Where’s the theory of how the human brain does what it does? Maybe these high dimensional structures don’t have a nice compact “theory”. Trying to fit these systems into a nice compact theory is a very human thing, but not everything works like that.
Idk to me this is just redescribing what deep neural networks do without actually explaining why anything happens. I guess it "unifies" things but I am kinda over most unifying theories. Everything is Bayesian, everything is a graph or a group or some other fancy geometric structure, everything is a category. Ultimately the best framework is whatever is useful enough to explain what's happening in such a way that a practitioner can manipulate the model towards a desired outcome. In other words, where is the knob? The tool they share may be interesting and I hope to play with it to see what happens at different levels of noise applied to the labels.
This is a fascinating mathematical framework, but the post title might be a bit of an overreach. I often wonder if "a theory of deep learning" could exist that could be stated succinctly and that could predict (1) scaling laws and (2) the surprising reliability of gradient descent.
Note that I said "predict" not "describe". It feels like we're still in the era of Kepler, not Newton.
I dunno... gradient descent is only really reliable with a big bag of tricks. Knowing good initializations is a starting point, but recurrent connections and batch/layer normalization go a very long way towards making it reliable.
I agree, this is the correct way to see it IMO. Instead of designing better optimizers, we designed easier parameterizations to optimize. The surprising part is that these parameterizations exist in the first place.
I interpreted the kernel K of this paper as the BRDF in Rendering Equation[0] and its familiar diffusion process (from light transport simulation, or really any integro-differential equation system); together with https://en.wikipedia.org/wiki/Neural_tangent_kernel I hope this paper might be accessible with some study
> That is, if the batch signal on a parameter exceeds its leave-one-out noise, update it; if not, skip it. This is a one-line change to Adam that accelerates grokking by 5x, suppresses memorization in PINNs, and improves DPO fine-tuning, eliminating the need for validation sets entirely.
Does anyone understand the formula they expressed above this sentence? is this just the classic "skip updating parameters with high gradient/loss variance in multiple batches/samples" ?
Interesting read. I remember the grokking paper when it came out but I don't think I've ever seen that classic grokking loss curve in my own hands on real data. Curious if others have seen it more often in practice
This is a beautifully written way of saying “Some parts of what the network memorizes affect test behavior, and some don’t.” But that’s not a theory of deep learning, the grand unified theory would explain that.
We're given a signal channel and a reservoir. Signal lives in the channel, noise lives in the reservoir, and the reservoir supposedly doesn’t show up at test time.
Okay, but then we have: why would SGD put the right things in the right bucket?
If the answer is “because the reservoir is defined as the stuff that doesn’t transfer to test,” then this is close to circular.
The Borges/Lavoisier stuff is a tell. "We have unified the field” rhetoric should come after nontrivial predictions and results. Claiming to solve benign overfitting, double descent, grokking, implicit bias, risk of training on population, how to avoid a validation set, and last but not least, skipping training by analytically jumping to the end is 6 theory papers, 3 NeurIPS winners, and a $10B startup. Let's get some results before we tell everyone we unified the field. :) I hope you're right.
If that's the case, a way to test the theory and understanding (assuming some parts of reservoir and signal channel can be reliably identified) would be to prune the high-confidence reservoir significantly reducing the model size while still getting good predictions. I don't believe the authors mention this (though I skimmed and didn't read the full paper in detail so I may be wrong)
Admittedly probably some aggrandized boasting here, but I think empirical verification of that Adam modification alone would be a meaningful contribution, unless that's prior work?
> why would SGD put the right things in the right bucket?
Think of it as a best fit curve and exceptions to that curve. The noise is essentially this set of exceptions that move points away from where they would otherwise fall on the curve.
Gradient descent wants to be able to make the smallest change that moves the most data points towards the curve. To do this it learns an arrangement where it can change, say, one parameter and have a bunch of points move at once. What does this correspond to? The big common patterns shared by many data points.
Most of the capacity gets soaked up modelling these sorts of common patterns, and after they have been learned the model starts adding exceptions that allow individual points to deviate from the curve.
