Case Studies


Scalable deep learning

Check out the talk on scalable deep learning introducing Hyperband for hyperparameter optimization and Paleo, an analytical performance model for deep learning.

Addressing the challenges of massively parallel hyperparameter optimization

Follow how one engineer views and navigates another teammate’s experiments, and continue training from a previously trained model.

Random search and reproducibility for neural architecture search

Watch Ameet Talwalkar’s talk on his recent research demonstrating both the potential promise of Neural Architecture Search (NAS) along with the current immaturity of the field.

Taming the deep learning workflow

Learn about why we're in the dark age of AI infrastructure, and how to dramatically improve deep learning workflows via novel algorithmic and software solutions.