Grokking is induced by any small nonzero regularizer whose favored solutions generalize, with a delay that scales like one over the learning rate times the regularization strength.
Then for all 0 ≤ j ≤ k − 1, |Rj| ≤ ϵα∥∇g(x(j))∥2 2 if α ≤ 2(ϵ − γ)/L2p (ϵ + γ)α∥∇g(x(j))∥2 2 if α ≤ 2ϵ/L2p (37) Proof
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Grokking Beyond the Euclidean Norm of Model Parameters
Grokking is induced by any small nonzero regularizer whose favored solutions generalize, with a delay that scales like one over the learning rate times the regularization strength.