v1.0.5

Fix rlgym_learn.pyi not working on some systems where typeshed types don't get patched in at runtime

v1.0.4

• Add Python 3.13 support
• remove "device" parameter from learning coordinator config model (it should be decided by agent controllers' config)
• update example scripts for rlgym-learn-algos 0.2.0

v1.0.3

• update and correct bindings
• add method of instantiating timestep outside of Rust
• rename config model parameter in the generate_config method
• fix state setting
• update tests and quick start guide to use correct kwargs
• minor error message improvements

v1.0.2

• Fix RLViser hard-coded integration in env process
• update pyo3 and numpy versions
• fix examples

v1.0.1

Update quick_start_guide.py and fix .pyi bindings for numpy serde type

v1.0.0

First major release

v0.1.8

• refactor ppo implementation to rlgym-learn-algos
• implement Rocket League helper functions + classes
• improve flexibility of state info and timestep data with addition to EnvActionResponse to choose when to use the state serde to receive the state from the given env process
• slight QOL changes

v0.1.7

various optimizations

v0.1.6

• switch over to shared info instead of state metrics
• allow setting shared info keys as part of env actions
• github issue fixes

v0.1.5

update pyany-serde to support linux