What's Changed
- support SMACv2 and add ExpoComm by @TangY1fan in #203
- Fix model loading and test (#204) by @KKTQawa in #205
- Fix crashes in torch PDQN/MPDQN/SPDQN and Platform env compatibility by @xross223 in #206
- Renamed agent.action() to agent.get_actions() to avoid confusion with the agent.actions variable.
- Introduced tensor memory to improve GPU utilization and training efficiency.
- Fixed issues in the MARL replay buffer caused by incorrect variable name references.
New Contributors
Full Changelog: 1.4.3...1.4.4