API for reinforcement learning environments. Each environment exposes a Gymnasium like interface. Specifically:
GET /{idx}/is-alive
POST /{idx}/close
POST /make
POST /{idx}/reset
POST /{idx}/step
There is also a limited API for Tensorboard:
POST /init
POST /close
POST /add-text
POST /add-scalar
POST /add-scalars
The API is based on FastAPI. You can launch a server using the entrypoint.sh script.
Alternatively, you can use Docker. You can find examples with C++ at: bitrl