Increasing L2 regularization produces additional first-order accuracy transitions beyond the known onset-of-learning, and the paper interprets grokking as hysteresis across these transitions.
Information bottleneck for Gaussian variables,
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
-
Phase Transitions between Accuracy Regimes in L2 regularized Deep Neural Networks
Increasing L2 regularization produces additional first-order accuracy transitions beyond the known onset-of-learning, and the paper interprets grokking as hysteresis across these transitions.