a) A gradient-based optimization algorithm
b) A module for data augmentation
c) A module for regularization
d) None of the above
Answer: a) A gradient-based optimization algorithm
- What is Transfer Learning?
a) A training technique where a pre-trained model is used as a starting point for a new model
b) A technique to transfer data from one device to another
c) A technique to transfer data between different neural networks
d) None of the above
Answer: a) A training technique where a pre-trained model is used as a starting point for a new model
- What is the purpose of learning rate in neural network training?
a) To define the number of layers in the neural network
b) To define the activation function of the neural network
c) To define the step size during gradient descent optimization
d) None of the above
Answer: c) To define the step size during gradient descent optimization
- What is Batch Normalization?
a) A technique to normalize data before feeding it to a neural network
b) A technique to avoid overfitting during neural network training
c) A technique to speed up neural network training
d) None of the above
Answer: c) A technique to speed up neural network training
- What is L1 regularization?
a) A regularization technique that adds the sum of absolute values of all weights to the loss function
b) A regularization technique that adds the sum of squares of all weights to the loss function
c) A regularization technique that randomly switches off some neurons during training
d) None of the above
Answer: a) A regularization technique that adds the sum of absolute values of all weights to the loss function
- What is the purpose of ReLU activation function?
a) To normalize data before feeding it to a neural network
b) To add non-linearity to a neural network
c) To make neural network training faster
d) None of the above
Answer: b) To add non-linearity to a neural network
- What is Data Parallelism in PyTorch?
a) A concept of parallel computation across multiple machines
b) A technique to parallelize computations within each batch
c) A technique to parallelize computations across multiple GPUs/CPU cores in a single machine
d) None of the above
Answer: c) A technique to parallelize computations across multiple GPUs/CPU cores in a single machine
- What is the purpose of MaxPooling layer in PyTorch?
a) To reduce the number of parameters in a neural network
b) To add non-linearity to a neural network
c) To avoid overfitting during neural network training
d) None of the above
Answer: a) To reduce the number of parameters in a neural network
- What is the purpose of Dropout layer in PyTorch?
a) To randomly switch off some neurons during training to avoid overfitting
b) To add non-linearity to a neural network
c) To reduce the number of parameters in a neural network
d) None of the above
Answer: a) To randomly switch off some neurons during training to avoid overfitting
- What is the purpose of Cross Entropy Loss function in PyTorch?
a) To measure the difference between predicted and actual labels in a classification task
b) To minimize the difference between predicted and actual values in a regression task
c) To add regularization to a neural network
d) None of the above
Answer: a) To measure the difference between predicted and actual labels in a classification task
- What is Categorical Cross Entropy Loss function in PyTorch?
a) A loss function for binary classification problems
b) A loss function for multiclass classification problems
c) A loss function for regression problems
d) None of the above
Answer: b) A loss function for multiclass classification problems
- What is the purpose of Adam optimizer in PyTorch?
a) To perform backpropagation and update the weights
b) To initialize the weights of a neural network
c) To regularize the weights of a neural network
d) None of the above
Answer: a) To perform backpropagation and update the weights
- What is ELU activation function in PyTorch?
a) A linear activation function
b) An activation function that is similar to ReLU but has a non-zero output for negative inputs
c) An activation function that is similar to sigmoid function
d) None of the above
Answer: b) An activation function that is similar to ReLU but has a non-zero output for negative inputs
- What is the purpose of Tanh activation function in PyTorch?
a) To normalize data before feeding it to a neural network
b) To add non-linearity to a neural network
c) To make neural network training faster
d) None of the above
Answer: b) To add non-linearity to a neural network
- What is the purpose of Cross-validation in PyTorch?
a) To split data into training and testing sets
b) To validate the performance of a neural network model
c) To regularize the weights of a neural network
d) None of the above
Answer: b) To validate the performance of a neural network model
- What is the purpose of Residual Networks (ResNets) in PyTorch?
a) To avoid vanishing gradients problem during neural network training
b) To increase the accuracy of a neural network model
c) To speed up neural network training
d) None of the above
Answer: a) To avoid vanishing gradients problem during neural network training
- What is the purpose of Softmax activation function in PyTorch?
a) To squash the output of a neural network to a range between 0 and 1
b) To make the output of a neural network to sum up to 1
c) To add non-linearity to a neural network
d) None of the above
Answer: b) To make the output of a neural network to sum up to 1
- What is the purpose of Normalization in Deep Learning?
a) To reduce the number of parameters in a neural network
b) To avoid overfitting during neural network training
c) To speed up neural network training
d) None of the above
Answer: b) To avoid overfitting during neural network training
- What is Early Stopping in PyTorch?
a) A technique to stop neural network training once it starts overfitting
b) A technique to stop neural network training after a fixed number of epochs
c) A technique to stop neural network training once it reaches a certain accuracy
d) None of the above
Answer: a) A technique to stop neural network training once it starts overfitting
- What is the purpose of Precision and Recall in classification tasks?
a) To measure the accuracy of a model
b) To measure the effectiveness of a model in detecting positive samples
c) To measure the effectiveness of a model in detecting negative samples
d) None of the above
Answer: b) To measure the effectiveness of a model in detecting positive samples
- What is the purpose of F1 Score in classification tasks?
a) To measure the accuracy of a model
b) To measure the effectiveness of a model in detecting positive samples
c) To measure the effectiveness of a model in detecting negative samples
d) None of the above
Answer: d) None of the above
- What is the purpose of Precision-Recall Curve in classification tasks?
a) To visualize the trade-off between precision and recall for different threshold values
b) To visualize the accuracy of a model
c) To visualize the effectiveness of a model in detecting negative samples
d) None of the above
Answer: a) To visualize the trade-off between precision and recall for different threshold values
- What is the purpose of Learning Rate Scheduler in PyTorch?
