xrisk

that doesnt say what cards the reference number used.

For comparison, rental prices seem to be ~$2.75/ per H-100-hour [1]. Meanwhile, Deepseek says "38 USD per hour per card”. They haven’t specified which card as far as I can tell.

[1] https://www.ft.com/content/a1815bc5-b5d5-47ff-bbbc-e1a689929...

My hunch is that you would need some post training. On top of that; I don’t think the llm itself can read inside the transformer state although I can see how that could be enabled. (I feel that would open up yet another class of exfiltration opportunities)

This seems like something that could exist as free software and can be easily vibecoded in the form of a pile of scripts. Thanks for the idea!

Truly built on vibes.

pyenv has been standard tooling for far longer than uv. depending on package manager supplied Python packages only makes sense if you’re running rhel or Debian or something and your application is packaged/deployed/the maintenance path uses dnf/apt. Otherwise you should always use a venv and use an out of package manager update mechanism. Like, in a broader sense, vendoring dependencies only makes sense if you’re shipping an application, not on a dev box.

Might be an indication that your task unit is too unstructured or your code base is a mess.

Is your actual code doing something complex or is this incidental complexity?

Relying on the model’s “intelligence” to patch over these issues hasn’t proven to be a reliable strategy for me. Of course, this might not apply to you, just my 2 paisa.

tfa: "Results were adjusted for age, sex, ethnicity, and sociodemographic, lifestyle, and health factors"

Would like to see wall times. I feel that’s the part that annoys me most, my tasks aren’t particularly challenging I want them done fast

What models?

Could you point me towards some resource that would help me understand what you wrote? Genuinely curious about how this stuff works

Assuming that the prediction market is perfectly priced right? How accurate is that assumption, or are you counting that as an “inefficiency”?

Stockfish uses neural nets for its evaluation function, I don’t see how it’s unfair to call it “AI”.

Maybe explainable via the fact that these tests are part of the LLM training set?

Only books published by American publishers yeah?

hmm there are apps produced by your script that claim to be fixed according to https://avarayr.github.io/shamelectron/ (Signal, Discord, Notion, etc). And I checked that those apps are updated. Which one’s correct?

Did switching your deliverer to SES have any effect on how clients like Gmail “tagged” your email? (Promotional category or something IIRC)

If your pds refuses to serve you your CAR file I don’t think you can do anything about it, can you?

That’s a cool result, thanks for the link!

While I don’t know if the claim is true, you’ve linked a post from 2012…

Yeah I was looking at that page. Found this btw: https://github.com/darkrain42/tailscale-initramfs

This is not the same thing is it? Arch Wiki mentions something about having to install a separate ssh server into initramfs to support ssh’ing into fully encrypted systems.

systemd-cryptenroll seems to be about storing encryption keys into the TPM so that they can be decrypted automatically at boot (?)

Apologies if I misunderstood something.

This is such a hilarious failure mode. I would never have imagined something like this to a problem.

In the case of SSH though, I assume retrying after a second or so would be enough. You probably have some sort of retry mechanism to deal with network failures anyway.

actual title: "macOS 26.0 Tahoe build 25A354 is incompatible with Mac Studio M3 Ultra"

Yeah I’m not buying it. If the corporations wanted to, they would.

you are aware that the app you just wrote with Claude pulls in dependencies, yes?

Seems to be a separate incident?

That is, undoubtedly, a problem created by Google itself. See for example: Kagi’s small web (https://blog.kagi.com/small-web)

> I’m not saying that one should believe everything at face value

It’s for (often implicit) communities to decide; communities whose members share a certain set of norms.

Further, legality does not imply correctness.

For example, it’s probably legal to call somebody a transphobic slur in many parts of the world but to suggest that trans people shouldn’t attempt to avoid or “cancel” such people is ridiculous.

And if you sincerely think that the only acceptable action to take is make a petition to change the law, I would suggest you go out and touch some grass. The law doesn’t work that way.