biot

This is the LLM equivalent of a riddle, eg: “A farmer has 17 sheep. All but 9 die. How many are left?”

Depending on how you define 3D printing: braille.

The article says:

The full menagerie of shapes is too diverse to get a handle on, so mathematicians tend to focus on convex polyhedra

The phrase "tend to focus on" suggests it's not an exclusive thing. However, you're right -- it appears that the Rupert property only applies to convex polyhedra, so the article title and text is at the very least incomplete given that a sphere is a shape.

Presumably a simple sphere would trivially qualify as being unable to pass through itself.

Serverless Advocate: Yes, but instead of paying for infrastructure overhead and hiring 5–10 highly specialized k8s engineers, you pay AWS to manage it for you.

This argument lost me. If you’re running your own k8s install on top of servers, you’re doing it wrong. You don’t need highly specialized k8s engineers. Use your cloud provider’s k8s infrastructure, configure it once, put together a deploy script, and you never have to touch yaml files for typical deploys. You don’t need Lambda and the like to get the same benefits. And as a bonus, you avoid the premium costs of Lambda if you’re doing serious traffic (like a billion incoming API requests/day).

Every developer should be able to deploy at any time by running a single command to deploy the latest CI build. Here’s how: https://engineering.streak.com/p/implementing-bluegreen-depl...

If you buffer starting with the ( character, then you'd still send the [text] part of the link, and worst case is that with no matching ) character to close the link, you end up buffering the remainder of the response. Even still, the last step is "flush any buffered text to the client", so the remainder of the response will be transmitted eventually in a single chunk.

There are some easy wins that could improve this further: line endings within links are generally not valid markdown, so if the code ever sees \n then just flush buffered text to the client and reset the state to TEXT.

There are a few things in our implementation that make a more general solution unnecessary. We only need the output to support a limited set of markdown which is typically text, bullet points, and links. So we don't need code blocks (yet).

However, the second thing (not mentioned in the post) is that we are not rendering the markdown to HTML on the server, so []() markdown is sent to the client as []() markdown, not converted into <a href=...>. So even if a []() type link exists in a code block, that text will still be sent to the client as []() text, only sent in a single chunk and perhaps with the link URL replaced. The client has its own library to render the markdown to HTML in React.

Also, the answers are typically short so even if OpenAI outputs some malformed markdown links, worst case is that we end up buffering more than we need to and the user experiences a pause after which the entire response is visible at once (the last step is to flush any buffered text to the client).

This has long been my favorite entry! Still on my todo list to go through and fully understand it.

Maybe as an eventually consistent life form using extremely slow message passing. Though gravity becomes a major factor that would limit the size unless it’s incredibly sparse.

One of my favorite episodes of Love, Death, & Robots is “Swarm”. Worth a watch.

While the search offered is handy, I watch logs on multi-pod workloads via:

    kubectl logs -f -l app=api --max-log-requests=50

    kubectl logs -f -l app=api --max-log-requests=50 | grep ERROR

You can run ad blockers in mobile Safari. This one works great: https://apps.apple.com/us/app/firefox-focus-privacy-browser/... -- it's both a browser itself as well as acting as a content blocker for Safari and does a fantastic job at blocking ads.

The revenue milestones appears to show annual revenue goals rather than monthly. For example, if you have only 1 plan at $1/month, the $1000/month target shows you need 84 subscribers.

Fun fact: it's a different device, but the principle is the same as the device used in the documentary Tim's Vermeer. It results in the images overlapping between your left and right eye and you simply paint until the difference goes away.

However, if it's inside a plastic case that blocks light, it doesn't.

You want an ordinary diode to allow current to flow easily when it senses light? Simple: shine a powerful laser at the plastic-encased diode and it will melt the plastic and liquify the metal, fusing it together and allowing current to flow again. See? You just needed to try harder.

Seems like all these other factors should matter too!

The way I read it was that there are some minerals that are no longer there to begin with because they're simply not in the soil. For nutrients (like vitamin C) created as part of the growing process, speed is the only factor that matters.

So yes, it does matter but unless you have a way of knowing which farm(s) the produce came from and what their soil mineral content is, that's not something you can factor into your purchasing decision.

I think an IDE, a keyboard, and a projector are better than printing code blocks at a specific revision which is identified by a computer-readable id, and which must be given a new ID and a new printed page every time you want to try executing a new version.

You're making several incorrect assumptions here:

1. That you can't interactively try out the code as you're editing it.

2. That the system as implemented is the final vision of how the system ought to work forever.

From https://youtu.be/5Q9r-AEzRMA?t=150 "Anyone can change any program at any time and see the changes immediately", which demonstrates live editing the code and seeing the projected flow lines change color. So you can keep editing and iterating on the program, trying out your changes without ever having to print anything. Once you are satisfied with your improvements, you then "commit" the code, which results in the system printing out a new page with its new identifier.

And if any part of your expectations isn't how things work, it's likely because this is a research project and nobody has written the code to make it behave the way you'd like. Since Realtalk is built on itself, one would only need to make the appropriate changes to improve the system.

From the article:

"This is a big object. An object of this size is going to have the equivalent impact energy in the hundreds of megaton approaching a gigaton," Brown said. "That'd be a regional impact. It's the sort of thing that if it hit the east coast of the U.S., you would have catastrophic effects over most of the eastern seaboard. But it's not big enough to affect the whole world."

The interviewee touches on this:

And so I would say, in response to what’s often asked, which is: If depression isn’t simply an abnormality in serotonin, why would you prescribe an SSRI? And I think the answer is that through serotonin and norepinephrine, we can promote the brain’s capacity for resilience against the adverse effects of stress and depression on the brain.

