krackers

4 ms·

1.
by krackers 8d ago
You could combine it with existing products like DevonThink that are meant for researchers organizing documents and provide tagging, semantic search, similarity graphs, and so on.
2.
by krackers 10d ago
Yes, updated the submission title to say "post-training" to hopefully prevent further confusion
3.
Xiaomi Mimo 2.6 live post-training dashboard(mimo.xiaomi.com)
547 points by krackers 10d ago|154 comments
4.
by krackers 11d ago
So an encoder-only model with a classifier trained on the heads or something? DeepSeek recently switched to an encoder-decoder architecture in an attempt to get the best of both worlds (fast prefill while preserving generation capability),
5.
by krackers 13d ago
https://github.com/uBlockOrigin/uAssets/discussions/34325#di... I think that implies mv3 does support it. Seeing as it can inject scripts, I don't see why it wouldn't be able to.
6.
by krackers 14d ago
I sure hope not: https://en.wikipedia.org/wiki/Train_to_the_End_of_the_World
7.
by krackers 15d ago
>This spills over into other processes wanting to use the GPU, namely the WindowServer. Why does this spill over? Unlike CPU which is multiplexed by the kernel's scheduler (so infinite loops can't lock out other programs), is t
8.
by krackers 17d ago
How do you know that what is being proved in the lean code is the same as the millennium prize criteria though?
9.
by krackers 19d ago
https://map-projections.net/compare.php?p1=equalearth&p2=rob...
10.
by krackers 19d ago
>For all trains to be self-driving Given that we have self-driving cars, isn't this easier if someone really wanted? I guess compared to cars the marginal savings is not worth it though.
11.
by krackers 19d ago
I've never quite understood the hate either. But the way I see it is that it's not fundamentally inferior to explorer, but rather just has annoyances and warts that Apple has bizarrely not fixed for decades which makes tedious to
12.
by krackers 20d ago
Elliot Glazer (FrontierMath lead) traces how it snowballed over time https://x.com/ElliotGlazer/status/2096298696438906934
13.
by krackers 21d ago
For it to be memory safe, do you have to disable the JIT?
14.
by krackers 21d ago
It mentions a type confusion in V8. Is it possible to trigger that without JS enabled? The "all chromium versions" part of the title is also misleading, most browser CVEs do not distinguish between "untested lower bound"
15.
by krackers 21d ago
Automatic Captions are not ADA compliant. https://www.speechpad.com/blog/ada-compliant-captions-vs-aut... If you have a risk of getting sued because someone notices a transcription error versus just not uploading video
16.
by krackers 25d ago
How? Can't apps link you to the app store where you have to make an in app purchase?
17.
by krackers 27d ago
>almost nothing is safe That's a bit of an overstatement? There's a list of things that you _can_ call, and fairly useful ones too like `write` https://man7.org/linux/man-pages/man7/signal-safety.
18.
by krackers 28d ago
Feels weird that after their attempts to shut down corellium they themselves shipped the necessary bits to allow this https://github.com/wh1te4ever/super-tart-vphone-writeup
19.
by krackers 1mo ago
https://eqbench.com/results/creative-writing-v3/hybrid_parsi...
20.
by krackers 1mo ago
They laughed at objective-c back then, now the shoe's on the other foot
21.
by krackers 1mo ago
The previous HN thread https://news.ycombinator.com/item?id=49374873 Apparently the open question is whether you can achieve arbitrarily high ranks. And the username is "ranksunbounded" so presumably the answer is
22.
by krackers 1mo ago
https://news.ycombinator.com/item?id=48864882
23.
by krackers 1mo ago
>It's crazy, because 15.ai could have had the venture-scale outcomes if he'd fundraised and turned it into a SaaS product Or equivalently could have gone down as a "deepseek" moment in TTS if the model was open-source
24.
by krackers 1mo ago
Another benefit of open models is that you can mask out out words from logits directly when sampling. I wonder if anyone has put together a "desloppifier" for e.g. DeepSeek
25.
by krackers 1mo ago
There might be precedent for being able to detect changes in magnetic field https://www.pbs.org/newshour/science/dogs-poop-in-alignment-... but then again maybe not https://www.sciencedirect.com/sc
26.
by krackers 1mo ago
Or take the ribbonfarm-pill and jump straight to Gervais Principle https://news.ycombinator.com/item?id=46657678
27.
by krackers 1mo ago
ref: https://www.youtube.com/watch?v=IU4ByUbDKNc
28.
by krackers 1mo ago
>fails to understand the gumbel softmax technique I think in this case it doesn't help that there are multiple watermarking schemes, and the easiest for people to understand is the red/green scheme by Kirchenbauer et al. ( http
29.
by krackers 1mo ago
I don't understand Gruber's points either, I wonder if there is some fundamental technical misunderstanding. Does he think that the logits should be sampled from in a "pure" manner without introducing any other bias? Doe
30.
by krackers 1mo ago
A similar project (LLM trained only on vintage material): https://talkie-lm.com/introducing-talkie

More ›