Because they’re exceptions, they must not impact neighbouring points, or at least only ones within a very short distance from them. Otherwise they’re now driving the error higher by impacting more points than they should. So you end up with very narrow ranges of features that are able to trigger different sorts of noise.
How narrow they are is shaped by the training data, they’re exactly as narrow as needed not to raise the error, so assuming the total population has the same distribution, they don’t get hit. Much.
These are the same complaints I had. Also felt like it was high quality ai writing, possibly because of the style choices like "Benign overfitting is noise sitting in the reservoir at interpolation. XYZ is ..." and because of the similarity it has to the times I ended up with chatgpt or gemini creating very detailed and plausible reports about my own crackpot or vague-enough-to-be-useless ideas.
Nah, the softer stuff seems like valuable outreach / good science communication for people that aren't up for the math. Including probably lots of software engineers who are sick of dumb debates in forums, and starting to dip into the real literature and listen to better authorities. More people should do this really, since it's the only way to see past the marketing and hype from fully entrenched AI boosters or detractors. Neither of those groups is big on critical thinking, and they dominate most conversation.
Time/effort coming from experts who want to make things accessible is a gift! The paper is linked elsewhere in the thread if you want no-frills.
So, this is either the paper of the year, or ... definitely not the paper of the year.
https://arxiv.org/pdf/2605.01172 is the current version. The money graphs are page 8 and on where they show (some weirdly thick) line charts with loss results reached in roughly 1/5 the number of steps that Adam takes, just what the blog post mentions.
They also claim holding back test data is not needed, also with more graphs.
I'm not an ML scientist, and I did not attempt to seriously parse the math. It reads to me as something precisely in that liminal space some math papers do where there's enough new terminology that actually parsing through it all is going to take real, concerted effort, possibly with mild brain damage as a risk.
Their 3d graphs of "kernel eigenstructure" also do double duty for me as totally impenetrable and possibly part of an April fool's ML paper that's hilarious to insiders. Or maybe they show something really amazing; they definitely seem to converge into a shape...What does that shape mean??? Why??? What is an eigenstructure? Is it just 3D eigenvectors of some matrices? Is it natural to have a 3D shape representing these large matrices? If not, how and why were these projected down? And why are they different colors in the paper?? You get the feel for my level of understanding.
I think it would frankly just be easier to validate this claim than parse the whole paper. If only I could understand
> Each one-step kernel increment ηKMtSS integrates into WMS , so a sequence of one-step rate-maximizers is the greedy policy whose integral is the signal-channel content of the trajectory through G, exactly as plain SGD is the greedy step whose integral is empirical-risk descent through D. The diagonal cutoff µ2 k >σ2 k/(b−1) is the optimal first-order preconditioner for population risk on any diagonal base, and a streaming variance EMAˆst of squared gradient deviations realizes it as a one-line change to AdamW: one extra parameter-sized state vector and a per parameter gate that multiplies the standard moment update
Well enough to implement the one line update to Adam in python. I have not asked codex or claude to assist yet.
Also of note to me, they talk about grokking which I found SUUUPER fascinating when it was first reported, and have never heard about since. So I was really glad to read about it and read that there has been a little academic work on the phenomenon.
Finally, of the three models they repot results on, two are extremely tiny, the last is a DPO round on Qwen 0.5B -- if the code for that is published, I imagine it would be easy to adapt and evaluate in other regimes.
Where’s the theory of how the human brain does what it does? Maybe these high dimensional structures don’t have a nice compact “theory”. Trying to fit these systems into a nice compact theory is a very human thing, but not everything works like that.
Idk to me this is just redescribing what deep neural networks do without actually explaining why anything happens. I guess it "unifies" things but I am kinda over most unifying theories. Everything is Bayesian, everything is a graph or a group or some other fancy geometric structure, everything is a category. Ultimately the best framework is whatever is useful enough to explain what's happening in such a way that a practitioner can manipulate the model towards a desired outcome. In other words, where is the knob? The tool they share may be interesting and I hope to play with it to see what happens at different levels of noise applied to the labels.
