This repository is a MUSA-adapted version of the Jittor deep learning framework, intended to run Jittor workloads on Moore Threads MUSA devices.
Jittor is a just-in-time compiled deep learning framework. This fork keeps the core Jittor programming interface and adds/adapts backend support for Moore Threads MUSA hardware.
Install from the repository root:
pip install .After installation, you can import Jittor as usual:
import jittor as jt
print(jt.__version__)- This project is based on Jittor and is not the official upstream Jittor repository.
- Please make sure your Moore Threads driver, MUSA runtime, and related toolchain are installed correctly before using the MUSA backend.
- The package name remains
jittor, following the original project layout.
If you use this project in academic work, please cite the original Jittor project:
@article{hu2020jittor,
title={Jittor: a novel deep learning framework with meta-operators and unified graph execution},
author={Hu, Shi-Min and Liang, Dun and Yang, Guo-Ye and Yang, Guo-Wei and Zhou, Wen-Yang},
journal={Science China Information Sciences},
volume={63},
number={222103},
pages={1--21},
year={2020}
}This project is derived from and deeply indebted to the original Jittor/Jittor project. Thanks to the Jittor authors and contributors for their excellent open-source work.
This repository follows the license terms of the original project. See LICENSE.txt for details.