TomatoCo

Well, I'm under the impression that the relations are recursive but you skip computing them recursively by caching B's relations. However if you look at your sibling comments you'll see that I may not have the right grasp on the matter.

Oh, of course, but I mean for moving people who want to move around. Someone else observed that ocean liners are less efficient than planes, so if you wanna go overseas there's no alternative.

Except, of course, for business meetings which can be virtual (or just an email). I'm asking what's a practical alternative assuming we still want people to be able to go overseas. I agree for land routes high speed rail is better, too. I don't really understand my downvotes just for asking about alternatives.

I was under the impression that each new token attends to one previous token per attention head, and that the slowdown observed was more because those attended to tokens are more spread out in memory and get less memory-architecture-style cache hits.

What's the alternative?

Yes, except no with the KV cache. Because tokens aren't modified by future tokens you can cache the meaning of previous tokens. This makes the total effort linear over the entire context (or constant per forward pass).

Right! When I wrote that I made a mistake and thought that there were filesystems that used multiple drives but still had the meaningful distinction of "what drive is that file on?"

Like, yeah, RAID, several drives, one filesystem, right? Except no, because it's just one virtual device.

> By the time the humans even became aware of the problem

Except, let me quote part of the report:

> July 4: Artifactory outage caused by agent interactions overloading the system.

> July 6: New artifactory instance is patched and put back online, wiping out the previous message board and method of internet access. A team at OpenAI was aware of the previous message board, but key decision makers were not.

> July 7: Experiments are restarted.

There was a very clear point where humans were aware and they went "Eh, whatever" and turned the bots back on! That's not the lack of any battle of wits that people are expecting. You're saying that the bots attack should be fait accompli, but it was noticed, shrugged at, and expressly allowed to proceed.

Right? If you move between file system or device I would expect "move" to be impossible. At best you could copy a logical block and then delete a logical block. If you share a filesystem then perhaps a logical block is a few hundred kilobytes. If you don't then I'd expect the smallest logical block to be the file itself.

Well, it's hard to convey, but he gave off a strong sense of "hitting on her" and she gave off a strong sense of "I'm in customer service and cannot tell him to stop".

But, I was in my latest teens, earliest twenties; I could have been misreading the situation. All I can say for certain is I was uncomfortable.

Another way to think of this is that everybody occasionally needs to do something unusual. But some people, due to their life circumstances, always need to do something unusual. And making things better for the always-odd folk is also a positive to the occasionally-odd folk.

Which like, we're all eventually gonna be odd folk! That's what age does to us. But we should understand these accommodations as being a net positive, not a cost, to society.

This reminded me of something that happened to me maybe 15 years ago. I was helping my Grandpa get his first smart phone. The saleswoman was maybe mid twenties and had a kinda hoarse voice. I was a little bit younger than her and I thought it was cute in its own way but of course I wasn't gonna say anything. She apologized for being hard to understand and my grandpa said no, it was fine, her voice was sexy.

Naturally I was mortified.

But I wonder how much of it is just "Fuck it, I'm an old man, I'm gonna let my creep flag fly."

I hadn't read carefully enough to recognize that! That's a wacky coincidence that I love. Thank you for your writing.

You might appreciate the first author on this paper, https://oeis.org/A180632/a180632.pdf

https://en.wikipedia.org/wiki/Superpermutation#Lower_bounds,...

There's a tiny bit of play, like sliding window attention. As tokens leave the sliding window you can keep them or discard them. If you keep them, you can freely truncate the context and resume generation from an earlier point. If you discard them, you have to recompute the KV cache up to that point.

Llama.cpp checkpoints and moves snapshots of the cache to main RAM for faster resumption after truncation.

Most pi pico's come with 16mb of flash. I wonder what kind of performance that can eek out.

This author also has other things like: A compiler that emits only `mov` instructions and another compiler that deliberately messes with the control flow so that, if disassembled, common debuggers will draw symbols like skulls or threats. https://github.com/xoreaxeaxeax/repsych

There's plenty of arguments that mass scrapers have compute to spare but it seems to me that if Anubis makes it 100x more expensive to scrape then, for any given scraping budget, that means you get scraped 100x less. Which is the difference between your server buckling under the load or continuing to serve reliably.

I've mostly been running LLMs on my own hardware so the phrase "been at it for about 20 minutes" gave me pause. My first instinct was to ask "but on what hardware?" but I suppose one advantage of closed models is that there's a somewhat-consistent cognitive-effort-to-time ratio.

I've usually been more annoyed at the surprise dissonance of "was that an anime girl on kernel.org?" than annoyed at the delay.

I'm picturing something like RUST servers, where the admins get to set their own rules and resource multipliers to facilitate the particular part of gameplay that they want.

On one hand, sure, the entire point of a programming language is to make complex ideas able to be expressed in simpler abstractions. On the other hand, we can damn well try.

In the USA you'll get buttloads of mail urging you to do things such as confirm your home warranty at risk of not being covered. With addresses that says "RE: (your mortgage provider)" to make it look like it's from them.

You'll see some scripting languages (ab)use this. Where the native "number" type is a 64 bit float and only one NaN bit pattern is a real NaN. The others smuggle a pointer to an object in the lower bits. This way you don't spend any memory overhead indicating if a given variable contains a primitive or an object.

Vaguely reminds me of that planet in Hitchhiker's Guide that started to lose too much mass to tourists leaving. So,

> Thus today the net balance between the amount you eat and the amount you excrete while on the planet is surgically removed from your body weight when you leave; so every time you go to the lavatory there, it is vitally important to get a receipt.

On a metal as soft as copper I imagine that texture'll last about 30 minutes after it's issued to the soldier.

One of my favorite clips to give a sense of scale for rockets is this one. https://www.youtube.com/watch?v=70u748VALt4

I show someone and then I tell them, that's not the rocket exhaust. That's the exhaust for the engine that runs the fuel pump for the rocket.

This road seems to lead to the exclusion of third party app stores and/or the ability to load apps that aren't signed by Google/Apple.

Going by Fabien Sanglard's cheat sheet (who I trust uncritically) https://fabiensanglard.net/usbcheat/index.html it looks like 3.2 actually is a broader term than expected. Maybe there was some awful attempt at backwards compatibility? Or forwards?

It reminds me of how vinyl records are fairly lossy, but they provide a superior experience in some cases because those limitations have been accounted for during the mastering process.

It's an entire pipeline from photomultiplier to recording medium to the inverse process and everything is optimized not for any particular mathematical truth but for the subjective experience.

The term I've seen a lot is https://en.wikipedia.org/wiki/Unit_in_the_last_place

So I'd probably rewrite that code to first find the ulp of the larger of the abs of a and b and then assert that their difference is less than or equal to that.

Edit: Or maybe the smaller of the abs of the two, I haven't totally thought through the consequences. It might not matter, because the ulps will only differ when the numbers are significantly apart and then it doesn't matter which one you pick. Perhaps you can just always pick the first number and get its ULP.