TY  - RPRT
TI  - LORENZA: Enhancing Generalization in Low-Rank Gradient LLM Training via Efficient Zeroth-Order Adaptive SAM
AU  - Yehonathan Refael
AU  - Iftach Arbel
AU  - Ofir Lindenbaum
AU  - Tom Tirer
PY  - 2025
UR  - https://arxiv.org/abs/2502.19571
ID  - 2502.19571
ER  -