benchmarl.environments.PettingZooTask

class PettingZooTask(value)

Bases: Task

Enum for PettingZoo tasks.

MULTIWALKER = 1

WATERWORLD = 2

SIMPLE_ADVERSARY = 3

SIMPLE_CRYPTO = 4

SIMPLE_PUSH = 5

SIMPLE_REFERENCE = 6

SIMPLE_SPEAKER_LISTENER = 7

SIMPLE_SPREAD = 8

SIMPLE_TAG = 9

SIMPLE_WORLD_COMM = 10

static associated_class()

The associated task class

classmethod env_name() → str

The name of the environment in the benchmarl/conf/task folder

get_task(config: Dict[str, Any] | None = None) → TaskClass

Get the TaskClass object associated with this enum element by passing it the task name and config.

If no config is given, it will be loaded from benchmarl/conf/task/self.env_name()/self.name using Task.get_from_yaml().

Parameters:

• config (dict) – Optional configuration of the task.

• provided (If not)

• yaml. (the default configuration will be loaded from)

Returns:

The TaskClass object for the task.

get_from_yaml(path: str | None = None) → TaskClass

Load the task configuration from yaml

Parameters:

path (str, optional) – The full path of the yaml file to load from. If None, it will default to benchmarl/conf/task/self.env_name()/self.name

Returns:

the TaskClass with the loaded config

property config

update_config(config: Dict[str, Any]) → Task

supports_continuous_actions() → bool

supports_discrete_actions() → bool

max_steps(env: EnvBase) → int

has_render(env: EnvBase) → bool

group_map(env: EnvBase) → Dict[str, List[str]]

observation_spec(env: EnvBase) → Composite

info_spec(env: EnvBase) → Composite | None

state_spec(env: EnvBase) → Composite | None

action_spec(env: EnvBase) → Composite

action_mask_spec(env: EnvBase) → Composite | None

static log_info(batch: TensorDictBase) → Dict[str, float]

get_reward_sum_transform(env: EnvBase) → Transform

get_env_transforms(env: EnvBase) → List[Transform]

get_replay_buffer_transforms(env: EnvBase, group: str) → List[Transform]

static render_callback(experiment, env: EnvBase, data: TensorDictBase)