cnine¶
cnine is a lightweight C++ tensor library with a CUDA backend. The primary purpose of cnine is to provide provide direct control over GP-GPU functionality to Python and C++ developers. This document gives a high level introduction to cnine’s PyTorch interface.
cnine is developed by Risi Kondor and released for noncommercial use only under a custom license that can be found in the file LICENSE.TXT
Not all features in the C++ library are available through the PyTorch interface.
The documentation of the C++ API can be found in pdf format in the package’s doc directory.