15_OptimizeAIML

Optimizing AI/ML Workflows in Python for GPUs

Optimizing AI/ML Workflows in Python for GPUs is the fourteenth workshop in this series and was presented by Daniel Howard (CISL's Consulting Services Group) on August 25th. The video recording for this session will be uploaded here. Topics covered include:

1. Background on machine learning research in Earth sciences
2. Setting up Python virtual conda environments
   • The RAPIDS AI software suite
   • GPU enabled TensorFlow and PyTorch
3. Enabling tuning and profiling with TensorFlow and PyTorch
4. Profiling with DLProf and performance optimizations for NVIDIA Tensor Cores

This session's notebook and a html/PDF version of the notebook are available for use and reference in this directory.