Search Results for author: Adrià Armejach

Found 2 papers, 0 papers with code

A Dynamic Approach to Accelerate Deep Learning Training

no code implementations25 Sep 2019 John Osorio, Adrià Armejach, Eric Petit, Marc Casas

The first approach achieves accuracy ratios slightly slower than the state-of-the-art by using half-precision arithmetic during more than 99% of training.

Cannot find the paper you are looking for? You can Submit a new open access paper.