@misc{indiciae97ba18f00548,
  title = {Where Rounding Happens Matters: Rounding-Location Alignment Stabilizes LLM Reinforcement Learning},
  author = {Yijing Ke and Tianle Cai and Ruosong Wang},
  year = {2026},
  url = {https://arxiv.org/abs/2610.32951},
  note = {Source identifier: 2610.32951}
}