@article{featurehedgingcorrelatedfeaturesbreak,
  title = {Feature Hedging: Correlated Features Break Narrow Sparse Autoencoders},
  author = {David Chanin and Tomáš Dulka and Adrià Garriga-Alonso},
  year = {2025},
  eprint = {2505.11756},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2505.11756v1},
}