1
CARE-RL: Capability-Aware Reinforcement Learning for Mitigating Cross-Domain Conflicts
提出CARE-RL框架，通过能力感知机制缓解强化学习中的跨域冲突，实现多任务协同优化。
arXiv:2606.00609v1 Announce Type: new Abstract: Reinforcement learning (RL) with verifiable rewards has achieved strong progress in reasoning-oriented…