This project aims to deepen our understanding of neural networks by implementing them without any support from popular deep learning or machine learning libraries like TensorFlow, PyTorch, or Keras. Instead, we will rely on the fundamental numerical library, NumPy, to build and train neural networks from the ground up.
Gain in-depth knowledge of neural network architecture, forward and backward propagation, and gradient descent optimization. Develop the ability to troubleshoot and debug complex machine learning models. Enhance your understanding of how machine learning libraries like TensorFlow and PyTorch work under the hood.
To get started with this project, follow these simple steps:
$ git clone https://github.com/<yourusername>/NNFS.git
$ cd NNFS
$ pip install -r requirements.txtTo train the neural network using the provided sample data, follow these steps:
-
Make sure you have completed the installation steps as outlined in the Installation section.
-
Navigate to the project folder if you're not already there:
cd NNFS- Run the train script using the following command:
python train.py-
Monitor the training process. The script will print information about the training progress, including loss and accuracy, to the console.
-
Inspect the results. Once the training is complete, you can evaluate the trained model on the provided sample data and inspect the results as needed.