v1.0.3

Changed

• Define is_focal is in scenario configs.
• Use chex.dataclass for dm-tree compatibility.

Fixed

• Use correct is_focal settings for team-vs-team games #16.

v1.0.2

Added

• Substrates and Scenarios now have ReactiveX observables.

Changed

• Don't add INVENTORY observation to scenarios that don't use it.
• Various updates to RLlib example.
• Improved performance of the component system for environments.

Fixed

• Simulation Speed #7
• Horizon setting in examples #9
• Error running example "self_play_train.py" #10
• build lab2d on m1 chip #13

Full Changelog: https://github.com/deepmind/meltingpot/commits/v1.0.2

v1.0.1

Submitted a number of fixes to ensure substrates and scenarios operate as intended. Other changes not expected to have any impact.

Fixed:

• Python seed is not forwarded to Lua #4
• Second episode using same seed as the first #5
• Bots receive incorrect timesteps in Scenario #3

Full Changelog: https://github.com/deepmind/meltingpot/commits/v1.0.1

v1.0.0

Initial release.

Full Changelog: https://github.com/deepmind/meltingpot/commits/v1.0.0