View profile

Deep Learning Weekly - 🤖 - Issue #83: Nature ML, PyTorch 1.0, TPU vs V100, L2 vs Weight Decay, MLPerf Benchmark and Swift for TensorFlow...

Revue
 
Welcome back to another tempestuous week in Deep Learning! There was a large discussion around the up
 
May 9 · Issue #83 · View online
Deep Learning Weekly
Welcome back to another tempestuous week in Deep Learning!
There was a large discussion around the upcoming Nature ML journal, PyTorch 1.0 was announced, RiseML compared TPUv2 and V100, Google promises free GCP credits for researchers and a new ML benchmark was released from some major players. Ian Goodfellow talked about failure in an interesting interview, we found an interesting article on L2 vs weight decay, as well as an overview of stochastic weight averaging and the introduction of Swift for TensorFlow.
As always, we hope you’ll enjoy reading as much as we did and would appreciate you sharing this newsletter with friends and colleagues.
See you later this week!

Industry
Statement on Nature Machine Intelligence
Comparing Google’s TPUv2 against Nvidia’s V100 on ResNet-50
Google Cloud Platform announces new credits program for researchers
MLPerf – Will New Machine Learning Benchmark Help Propel AI Forward?
Join the fastest growing Deep Learning Developer Community
Learning
How I Fail - Ian Goodfellow
Weight decay vs L2 regularization
Stochastic Weight Averaging — a New Way to Get State of the Art Results in Deep Learning
Libraries & Code
Announcing PyTorch 1.0 for both research and production
Introducing Swift For TensorFlow
Tacotron 2 - PyTorch implementation with faster-than-realtime inference
Papers & Publications
AI Safety via Debate
DeepMind papers at ICLR 2018
Did you enjoy this issue?
If you don't want these updates anymore, please unsubscribe here
If you were forwarded this newsletter and you like it, you can subscribe here
Powered by Revue