HackerTrans
TopNewTrendsCommentsPastAskShowJobs

aeve890

no profile record

comments

aeve890
·il y a 9 jours·discuss
>Obviously, the best move would be to keep the data in Europe instead.

While 80% of Europe is subservient to the US?
aeve890
·il y a 9 jours·discuss
You have my curiosity
aeve890
·il y a 10 jours·discuss
You mean like Goodhart's Law? (Unless you're being sarcastic)
aeve890
·il y a 11 jours·discuss
DoD in short, to make it look like serious business
aeve890
·il y a 13 jours·discuss
It's absolutely not the same. If you think llms and brains works the same way you clearly don't know how either works.

For a LLM learning what you wrote your last session would be update the weights with the new relationships and factual knowledge created in the session. That doesn't happen. The weights are static and fixed after training. There's no online training in the transformer architecture or any variant. If the weights don't update, the network doesn't learn. Period.
aeve890
·il y a 14 jours·discuss
>It definitely learns things.

By the very way this technology works they can't learn anything after training. What you think is "learning" it's just a session log written back to the context when you resume the session.
aeve890
·il y a 14 jours·discuss
You're free to write your own media-conglomerate boot-licking post if don't agree.
aeve890
·il y a 14 jours·discuss
Is this article targeted just to an American audience? Because the US isolationism will only hurt them. Unless they truly have achieved AGI or ASI, the Chinese models soon will catch up. I'm pretty sure we will have an open weights frontier model this year.
aeve890
·il y a 14 jours·discuss
>I wonder if our common expectation that true theories somehow had to be beautiful and elegant is going to survive the coming century.

That's the layman's idea of physics theories. They are beautiful and elegant only on the surface, that's why they're technically models and approximations of the real world. The standard model renormalization techniques are a mess of patches and ad-hoc heuristics, pretty far from the "this lagrangian literally contains all physics". Generally you just _ignore_ higher order terms and just call it a day. The famous E=mc^2 it's just the first term of a Taylor expansion. The beautiful form of physics it's what you would call "good enough" and often just a pedagogical tool.
aeve890
·il y a 14 jours·discuss
>I wonder if there is any way to counter that.

Not buying invasive shit?
aeve890
·il y a 16 jours·discuss
> ship around 3x to 5x faster

Web apps and CRUDs, if may I ask? Or is AI helping you with something that you couldn't ever do by yourself? I have mixed results across different technologies like frontend, backend, infra and hardware.
aeve890
·il y a 16 jours·discuss
>At AIs absolute best (rarely) it builds software like a very competent 9-5er. It's fine, it works, it's largely inoffensive.

If only middle management and c-suite knew that
aeve890
·il y a 16 jours·discuss
The submitted site runs smoother, has less annoying popups, don't use the entire screen (on mobile) to show info of the selected startup and overall I think it has better UX than yours.
aeve890
·il y a 18 jours·discuss
I bet (pun intended) that's pretty good for the owners pockets
aeve890
·il y a 18 jours·discuss
>massive, historic accomplishment in creating digital scarcity

accomplishment? like, something big and good?
aeve890
·il y a 19 jours·discuss
>TLDR it is quite a bad article

You can write a rebuttal to address what's wrong with the article, from your point of view. Maybe I'm old but the whole "live reaction in twitch" thing doesn't help how the scientific community perceives your area of expertise.
aeve890
·il y a 19 jours·discuss
Ha. When I found that problem I draw the grids and paths from the example, left for a coffee and when came back I just look at the drawings at an angle and thought "well this is just Pascal's triangle". And the solution was obvious.
aeve890
·il y a 21 jours·discuss
>If you have a ton of capital

That's my point. This "open source" doesn't feel like the real open source. It's open just for the few ones with ton of capital, and mostly in the US, or US adyacent markets. It's like if SpaceX publish an open source rocket design and people celebrating like it's the new Linux. Feels more like a goodwill gesture than something with real impact for the benefit of mankind, like the spirit of open source software as commonly understood.
aeve890
·il y a 21 jours·discuss
Please correct me if I'm wrong, I'm totally out of my field here but what's the point of sota models that can be run only by hyperscalers? I mean, glm-5.2 is open source but with 1.5TB in weights who can run it really? It still needs dozens of H100s. Those 753B quantized down to Q4 (~400Gb) would require datacenter levels of hardware. Down to Q2 still would require serious hardware, way out of reach for most users, and you'll be far from the sota benchmark of the full precision model. I get it, it's open source but not quite democratizing LLM for everyone except compute providers. It's no like, let's say, Kubernetes. I can run k8s fully in my shitty homelab, without "quantization" exactly like Google does in their datacenters.
aeve890
·il y a 26 jours·discuss
>Social credit bonus

People does still believe this shit?