Neural_Networks_from_Scratch
Neural Networks From Scratch - Using only Numpy
Paper: The Core Mathematics of a Neural Network | Video | Data
Get started
You can set up the environment with all dependencies like so:
conda create --name NN-FS python=3.8
conda activate NN-FS
pip install numpy pandas
High-Level structure
- Paper: Explores in more detail the core mathematics of the neural network.
- data: MNIST Dataset of handwritten digits
- model: Simple 3-Layer Neural Network
How to Run
-
Please download the python or jupyter-notebook file
-
Dowload the dataset - train.csv from (https://www.kaggle.com/competitions/digit-recognizer/data?select=train.csv), and place into a folder named mnist_data
- At this point your project structure should look like :
├── Neural_Network_from_Scratch.ipynb
├── Neural_Network_from_Scratch.py
└── mnist_data
└── train.csv
- Then either run the cells of jupyter notebook or run the python file using
python Neural_Network_from_Scratch.py
Disclaimer(s)
- The result produced by this code might be slightly different when running on a different GPU.