hodgehog11

4 ms·

1.
by hodgehog11 6d ago
A group already (sort-of) has for the broader theory: https://arxiv.org/abs/2604.21691 It's a good list, but "alignment" is much more targetted. I attended a workshop with researchers from OpenAI and Ant
2.
by hodgehog11 8d ago
This honestly sounds like holy grail tech at this point; what kind of battery life are you looking at?
3.
by hodgehog11 10d ago
I don't understand what you mean. Test error is literally out of sample error. Marginal likelihood is designed to estimate out of sample error. The whole discussion is about out of sample; nothing has been about in-sample error. The in
4.
by hodgehog11 11d ago
Ah, sorry for the misunderstanding, I can see how my comment reads that way. That is done in the Gaussian process context, not in my first example, and yes, it's a dirty idea, but you can justify it using differential privacy arguments
5.
by hodgehog11 11d ago
Just to add on top of the quality reference provided by srean, I like to first drill in Bayesian principles and then use this article to derive PAC-Bayes from that: https://arxiv.org/abs/1605.08636 Regular PAC falls ou
6.
by hodgehog11 11d ago
No, I am talking about out of sample error and estimates thereof. It is "overfitting" to data, but it also has lower out of sample error than the case where you do not "overfit". This is why the notion of overfitting is
7.
by hodgehog11 11d ago
I agree that this is a good nuanced take. However, I find that students who have learned PAC (which usually takes quite some time) often have to unlearn certain principles to do PAC-Bayes, so my comments come from a fair amount of frustrati
8.
by hodgehog11 11d ago
You can choose the prior according to any selection rule that does not see the data (actually, you can do more, but justifying this is the realm of empirical Bayes and requires some more precise arguments). In this case, you can choose it a
9.
by hodgehog11 11d ago
Apologies, I'm skipping details, because that's how I speak with my colleagues, but I realize this is an external environment without context. No references since this is folklore (you can look at Hastie et al's Surprises in
10.
by hodgehog11 11d ago
Absolutely not. This link is a reference on PAC learning, which is thoroughly misleading in the land of deep learning and inevitably leads to vacuous bounds. This is common knowledge in deep learning. I would not recommend that any studen
11.
by hodgehog11 11d ago
This is provably not true, and you can use the marginal likelihood / PAC-Bayes to prove it (or any other framework for measuring model quality). Increase the number of parameters in a linear model way beyond the point of interpolation,
12.
by hodgehog11 18d ago
This is an insane thing to read. Bubeck had a reputation even before he started with OpenAI. Of course it was him that was involved in this drama. This is such a sad mess, and it really didn't have to be this way.
13.
by hodgehog11 18d ago
It's literally a toggle in the options for ChatGPT, one which is on by default and most researchers probably have on without realising it. So to say that it is unlikely is extremely suspicious. No, they did not literally pull user da
14.
by hodgehog11 18d ago
Yes, mathematicians are. And yes, most of my colleagues did not even know the opt-out was an option.
15.
by hodgehog11 20d ago
Touche, I used to work in pure probability where that ridiculous Hardy-Littlewood rule used to cause all sorts of problems, but now work in statistics, where it is no longer an issue. To be clear, the colleagues I am referring to mostly wor
16.
by hodgehog11 20d ago
Uh, care to explain? I have several colleagues that stopped submitting to journals once they reached full professor. They only submit papers from their students for the benefit of their careers. First-author papers, not so much.
17.
by hodgehog11 20d ago
Exactly, and the advantage is that checking that the problem is "formalized" here is essentially isolated to verifying that the final theorem statement matches the claim. If there are no 'sorry's and the program compiles
18.
by hodgehog11 20d ago
You can prove that doing this will spiral training into a fixed point. There was a lot of research into getting this to work in the past, but it never truly worked well. The hope was that if RLVR was used quite a bit, and the general perfor
19.
by hodgehog11 20d ago
Why would they send it out for "expert review"? Every time, they have just made the AI generate a Lean proof. In fact, it seems like the most plausible direction to NS is computationally assisted detection of a blowup solution, wh
20.
by hodgehog11 20d ago
It basically is a formality at this level. Many top math researchers now hardly even submit to journals at all and just put up a preprint. At this scale, peer review happens by the audience. They don't need a journal to get people re
21.
by hodgehog11 25d ago
I also believe this. Post-training LLMs with vague metrics can only be achieved with RLHF, which is not impossible, but extremely costly and difficult. Instead, companies will opt for RLVR, focusing on math and programming tasks. This pushe
22.
by hodgehog11 26d ago
I agree that this should be something that researchers reflect on. GPT-2 is one of the primary models to research on nowadays, and many recent developments have come from studying it as a test bench. Imagine if CRISPR was considered "t
23.
by hodgehog11 1mo ago
It's good to see validated numerical proofs seeing a resurgence now that they are substantially easier to achieve. Others might be able to chime in, but my experience is that AI is effectively taking proofs that were once iterative (pu
24.
by hodgehog11 1mo ago
In a frontier scientific research environment, funding is often limited, so personal subscriptions are more common. Fable can hit a 5-hour usage limit on the Max subscription tier before it finishes a single complex math prompt . Most of t
25.
by hodgehog11 1mo ago
I hope you understand the context in which that was said. The point of that statement is that the only way to rigorously verify correctness of a program is by using formal methods. Those are often too difficult to achieve by humans, which i
26.
by hodgehog11 1mo ago
No it really is about the test suite, and provably so. As another poster pointed out, speed is a superoptimization problem and the test suite provides the constraints. If the constraints are appropriately set, even a naive genetic algorithm
27.
by hodgehog11 1mo ago
The Chinese labs are picking up on the low hanging fruits on efficiency, and no, you do not need to abandon transformers, you just need to push them closer to the more computationally efficient architectures of the past. OpenAI and Google
28.
by hodgehog11 2mo ago
> The game often has its own DRM though which will stop you I think you missed the "know where to look" part. It's called a Steam emulator, for starters. Note that I speak about this strictly for preservation purposes, as
29.
by hodgehog11 2mo ago
Even if you get all of your games via Steam, provided you have them downloaded, you can still run them without Steam if you know where to look. Obviously GOG is far better in this regard, but preservation is not a concern on PC, outside of
30.
by hodgehog11 2mo ago
My expertise lies in deep learning theory, and yes, the "intelligence" is coming primarily from scaling up, among other things. There are good reasons for this, but essentially it comes down to taking advantage of a narrow statist

More ›