TY  - RPRT
TI  - RoiRL: Efficient, Self-Supervised Reasoning with Offline Iterative Reinforcement Learning
AU  - Aleksei Arzhantsev
AU  - Otmane Sakhi
AU  - Flavian Vasile
PY  - 2025
UR  - https://arxiv.org/abs/2510.02892
ID  - 2510.02892
ER  -