Build a neural network using numpy only.
-
MNIST Dataset
- I use the MNIST database which consists of images of handwritten numbers and their labels.
-
Model Architecture
- Input layer: 768
- hidden layer: 64
- output layer: 10
-
Hyperparameters
- You can modify hyperparameters such as batch size and learning rate in train.py.
-
Prerequisite
pip install -r requirements.txt
python3 train.py
Thank Chia-Hung Yuan and Yu-Rong Zhang for helping me solve bunch of problems.
Take a look at Chia-Hung's repo: Neural-Networks-from-Scratch