TY  - RPRT
TI  - DeepCompress: A Dual Reward Strategy for Dynamically Exploring and Compressing Reasoning Chains
AU  - Tian Liang
AU  - Wenxiang Jiao
AU  - Zhiwei He
AU  - Jiahao Xu
AU  - Haitao Mi
AU  - Dong Yu
PY  - 2026
UR  - https://arxiv.org/abs/2510.27419
ID  - 2510.27419
ER  -