Fetching the paper…

Omni-Thinker: Scaling Multi-Task RL in LLMs with Hybrid Reward and Task Scheduling · Around