Stabilizing RLHF through Advantage Model and Selective Rehearsal
- 2023-09-18
- 22
@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 Yu Dong},
year = {2023},
eprint = {2309.10202},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2309.10202},
}