HNHacker News
TopNewBestAskShowJobs

razodactyl

563 karma · joined September 8, 2014

submissionscomments
razodactyl··on Who Makes Code-Supernova?
As per title: Seems like a company with frontier capabilities and ability to serve at such scale for Kilocode, Windsurf, Cursor with a galactic themed model would speculatively be xAI no?

For reference: I don't care about politics in the scope of standing on the shoulders of giants.

This seems like a massive effort to collect perpetual coding data to ensure longevity of coding abilities.

razodactyl··on Apple Graveyard – Products that it killed
Of all things to put absolutely no effort into... why...?
razodactyl··on Sora 2
How dare you be critical about a service offering in favour of a better end-user experience! /s
razodactyl··on Claude Sonnet 4.5
You're absolutely right!
razodactyl··on Zed's Pricing Has Changed: LLM Usage Is Now Token-Based
Thoughts on https://nova.app/ ?
razodactyl··on GNU Midnight Commander
xtgold?
razodactyl··on We regret but have to temporary suspend the shipments to USA
I'm not sure where to stand on this statement so I can only sit back and watch: I thought we would meet another AI winter ~2015 when we saturated computer vision and classification problems - then things started taking off for a little bit but settled but then things really took off...

You're still likely right as things must always come down but what if it doesn't.

razodactyl··on Ask HN: Why does the US Visa application website do a port-scan of my network?
There should be settings to emulate the bad network conditions: 2G in the field is nothing like the browser throttling.
razodactyl··on From GPT-4 to GPT-5: Measuring progress through MedHELM [pdf]
A viable consideration is that the models will hone in on and reinforce an incorrect answer - a natural side effect of the LLM technology wanting to push certain answers higher in probability and repeat anything in context.

Regardless of being in conversation or thinking context this doesn't prevent the model from speaking the wrong answer so the paper on the illusion of thinking makes sense.

What actually seems to be happening is a form of conversational prompting. Of course with the right conversation back and forth with an LLM you can inject knowledge in a way that causes the natural distribution to shift (again - side effect of the LLM tech.) but by itself it won't naturally get the answer perfect every time.

If this extended thinking were actually working you would expect the LLM to be able to logically conclude an answer with very high accuracy 100% of the time which it does not.

razodactyl··on DOS Games in JavaScript
I'm actually amused how not-awkward Doom is to play with a mouse.
razodactyl··on DOS Games in JavaScript
Feeling like a high-schooler giving out links to the good stuff.
razodactyl··on What went wrong inside recalled Anker PowerCore 10000 power banks?
This scrutiny actually makes me more inclined to buy Anker products. There's more trust in a company that will transparently communicate, correct mistakes and raise the safety bar for future products. It also brings to light issues with other manufacturers in the supply chain.
razodactyl··on OpenAI delays launch of open-weight model
There's a bias that smart people have in underestimating the stupidity of those on the left side of the intelligence curve.

Not even that, children believe anything and more so that of a computer designed to be "harmless".

razodactyl··on Get the hell out of the LLM as soon as possible
I hate that I understand the internals of LLM technology enough to be both insulted and in agreement with your statement.
razodactyl··on Meta puts stop on promotion of tell-all book by former employee
...Streisand effect about to hit.
razodactyl··on Meta puts stop on promotion of tell-all book by former employee
A gag order on a book?
razodactyl··on Crossing the uncanny valley of conversational voice
This isn't getting the attention it deserves. First AI I've had a real conversation with.
razodactyl··on Goku Flow Based Video Generative Foundation Models
Why not link to the paper instead of the GitHub? The model in this repo is for an LLM.
razodactyl··on AI is stifling new tech adoption?
Not entirely sure it's a hard fact but this is definitely an example of bias in an AI system.

More reason to decouple and think for ourselves.

razodactyl··on Musk-led group makes $97B bid for control of OpenAI
It's an English term for mocking, teasing, ridiculing. I believe from the link of "piss" being alcohol. "He's on the piss again."
razodactyl··on Team 17 Co-Founder Martyn Brown Has Passed Away Aged 57
57 years young. Tragic loss.
razodactyl··on GPT-5 is behind schedule
Two thoughts:

1. Even if LLM architecture doesn't work out, it's wise to remember that it's the quality of training data which is a deciding factor. A pivot is easily doable since this is a constant and disconnected from the technology itself.

2. This is clearly a moonshot project but it still feels wasteful especially with context of previous iterations of models and their shortcomings-it feels like the tremendous amount of money available is being used simply because it's available.

razodactyl··on OpenAI O3 breakthrough high score on ARC-AGI-PUB
No I think we're just tired and depressed as a species... Existing systems work to a degree but aren't living up to their potential of increasing happiness according to technological capabilities.
razodactyl··on OpenAI O3 breakthrough high score on ARC-AGI-PUB
Is this comment voted down because of sentiment / polarity?

Regardless the critical aspect is valid, AGI would be something like Cortana from Halo.

razodactyl··on OpenAI O3 breakthrough high score on ARC-AGI-PUB
Great. Now we have to think of a new way to move the goalposts.
razodactyl··on The surprising effectiveness of test-time training for abstract reasoning [pdf]
You're on track in your arguments but don't underestimate how hard the puzzles in ARC actually are.

It takes a considerable amount of depth in reasoning to see and reason about the patterns / problems / solutions.

Try doing a few of them by hand to see what I mean.

Simulated worlds are complex enough to hide their own flaws just like LLMs are complex enough to lead us to believe they can reason when most of the time they are pattern matching.

razodactyl··on The surprising effectiveness of test-time training for abstract reasoning [pdf]
Majority of ARC can be gamed / hard-coded, no doubt about it.

The real pressure is the private hold-out set and the variations that can be added to counter this aspect.

A true AGI would be able to solve anything thrown at it which is where the authors are trying to lead AI engineering towards since LLMs have pretty much taken over.

If it starts getting too easy, they just reconsider and add harder problems.

It's like how we don't talk about the Turing Test anymore as it's no longer the best metric to determine real intelligence.

The authors are signalling to the industry that new ideas are needed and the monetary aspect is to show how serious they are about it.

It's good because as per above we have research being thrown at it which means we can iterate until we perhaps find another breakthrough.

razodactyl··on AMD Open-Source 1B OLMo Language Models
I appreciate this comment keeping us in line.
razodactyl··on Ask HN: Has the LLM/transformer architecture hit its limit?
This is a great point for what I was trying to explain about where we currently stand with these models - we seem to be hardware constrained in relation to the trade off between rote learning, full memorisation of information and the model's ability to use what it knows and generalise to achieve an outcome.

Your experience aligns with what I've seen: 4o and o1 are better at filling gaps from each other where there should be an upward trend of intelligence instead of having to jump around as we do currently.

So the real question is: are these models forgetting their skills, do we need to make them larger? Does distillation and generalisation work?

So many questions that might end up as LLMs being unable to escape the problem of catastrophic forgetfulness.

razodactyl··on Ask HN: Has the LLM/transformer architecture hit its limit?
I think there's still a lot of room "relatively" to move around but my current opinion is that hardware isn't where we would want it to ideally be to have next level LLMs everywhere.

We've seen the trend of distilling models at what seems to be the cost of more nuanced ability to iterate and achieve correct results.

I'm very convinced LLMs can go much further than we've achieved so far but I'm very welcoming of newer techniques that will improve accuracy / efficiency and adaptability.

← PreviousPage 5 of 13Next →