Data Science platform

• This is a directory tree for documenting all about developing data science workflows on KSL platforms

• a bespoke quickstart guide to work with datasceince workloads

• We will list the information about the meta modules and their contents.

• We will list about the details on how to install software as self-service model using pacake managers like conda, spack, containers etc.

• Cray’s Machine Learning development enviroment will also be discussed here

• Hyperparameter libraries will be disucssed here

• Example jobscript related to distributed frameworks

• A high level introduction to NGC container registry

• parallel data processing tools and techniques will be discussed, e.g. dask and rapids on CPUs and GPUs

• Accelerating machine learning using multithreaded ScikitLearn

  • porting ML from CPUs to GPUs

• Deep Learning on AMD Genoa CPUs – for those workloads which don’t qualify for GPUs or are developing models and modest sized datasets