@article{saifasparseautoencoderframeworkfor, title = {SAIF: A Sparse Autoencoder Framework for Interpreting and Steering Instruction Following of Language Models}, author = {Zirui He and Haiyan Zhao and Yiran Qiao and Fan Yang and Ali Payani and Jing Ma and Mengnan Du}, year = {2025}, eprint = {2502.11356}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2502.11356v1}, }