Scalable and Efficient URLLC Resource Allocation Based on Offline Graph Reinforcement Learning | IEEE Conference Publication | IEEE Xplore