pyronn-layers

This is a Python wrapper around PYRO-NN-Layers. Install it via

pip install pyronn-layers

Or if you downloaded this repository

pip install -e .

It uses pystencils-autodiff to automatically compile the C++/CUDA source files. The code gets automatically recompiled if you make changes to the source files in the folder src/pyronn-layers/PYRO-NN-Layers.

If you prefer pre-build versions of PYRO-NN-Layers, go to the release page of PYRO-NN-Layers or download the latest artifacts from our CI (links only working if pipeline badge above is green):

• pyronn-layers.so for Ubuntu, tensorflow_gpu-1.14.0-cp37-cp37m-manylinux1_x86_64.whl
• pyronn-layers.so for Ubuntu, tf-nightly-gpu-2.0-preview.*-cp37-cp37m-manylinux1_x86_64.whl

Requirements

You have a functioning toolchain for NVCC installed. nvcc must be available in PATH. If compilation fails, you can change the selected compiler: http://pycodegen.pages.walberla.net/pystencils/sphinx/configuration.html?highlight=compiler