a) To adjust the learning rate during neural network training
b) To adjust the number of layers in a neural network
c) To adjust the activation function of a neural network
d) None of the above
Answer: a) To adjust the learning rate during neural network training
- What is the purpose of Weight Initialization in PyTorch?
a) To reduce the number of parameters in a neural network
b) To speed up neural network training
c) To avoid vanishing and exploding gradients during neural network training
d) None of the above
Answer: c) To avoid vanishing and exploding gradients during neural network training
- What is the purpose of Tensors in PyTorch?
a) To store data in multi-dimensional arrays
b) To define convolutional neural network layers
c) To define activation functions
d) None of the above
Answer: a) To store data in multi-dimensional arrays
- What is the purpose of Backpropagation in PyTorch?
a) To initialize the weights of a neural network
b) To regularize the weights of a neural network
c) To perform gradient descent optimization during neural network training
d) None of the above
Answer: c) To perform gradient descent optimization during neural network training
- What is the purpose of L2 regularization in PyTorch?
a) A regularization technique that adds the sum of absolute values of all weights to the loss function
b) A regularization technique that adds the sum of squares of all weights to the loss function
c) A regularization technique that randomly switches off some neurons during training
d) None of the above
Answer: b) A regularization technique that adds the sum of squares of all weights to the loss function
- What is the purpose of Splitting Data in PyTorch?
a) To normalize data before feeding it to a neural network
b) To validate the performance of a neural network model
c) To reduce the number of parameters in a neural network
d) None of the above
Answer: b) To validate the performance of a neural network model
- What is the purpose of the transform parameter in the Dataset class in PyTorch?
a) To specify the type of neural network architecture to use
b) To specify the learning rate during training
c) To specify the data augmentation techniques to apply to the data
d) None of the above
Answer: c) To specify the data augmentation techniques to apply to the data
- What is the purpose of Batch Size in PyTorch?
a) To specify the number of epochs to train a neural network
b) To specify the number of training samples to feed to a neural network at once
c) To specify the size of the input layer in a neural network
d) None of the above
Answer: b) To specify the number of training samples to feed to a neural network at once
- What is the purpose of Sequential class in PyTorch?
a) To define convolutional neural network layers
b) To define activation functions
c) To define the neural network architecture by chaining individual layers
d) None of the above
Answer: c) To define the neural network architecture by chaining individual layers
- What is the purpose of Inception module in PyTorch?
a) To avoid vanishing gradients during neural network training
b) To reduce the number of parameters in a neural network
c) To increase the accuracy of a neural network model
d) None of the above
Answer: b) To reduce the number of parameters in a neural network
- What is the purpose of Recurrent Neural Networks (RNNs) in PyTorch?
a) To process sequential data such as time series or natural language data
b) To reduce the number of parameters in a neural network
c) To speed up neural network training
d) None of the above
Answer: a) To process sequential data such as time series or natural language data
- What is the purpose of Convolutional Neural Networks (CNNs) in PyTorch?
a) To process sequential data such as time series or natural language data
b) To reduce the number of parameters in a neural network
c) To process image data
d) None of the above
Answer: c) To process image data
- What is the purpose of Mean Squared Error (MSE) Loss function in PyTorch?
a) To measure the difference between predicted and actual labels in a classification task
b) To minimize the difference between predicted and actual values in a regression task
c) To add regularization to a neural network
d) None of the above
Answer: b) To minimize the difference between predicted and actual values in a regression task
- What is the purpose of Mean Absolute Error (MAE) Loss function in PyTorch?
a) To measure the difference between predicted and actual labels in a classification task
b) To minimize the difference between predicted and actual values in a regression task
c) To add regularization to a neural network
d) None of the above
Answer: b) To minimize the difference between predicted and actual values in a regression task
- What is the purpose of Learning Rate in Gradient Descent algorithm?
a) To adjust the step size during each iteration of gradient descent
b) To adjust the number of iterations during gradient descent
c) To adjust the gradient during each iteration of gradient descent
d) None of the above
Answer: a) To adjust the step size during each iteration of gradient descent
- How does Gradient Descent algorithm work in PyTorch?
a) It calculates the derivative of loss function w.r.t. weights and updates the weights in the direction of negative gradient
b) It randomly initializes the weights of a neural network and updates the weights in a random direction
c) It calculates the sum of squared errors between predicted and actual values and updates the weights in the direction of negative gradient
d) None of the above
Answer: a) It calculates the derivative of loss function w.r.t. weights and updates the weights in the direction of negative gradient
- What is the purpose of Dropout regularization in PyTorch?
a) To randomly switch off some neurons during training to avoid overfitting
b) To add non-linearity to a neural network
c) To reduce the number of parameters in a neural network
d) None of the above
Answer: a) To randomly switch off some neurons during training to avoid overfitting
- What is the purpose of Sigmoid activation function in PyTorch?
a) To normalize data before feeding it to a neural network
b) To add non-linearity to a neural network
c) To make neural network training faster
d) None of the above
Answer: b) To add non-linearity to a neural network
- What is the purpose of Embedding layer in PyTorch?
a) To encode categorical data as numerical vectors
b) To add non-linearity to a neural network
c) To avoid overfitting during neural network training
d) None of the above
Answer: a) To encode categorical data as numerical vectors
Sr. Software Engineer at Cotocus Private Limited
There is no end to education. It is not that you read a book, pass an examination, and finish with education. The whole of life, from the moment you are born to the moment you die, is a process of learning.
Latest posts by Ashwani Kumar (see all)
- Top 10 DevOps Trainers in the world - October 7, 2023
- What is Cookies and Why it is Used? - May 24, 2023
- TOP trends of transitions in TikTok - May 10, 2023