nightski

7 ms·

1.
by nightski 5d ago
Maybe Enron eh?
2.
by nightski 6d ago
Seeing as it adds a second PSU, I'm guessing the workstation one.
3.
by nightski 7d ago
No, I'd have to guess that new model releases have either new from scratch or continued pre-training. This is not the same as continual learning. Starting a new pre-training session is a dramatically different affair and involves util
4.
by nightski 7d ago
I disagree about your last statement. That only covers the initial creation/generation phase. It ignores the engineering aspect of maintaining that design through the entire lifecycle of the product. How do you evolve that table des
5.
by nightski 7d ago
Alright let's assume your premise is true, that transformers can learn from interaction with the world by updating their weights - then why isn't this done? Because backprop fundamentally wants the entire data set in every pass.
6.
by nightski 7d ago
So you enjoy reviewing your entire code base and relying on an extremely detailed regression suite just to make simple changes? I don't think AI changes good engineering at all, it just changes who or what is doing it.
7.
by nightski 8d ago
I was not saying that they are deterministic, rather that the distributions (aka weights) are fixed. A model as deployed today at anthropic/open ai/etc is not learning beyond the context as far as I know. What prevents continuous
8.
by nightski 8d ago
It's a little different than that. Your bundle of nerves and meat is not static. It changes over time. To me the heart of the "next token predictor" is that the distributions are static. You can manipulate what you feed in
9.
by nightski 8d ago
Do you use an operating system? What about databases? Or the myriad of other software used which we build upon. I can't imagine a professional coding context where we don't outsource intelligence to some degree, if not a major de
10.
by nightski 9d ago
Are you a member of a team? Then you are the outsourced intelligence. Your manager is not writing everything by hand, they are outsourcing it to you. Just understand that may change (I mean I am already mentally prepared for it to be a th
11.
by nightski 13d ago
So you spent the cost of my entire cluster in one day on Sol tokens lol. I have no need for that many tokens, a few million a day is perfectly acceptable. But if you do, then yeah Sol is probably your bet. Have fun :)
12.
by nightski 13d ago
My DGX spark cluster is humming along.
13.
by nightski 21d ago
I mean it's entirely a personal decision. I didn't mean to come across judgemental, if anyone uses the frontier models I don't hold it against them. But for me personally I am very against the frontier labs in general. This
14.
by nightski 21d ago
Not everything is about pure cost. Maybe I don't want to sell my soul supporting the frontier labs because they are straight up pure evil?
15.
by nightski 21d ago
Remember when big co used to just donate to open source projects to help them out instead of acquiring/dominating them?
16.
by nightski 22d ago
That is amusing because to me one of the largest abominations about mac is the keyboard layout and mappings. It's terrible for anyone with decent sized hands to accomplish anything of even medium complexity.
17.
by nightski 23d ago
I am not the parent but I took it as them saying the premise was wrong to begin with, which I very much agree with. Learning should not be primarily directed by job availability.
18.
by nightski 1mo ago
All of my code repos are on gitea (personal and professional). The instance is not exposed to the internet (wireguard access). It's fantastic honestly, simple and just works.
19.
by nightski 2mo ago
Not the parent, but here is one location that has rates in that range in the U.S. https://casscountyelectric.com/rates
20.
by nightski 2mo ago
These posts seem more like emotional therapy for the founder, helping them cope. It's not about the people being let go at all.
21.
by nightski 2mo ago
No you don't. I use the same passkeys for all my devices sync'd with bitwarden.
22.
by nightski 2mo ago
It's simple, the 395+ Max Strix Halo you bought for $1800 is now a ~$4000 build (at least the AMD AI dev unit). If only we had time travel right? Either way, the Nvidia unit comes with Connect-X 7. That may or may not matter to you,
23.
by nightski 2mo ago
The baseline is the market average, not 0.
24.
by nightski 2mo ago
I mean I pay for a domain for my homelab. No one else uses it.
25.
by nightski 2mo ago
Except LLMs are not stochastic in nature. Correct me if I am wrong, but it's just the final layer which outputs a distribution across the output tokens. In reality, the rest of the model is deterministic. This is not like a bayes mod
26.
by nightski 2mo ago
I mean, you are using BaseUI directly... The components on top live in your code base.
27.
by nightski 2mo ago
Only Asus from what I have seen. Acer/Dell units seem to run cooler than the Nvidia unit, by a noticeable margin. That said I bought the OEM units for the Gen 5 4TB SSD and the fact that it was on sale.
28.
by nightski 2mo ago
Open source isn't really open any more. It's just pre-acquisition. I'm happy to the creators for their payday but honestly just happy I opted out of BetterAuth building my latest product.
29.
by nightski 2mo ago
Meta's FAIR has several R&D offices in the EU, yes. So you are saying their labs can conduct R&D on models in the EU, potentially even train them there, they just can't have production LLM inference serving or release the
30.
by nightski 2mo ago
So Deepmind/Meta/etc... would all have to cut off their EU offices even though they hold incredible talent? I'm just not finding this scenario likely.

More ›