7734128

7 ms·

1.
by 7734128 3d ago
If they were not using actual Norton trademarks I would have thought this was a parody. Why would anyone use Norton? It's just malware.
2.
by 7734128 6d ago
They "burn" a lot when they do benchmarks, while these runs can become valid roll outs for training. Perhaps less efficient than other data creation, but hardly burned in the same way.
3.
by 7734128 7d ago
I'm not sure how viable it still is. Perhaps it's still possible, and perhas that's exacly what they did in wtich case my objection falls, but I don't know.
4.
by 7734128 7d ago
They would know the specific thoughts released by the "stolen thought" paper, which became part of the public internet on August 10. Unfortunately those are the only thought examples you can use to perform this experiment, as no o
5.
by 7734128 7d ago
This is probably the 20th time the german wiki is posted here. It's not a new incident, nor a new discovery of an old incident.
6.
by 7734128 7d ago
Sure, but TFA is trying to use Qwen's reaction to the thoughts as proof that they did indeed extract thoughts to train on. My point is that any model trained after August 10 will know of those specific thoughts.
7.
by 7734128 7d ago
The problem with this is obviously that the only GPT 5.5 thoughts that we have access to are from stolen thought. Qwen 3.8 0902 was trained after the release of the paper on August 10, so it should have seen those specific thoughts.
8.
by 7734128 7d ago
How the hell could that cookie consent form ever be legal? Fuck them.
9.
by 7734128 9d ago
The correct response should be to actively disregard all IP owned by corporations participating in that behavior.
10.
by 7734128 14d ago
It seems that all competitors are rushing to release before Astra, which would suggest that they think it's going to be major.
11.
by 7734128 14d ago
Practically free for "contributors" at 0.2 usd/mtok. That's going to be hard to say no to for hobbyists.
12.
by 7734128 21d ago
They're absolutely right – and this is is why it's a load bearing observation that cuts to the heart of the issue.
13.
by 7734128 21d ago
There have been many blogs like this over the last years. Yes, embeddings are computationally heavy, but they are not at all complicated and they provide a lot of benefit. 90% of "document" based RAG projects should view semantic
14.
by 7734128 23d ago
No, it's actually 200 000 trillion for both countries. Source: same as yours
15.
by 7734128 24d ago
This forwards me to a Youtube channel page.
16.
by 7734128 26d ago
Why are you spreading such hateful garbage? A login wall after spending minutes reading is not acceptable.
17.
by 7734128 29d ago
The door mats are mostly named after places in Denmark, as they should be.
18.
by 7734128 2mo ago
I'm quite often working in a Jupyter notebook where the imported modules are reloaded. I can have x in memory, change method y of X and then call x.y() and get the new definition. It feels almost criminal that it works, but it's q
19.
by 7734128 2mo ago
I really wonder at the bounce rate for articles with a sign-in wall. I can't imagine the kind of person who actually creates an account when faced with that wall, especially when the website is also promising to send spam emails.
20.
by 7734128 2mo ago
My first ever programming experience was a Minecraft computer running Forth https://technicpack.fandom.com/wiki/Forth_language I built a massive mining platform that repeated a series of about 200 movements to vacuum u
21.
by 7734128 2mo ago
I wonder when one of these outages will be the result of OpenAI's testing going wrong again.
22.
by 7734128 2mo ago
Why are people still using Twitter? I've always hated it, but previously it at least let guests view some of the content, but now it's just a massive "install app" banner.
23.
by 7734128 2mo ago
I think that's partly because it's a fairly long conversation and partly because of all the LaTeX. There's enough formulas there to fill a small text book, and I doubt their renderer is all that efficient.
24.
by 7734128 2mo ago
Fun. For all people with VR headsets I'd strongly recommend the game Windlands 2. It's an awesome swinging game. The default settings are more "grapple hook" than swing, but it can be changed.
25.
by 7734128 2mo ago
No, you can't divide the entire size by the expert count. A lot of weights are constant for all tokens, so total active count is ((2800-(shared)/896)*16 + (shared))
26.
by 7734128 2mo ago
n=1 or p=0
27.
by 7734128 3mo ago
Oh, I'm an idiot who didn't see that at all. Thanks.
28.
by 7734128 3mo ago
Does anyone know what would the price of this be, roughly of course, given that a customer ordered some thousands or tens of thousands? I have no grasp of even the magnitude of the price for something like this.
29.
by 7734128 3mo ago
I have a hard time imagining an actual human being as hateful as you are. Why would you spending so much energy trying to make people feel bad about an issue they have? I spent every ounce of effort I could muster last year to lose weight a
30.
by 7734128 3mo ago
Why do you say "voluntary"?

More ›