pja

Emacs will happily run in the background.

This was all dropped at some point - the only surviving relic being the name of the character.

Arguably every unused variable in the code base is a potential bug like this waiting to chomp on some poor user’s data though, so arguably they should be turning it on & dealing with the consequences?

Fair warning: when you turn over some of the rocks here you find squirming, slithering things that should not be given access to the light.

Yeah, this surprises absolutely nobody.

They’ve already had a splinter rationalist group go full cult, right up to & including the consequent murders & shoot-out with the cops flameout: https://en.wikipedia.org/wiki/Zizians

Sounds like ... programming.

Program specification is programming, ultimately. For any given problem if you’re lucky the specification is concise & uniquely defines the required program. If you’re unlucky the spec ends up longer than the code you’d write to implement it, because the language you’re writing it in is less suited to the problem domain than the actual code.

> Manhattan needs a serious demolishing to become sane

Ah yes, lets demolish one of the most economically productive regions of the USA, both in GDP / capita and GDP / km^2 in order to make it easier for people to drive through it.

Listen to yourself, this is deranged.

Even if you focus only on emissions this completely ignores the cost of congestion, which is huge.

Your complaint about self driving buses makes no sense either. If the most polluting part of the bus is the driver then removing the driver makes the bus far, far less polluting.

When your arguments don't even make sense on their own terms it suggests that you're making them from an emotional position instead of a rational one. That's ok: if you don't like buses just say so, but be honest about it instead of making spurious arguments.

But it also compromised model accuracy & performance at the same time: The more you tune to eliminate or reinforce specific behaviours, the more you affect the overall performance of the model.

Hence my speculation that Anthropic is using a chain-of-thought model that has not been alignment tuned to improve performance. This would then explain why you don’t get to see its output without signing up to special agreements. Those agreements presumably explain all this to counter-parties that Anthropic trusts will cope with non-aligned outputs in the chain-of-thought.

It would make sense if the model used for train-of-though was trained differently (perhaps a different expert from an MoE?) from the one used to interact with the end user, since the end user is only ever going to see its output filtered through the public model the chain-of-thought model can be closer to the original, more pre-rlhf version without risking the reputation of the company.

This way you can get the full performance of the original model whilst still maintaining the necessary filtering required to prevent actual harm (or terrible PR disasters).

Try it out! Benchmarks here: https://github.com/ggml-org/llama.cpp/discussions/10879

(ollama doesn’t support vulkan for some weird reason. I guess they never pulled the code from llama.cpp)

If you (for example) extend Peano numbers with extra axioms that state things like “hey, here are some hyperreals” or “this Goedel sentence is explicitly defined to be true (or false)” it’s unsurprising that you can end up in some weird places.

Are such objects not inevitably isomorphic to the natural numbers?

Can you give an example of a formal definition that leads to something that obviously isn't the same as the naturals?

If you do this you are creating a rod for your own back: You need management to see the failures & the time it takes to fix them, otherwise they will assume everything is fine & wonderful with their new toy & proceed with their plan to inflict it on everyone, oblivious to the true costs + benefits.

You’d think a phone could do cancellation between the bud microphone & it’s own microphones; maybe once the audio data from the bud has been pushed through bluetooth audio compression there’s no longer enough information to do that effectively?

The steering wheel on modern Tesla's has those godawful turn buttons instead of a proper indicator stalk.

Fine for (most of) the US market but wildly impractical for anywhere you need to use the turn signals mid turn. Anywhere the road system uses roundabouts for instance.

To some extent this is just US parochialism leaking out into the rest of the world, but it's typical of Musk style design to not really think through the reasons for existing design choices.

Or pick up a Framework 13 which has a 13.5" 2880x1800 16:10 screen: https://frame.work/products/laptop-diy-13-gen-amd/configurat...

May as well go the whole hog & jump to 48V.

(50V is as high as you can go whilst still being inside the “low voltage” electrical safety regime in most countries IIRC.)

(edit) This is probably the case paper I was thinking of: https://pmc.ncbi.nlm.nih.gov/articles/PMC2190741/ It was her ovaries that were 99/1% proportions. Other parts of her body had varying levels of mosaicism.

Given that, it’s plausible that sloppy lab handling procedures led to someone being infected with multiple different viruses that were present in the same part of the lab.

None of this proves the lab leak hypothesis of course, but a lab worker being infected with multiple variants simultaneously (or separately) is a perfectly plausible outcome.

That said, I’ve had no issues with either of my recent Pixel phones & haven’t heard of widespread problems with either the Pixel 8 or 9 online. Google themselves seem to have been fairly good at recalling phones for actual problems. I don’t know whether they’re better or worse than, say, Apple though.