Is it possible to query the AutoAscend agent at a given state?

[agarwl]

If I want to do behavior cloning on this agent, is it possible to do something like Dagger where I collect the data using an RL agent but query the teacher (AutoAscend) to get what action it would have taken instead?

[roger-creus]

@agarwl 3 years later, I still think this would probably close the gap between symbolic and neural policies in nethack... especially seeing this paper.

I have been trying to re-write the code to allow for Dagger without success... this would be great :)