/neural-bitmapper

Primary LanguagePythonMIT LicenseMIT

The idea

What if I used stochastic gradient updates on an XNOR+bitcount binary neural network and trained it to implicitly represent an image?

Requirements

I develop inside of the January 2024 edition of the Nvidia PyTorch Docker image. docker run -it -d --gpus all -v /workspace:/workspace nvcr.io/nvidia/pytorch:24.01-py3