Skip to content

Latest commit

 

History

History
8 lines (5 loc) · 481 Bytes

README.md

File metadata and controls

8 lines (5 loc) · 481 Bytes

cnine

Lightweight C++ tensor library

Cnine is a simple C++/CUDA tensor library developed by Risi Kondor's group at the University of Chicago. Cnine is designed to make some of the power of modern GPU architectures accessible directly from C++ code, without relying on complex proprietary libraries.

Documentation for the Python/PyTorch API is at https://risi-kondor.github.io/cnine/.

Cnine is released under the custom noncommercial license included in the file LICENSE.TXT