TY  - RPRT
TI  - When Pruning Meets Interpretability: Preserving Sparse Autoencoder Robustness in LLMs
AU  - Suchit Gupte
AU  - Xueru Zhang
AU  - Mohammad Mahdi Khalili
PY  - 2026
UR  - https://arxiv.org/abs/2608.25941
ID  - 2608.25941
ER  -