@article{omnithinkerscalingmultitaskrlinllmswithh, title = {Omni-Thinker: Scaling Multi-Task RL in LLMs with Hybrid Reward and Task Scheduling}, author = {Derek Li and Jiaming Zhou and Leo Maxime Brunswic and Abbas Ghaddar and Qianyi Sun and Liheng Ma and Yu Luo and Dong Li and Mark Coates and Jianye Hao and Yingxue Zhang}, year = {2025}, eprint = {2507.14783}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2507.14783}, }