Unveiling Decision-Making in LLMs for Text Classification : Extraction of influential and interpretable concepts with Sparse Autoencoders

@misc{dentan2025unveilingdecisionmakingllms,
      title={Unveiling Decision-Making in LLMs for Text Classification : Extraction of influential and interpretable concepts with Sparse Autoencoders},
      author={Jérémie Dentan and Davide Buscaldi and Sonia Vanier and Mathis Le Bail},
      year={2025},
      eprint={2506.23951},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2506.23951},
}