I suspect a lot of people overestimate how special their code is. Also, if your code exists in a private repo on GitHub, then you're already trusting the same third party when using GitHub Copilot.

If some company’s alcoholic beverage doesn’t get approved, you wouldn’t say “finally some faith in the firearms system” just because the Bureau of Alcohol, Tobacco, Firearms, and Explosives handles both.

From the paper’s introduction:

Machine learning (ML) models for materials science have the potential to vastly expedite the computational discovery process. State-of-the-art ML models can predict the results of physics-based quantum mechanical calculations but are several orders of magnitude faster, making them ideal for predicting general material properties [5–7]. In addition to direct property prediction, combining universal ML potentials such as M3GNet [8], CHGNet [9], and GNoME [10] has made it possible to perform geometric optimization, and hence evaluate thermodynamic stability, for arbitrary combinations of elements and structures. The significant speed advantage of ML-based techniques over direct simulation has made it possible to explore materials across a vast chemical space that greatly exceeds the number of known materials.

Models mentioned:

M3GNet: M3GNet is a new materials graph neural network architecture that incorporates 3-body interactions. A key difference with prior materials graph implementations such as MEGNet is the addition of the coordinates for atoms and the 3×3 lattice matrix in crystals, which are necessary for obtaining tensorial quantities such as forces and stresses via auto-differentiation.

Source: https://materialsvirtuallab.github.io/m3gnet/

CHGNet: A pretrained universal neural network potential for charge-informed atomistic modeling (see publication). Crystal Hamiltonian Graph neural Network is pretrained on the GGA/GGA+U static and relaxation trajectories from Materials Project, a comprehensive dataset consisting of more than 1.5 Million structures from 146k compounds spanning the whole periodic table.

Source: https://github.com/CederGroupHub/chgnet

GNoME: GNoME Is A State-Of-The-Art Graph Neural Network (GNN) Model. The Input Data For GNNs Take The Form Of A Graph That Can Be Likened To Connections Between Atoms, Which Makes GNNs Particularly Suited To Discovering New Crystalline Materials.

Source: https://gmnomeai.cloud/

I didn't like any of the sardukar fighting scenes, or how the ships exploded

I thought the ship explosion effect was brilliant and realistic in terms of how it would actually work if energy shields were a thing in real life. Just like how the slow blade penetrates a personal shield, the slow missile penetrates the ship’s shield. The movie shows this effect and the consequences. Once the missile gets past the shield, it explodes and the explosion is initially contained by the shield. However, the explosion spreads internally and it takes a second to take out the shield generator, at which point the explosion is no longer contained.

Just as an energy shield contains external explosions to the outside, it would also have the same effect on the inside for the brief moment the shield generator still operates.

suggestion: when you click to loop a sound, the URL should update with the current state of everything you have selected. Then you could send the URL to other people and they could listen to your garbage.

The bike shown in the first image is Specialized Sirrus X. You can make out from the image of the manual that it says "spacer/axle/bolt specifications". Searching for this yields the following Specialized bike manual which is similar: https://www.manualslib.com/manual/1974494/Specialized-Epic-E... -- there are some notable differences, but the Specialized Sirrus X manuals that are online aren't in the same style.

The prior page (8) shows "SEAT COLLAR 4mm HEX" and, based on looking up seat collar in an image search, the part in question matches.

In terms of the toolbox, note that it only identified the location of the Allen wrench set. The advice was just "Within that set, find the 4 mm Allen (Hex) key". Had they replied with "I don't see any sizes in mm", the conversation could've continued with "Your Allen keys might be using SAE sizing. A compatible size will be 5/32, do you see that in your set?"

Much like how "garbage collection" is industry terminology referring to a specific feature of memory management in programming languages, "space junk" is also industry terminology that's specific in its meaning:

Space debris (also known as space junk, space pollution, space waste, space trash, space garbage, or cosmic debris) are defunct human-made objects in space – principally in Earth orbit – which no longer serve a useful function.

Source: https://en.wikipedia.org/wiki/Space_debris

If it's no longer in space, it's no longer "space junk".

This appears to build off of earlier research: https://www.nature.com/articles/s41586-020-2010-9

"Here we show that thin-film devices made from nanometre-scale protein wires harvested from the microbe Geobacter sulfurreducens can generate continuous electric power in the ambient environment. The devices produce a sustained voltage of around 0.5 volts across a 7-micrometre-thick film, with a current density of around 17 microamperes per square centimetre. We find the driving force behind this energy generation to be a self-maintained moisture gradient that forms within the film when the film is exposed to the humidity that is naturally present in air. Connecting several devices linearly scales up the voltage and current to power electronics."

If something is E2EE it is deeply inconvenient and fragile. If something is easy to use, it cannot be E2EE.

The only difficult part of E2EE is establishing trust. As in: how do I securely exchange encryption keys with this other entity in a way I know isn’t being spoofed?

The online banking I do all the time is E2EE from my device to my bank and it’s convenient and not fragile at all. I message friends over iMessage and there’s nothing inconvenient or fragile about that.

If a developer can’t put together a solution that’s convenient and robust, it’s likely due to lack of experience, not anything inherent to E2EE.

When a collector purchases the Ship of Theseus, they expect to get the Ship of Theseus, not some mere copy.

Prior research in 2004: https://www.semanticscholar.org/paper/Keyboard-acoustic-eman...

"We show that PC keyboards, notebook keyboards, telephone and ATM pads are vulnerable to attacks based on differentiating the sound emanated by different keys. Our attack employs a neural network to recognize the key being pressed. We also investigate why different keys produce different sounds and provide hints for the design of homophonic keyboards that would be resistant to this type of attack."

Which model vent is it?