UniformSampler

The UniformSampler class provides functionality to sample environments and agents uniformly for simulation purposes. It extends from the BaseSampler class.

Methods

sample

def sample(
    self,
    agent_classes: Type[BaseAgent[ObsType, ActType]]
    | list[Type[BaseAgent[ObsType, ActType]]],
    n_agent: int = 2,
    replacement: bool = True,
    size: int = 1,
    env_params: dict[str, Any] = {},
    agents_params: list[dict[str, Any]] = [{}, {}],
) -> Generator[EnvAgentCombo[ObsType, ActType], None, None]:

Sample an environment and n_agent agents.

Parameters:

• agent_classes (Type[BaseAgent[ObsType, ActType]] | list[Type[BaseAgent[ObsType, ActType]]]): The agent class or list of agent classes to be sampled.
• n_agent (int): Number of agents to sample. Defaults to 2.
• replacement (bool): Indicates if sampling is with replacement. Defaults to True.
• size (int): Number of samples to generate. Defaults to 1.
• env_params (dict[str, Any]): Parameters for the environment. Defaults to an empty dictionary {}.
• agents_params (list[dict[str, Any]]): List of parameters for each agent. Defaults to [{}, {}].

Returns:

• Generator[EnvAgentCombo[ObsType, ActType], None, None]: A generator yielding tuples of environment and agents.

Runtime Checks:

1. If agent_classes is a list, it must have a length of n_agent.
2. agents_params must also be a list of length n_agent.
3. Uniform sampling without replacement is not supported due to inefficiencies in sequential sampling and rejection sampling.

Usage Example

from sotopia.agents.some_agent import SomeAgent
from sotopia.samplers.uniform_sampler import UniformSampler
 
sampler = UniformSampler()
agent_classes = [SomeAgent, SomeAgent]
agents_params = [{"param1": "value1"}, {"param2": "value2"}]
 
for env, agents in sampler.sample(agent_classes, n_agent=2, agents_params=agents_params):
    print(env)
    for agent in agents:
        print(agent)