arXiv · 2026 · Preprint

On the Implicit Reward Overfitting and the Low-rank Dynamics in RLVR

Hao Ye, Jisheng Dang, Jun-Feng Fang, Bimei Wang, Yizhou Zhang, Ning Lv, Wen-Can Zhang, Hong Peng, Bin Hu, Tat-Seng Chua

Date
2026-05-07
Citations
2
arXiv
2605.06523
Cite
@misc{ye2026implicit,
  title = {On the Implicit Reward Overfitting and the Low-rank Dynamics in RLVR},
  author = {Hao Ye and Jisheng Dang and Jun-Feng Fang and Bimei Wang and Yizhou Zhang and Ning Lv and Wen-Can Zhang and Hong Peng and Bin Hu and Tat-Seng Chua},
  year = {2026},
  eprint = {2605.06523},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.06523},
}