Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Picture for Abhinav Venigalla

Representation range needs for 16-bit neural network training


Apr 06, 2021
Valentina Popescu, Abhinav Venigalla, Di Wu, Robert Schreiber

* 7 pages, 3 figures, 4 tables; conference 

  Access Paper or Ask Questions

Adaptive Braking for Mitigating Gradient Delay


Jul 10, 2020
Abhinav Venigalla, Atli Kosson, Vitaliy Chiley, Urs Köster

* In Beyond First Order Methods in ML Systems workshop at the 37th International Conference on Machine Learning, 2020 

  Access Paper or Ask Questions

Pipelined Backpropagation at Scale: Training Large Models without Batches


Mar 25, 2020
Atli Kosson, Vitaliy Chiley, Abhinav Venigalla, Joel Hestness, Urs Köster


  Access Paper or Ask Questions