Will compression be machine learning's killer app? https://t.co/TMtCYzyT7U
— Pete Warden (@petewarden) October 16, 2018
Will compression be machine learning's killer app? https://t.co/TMtCYzyT7U
— Pete Warden (@petewarden) October 16, 2018
good post & links! Touches on gradient accumulation, gradient checkpointing (no, not the normal checkpointing), the nearly unambiguous superiority of distributed data parallel container in PyTorch, and the overall importance of understanding what's under the hood. https://t.co/2WYZRz9a2X
— Andrej Karpathy (@karpathy) October 16, 2018
Here is the same dynamic RNN implemented in 4 different frameworks (TensorFlow/Keras, MXNet/Gluon, Chainer, PyTorch). Can you tell which is which? pic.twitter.com/nsfuTULlKS
— François Chollet (@fchollet) October 16, 2018
These are extremely important techniques that I haven't seen written up elsewhere before.
— Jeremy Howard (@jeremyphoward) October 15, 2018
Many people still think batch size is limited by gpu ram, but that's not true. https://t.co/1WqizgTSJP
Huawei Goes the AI Way: this week's #ChinAI translation unpacks Huawei's newly announced AI strategy as well as its major moves into China's fast-growing surveillance tech market: https://t.co/PzcP8sVRXk
— Jeffrey Ding (@jjding99) October 15, 2018
It was a pleasure being interviewed by @AnalyticsVidhya for an episode of their DataHack Radio. We talked about all things #NLProc. See the post for a summary and the full interview. https://t.co/xj7C9w6frn
— Sebastian Ruder (@seb_ruder) October 15, 2018
Interesting, like most real-world wins for ML, it appears not to involve the technology that DeepMind has invested in developing (mostly deep RL) and instead consists of simply training a supervised learning system to forecast some variable of the subsequent time period
— Zachary Lipton (@zacharylipton) October 14, 2018
Amazing interactive graphic deconstructs the newest iPhone piece by piece and tells you what each component does. At the end of the article it is revealed that the $1099 phone costs about $390 in parts. That's a big fat margin! Interactive graphic: https://t.co/Fuj5nqgLwy pic.twitter.com/f5OvHoSVqI
— Simon Kuestenmacher (@simongerman600) October 13, 2018
There must be tens of thousands of PhD students working on AI/ML research problems. A large portion of them will be entering the workforce in a few years trying to solve real-world problems.
— hardmaru (@hardmaru) October 13, 2018
"This is one of the very first times people have tried to make common sense measurable” – #AI2 CEO @etzioni on new initiative with @DARPA to define, measure, and inspire research progress in common sense for #AI: https://t.co/XDwpdpMZ5p #CommonSenseAI
— Allen Institute for Artificial Intelligence (AI2) (@allen_ai) October 12, 2018
Interesting idea - one of our students has created a "license plate recognition as a service" company. https://t.co/stgTbyvb9Q pic.twitter.com/QBrXXSm5po
— Jeremy Howard (@jeremyphoward) October 12, 2018
You can call yourself an AI company when…. your notebook infrastructure is as complex as this! pic.twitter.com/FcXXhbmK4r
— Denny Britz (@dennybritz) October 12, 2018