Assessment: Deep Learning

Explain

• What happens during forward propagation?
• What happens during backpropagation?
• Why do activation functions matter?
• What causes vanishing gradients?
• Why does regularization help?

Build

• Design a classifier training pipeline.
• Explain how you would choose batch size and learning rate.
• Explain how you would export and serve a model.

Debug

• Training loss decreases but validation loss increases. What is happening?
• Loss does not decrease. What do you check first?
• Inference output differs from training validation output. What could be wrong?

Completion Gate

• Can train a small model.
• Can debug loss curves.
• Can serve the model through an API.