Permutation symmetry alone forces the leading behavior of a neural layer into a quadratic form, and the resulting dynamics unify sudden learning and training-time scaling laws.
Parameter symmetry potentially unifies deep learning theory, 2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Neural Quadratic Forms: A Unified Minimal Model for Sudden Learning and Scaling Laws
Permutation symmetry alone forces the leading behavior of a neural layer into a quadratic form, and the resulting dynamics unify sudden learning and training-time scaling laws.