Adding computer scientists to the list, and especially working in "algorithmic fairness." If you actually care about discrimination and injustice, now is your chance to do something. Black lives are not your intellectual curiosity. https://t.co/UvNU7Ltlyu
— Rediet Abebe ረድኤት (@red_abebe) May 30, 2020
After going through GPT-3 paper, I have to remind myself that we can also do amazing things with small compute. https://t.co/pgWqQqgsJf
— hardmaru (@hardmaru) May 29, 2020
I've been having fun with SIMD of late, specifically in Rust, so if you're interested in Rust, SIMD, ISPC, and how to benchmark them I've put together an introduction.https://t.co/nWGUDZAEXy
— Smerity (@Smerity) May 28, 2020
Tips for Research and a PhD by @seb_ruder
— hardmaru (@hardmaru) May 27, 2020
TL;DR:
• Read broadly
• Work on 2 things
• Be ambitious
• Collaborate
• Be proactive
• Blog
• Keep a source of positive energy
• Play to your strengths
• Intern / visit a university
• Play the long game https://t.co/jk5wpzf9dB
LaTeX.css is a CSS library that makes your website look like a LaTeX documenthttps://t.co/AHqsSarpbZhttps://t.co/OWaqRmYKmU https://t.co/kGNSfihWt9
— hardmaru (@hardmaru) May 25, 2020
”Most of the folks we interview know very little about basics. They just know how to use the packages on already prepared data.”
— hardmaru (@hardmaru) May 23, 2020
“When they do get a job at a descent company with good pay, but are asked to produce ‘mediocre’ models, they feel like losers.”https://t.co/a1X5aVwBit
Almost forgot about this but @YouTube's transcription sync is a really nice example of complementary human <-> ML interaction 💯
— Ines Montani 〰️ (@_inesmontani) May 22, 2020
1️⃣ Let human do human work (transcribe).
2️⃣ Assist human with efficient UI.
3️⃣ Let machine do what's easy once human work is done: align text + audio. https://t.co/5aYyKcGvbq pic.twitter.com/N15sbCJYFI
Many thanks to @MaxLenormand for this insightful article on optimizing GPUs for computer vision projects and competitions on Kaggle, via @TDataScience | [READ] https://t.co/Gjk7ldhMxj
— Kaggle (@kaggle) May 21, 2020
There are two kinds of people:
— Soumith Chintala (@soumithchintala) May 20, 2020
A) ones who get incrementally ambitious
B) ones who have a confident grand vision
I'm definitely (A), and most controversial people in AI are definitely (B).
Most of these grand visions are wrong, but they are there to guide and focus oneself.
Ethics in data science in the real world. Rebekah Jones is a hero. https://t.co/TKkNFb2DF1
— Chris Albon (@chrisalbon) May 20, 2020
We were all inexperienced & unskilled at some point. We have all made mistakes, and we will all make more mistakes in the future. But being a jerk is a choice. Don't be a jerk.
— François Chollet (@fchollet) May 19, 2020
TIL that If you typehttps://t.co/OT38A6XRmx
— Rafael Irizarry (@rafalab) May 19, 2020
in your browser's address bar, a new google doc starts up!
Works for https://t.co/PzkcsZN6mE https://t.co/SWsJAHiIOx https://t.co/qh3ghq1gzm https://t.co/pO1GIZWzhq as well.
It's been available for over a year.
