@misc{indiciae5f7ac35a291f,
  title = {Next-token pretraining implies in-context learning},
  author = {Paul M. Riechers and Henry R. Bigelow and Eric A. Alt and Adam Shai},
  year = {2025},
  url = {https://arxiv.org/abs/2505.18373},
  note = {Source identifier: 2505.18373}
}