@misc{indiciae63eb762e55a1, title = {DeepCompress: A Dual Reward Strategy for Dynamically Exploring and Compressing Reasoning Chains}, author = {Tian Liang and Wenxiang Jiao and Zhiwei He and Jiahao Xu and Haitao Mi and Dong Yu}, year = {2026}, url = {https://arxiv.org/abs/2510.27419}, note = {Source identifier: 2510.27419} }