Welcome to Metinor, your one-stop framework for neural network development, optimization, and visualization. Metinor is built on top of PyTorch and PyTorch Lightning, leveraging their flexibility and ease-of-use.
The name "Metinor" is a fusion of "Metis," the Greek Titaness of wisdom, and "Valinor," the undying lands in J.R.R. Tolkien's literary universe where eternal wisdom and beauty prevail. Metinor aims to be the wise and enduring sanctuary for all your neural network needs, offering a suite of powerful tools and capabilities.
Metinor is an extensible framework that allows you to easily develop, visualize, and optimize neural networks. It offers the following features:
- A modular, object-oriented design that allows you to easily extend the framework with your own modules.
- A powerful visualization tool that allows you to visualize the structure of your neural network.
- A set of optimization algorithms that allow you to prune, quantize, and compress your neural network.
To install Metinor, follow these steps:
- Clone the repository to your local machine.
- Create a new virtual environment using
python -m venv venvand activate it. - Install the required dependencies using
pip install -r requirements.txt. - Install OpenMMLab packages using
mim install -r requirements-mim.txt. - Install MMPose using
pip install "mmpose==1.1.0" - Install the package using
pip install ..
To contribute to Metinor, follow these steps:
- Clone the repository to your local machine.
- Create a new virtual environment using
python -m venv venvand activate it. - Install the required dependencies using
pip install -r requirements.txt. - Install OpenMMLab packages using
mim install -r requirements-mim.txt. - Install MMPose using
pip install "mmpose==1.1.0" - Install the development dependencies using
pip install -r requirements-dev.txt. - Install the package using
pip install -e ..
To run the unit tests, run pytest in the root directory of the repository.
Metinor is based on several open-source projects, including: