An interesting history of snake oil https://t.co/PiEfYzAtXD pic.twitter.com/NDuRVF1Zs6
— Rachel Thomas (@math_rachel) April 16, 2019
An interesting history of snake oil https://t.co/PiEfYzAtXD pic.twitter.com/NDuRVF1Zs6
— Rachel Thomas (@math_rachel) April 16, 2019
📝 Great write-up on types and uses of documentation.
— Mara Averick (@dataandme) April 16, 2019
❓ "What Docs When" by @gvwilsonhttps://t.co/Ju1jdT6TGU
/* 🖍 scribbles mine */ pic.twitter.com/sXNYC3e2y3
Why software projects take longer than you think – a statistical model https://t.co/RMinSWF7ga (blog post)
— Erik Bernhardsson (@fulhack) April 16, 2019
The heart of a @PyTorch training loop with callbacks. By aligning the training code and callback code, you can see exactly what's going on in each.
— Jeremy Howard (@jeremyphoward) April 14, 2019
Formatting code for understanding is too important to leave to automated tools or hard and fast rules. pic.twitter.com/hO5FXgTUC4
Surveillance is huge business in China and no surprise that it is the focus for computer vision startups such as Sensetime and Face++. IMO, these two companies may have some of the best ML research and engineering talent in the world, and it is worrying what they will do with it. pic.twitter.com/pFUKWK6Rg3
— hardmaru (@hardmaru) April 14, 2019
Fighting misinformation feels so depressing most of the time. No wonder so many fact-checkers experience burnout, and why any kind of technical assistance is god sent. https://t.co/6FgWyjfKH9
— Delip Rao (@deliprao) April 14, 2019
90% of the time, when an engineer (or researcher) uses X to solve Y, it's a case of "I kind of know X" rather than "I have evidence that X is an appropriate solution in this case"
— François Chollet (@fchollet) April 13, 2019
I have a deep (perhaps naive) belief that we could significantly improve US AI policy by having way more nerds come to Washington to talk about technical progress and the truth of it (both good and bad). I think sending lobbyists to tell pollyana-esque 'future is GREAT' = mistake
— Jack Clark (@jackclarkSF) April 12, 2019
Why is our entire employee handbook out in the open for anyone to read? Because evolving and explaining how we run our company is a public good, not a private act. Here are some recent updates, and some of the why’s behind the changes —> https://t.co/E7f80cWut2
— Jason Fried (@jasonfried) April 12, 2019
So glad they've beefed up the HR team…
— Mara Averick (@dataandme) April 12, 2019
Victim-blaming, sew FUD, don't leave a paper trail (Let's have a phone call…), try to keep dissidents from knowing about each other (One-way webinar, anyone?)
👏 Playbook material.https://t.co/VonptEg54j
Kaiming He's original residual network results in 2015 have not been reproduced, not even by Kaiming He himself. https://t.co/piSvPx9nDz
— /MachineLearning (@slashML) April 12, 2019
Several reviews of Deborah Mayo’s new book, Statistical Inference as Severe Testing: How to Get Beyond the S https://t.co/pwamtK9fvs
— Andrew Gelman (@StatModeling) April 12, 2019