How to reproduce the results of Figure 3 in the paper "Interpretable and Explainable Logical Policies via Neurally Guided Symbolic Abstraction"

[hualuoyu]

Dear developer,
Thanks for your open source code of "Interpretable and Explainable Logical Policies via Neurally Guided Symbolic Abstraction". In the Readme.md in the code, python train.py -c in/config/my_config.yaml for training, but the algorithm in the yaml is ppo or logic. How to set the Neural PPO, Neural Logic RL, and the most important thing about how to write a Python command is to train the NUDGE algorithm. Besides, after the command "python train.py -c in/config/my_config.yaml" , how to set command to test the algorithm's performance to collect returns for draw the figure 3. Hope you can help me, Thanks a lot.