search

Unity-Technologies / ml-agents

The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents using deep reinforcement learning and imitation learning.
https://unity.com/products/machine-learning-agents
Other
17.12k stars 4.15k forks source link

While learning in curriculum learning, it would be nice to be able to see that the environment parameter changes as the lesson changes. #4464

Closed jwson97 closed 2 years ago

jwson97 commented 4 years ago

During the training step, reward and mean reward are now displayed. It would be nice to show environment parameter changes as well. And I would like to show the status of the lessons currently in progress.

andrewcoh commented 4 years ago

Hi @jwson97,

Just to be sure I understand, do you mean you'd like to see the actual value of the environment parameter during training in addition to the lesson number which is currently reported on tensorboard?

Please correct me if I'm misunderstanding. I'll bring this up to the team during the next meeting.

jwson97 commented 4 years ago

Yes, that's what I want. Thanks.

andrewcoh commented 4 years ago

This may not be a feature we can support in general. Environment parameters are not always float values, they can also tuples that describe a sampler i.e. https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Training-ML-Agents.md#supported-sampler-types. It's not clear how we could generalize to this and future controllers for environment parameters.

github-actions[bot] commented 1 year ago

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.