Closed marlon360 closed 5 years ago
I need to adapt the ML-Agents curriculum to work with yaml configuration files, this won't work at the moment. Let me have a look and get back to you with an ETA for this.
Could you describe what you were trying to use as a curriculum? Did you want to be able to reset arenas with new configurations each time an agent reaches some threshold on a given metric?
Yes exactly. The documentation says that this is easy to implement, but I just know the ML-Agents way.
Some update for you: I've adapted the MetaCurriculum and Curriculum to work with the yaml configuration files, I'm running some tests and writing documentation. I plan on releasing this in v1.1 on Monday.
I created a MetaCurriculm and added it to the TrainerController. I got an error on the environment reset.
I modified the trainMLAgents.py:
And this is the error: