SODA unifies several modern optimizers under optimistic dual averaging and supplies a 1/k decay wrapper that improves performance without weight decay tuning.
Logarithmic- time schedules for scaling language models with momentum.arXiv preprint arXiv:2602.05298
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
-
Optimistic Dual Averaging Unifies Modern Optimizers
SODA unifies several modern optimizers under optimistic dual averaging and supplies a 1/k decay wrapper that improves performance without weight decay tuning.