A lightweightgan trained briefly on
https://huggingface.co/datasets/johnowhitaker/colorbs
See
https://huggingface.co/johnowhitaker/orbgan_e1 for training script and so on, since this was basically just copying that and running on a new dataset.
Note: lightweightgan code was updated between training orbgan_e1 and this one, so if you're trying to run the CPU inference notebook you'll get errors. See an updated version running this model on a CPU here:
https://colab.research.google.com/drive/16XKJ7XZeSI0rvUf1GU6m9qrmwr1pMRWy?usp=sharing