TY  - RPRT
TI  - Freeze Deep, Train Shallow: Interpretable Layer Allocation for Continued Pre-Training
AU  - Yu-Hang Wu
AU  - Qin-Yuan Liu
AU  - Qiu-Yang Zhao
AU  - Bo Jiang
AU  - Jiang-Feng Yang
AU  - Qing-Wei Cong
PY  - 2026
UR  - https://arxiv.org/abs/2605.11416
ID  - 2605.11416
ER  -