Jerrrry's comments

On: OpenAI says it has evidence DeepSeek used its model to train competitor

[Jerrrry]

No one knows if the pigeon-hole principle applies absolutely exclusive to the ability to generalize outside of a training set.

That is the existential, $1T question.

On: Questions censored by DeepSeek

[Jerrrry]

This is day 1 jailbreaking common sense

On: How has DeepSeek improved the Transformer architecture?

[Jerrrry]

False, facts only need to be seen once, and one mis-step in reasoning and your CoT is derailed.

[whimsicalism → Jerrrry]

> one mis-step in reasoning and your CoT is derailed.

tell me you've never seen reasoning traces without telling me

On: How has DeepSeek improved the Transformer architecture?

[Jerrrry]

  > the open-domain Frame Problem is equivalent to the Halting Problem and is therefore undecidable.

Thank you, Code as Data problems are innate to the von Nuemman architecture, but I could never articulate how LLMs are so huge they are essentially Turing-complete and equivalent computationally.

You _can_ combinate through them, just not in our universe.

[adgjlsfhk1 → Jerrrry]

this is very wrong. LLMs are very much not Turing complete, but they are algorithms on a computer so they definitely can't compute anything uncomputable

[Jerrrry → adgjlsfhk1]

"Essentially"

This is the #1 pedantic thing HN frequenters flail over.

Turing built a limited machine, bound to a finite tape.

Bravo.

[nyrikki → Jerrrry]

Turing machines are typically described as having an infinite tape. It may not be able to access that tape in finite time, but the tape is not bound to a finite tape

But it doesn't matter, it is an abstract model of computation.

But it doesn't matter Church–Turing thesis states that a function on the natural numbers can be calculated by an effective method if and only if it is computable by a Turing machine.

It doesn't matter if you put the algorithm on paper, on 1 or k tapes etc...

Rice's theorem I mentioned above is like Scott–Curry theorem in Lambda calculus. Lambda calculus is Turing complete, that is, it is a universal model of computation that can be used to simulate any Turing machine.

The similar problems with 'trivial properties' in TMs end up being recursively inseparable sets in Lambda calculus.

On: Promising results from DeepSeek R1 for code

[Jerrrry]

Maybe those software engineers should "lrn2code", just as the journalists, artists, and trucker drivers had to.

On: CIA now favors lab leak theory to explain Covid's origins

[Jerrrry]

The hospital was full.

The institute would never be full.

There is no "bring your child to work day" in China.

Weird conflation, unlikely... uunless you mean to muddy the waters and sow dis-information.

[lazyeye → Jerrrry]

I would argue you are sowing disinfo and I honestly dont know what point you are trying to make. Spikes and/or significant reductions in activity as indicated by external data sources, and particular the timing thereof, will obviously be very useful for determining the sequence of events.

[Jerrrry → lazyeye]

The hospital, sure.

Why would Wuhan center being empty represent anything?

People don't rush to a random (unrelated) building when they get sick at the "market".

Coincidence, surely?

What are the chances of having a specialized infectious disease center next to where that type of disease spontaneously emerged?

More or less likely than a bat having sex with a panda possum?

No one, not even the cowardly academics, believed it.

On: CIA now favors lab leak theory to explain Covid's origins

[Jerrrry]

Lol people were getting called racist for suggesting it was anything less than the spawn of a bat and an asian water racoon.

It still isn't even acceptable to acknowledge the (blatant) possibility that Omnicron was intentionally leaked because of low vaccine effectiveness.

[rUsHeYaFuBu → Jerrrry]

Let's be honest, if their is a global conspiracy to spread disease I think it's to kill off the masses due to AI replacing jobs and lowering the amount of green houses gasses people produce.

On: Show HN: Compile C to Not Gates

[Jerrrry]

I'm guessing by not jumping into a terminating/ halting NOOP.

The logic is within the branching.

On: Transformer^2: Self-Adaptive LLMs

[Jerrrry]

Sieve the temporary backprop interim weights as a function of its loss of varentrophy relative to its place in the revalued weights.

Remove the bottom weights dynamically based on the local gradient in varentrophy so that internal dissonance ("doubt") can be selected against.

"Preference Optimization" but with more opportunities for meta-optimization.

On: Ask HN: Are there any real examples of AI agents doing work?

[Jerrrry]

"Workflow automation" is gonna be the goalpost to beat