Toggle navigation
OpenReview
.net
Login
×
Go to
DBLP
homepage
Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization
Taishi Nakamura
,
Takuya Akiba
,
Kazuki Fujii
,
Yusuke Oda
,
Rio Yokota
,
Jun Suzuki
Published: 01 Jan 2025, Last Modified: 30 Jun 2025
ICLR 2025
Everyone
Revisions
BibTeX
CC BY-SA 4.0
Loading