barrkel

Debian by default has had ancient versions of stuff, and if you want third party software, Ubuntu has more mindshare so it was more likely to be tested - anything from graphics card drivers to Steam games.

People are emotionally invested in a deflation and slowdown in AI because they feel threatened by AI. They'll cling to reeds of hope and wilfully misread financial stats to make themselves feel better.

That is from the increase in the share price, because they owe that in shares and the share price is higher now. On the other side is the IPO, it balances the books but it doesn't count as an asset today, whereas the promised shares count as a debt today.

They are cash flow positive, $3.4 billion in H1.

That means they're making money.

Their S-1 says they're not operating at a loss. If they lied in the S-1, they'll go to jail for a decade or more.

It names Mamo Vangelio Moscote Alberto and Saka Rosa Mamatacán Sauna; they are also identified in a predating cultural document; https://static1.squarespace.com/static/661d435c11d79f60042ba...

There's a separate account that describes it as the result of four days of consultation: https://www.wildsidejourneys.com/post/a-message-from-the-eld...

Obviously it has been written in collaboration with someone with good knowledge of AI, but it's written to that audience, so why wouldn't they collaborate?

It would be a worse kind of cultural imperialism to deny these people the right to say something by criticizing how they say it.

Note that it is also a mod, to support the data model.

Also, it isn't just a map. It's also a mod to enable better rendering of the data at distance, with a couple of modes; faster flight; layered read mechanism; etc.

No, you download it then get an invoice in the mail with the bill, 30 days to pay.

An ETH student's project apparently. Despite the naysayers, I think it's great someone had the impetus to get this done, and it has my son quite excited.

Would the time you spend getting a model subscription, giving it an environment to work in, and figuring out how to enable it to operate safely, be less than 15 francs of your time?

Minimum wage in Zurich city is 23.90 chf/hour. About 37 minutes of time, at most.

Short term rentals universally include a cleaning fee.

You cannot hotlink to jwz; he will redirect you to a testicle in an eggcup based on referer header.

You're not missing anything. Most devices these days are extremely limited without internet access.

Tailscale is the key ingredient.

This isn't the behavior I observe on my phone.

All my SIMs I've used in my phone seem to have been added as verified phone numbers to my primary account, but my secondary accounts don't have any phone numbers still.

It's still somewhat obnoxious since it means random travel sims that could get reused after they expire are "verified" for my account.

What is the point in being wealthy if it is not to avoid tedious inconveniences?

Thing is, X has actually paid people to tweet.

(It made everything worse if you ask me but they did do it.)

PowerShell works on Windows because of WMI, to a first approximation.

Because Windows is tied together through APIs and databases and not text files (unlike Linux), everything is text is not as useful as something that can talk objects. And a lot of Windows is object oriented, from the window message dispatching system through COM and down to NT's object manager.

I guess it also has access to the cookies for all your logins.

If you think about it there's overlap in what the VAE encodes and main model encodes. Objects at a distance resemble texture and textures zoomed in gain structure. The VAE makes textures more efficiently representable at the cost of reducing the representable space of pixels. So things like tiny text become nonsense scribbles. Working in pixel space, especially with something with recursive or cascaded structure, opens the possibility of using the learnt structure of real writing at a higher level to perfect tiny details that actually cannot be approximated without being obviously wrong.

Time is "just" another dimension. There's temporal continuity between frames, a video VAE would be learning and representing those temporal shifts, but there's nothing to say that e.g. a recursively applied generative model at the pixel level also doesn't learn and represent those things.

As ever, figuring out how to train the thing is the hard bit I expect.

(Handwaving over "textures" here, VAEs encode more like somewhat macro blocks of image whose content is also conditioned on surrounding blocks, rather than tiny patches of patterned pixels.)

(And yes I'm a total imposter layman here, I just see VAEs as seeming to be a crutch that reduce data size - super super helpful of course - but being strictly speaking redundant and inhibiting correct fine detail.)

