ml.tasks.environments.base
Defines a generic reinforcement learning environment class.
- class ml.tasks.environments.base.Environment[source]
Bases:
ABC,Generic[RLState,RLAction]- abstract reset(seed: int | None = None) RLState[source]
Gets the initial environment state.
- Parameters:
seed – The initial random seed to use
- Returns:
The initial state of the environment.
- abstract render(state: RLState) ndarray | Tensor[source]
Renders the environment.
- Parameters:
state – The state to render
- Returns:
The rendered environment as a single frame, as an image array.
- abstract sample_action() RLAction[source]
Samples an action from the environment’s action space.
- Returns:
The sampled action.
- abstract step(action: RLAction) RLState[source]
Performs a single step in the environment.
- Parameters:
action – The action to perform in the environment.
- Returns:
The next state of the environment.
- abstract terminated(state: RLState) bool[source]
Checks if the environment has finished.
- Parameters:
state – The most recent state
- Returns:
If the environment has finished
- property fps: int