Notebooks to provide trade studies and visualizations for developing the final OPERA dist-s1 algorithm
mamba env update -f environment.yml
conda activate dist-s1
python -m ipykernel install --user --name dist-s1
Use the environment environment-gpu.yml. Ostensibly, it removes some of the leafmap/flask dependencies and adds pytorch-cuda. I found that conda-forge distributions were most reliable for ensuring cuda compatibility (i.e. cuda driver from GPU with pytorch). Still, the pytorch and nvidia channels are prioritized, but below conda-forge. This is WIP.
To use the well-known denoiser, please use Rosetta.
CONDA_SUBDIR=osx-64 conda create -n dist-s1-intel
conda activate dist-s1-intel
python -c "import platform;print(platform.machine())" # Confirm that the correct values are being used.
conda config --env --set subdir osx-64
Create your own directory with your last name and do as you please. Don't mess with other people's work. This is a poorly versioned controlled repository and meant to provide sample code and prototypes that can be distilled down later.