Riemannian optimization on low-rank attention parameters yields no conclusive improvement over AdamW after hyperparameter tuning.
MoLAE: Mixture of latent experts for parameter-efficient language models
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
-
Riemannian Gradient Descent for Low-Rank Architectures
Riemannian optimization on low-rank attention parameters yields no conclusive improvement over AdamW after hyperparameter tuning.