@misc{lit_2511_16886,
  title = { {Latent Reasoning in TRMs is Secretly a Policy Improvement Operator} },
  author = {Arip Asadulaev and Rayan Banerjee and Fakhri Karray and Martin Takac},
  year = {2025},
  note = {arXiv preprint},
  eprint = {2511.16886},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2511.16886}
}
