Flax Models
A collection of pretrained models in Flax.
About
The goal of this project is to make current deep learning models more easily available for the awesome Jax/Flax ecosystem.
Models
- GPT2 [model]
- StyleGAN2 [model] [training]
- ResNet{18, 34, 50, 101, 152} [model] [training]
- VGG{16, 19} [model] [training]
- FewShotGanAdaption [model] [training]
Installation
You will need Python 3.7 or later.
- For GPU usage, follow the Jax installation with CUDA.
- Then install:
> pip install --upgrade git+https://github.com/matthias-wright/flaxmodels.git
For CPU-only you can skip step 1.
Documentation
The documentation for the models can be found here.
Checkpoints
The checkpoints are taken from the repositories that are referenced on the model pages. The processing steps and the format of the checkpoints are documented here.
Testing
To run the tests, pytest needs to be installed.
> git clone https://github.com/matthias-wright/flaxmodels.git
> cd flaxmodels
> python -m pytest tests/See here for an explanation of the testing strategy.
Acknowledgments
Thank you to the developers of Jax and Flax. The title image is a photograph of a flax flower, kindly made available by Marta Matyszczyk.
License
Each model has an individual license.
