@article{lu2026scalerlt,
  title         = {{ScaleRLT}: Exact Log-Probabilities and Fast Deferred Rollouts for Reinforcement Learning on Looped Transformers},
  author        = {Lu, Dezhen and Ni, Zixiang and Li, Runguo},
  year          = {2026},
  eprint        = {2610.34427},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  url           = {https://arxiv.org/abs/2610.34427}
}
