Hi,
after going thouroughly through part 1 of the course I am now doing a review from the begining and I decided to make a kind of review document for lesson 3, the MNIST dataset to help me out. Its a jupyter notebook, the first part is the Code Summary part, which is basically just the code for each of the 3 models in the chapter 4 of the book; the pixel similarity, the linear model and the simple neural network. There is also a Detailed Summary section below that, where it’s a review of the entire chapter 4 of the book with code stuff, summaries of sections and extra details on other stuff. Its not really a summary because its pretty detailed, but writing it out did help me with the theory.
I also answered to most of the questions at the bottom.
If anyone is interested you can see it on kaggle or github (ipynb or pdf):