Be notified of new releases
Create your free GitHub account today to subscribe to this repository for new releases and build software alongside 31 million developers.Sign up
We have reorganized the project repository. Please see Migrating from v0.4 to v0.5 documentation for more information. Highlighted changes to repository structure include:
pythonfolder has been renamed
ml-agents.It now contains a python package called
unity-environmentfolder, containing the Unity project, has been renamed
- The protobuf definitions used for communication have been added to a new
- Example curricula and the trainer configuration file have been moved to a new
To learn more about new and improved environments, see our Example Environments page.
The following environments have been changes to use Multi Discrete Action:
The following environment has been modified to use Action Masking:
[Gym] New package
gym-unitywhich provides gym interface to wrap
UnityEnvironment. More information here.
[Training] Can now run multiple concurent training sessions with the
--num-runs=<n>command line option. (Training sessions are independent, and do not improve learning performance.)
[Unity] Meta-Curriculum. Supports curriculum learning in multi-brain environments.
[Unity] Action Masking for Discrete Control - It is now possible to mask invalid actions each step to limit the actions an agent can take.
[Unity] Action Branches for Discrete Control - It is now possible to define discrete action spaces which contain multiple branches, each with its own space size.
- Can now visualize value estimates when using models trained with PPO from Unity with
- It is now possible to specify which camera the
- Console summaries will now be displayed even when running inference mode from python.
- Minimum supported Unity version is now 2017.4.
Fixes & Performance Improvements
- Replaced some activation functions to
- Visual Observations use PNG instead of JPEG to avoid compression losses.
- Improved python unit tests.
- Fix to enable multiple training sessions on single GPU.
- Curriculum lessons are now tracked correctly.
- Ending training early using
CTL+Cdoes not save the model on Windows.
- Sequentially opening and closing multiple instances of
UnityEnvironmentwithin a single process is not possible.