@article{reasoningcachecontinualimprovementoverlo, title = {Reasoning Cache: Continual Improvement Over Long Horizons via Short-Horizon RL}, author = {Ian Wu and Yuxiao Qu and Amrith Setlur and Aviral Kumar}, year = {2026}, eprint = {2602.03773}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2602.03773}, }