TY  - RPRT
TI  - Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture
AU  - Tianhao Fu
AU  - Xinxin Xu
AU  - Weichen Xu
AU  - Jue Chen
AU  - Ruilong Ren
AU  - Bowen Deng
AU  - Xinyu Zhao
AU  - Jian Cao
AU  - Xixin Cao
PY  - 2026
UR  - https://arxiv.org/abs/2511.07110
ID  - 2511.07110
ER  -