It's fun to run a model locally, but I don't think the economics make sense for anyone just trying to use models atm. It's absurdly cheap to use the same model via openrouter in comparison.
Seriously, just put $10 into openrouter and play with models that are cheap but bigger than what you'd reasonably be able to run locally like deepseek v4 flash (unquantized). You'll be surprised by how far that $10 goes for a model better than what you'd be able to run. Even further on the model you would be able to run locally. Then think of how many long it would take to match the cost of spend + power on doing it locally...
In my anecdotal experience of reacting to “wow this espresso is good” it’s often been a Slayer machine. It’s been a rough indicator of where to get good coffee for me.
I’m looking closely and I’m not seeing IP laws being relevant. Maybe regarding windows 11 keys? Can you elaborate?
I think the claim isn’t totally inaccurate. Claiming that companies prioritize making products last just beyond return windows because that maximizes profits is a critique of capitalism IMO. Since capitalism is about profit (capital) being the dominant signal as far as I understand it.
In my opinion, a lot of frustration about capitalism nowadays can be tied to Goodhart’s law. Profit being the measure and companies getting so efficient at optimizing it, that we’re starting to harm the things that profit is a proxy for.
I don’t find it fair that you point out straw man in your parent comment and then use ad hominem in this comment. I would love to see you post some examples. I think you’d have a chance of persuading several readers to at least be more open minded.
I think a lot of people in tech unintentionally blind themselves to reality by obsessing over technology, and then thinking that is all that matters. It’s a behavior that is incentivized because it helps those with power utilize the tech-blind’s skills for their bidding.
It seems that there’s still unavoidable subjectivity in making the choice of prior distribution? I get how it’s objective for a fixed choice, but my understanding is that you need to first make that choice in order to be objective. Is it actually that making your choice of prior is obvious (or there is some objectively optimal way to pick a prior), which rules out any subjectivity in the choice of prior?
The latter quarter of the article goes into depth about how one can accidentally find oneself in an inner ring, but it is through a wholesome pursuit and no ulterior motive. The other reply to this comment gets it write: people here like interacting with the things on HN.
In other words, to comment on HN does not mean you are striving to be in the inner circle.
Recently, I finished the making of the atomic bomb by Richard Rhodes. It’s quite long, but a fascinating view on physics, weapon development, and then the politics of the bomb. It’s a good exercise to compare with the current state of AI and see what things are similar and what are different.
I thought of these as well while reading the article as they convey the fun aspects of archival visits well (I’ve done a bit of my own and really enjoyed the random “side quest” things you discover along the way).
If anyone else has recommendations for things like these (text or video) then I would love to see more!
Yea, this is what’s really going on here and feels like it’s been shrouded in language to make it seem more grandiose. That being said, I would believe generalization to occur from minimum norm solutions in some sense, but whether that corresponds to minimum norm weights or not is a different question, and one you probably won’t know a priori (not to mention even knowing which norm to choose).
OP, have any doctors mentioned that it may be epilepsy? Seizures present in many ways and can sometimes be described as dreamlike experiences. Might be worth asking about.
I sympathize with the author’s (OP I believe based on the last name) feelings here and also find myself wanting to leave somethings not understood. Maybe we’ll be fortunate and it literally will be the case that some things always will be a mystery.
One thing I would like clarified: I don’t follow the refutation of epiphenomenalism. Is the intent that I can raise my hand only as a result of my conscious experience? Couldn’t one imagine instead a LLM hooked up to a robotic arm that also is trained to raise its hand it reaction to similar text? I feel like that or something similar would refute this. How can you say was is causal in sending the signals to raise the arm? Why can’t the conscious experience just be a theater?
What’s more weird to me about epiphenomenalism is that consciousness seems like it need not exist - if it performs no function than why bother with it? So why do we experience it? Sort of seems that its very existence implies that it should be more than just a theater.
I’m a total amateur about this, but would appreciate recommendations if anyone resonates with this.
> Here's a new trend happening these days. Upon releasing new non-fiction books to the general public, authors are simultaneously offering an LLM-based chatbot box where you can ask the book any question.
Seriously, just put $10 into openrouter and play with models that are cheap but bigger than what you'd reasonably be able to run locally like deepseek v4 flash (unquantized). You'll be surprised by how far that $10 goes for a model better than what you'd be able to run. Even further on the model you would be able to run locally. Then think of how many long it would take to match the cost of spend + power on doing it locally...