Questions

Let’s test what we’ve learned in this chapter using the CIFAR-10 notebook:

1. Build a neural network using our three-step approach.
2. Increase the number of neurons from 5 to 100 in the hidden layer.
3. Use a custom callback to stop training when the training accuracy is 90%.
4. Try out the following learning rates: 5, 0.5, 0.01, 0.001. What did you observe?