Problem
Although the accuracy level is pretty high, can you improve it further? Even for a tiny fraction? To improve the result, experiment with:
i. Increase the number of epochs and then experiment with or without GPU support on Google Colab. What do you notice?
ii. Adding additional layers: 1, 2, and 3 layers. Make sure that your model runs. This maybe a bit tricky to do. Send any questions that you may have.
iii. What happens to accuracy and training Time after adding one, two, and three layers?