SafeMIL: Learning Offline Safe Imitation Policy from Non-Preferred Trajectories

OffRLRe-assign community