Simon @spoltier@qoto.org

Llama 3.2 is out, and it's a much more substantial release than the 3.1 to 3.2 version bump might indicate

Four new models, including Meta's first two vision models (11B and 90B) and two new text-only small models (1B and 3B)

My notes so far: https://simonwillison.net/2024/Sep/25/llama-32/

RTO or GTFO? Last week, Amazon announced that starting in 2025, all employees will be in a physical office 5 days a week. While putatively to strengthen culture (as "the world's largest startup"), this is proving unpopular with many. Will other companies follow suit? On today's episode, @ahl and I are going to talk about remote work, RTO mandates -- and the physicality of both innovation and organization. Join us and share your perspective, 5p Pacific!

https://discord.gg/QrcKGTTPrF?event=1287787453599584327

@TomF I'm right handed and started having minor issues with my wrist while using a mouse - so about 15 years ago I started switching my mouse hand on my work machine to left and (back when we still went to offices) my home machine stayed mouse-right.

It took a few weeks to really get used to it, but it wasn't bad. And now I'm mouse-ambidextrous and generally free of wrist troubles.

In the course of my unachievable hobby project working on SAT model counting I've figured out a neat trick, which is a reasonably efficient method for doing set operations on SAT problems in CNF form (without increasing the number of variables). e.g. taking the intersection or set difference of two SAT problems.

Is this a known thing? I've not seen it before but it seems like it must be.

@sandofsky

"The fairy photon enters the aperture and sings as it makes its way to the CMOS chip.

What's that? The permission troll just crawled out from under the Broadcom WiFi chip where it spends most of its time punching mean Internet traffic, and swatted the fairy photon away!

Oh no!

Halide needs your help, fair wise iPhone owner.

Close your eyes and make a wish and tap your heels together three times.

Tell the permission troll to stay away! Let the fairy photons come home to CMOS."

Dickens: It was the best of times, it was the worst of times.

Schrödinger: Nice.

Here’s a statement from Bloom’s 1986 /The Closing of the American Mind/, with which I’m unhealthily obsessed:

“The best point of entry into the very special world inhabited by today’s students is the astonishing fact that they usually do not, in what were once called love affairs, say, ‘I love you.’

Were you a student in 1986? Did you, at that time, say “I love you” to a romantic partner?

Mastodon instance with a "scramble suit" functionality where every time you make a post it comes from a random account handle with a random display name and a random avatar

Ok I made this into a blog post, you're welcome

https://hazelweakly.me/blog/cache-me-not-cache-me-cache-me-not/

But seriously, caching is hard, it's really hard, but you can make life WAY easier for yourselves when building an SPA if you do this super simple thing.

Break down all your content into two axis:
1. push vs pull
2. owned vs user

Whenever possible, turn your pull assets into push assets and all of your user assets into owned assets.

What do I mean by that? Let's break it down further 👇

If you’re generating code, and you’re *not* doing it with an LLM, is it reasonable to use metrics like F1 and recall to measure how well the tools you use are doing? This is bothering me because it feels a bit weird to apply metrics like this to static analyses, build tooling frameworks, or things that just plain don’t have any recall to begin with.

Earlier this year, I worked on a side project to hack a car in JavaScript and finally found the energy to write the blog post about it! 🚙 📡

https://charliegerard.dev/blog/replay-attacks-javascript-hackrf

I have played a little bit with OpenAI's new iteration of GPT, GPT-o1, which performs an initial reasoning step before running the LLM. It is certainly a more capable tool than previous iterations, though still struggling with the most advanced research mathematical tasks.

Here are some concrete experiments (with a prototype version of the model that I was granted access to). In https://chatgpt.com/share/2ecd7b73-3607-46b3-b855-b29003333b87 I repeated an experiment from https://mathstodon.xyz/@tao/109948249160170335 in which I asked GPT to answer a vaguely worded mathematical query which could be solved by identifying a suitable theorem (Cramer's theorem) from the literature. Previously, GPT was able to mention some relevant concepts but the details were hallucinated nonsense. This time around, Cramer's theorem was identified and a perfectly satisfactory answer was given. (1/3)

Omg I'm laughing so much 😂😂😂😂
https://onerpm.link/EatingTheCats

#Music #Song

"I don't want to live in a world where five companies dictate everything we do."

#Nextcloud founder kicking off the conference