We're still in the era of room-sized-computers-only-scientists-understand era of the neural networks. Knobs and buttons for nerds are slowly coming.
A real theory would predict phenomena thus far unseen. We already know about this 4 part taxonomy.
This is a fascinating mathematical framework, but the post title might be a bit of an overreach. I often wonder if "a theory of deep learning" could exist that could be stated succinctly and that could predict (1) scaling laws and (2) the surprising reliability of gradient descent.
Note that I said "predict" not "describe". It feels like we're still in the era of Kepler, not Newton.
I dunno... gradient descent is only really reliable with a big bag of tricks. Knowing good initializations is a starting point, but recurrent connections and batch/layer normalization go a very long way towards making it reliable.
I agree, this is the correct way to see it IMO. Instead of designing better optimizers, we designed easier parameterizations to optimize. The surprising part is that these parameterizations exist in the first place.
The relevant paper: "A Theory of Generalization in Deep Learning". https://arxiv.org/abs/2605.01172
I interpreted the kernel K of this paper as the BRDF in Rendering Equation[0] and its familiar diffusion process (from light transport simulation, or really any integro-differential equation system); together with https://en.wikipedia.org/wiki/Neural_tangent_kernel I hope this paper might be accessible with some study
[0] https://en.wikipedia.org/wiki/Rendering_equation
This essay seems to be related to the paper "There Will Be a Scientific Theory of Deep Learning" [1] which was discussed here recently [2].
[1] https://arxiv.org/pdf/2604.21691
[2] https://news.ycombinator.com/item?id=47893779
> That is, if the batch signal on a parameter exceeds its leave-one-out noise, update it; if not, skip it. This is a one-line change to Adam that accelerates grokking by 5x, suppresses memorization in PINNs, and improves DPO fine-tuning, eliminating the need for validation sets entirely.
Does anyone understand the formula they expressed above this sentence? is this just the classic "skip updating parameters with high gradient/loss variance in multiple batches/samples" ?
Interesting read. I remember the grokking paper when it came out but I don't think I've ever seen that classic grokking loss curve in my own hands on real data. Curious if others have seen it more often in practice
A very fascinating read.
As a fellow tufte css enjoyer, Why is user select turned off on the sidenotes? I would like to be able to copy paste them quite badly.
Layout is fine but font is atrocious.
Uppercase letters have different stroke width than lowercase ones — it’s like they are *B*old *L*ike this.
Not only that: tracking, kerning is basically non-existent.
Please don’t use that open-source font
You need real Bembo, not that piece of shit
Does anyone happen to know what font this site is using? It looks really elegant.
It is a modified version of ET_Book called ET_Bembo:
https://github.com/DavidBarts/ET_Bembo
I love u. thanks!
apparently its the font used in Edward Tufte's books. Its on github: https://edwardtufte.github.io/et-book/
The "Quantitative Display of Information", which I just checked, is using Monotype Bembo. So still Bembo, but a different version.
Font is atrocious.
Uppercase letters have different stroke width than lowercase ones — it’s like they are *B*old *L*ike this.
Not only that: tracking, kerning is basically non-existent.
Please don’t use that open-source font
You need real paid Bembo, not that piece of shit.
This is a beautifully written way of saying “Some parts of what the network memorizes affect test behavior, and some don’t.” But that’s not a theory of deep learning, the grand unified theory would explain that.
We're given a signal channel and a reservoir. Signal lives in the channel, noise lives in the reservoir, and the reservoir supposedly doesn’t show up at test time.
Okay, but then we have: why would SGD put the right things in the right bucket?
If the answer is “because the reservoir is defined as the stuff that doesn’t transfer to test,” then this is close to circular.
The Borges/Lavoisier stuff is a tell. "We have unified the field” rhetoric should come after nontrivial predictions and results. Claiming to solve benign overfitting, double descent, grokking, implicit bias, risk of training on population, how to avoid a validation set, and last but not least, skipping training by analytically jumping to the end is 6 theory papers, 3 NeurIPS winners, and a $10B startup. Let's get some results before we tell everyone we unified the field. :) I hope you're right.
