Hacker Newsnew | past | comments | ask | show | jobs | submit | pllbnk's commentslogin

The guy on the left in that same photo only has three fingers (and a thumb, I suppose). I thought image generation has already outlived that.

Edit: I feel stupid I didn't see the original OP already mentioning three finger issue. I'll just leave it here.


Do all Silicon Valley corporations use the same jingle for their product promotional videos? The creator must be really rich by now.

If they all use it, then it is free

> Can you do murder?

Only if it kills many people over the long time.


In the 90s a bunch of charities installed wells in Africa but didn't get them tested for water quality and poisoned whole villages.

Intent is such an interesting moral stance as opposed to actual outcomes.


I think the "intent" ( pun not intended) for "intent" to be a moral stance was to not discourage people from trying to do good things for fear they may do something bad and get in trouble or whatever.

In an ideal world this would be good but then of course it's ripe for manipulation because it's hard to prove one's intent


Stochastic murder, like selling oxycontin.

Long time ago I had one particular hardware issue, which Opus 4.6 found a workaround to fix. I don't remember the workaround and being careless (I thought I could ask an LLM again if I needed) I lost that solution. Some time passed and I needed it again - neither one of the newer models is able to come up with a nice solution I had back then. They can solve the issue and find a different workaround eventually, but not as nice. On a side note, I could use 4.6 again of course and try to reproduce it, somehow I only thought about it now writing this comment.

Anyway, my point is that I think these frontier companies are advertising their one-shot model abilities, but underneath the models aren't getting so much better as they try to make you believe.


Shouldn't this new reduction in thinking output make the model cheaper to operate? Could it be that the model is the same old LLM, with a bit newer architecture but still doing the same things, including huge amounts of thinking, just that its thinking is less clear to human readers?

Those tend to be under so-called ‘max’ modes, or ‘ultra’ - where you scale up inference time compute and then choose amongst your answers. Astra is new weights.

Thinking output is already a relatively small part of the total input compared to raw code/text files.

You still need the models to be able to perform web searches, don't you? In which case the data goes in and out of your machine and there is risk for prompt injection attacks. I think it's needed at least for documentation purposes.

Oh, I ask for help with code, yes, but the actual data I'm working on doesn't ever leave the environment.

Why would they sound the alarm if they were not trained (reinforced) to do that? I hope we don't expect sudden emersion of moral values from statistical models.

Just a couple days ago I learned about ninfer (https://github.com/Neroued/ninfer) and on RTX 5090 I can now get ~200 tok/s and over 400 tok/s on concurrent requests which is plenty fast for a local model of this strength.

Ok, I need to try that. I'm getting 45tok/s with vLLM on my 6000. >600tok/s concurrent, but 45tok/s single request.

Even without ninfer I would get over 80 on LM studio with default settings, so it should be noticeably more on 6000. You might want to try different a different inference engine or settings.

Is there an equivalent but for 4090s?

The repository has many forks, suggesting that folks are trying to (vibe) code support for different GPUs. Might be worth a shot.

can't second ninfer enough. amazing tech

dang only for certain nvidia GPUs, had my hopes up

It's not that LLMs (I think that's what we are talking about when talking about AI) are not that useful. They are. But the most straightforward way to use them is to generate walls of text, which contain a lot of BS. In order to consume all this text and make sense of it, LLMs are used which reshuffle that BS into more BS but with less context. In the end, a human is presented with total BS that looks very plausible and if they don't critically evaluate it before forwarding to their higher-ups, they put their reputation at risk.

If everybody is rich, then no one is rich. In other words, if everyone has a lot of money, then the prices will be high enough to suck this money out of everyone.

America, and now much of the Western world, runs on debt and taking on debt is being instilled from the early years, so no wonder why people can't save when their income is spent on interest payments.

We know empirically that lower wealth inequality works because in the past periods with lower wealth inequality societies lived more secure lives financially, so wealth redistribution is the answer.


A tale of two cities both had access to the North Sea oil wealth one spent like a drunken sailor, the other side set up a sovereign fund guess who was doing better today. (and no population size has nothing to do with doing the right thing). Britain has a ton of excuses after 60 years.

And the point is to save and live within your means you can’t sit around and worry about what if the right thing to do is to save and live within your means you don’t sit around and worry about if everybody’s rich then no one‘s rich that sounds like an excuse not to do anything.

Many people who I worked with always had an excuse there’s no point saving or living within your means because inflation is gonna kill you or you’re going to be taxed.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: