whiRL 2.0 verification tool for deep reinforcement-learning systems

This repository includes the artifact for the whiRL 2.0 tool for verification of RL-driven systems.

This artifact includes the whiRL 2.0 tool's code along with original experiments, as elaborated in our paper "Towards Scalable Verification of Deep Reinforcement Learning" (FMCAD 2021): https://arxiv.org/abs/2105.11931

For specific instructions please read the README.txt file.

For technical questions - contact us at guyam@cs.huji.ac.il