Post-Training at the Edge of Detectability: A Game-Theoretic Approach to Fine-Tuning

@misc{harris2026posttraining,
  title = {Post-Training at the Edge of Detectability: A Game-Theoretic Approach to Fine-Tuning},
  author = {Keegan Harris and Brian Lee and Ian Waudby-Smith and P. Amortila and Nika Haghtalab and Michael I. Jordan},
  year = {2026},
  eprint = {2607.26358},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.26358},
}