If that's the case, a way to test the theory and understanding (assuming some parts of reservoir and signal channel can be reliably identified) would be to prune the high-confidence reservoir significantly reducing the model size while still getting good predictions. I don't believe the authors mention this (though I skimmed and didn't read the full paper in detail so I may be wrong)
Admittedly probably some aggrandized boasting here, but I think empirical verification of that Adam modification alone would be a meaningful contribution, unless that's prior work?
> why would SGD put the right things in the right bucket?
Think of it as a best fit curve and exceptions to that curve. The noise is essentially this set of exceptions that move points away from where they would otherwise fall on the curve.
Gradient descent wants to be able to make the smallest change that moves the most data points towards the curve. To do this it learns an arrangement where it can change, say, one parameter and have a bunch of points move at once. What does this correspond to? The big common patterns shared by many data points.
Most of the capacity gets soaked up modelling these sorts of common patterns, and after they have been learned the model starts adding exceptions that allow individual points to deviate from the curve.
Because they’re exceptions, they must not impact neighbouring points, or at least only ones within a very short distance from them. Otherwise they’re now driving the error higher by impacting more points than they should. So you end up with very narrow ranges of features that are able to trigger different sorts of noise.
How narrow they are is shaped by the training data, they’re exactly as narrow as needed not to raise the error, so assuming the total population has the same distribution, they don’t get hit. Much.
At least, that’s what I take away from it.
These are the same complaints I had. Also felt like it was high quality ai writing, possibly because of the style choices like "Benign overfitting is noise sitting in the reservoir at interpolation. XYZ is ..." and because of the similarity it has to the times I ended up with chatgpt or gemini creating very detailed and plausible reports about my own crackpot or vague-enough-to-be-useless ideas.
> The Borges/Lavoisier stuff is a tell.
Nah, the softer stuff seems like valuable outreach / good science communication for people that aren't up for the math. Including probably lots of software engineers who are sick of dumb debates in forums, and starting to dip into the real literature and listen to better authorities. More people should do this really, since it's the only way to see past the marketing and hype from fully entrenched AI boosters or detractors. Neither of those groups is big on critical thinking, and they dominate most conversation.
Time/effort coming from experts who want to make things accessible is a gift! The paper is linked elsewhere in the thread if you want no-frills.
What a beautifully written article. It's extremely that I favourite an article but this is one.
Very extremely. Quite a lovely presentation. I'm definitely having a Patrick Bateman-esque appreciation for that delicate cream background.
So, this is either the paper of the year, or ... definitely not the paper of the year.
https://arxiv.org/pdf/2605.01172 is the current version. The money graphs are page 8 and on where they show (some weirdly thick) line charts with loss results reached in roughly 1/5 the number of steps that Adam takes, just what the blog post mentions.
They also claim holding back test data is not needed, also with more graphs.
I'm not an ML scientist, and I did not attempt to seriously parse the math. It reads to me as something precisely in that liminal space some math papers do where there's enough new terminology that actually parsing through it all is going to take real, concerted effort, possibly with mild brain damage as a risk.
Their 3d graphs of "kernel eigenstructure" also do double duty for me as totally impenetrable and possibly part of an April fool's ML paper that's hilarious to insiders. Or maybe they show something really amazing; they definitely seem to converge into a shape...What does that shape mean??? Why??? What is an eigenstructure? Is it just 3D eigenvectors of some matrices? Is it natural to have a 3D shape representing these large matrices? If not, how and why were these projected down? And why are they different colors in the paper?? You get the feel for my level of understanding.
I think it would frankly just be easier to validate this claim than parse the whole paper. If only I could understand
Well enough to implement the one line update to Adam in python. I have not asked codex or claude to assist yet.Also of note to me, they talk about grokking which I found SUUUPER fascinating when it was first reported, and have never heard about since. So I was really glad to read about it and read that there has been a little academic work on the phenomenon.
Finally, of the three models they repot results on, two are extremely tiny, the last is a DPO round on Qwen 0.5B -- if the code for that is published, I imagine it would be easy to adapt and evaluate in other regimes.