@misc{indiciaef3f56cfb9486,
  title = {Forecasting How RLHF Responds to Reward-Model Edits Before Training},
  author = {Zixiao Wang and Duzhen Zhang and Xingbo Du and Dongxia Wu and Le Song},
  year = {2026},
  url = {https://arxiv.org/abs/2611.03244},
  note = {Source identifier: 2611.03244}
}