A Highly Efficient Automatic Parallel Distributed Deep Learning System

Hetu is a high-performance distributed deep learning system targeting trillions of parameters DL model training, developed by DAIR Lab at Peking University. It takes account of both high availability in industry and innovation in academia, which has a number of advanced characteristics:

  • Applicability. DL model definition with standard dataflow graph; many basic CPU and GPU operators; efficient implementation of more than plenty of DL models and at least popular 10 ML algorithms.

  • Efficiency. Achieve at least 30% speedup compared to TensorFlow on DNN, CNN, RNN benchmarks.

  • Flexibility. Supporting various parallel training protocols and distributed communication architectures, such as Data/Model/Pipeline parallel; Parameter server & AllReduce.

  • Scalability. Deployment on more than 100 computation nodes; Training giant models with trillions of model parameters, e.g., Criteo Kaggle, Open Graph Benchmark

  • Agility. Automatically ML pipeline: feature engineering, model selection, hyperparameter search.

We welcome everyone interested in machine learning or graph computing to contribute codes, create issues or pull requests. Please refer to Contribution Guide for more details.