TY  - RPRT
TI  - Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs
AU  - Ziliang Wang
AU  - Kang An
AU  - Faqiang Qian
AU  - Jialu Cai
AU  - Cijun Ouyang
AU  - Yuhang Wang
AU  - Qibing Ren
AU  - Yichao Wu
PY  - 2026
UR  - https://arxiv.org/abs/2606.17735
ID  - 2606.17735
ER  -