You can find some examples from the grid-world in the notebooks folder.

  pip install -r requirements.txt

To learn the constraints run:

  python -m max_ent.examples.learn_hard_constraints

After learning, run the following to generate the reports (in ./reports/hard/ folder):

  python -m max_ent.examples.compare_hard_results

To learn the constraints run:

  python -m max_ent.examples.learn_soft_constraints

After learning, run the following to generate the reports (in ./reports/soft/ folder):

  python -m max_ent.examples.compare_soft_results

To run the transfer learning experiments and generate the results use:

  python -m max_ent.examples.transfer

The generated reports can be found in ./reports/transfer/ folder.

Run the notebook in ./notebooks/new_metrics.ipynb .

Also, you can set learn = True in ./max_ent/examples/orchestrator_exp.py then run:

  python -m max_ent.examples.orchestrator_exp

After that, set learn = False and run the above command again. The reports will be generated into ./reports/orchestrator/ folder.

This repository uses and modifies some codes from irl-maxent library.