I've written this code to understand the basics of neural networks. The initial file, mnist.py, is adapted from Michael Nielsen's book 'Neural Networks and Deep Learning' (http://neuralnetworksanddeeplearning.com). This version uses a single data point per optimization step, leading to slow optimization.
To address this, I've created mnist_batch.py, which updates weights in batches for faster optimization. Additionally, mnist_dropout.py demonstrates how to implement dropout within a neural network.
Thanks.