MNIST-CPP-and-python

Beginner-friendly repo on how to Code a Simple Neural network with backprop in C++, bind it to python and train MNIST!

Summary

ReLu and Linear layers are implemented in C++ following PyTorch's naming convention. Some functionals like softmax are also implemented. The code is bound to python using pybind11. The model is trained on MNIST dataset using a python script.

How to run

1. Download the data via make download_mnist.
2. Install the python dependencies via pip install -r requirements.txt.
3. Compile the C++ code via make.
4. Run the python script via python train.py.
5. (Optional) Compare against pytorch training via python train_pytorch.py.