Hacker News: Jerrrry

New comment by Jerrrry in "OpenAI says it has evidence DeepSeek used its model to train competitor"

Jerrrry — Wed, 29 Jan 2025 19:35:09 +0000

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.

New comment by Jerrrry in "How has DeepSeek improved the Transformer architecture?"

Jerrrry — Wed, 29 Jan 2025 14:50:12 +0000

"Essentially"

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

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

Bravo.

New comment by Jerrrry in "CIA now favors lab leak theory to explain Covid's origins"

Jerrrry — Wed, 29 Jan 2025 14:48:04 +0000

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.

New comment by Jerrrry in "1,156 Questions Censored by DeepSeek"

Jerrrry — Tue, 28 Jan 2025 22:47:17 +0000

This is day 1 jailbreaking common sense

New comment by Jerrrry in "How has DeepSeek improved the Transformer architecture?"

Jerrrry — Tue, 28 Jan 2025 20:37:54 +0000

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

New comment by Jerrrry in "How has DeepSeek improved the Transformer architecture?"

Jerrrry — Tue, 28 Jan 2025 20:36:27 +0000

  > 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.

New comment by Jerrrry in "Promising results from DeepSeek R1 for code"

Jerrrry — Tue, 28 Jan 2025 17:43:45 +0000

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

New comment by Jerrrry in "CIA now favors lab leak theory to explain Covid's origins"

Jerrrry — Mon, 27 Jan 2025 01:54:44 +0000

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.

New comment by Jerrrry in "CIA now favors lab leak theory to explain Covid's origins"

Jerrrry — Mon, 27 Jan 2025 01:51:53 +0000

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.

New comment by Jerrrry in "Show HN: Compile C to Not Gates"

Jerrrry — Fri, 17 Jan 2025 21:14:56 +0000

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

The logic is within the branching.

New comment by Jerrrry in "Transformer^2: Self-Adaptive LLMs"

Jerrrry — Wed, 15 Jan 2025 19:35:04 +0000

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.

New comment by Jerrrry in "Ask HN: Are there any real examples of AI agents doing work?"

Jerrrry — Sat, 11 Jan 2025 16:58:25 +0000

"Workflow automation" is gonna be the goalpost to beat

New comment by Jerrrry in "Bird-inspired drone uses legs to walk and jump into the air"

Jerrrry — Fri, 10 Jan 2025 18:03:25 +0000

Fruits want to be eaten, Veggies don't.

New comment by Jerrrry in "Agents Are Not Enough"

Jerrrry — Thu, 09 Jan 2025 19:38:20 +0000

  >solve for prompt injection attacks

It is essentially the same Code as Data problem as always.

New comment by Jerrrry in "Anthropic raising funding valuing it at $60B"

Jerrrry — Thu, 09 Jan 2025 15:17:48 +0000

Thank you, he has articulated some outcomes and signs I hadn't yet.

His thesis that AI is just distilled omega-capitalism is true, very plainly.

New comment by Jerrrry in "Anthropic raising funding valuing it at $60B"

Jerrrry — Wed, 08 Jan 2025 19:10:40 +0000

What an exisistentally terse moment for late stage capitalism; betting State-sponsored shitcoins on our future quasi-cybernetic giga-corp overlords to offset increasingly miniscule differences in individual economic starvation.

New comment by Jerrrry in "Anthropic raising funding valuing it at $60B"

Jerrrry — Wed, 08 Jan 2025 16:13:27 +0000

I would suggest being careful treading that thought experiment.

New comment by Jerrrry in "Anthropic raising funding valuing it at $60B"

Jerrrry — Wed, 08 Jan 2025 16:10:45 +0000

Productivity went up for those still employed.

New comment by Jerrrry in "Anthropic raising funding valuing it at $60B"

Jerrrry — Wed, 08 Jan 2025 16:01:08 +0000

This mindset allows one to donate their intellect to society and feel immortalized.

If privacy wasn't of particular concern, this is as near an "absolute good" idea as one can forge.

New comment by Jerrrry in "All clocks are 30 seconds late"

Jerrrry — Wed, 08 Jan 2025 14:24:14 +0000

Collective (un)concious, yes.

There is no parallel chains of unconnected thought.

We are barely, rarely more than the sum of our parts on average.

We have very convincing illusions of free will though - like the apparent exhibition of agency.