c7b

3 ms·

1.
by c7b 5d ago
Technically, Q is picked because it has the highest probability of all letters. But it makes a difference whether the probability for Q is barely above a uniform 1/26~3.8% or whether that one letter concentrates >50%. What I remembe
2.
by c7b 5d ago
Once the first letter is Q, the rest is probably pretty determined. Can you see the confidence for the first letter (don't want to accept the ToS to follow your link)?
3.
by c7b 6d ago
Interesting. Is there a paper that explains this in a bit more detail, like [0] for abliteration (underlying the Heretic software, afaik)? [0] https://arxiv.org/abs/2406.11717
4.
by c7b 7d ago
At last something that feels properly orthogonal to pelicans on bicycles.
5.
by c7b 7d ago
I mean, there's centuries' worth of mathematical prose to train on. But that's presumably already in the training data, so if it isn't good enough today, it might not get better fast enough to keep track with how fast th
6.
by c7b 7d ago
> Separately, our group has been exploring approaches along these lines to make such evaluations more scalable Actually, that sounds like an interesting idea for peer review in general, to include an interview between referees and author
7.
by c7b 7d ago
> we might imagine what it could look like to have an analog of the Millennium Prize Problems for open exposition problems The core idea seems to me that we should shift the standards for professional evaluation from generating proofs to
8.
by c7b 8d ago
Well, I hope you're going to be right. My gut feeling, however, is that putting all the little bits that make up the AVP experience together would already have been too much to ask from other manufacturers in the absence of any patents
9.
by c7b 9d ago
> We measure the actual symbol distribution of 29 ternary LLM models and find that zeros account for up to 51.5% of all weights. Motivated by this finding, we introduce BITCOS, a simple distribution-adaptive layout I honestly assumed tha
10.
by c7b 9d ago
In this case however, many of the patents have a hardware element to them. Gaze+tap is powered by an inside camera to track eye movements and a Lidar sensor at the bottom of the rim for motion detection. Not sure your playbook would work ou
11.
by c7b 11d ago
afaik the AVP has a dedicated Lidar sensor for gesture detection. I'd say Apple has managed to patent more basic stuff than that (pinch-to-zoom, rounded corners,...)
12.
by c7b 11d ago
The thing is, the AVP has a ton of quality-of-life features, eg relating to gestures, and most of them locked behind patents. Looks like we'll have to wait 20 years before other devices will be able to match it (or whatever the expiry
13.
by c7b 12d ago
Being able to comfortably sit back / lie down and work is an interesting proposition. The Apple Vision Pro is too bulky and expensive to carry that whole category on its own, they need an 'Android equivalent'. But I think App
14.
by c7b 12d ago
With formalized math, you only need to validate the problem statement (in theory, in practice agents have already managed to exploit Lean compiler bugs, but the incidence of those should decrease enough to be practically lusable for 'b
15.
by c7b 14d ago
Even funnier that it wouldn't even be the first time that happens: https://en.wikipedia.org/wiki/Grigori_Perelman
16.
by c7b 17d ago
I remember first hearing about NMS' concept and wishing that they would take the general idea but downsize it by at least 10 orders of magnitude. A universe with 1-100 millions of planets would still feel absolutely massive and way mor
17.
by c7b 17d ago
I don't think that follows from the published results. Would have been an interesting hypothesis to add though, and quite easy. Just throw the same setup at some benchmarks.
18.
by c7b 17d ago
I wasn't aware that we have access to raw reasoning tokens? I thought what you get is a kind of summary. Does the author have some kind of privileged access or was my assumption wrong? But for the question studied here it probably does
19.
by c7b 22d ago
The article clearly distinguishes between open source and open weights models, and the headline and parts of the article state that it's open source models that are having a moment. But it doesn't list any examples. The only model
20.
by c7b 22d ago
If you're doing it for fun anyway, why not use the language that gives you the most pleasure?
21.
by c7b 29d ago
Frankly, that's the one part I would have done differently. Mixing font sizes rarely looks good, but that's just my personal opinion. If you want to visually indicate intensity, I'd have used color, or rather saturation. Not
22.
by c7b 1mo ago
Congratulations! I was going to comment on the scroll field in particular when I saw this. I didn't even realize you had to hand-craft the component, but it's such a nice UI idea in general, the way the scrolling works and how the
23.
by c7b 1mo ago
You must be referring to SEPA Direct Debit, which would be very risky to use to siphon funds. There's an 8 week no-questions-asked refund policy, 13 months for unauthorized transactions, if the payment fails (eg due to insufficient bal
24.
by c7b 1mo ago
If you're mad at Bluesky for doing this, I think you're mad at the wrong party. Modifying intent is problematic and can be abused (although I think there's nothing particularly problematic about this particular instance). The
25.
by c7b 1mo ago
It's a noble cause, but there are probably bigger levers to pull than the model size if you care about environmental impact. If you're running Qwen3.8-27B on energy-efficient hardware like a Mac or a DGX Spark instead of an API (l
26.
by c7b 1mo ago
> I understand that to mean that they are the same architecture, just one version has a massive training set and the other has a very small subset. No. It means that the one model has 2.4 trillion parameters while the other has only 27 b
27.
by c7b 1mo ago
At a more technical level, what do you suggest? Training a small LLM on Python code exclusively? And then one on general CS/algorithms, which you'll also need? I don't think the current transformer architectures would compose
28.
by c7b 1mo ago
Sounds a bit like 'I want to make horses faster, surely I won't need mechanical engineering knowledge'. We don't know everything that we don't know, so it's hard to say what we don't need to know.
29.
by c7b 1mo ago
I've been wondering whether that is a feature of the foundation model or whatever finetuning they do on top. I remember this from the earliest versions of (pre Chat-) GPT I've been using, which would suggest it's a feature of
30.
by c7b 1mo ago
For those commenting on the long reasoning, it may be interesting to know that the reasoning effort is set to xhigh by default [0]. Other possible values are medium, low and none. Flag for changing it in llama.cpp below, but note that the l

More ›