ordinarily

I think most of Google's deep research projects were done in the pursuit of pure science, not monetization or productization. In hindsight sure, it looks like they missed an opportunity. But not everything needs to be about money.

Google literally invented the boat (transformers) to be fair.

I mean, looking at the state of things right now in the US, I'd have to strongly disagree with you.

I just mean ontologically, it does not surprise nor terrify me that a machine built to simulate human output also simulates the worst of us.

Doesn't seem that surprising or terrifying to me. Humans come equipped with a lot more internal biases (learned in a fairly similar fashion), and they're usually a lot more resistant to getting rid of them.

The truly terrifying stuff never makes it out of the RLHF NDAs.

The transients were pretty easy to replicate yes. The nuclear testing stuff was pretty inconclusive but they have a much better curated collection of plates that aren't available yet.

Awhile ago I built my own ML pipeline to automate scanning these plates, it was very revealing. Beatriz and her team were very helpful.

https://arxiv.org/abs/2604.04810

They're there before the tests though, and potentially more frequent around nuclear testing calendar days. The argument has never been "these only showed up after a nuclear test."

It's genuinely a great introduction to LLMs. I built my own awhile ago based off Milton's Paradise Lost: https://www.wvrk.org/works/milton

Nope, they're surprisingly hard to get ahold of. So I've resorted to being extremely noisy online.

Temporal accumulation: imagine you're observing a signal through a narrow window and you can only see a partial, noisy snapshot each time. "Temporal accumulation" is what happens when you let the observer remember previous snapshots and use them to improve its prediction of the next one. The persistence advantage P measures how much that memory helps, the difference in prediction error between an observer that accumulates across episodes and one that only uses the most recent snapshot.

For a black hole shadow (EHT), P ≈ 0: each snapshot already contains the full picture, memory adds nothing. For gravitational wave strain (LIGO), P is large and positive, the chirp evolves across snapshots, so memory is essential. The question the papers ask is: what determines how much memory helps? The answer turns out to be spectral entropy of the waveform, not mass.

I've been doing this for about a month. I also have wildly complicated ML pipelines working similarly in parallel. When Karpathy's 'autoresearch' came out I was surprised by how novel it was treated.

Awhile I made something really dumb: https://www.warpstream.com/etc/terminal you can enter a 'gibson' command.

This is awesome! We got some really technical deep dives here https://www.warpstream.com/blog-category/engineering

Horizontal space is still a premium regardless of monitor size when designing/building for responsive viewports. Vertical space is almost zero cost in terms of design constraints.

Even on large monitors you'd be surprised the number of people at 150% zoom with small windows opened instead of fullscreen.

I used SVG animations (and sites like https://www.svgator.com/) long before stuff like Rive or Lottie was commonplace. SVG animations are great.

I think the general tone is more of a warning than an endorsement.

The pieces are coming together quickly https://ai-2027.com/.

I have a variety of early printings of this. My favorite being a 1931 edition illustrated by Major Felten, its beautiful.

Was hoping you'd appreciate our efforts to retain your original quirky vision. We named the rice ball Geoff as a homage to you (intentionally spelled the silly way). (https://warpstreamlabs.github.io/bento/docs/about)/

Lol I was actually looking at the Wiki earlier https://en.wikipedia.org/wiki/Bento_(database)

Pretty good community, Slack has 600+ users.