PLUMAGE, an unbiased minimum-variance low-rank gradient estimator with optimizer-moment realignment, reduces the loss gap to full-rank training by about a third compared with GaLORE.
Title resolution pending
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
-
PLUMAGE: Probabilistic Low rank Unbiased Min Variance Gradient Estimator for Efficient Large Model Training
PLUMAGE, an unbiased minimum-variance low-rank gradient estimator with optimizer-moment realignment, reduces the loss gap to full-rank training by about a third compared with GaLORE.