The OCR example uses the MNIST dataset, which can be downloaded here. Save the dataset in a folder and let’s start coding.
Check also here for a presentation on this problem.