@article{stabilizingonpolicydistillationformllmre, title = {Stabilizing On-Policy Distillation for MLLM Reasoning with Global Normalization}, author = {Dongze Hao and Zhiwei Jin and Chen Chen and Haonan Lu}, year = {2026}, eprint = {2606.09091}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2606.09091}, }