Как я поймал Трансформер на читерстве: гроккинг, математика и Mechanistic Interpretability

Феномен  Grokking  и  Mechanistic Interpretability  — главные тренды в исследованиях лабораторий уровня OpenAI и Anthropic. Я решил потрогать эти концепции своими руками на уровне...

Avinav Sahoo(he/ him/ his) (@avinavsahoo) on X

At Pulseinnovas India we have done the impossible. Grokking used to take tens of thousands of epoch we have reduced it less than 500 epochs on the same data @elonmusk @sama @AnjneyMidha @bonatsos @a16z @deedydas @demishassabis @sundarpichai @anandmahindra reduced the cost by 70%

$\mathbf{Li_2}$: A Framework on Dynamics of Feature Emergence and Delayed Generalization

While the phenomenon of grokking, i.e., delayed generalization, has been studied extensively, it remains an open question whether there is a mathematical framework to characterize what kind of features emerge, how and in which conditions it happens from training, for complex structured inputs. We propose a novel framework, named $\mathbf{Li_2}$, that captures three key stages for the grokking behavior of 2-layer nonlinear networks: (I) Lazy learning, (II) independent feature learning and (III) interactive feature learning, characterized by the structure of backpropagated gradient $G_F$ across layers. In (I), $G_F$ is random, and top layer overfits to random hidden representation. In (II), the gradient of each node (column of $G_F$) only depends on its own activation, and thus each hidden node learns their representation independently from $G_F$, which now carries information about target labels, thanks to weight decay. Interestingly, the independent dynamics follows exactly the gradient ascent of an energy function $E$, and its local maxima are precisely the emerging features. We study whether these local-optima induced features are generalizable, their representation power, and how they change on sample size, in group arithmetic tasks. Finally, in (III), we provably show how hidden nodes interact, and how $G_F$ changes to focus on missing features that need to be learned. Our study sheds lights on roles played by key hyperparameters such as weight decay, learning rate and sample sizes in grokking, leads to provable scaling laws of memorization and generalization, and reveals the underlying cause why recent optimizers such as Muon can be effective, from the first principles of gradient dynamics. Our analysis can be extended to multi-layer architectures.

Hackers Exploit X’s Grok AI to Push Malicious Links Through Ads

Malicious actors have found a new way to slip harmful links into X’s promoted posts by tricking Grok, the platform’s AI assistant.

Command Line Handbook

e509 — Maverick and Marbles | Games At Work dot Biz

stories and discussion all around AI, LLMs, llamas, generated Quake, grokking, generalization and much more.

e509 — Maverick and Marbles | Games At Work dot Biz

stories and discussion all around AI, LLMs, llamas, generated Quake, grokking, generalization and much more.