Jittor: a novel deep learning framework with meta-operators and unified graph execution

Shi-Min Hu, Dun Liang, Guo-Ye Yang, Guo-Wei Yang, Wen-Yang ZhouScience China Information Sciences 2020 · OpenReview

Abstract

This paper introduces Jittor, a fully just-in-time (JIT) compiled deep learning framework. With JIT compilation, we can achieve higher performance while making systems highly customizable. Jittor provides classes of Numpy-like operators, which we call meta-operators. A deep learning model built upon these meta-operators is compiled into high-performance CPU or GPU code in real-time. To manage metaoperators, Jittor uses a highly optimized way of executing computation graphs, which we call unified graph execution. This approach is as easy to use as dynamic graph execution yet has the efficiency of static graph execution. It also provides other improvements, including operator fusion, cross iteration fusion, and unified memory.
BibTeX
@article{hu2020jittor,
  author = {Shi-Min Hu and Dun Liang and Guo-Ye Yang and Guo-Wei Yang and Wen-Yang Zhou},
  title = {{Jittor: a novel deep learning framework with meta-operators and unified graph execution}},
  journal = {Science China Information Sciences},
  year = {2020},
  url = {https://openreview.net/forum?id=E1xAxIHdT9},
}

Loading related content...

Figure