SABRE

Code for
Provable Safe Reinforcement Learning with Binary Feedback
Andrew Bennett, Dipendra Misra, Nathan Kallus
https://arxiv.org/abs/2210.14492

Code for running SABRE algorithm has been moved to new repository https://github.com/microsoft/Intrepid, which also provides code for several related RL algorithms