Sure, I'm only taking about innodb. The join strategies supported before 8.0 (5.7 and lower) were basically nested loop and that's it. Which only really works well for a couple of query patterns: point queries with a handful of joined tables all using indexed FKs, or a big sorted scan over one single table and everything joined to it, with pagination through where clause and sorting all on the primary key, and possibly a straight_join to make sure it doesn't start the execution plan anywhere other than the spine table.

MySQL was generally (pre 8) optimized for point queries on primary keys. So rows are stored in the PK index, the PK index is a clustered index. Everything more or less falls out of this.

10 litres alcohol is about 100 bottles of wine a year (exactly 100 at 13.33% alcohol content).

I drink a lot of wine, but 1 full bottle is too much to have regularly. A half bottle - two generous glasses - is more reasonable, but you'll still feel it the next day. And 10 litres of alcohol puts you at 8 generous glasses a week.

I reckon this is still quite a lot of alcohol.

Profitable compared to what, that provides base load?

If energy prices are too high, industry suffers. If you let the price rise, maybe China just eats your lunch.

What's the option value vs risk of gas price rises? Gas sets the marginal price of electricity in Italy today for 60% of the year [1]

[1] https://www.terna.it/DesktopModules/AdactoBackend/API/direct... "However, the electricity system remains significantly exposed to gas price volatility, which in 2024 determined electricity prices for more than 60% of the hours."

Sure, you have an LLM which can invoke functions. I will say that without storage and composition, you're asking for hallucination. LLMs are not deterministic and while they're good at regurgitating text - you can see how replies in an instruct model are structurally keyed off the question - they will rephrase, adjust, "correct" data they're schlepping from one call result to another call input. And one noisy MCP call and there goes your context.

You could build something with storage and composition out of MCP functions, but come on, have you seen how LLMs - particularly budget LLMs - try and invoke functions reliably? The amount of retries you have to hide, feedback you need to send back to the LLM about what it did wrong. Parameters get replaced with synonyms, arrays are passed for singular arguments and vice versa, structured inputs are flattened, etc.

So maybe you fine tune on interactions with your subset of MCPs, to improve reliability. But all you end up doing is reinventing a Unix-like command line, poorly.

Firecracker micro-VMs, gVisor, wasm sandboxes. There are ways to make this work that aren't heavyweight. Giving LLMs tools that they've seen how to use millions of times in training corpora just works better.

By OpenAI, I presume you mean Irregular - these guys https://www.irregular.com/about ?

These guys are the common factor, the guys running the evals that let all the AI agents out, it looks like.

Someone can be OpenAI/Anthropic/whomever.

If you don't have something running somewhere, you don't have an agent, you don't have a harness. You've got a token generator, an LLM from the 2024 era.

You hobble the expressiveness of the LLM and reduce its capability.

Think of an agentic harness as like a kind of body for the LLM. It gives it primitive inputs (read_file, web_search or whatever) and primitive outputs (edit file, respond to user, etc). Give it a command line environment (in a locked down sandbox, with as few or as many tools as you prefer), and you've given it a toolbox. It can do a whole lot more, faster and more efficiently. It can compose tools together. It makes fewer transcription errors manually shifting data around. It can tame verbosity with good protections in the harness and access to grep, sed and awk.

It's really up to you how useful you want your agent to be.

CLIs, run in a sandbox as tight as your preferred choosing, live in an ecosystem, where, via pipes and redirection, input and output can be easily manipulated. An agent can do similar things but more laboriously (and less token efficiently) via Python or similar but it would still live in a sandbox somewhere.

Going without the sandbox means hobbling the LLM. It can do things directly but is less able to construct ad-hoc programs to deal with looping, conditionality, tame verbosity, connect tools together, and so on.

It's a choice to not give the LLM an environment. As you say, it can be necessary if you're using dumb models. I don't find it particularly worth the trade most of the time.

What they do instead is scoop up execs when they travel through the US.

Peter Dicks, David Carruthers

State violence is used to ensure compliance, so that companies self-censor.

A CLI doing all this is still a better UI for the agent though.