Stabilizing RLHF through Advantage Model and Selective Rehearsal

@misc{peng2023stabilizing,
  title = {Stabilizing RLHF through Advantage Model and Selective Rehearsal},
  author = {Baolin Peng and Linfeng Song and Ye Tian and Lifeng Jin and Haitao Mi and Dong Yu},
  year = {2023},
  eprint = {2309.10202},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2309.10202},
}