23 - Quiz¶

• Your name:

• Your VT email:

Which of the following best describes hyperparameters?

• A. Parameters learned automatically during backpropagation
• B. Values computed from the training data before training
• C. User-defined settings that control the training process
• D. Outputs of the neural network during inference

Training for too many epochs can lead to:

• A. Overfitting
• B. Underfitting
• C. No learning
• D. Faster convergence without drawbacks

What is the main purpose of early stopping?

• A. To prevent overfitting and save computation
• B. To increase training time
• C. To stop training when gradients vanish
• D. To ensure the model reaches zero loss

Increasing batch size will most directly affect:

• A. The number of updates per epoch and compute efficiency
• B. The smoothness of the loss landscape seen by the optimizer
• C. The expressive capacity of the neural network
• D. The mathematical form of the objective being minimized