@misc{indiciaedbd05ff394c3,
  title = {Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment},
  author = {Harrish Thasarathan and Julian Forsyth and Thomas Fel and Matthew Kowal and Konstantinos G. Derpanis},
  year = {2026},
  url = {https://arxiv.org/abs/2502.03714},
  note = {Source identifier: 2502.03714}
}