TY  - RPRT
TI  - CLEANER: Self-Purified Trajectories Boost Agentic Reinforcement Learning
AU  - Tianshi Xu
AU  - Yuteng Chen
AU  - Meng Li
PY  - 2026
UR  - https://arxiv.org/abs/2601.15141
ID  - 2601.15141
ER  -