Finally, a special thank you to Jeremy Howard, Hamel Husain, and Wasim Lorgat for their work on nbdev. It’s an incredible tool that removes so many of the barriers to development work. Even the lawyers are coding now. Amazing!

Obiter is an open-source Python library of computational research tools. You can use the library to interact with @CanLii’s API to get metadata about each case in their database. Stay tuned: more features are coming. obiter.ai

What are programming things you think the data science community as a whole could use more or understand better to improve their programming ability?

Examples: decorators, higher order functions & callbacks, vectorizing, parallel processing, fastcore, etc. (whatever YOU think)

Amos is a new optimizer that we propose to pre-train large language models. It is more efficient and converges faster than AdamW: ≤ 51% memory for slot variables, and better valid loss within ≤ 70% training time!

New from Google Research. Preprint: arxiv.org/abs/2210.11693

I haven't disclosed this previously as I didn't want to get into the debate about the severity of COVID.

Yes, I had a BA.2 infection prior to BA.2 surging in the United States. I decided to meet a friend after they escaped Ukraine in early March.

🧵

Here my attempt to apply Random Forest on a practical Industrial issue: classify Aluminium scraps boxes. I've figured out to be a good baseline. Just combining the right models could drive to incredible results. b8ni.github.io/bottoni/fastai…

Interesting new podcast that discusses my tweet, questioning whether adding numpy-style broadcasting rules to APL would be a good idea

Magnus blunders a bishop on move 1!🙃

Completely agree. The search result for any vaguely consumer thing like 'What are some good ebikes?' or 'Where are fun places to go in Spain?' are useless piles of SEO'd gibberish.

Do the same search on Reddit and you get a ton of interesting results.

Please know your disabled/high risk friends are not ok.

You can’t collectively banish us from society and bombard us daily with messages that our lives don’t matter/society doesn’t need us/everyone’s better off without us wrecking their normal, & expect that to have no effect.

Just received the first copy of my book (3rd edition), woohoo! 🦎🎉😁
You can get it at:
homl.info/amazon3
Play with the notebooks at:
github.com/ageron/handson…
See what's new at:
homl.info/changes3
Huge tx to everyone who made it happen!!! 🙏
Hope you'll find it useful!

A dream come true 😍

Jarvislabs - Featured in the latest version of the fast.ai course as a 'platform that several fastai users love & built by an alumni'

JarvisLabsAI - Only startup in the list - bootstrapped 😀 & built in India 🇮🇳

🙏 for all ❤️

(a 🌶️spicy 🌶️ thread.) dear EAs who are trying to figure out if I'm either on-side or an uncharitable bad faith critic: I'm neither! I have a lot of EA-aligned beliefs but the vibes from EA are concerning to me. 1/

Here's a statistic that everyone should know.
wired.com/story/covid-19…

Now that ML is so much about scaling up on all fronts, and so little about ideas, scientific awards (e.g. 'NeurIPS oral') should be resource normalized.

Student and independent researcher led research should NOT be evaluated the same way as industry-backed research. Period.

Screenshot says

COWEN: As I understand your views, you're a fairly pure Benthamite utilitarian. Is that correct?
BANKMAN-FRIED: That's correct.
COWEN: Biven that that's the case, as I see it, the replacement costs of human life are pretty low, so you could spend a modest...

ITT: an OAI employee admits that the text-davinci API models are not from their papers.

Until OpenAI actually documents the connection between the models in their papers and the models released via APIs, #NLProc researchers need to stop using them to do research.