imitation Inverse Q-Learning (IQ-Learn) implementation

Hi, I am currently experimenting with a couple of imitation learning algorithms. I have recently found this paper and I was planning on giving a shot implementing it. Do you accept contributions? If yes, are there any guidelines ? ( I don't have too much experience with open source projects )

Jan 30 '22 07:01 hhroberthdaniel

Hi,

Yes, contributions are welcome! Especially as the reference implementation looks to not be free software, so having an open-source implementation of this would be valuable. Although this does mean you'll need to be careful to a) not copy the reference implementation at all, and b) include a warning to users that the software may be covered by a patent.

Please see our CONTRIBUTING.md for code style and testing. It looks like IQ-Learn learns from demonstrations, so it should subclass DemonstrationAlgorithm.

To merge your contribution we'd need to have unit tests covering the algorithm (we currently have quite high code coverage of 97% that we'd like to maintain), as well as benchmark results showing it matches the reference implementation and/or results reported on in the paper.

Hope this helps and let me know if you have any questions!

Adam

Jan 30 '22 21:01 AdamGleave

Hi,

How should the warning be included? Like actually thrown in code? Or in the documentation?

Jan 31 '22 10:01 hhroberthdaniel

I think prominently placed in the documentation should be sufficient.

Jan 31 '22 21:01 AdamGleave

Closing due to inactivity. Contributions here would still be welcome, though!

Oct 22 '22 02:10 AdamGleave

@hhroberthdaniel Hello, I am working on it as well. Were you able to implement it ?

Oct 22 '23 10:10 azafar1991

@hhroberthdaniel Hello, I am working on it as well. Were you able to implement it ?

Hi, didn't try, I had better than expected results with BC

Oct 22 '23 13:10 hhroberthdaniel