The regularized Transformer loss satisfies Villani's coercive energy criteria, yielding log-Sobolev constants C_LS ≤ λ^{-1} + d/λ² and finite-time convergence bounds for noisy SGD.
(non-) asymptotic prop- erties of stochastic gradient langevin dynamics
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Weight-Decay Turns Transformer Loss Landscapes Villani: Functional-Analytic Foundations for Optimization and Generalization
The regularized Transformer loss satisfies Villani's coercive energy criteria, yielding log-Sobolev constants C_LS ≤ λ^{-1} + d/λ² and finite-time convergence bounds for noisy SGD.