Minimalistic set of tools for Deep Learning. The project is used as part of video course on YouTube. (English|Russian)
!NB, this repository is for educational purposes and is not intended for production use. For production use PyTorch, Tensorflow, etc.
- Create a new environment
python -m venv venv
- Activate the new environment (for Windows the activate script will have a different name)
source venv/bin/activate - Install dependencies
pip install -r requirements.txt
There are 2 implementations of a feedforward neural networks so far: sequential and vectorized.
plaindl - Implements the training algorithm on pure python.
The purpose of this realisation is just to show, that there's no "magic" in the neural networks.
mindl - Implements the training algorithm using only numpy for matrix computations and code simplification.
You could either run the run.py script or start JupyterLab and check out more examples:
jupyter labIf the Lab is not opened automatically, then open http://localhost:8888/lab