@misc{indiciaed2a4ff8ef1b3,
  title = {Are LLM Uncertainty and Correctness Encoded by the Same Features? A Functional Dissociation via Sparse Autoencoders},
  author = {Het Patel and Tiejin Chen and Hua Wei and Evangelos E. Papalexakis and Jia Chen},
  year = {2026},
  url = {https://arxiv.org/abs/2604.19974},
  note = {Source identifier: 2604.19974}
}