Read and references will be updated soon.

DLIO and actual workloads may run on different python environments. The scripts are provided for Polaris@ALCF. Please modify accordingly for different supercomputers.

• DLIO python environment

  source setup_dlio_env.sh

• ML workloads python environment

  source setup_ml_env.sh

• UNet3D

  This is for medical image segmentation from the reference implementation

• ResNet50

  We have implementations for both TensorFlow and PyTorch

  • TensorFlow: ./resnet50_tf/resnet50_hvd.py
  • PyTorch: ./resnet50_pt/resnet50_hvd.py and ./resnet50_pt/resnet50_ddp.py,

• CosmoFlow

  This is a workload adopted from Nvidia's submission of MLPerf HPC.

• Deepcam

  PyTorch implementation for the climate segmentation benchmark, based on the Exascale Deep Learning for Climate Analytics codebase here: https://github.com/azrael417/ClimDeepLearn, and the paper: https://arxiv.org/abs/1810.01993