RAYRL Parser#

RLlib Argument Parsing Utilities Module#

This module provides utility functions for parsing command-line arguments for RLlib-based scripts. It is adapted from the RAYRL library to include custom configurations and parameters specific to the influencer games project.

The modifications are tailored to support multi-agent reinforcement learning experiments in the context of influencer games, with additional arguments and configurations unique to this domain.

Features:#

  • Adds RLlib-typical example script arguments.

  • Supports custom configurations for multi-agent reinforcement learning experiments.

  • Includes domain-specific parameters for resource distribution and influence modeling.

Dependencies:#

  • InflGame.utils

  • InflGame.domains

Usage:#

The add_rl_example_script_args function adds RLlib-typical example script arguments to a parser, allowing for customization of reinforcement learning experiments in the influencer games framework.

Example:#

import argparse
from InflGame.MARL.utils.my_parse import add_rl_example_script_args

# Create a parser and add RLlib arguments
parser = argparse.ArgumentParser()
parser = add_rl_example_script_args(parser)

# Parse arguments
args = parser.parse_args()

# Access parsed arguments
print("Algorithm:", args.algo)
print("Number of agents:", args.num_agents)
print("Domain type:", args.domain_type)
print("Initial positions:", args.initial_position)

Functions

InflGame.MARL.utils.my_parse.add_rl_example_script_args(parser=None, default_reward=100.0, default_iters=200, default_timesteps=100000)#

Adds RLlib-typical (and common) example script command line arguments to a parser.

Note

This function should be used by most of our example scripts, which already mostly have this logic in them (but written out).

Parameters:

  • parser (Optional[argparse.ArgumentParser]) – The parser to add the arguments to. If None, create a new one.

  • default_reward (float) – The default value for the --stop-reward option.

  • default_iters (int) – The default value for the --stop-iters option.

  • default_timesteps (int) – The default value for the --stop-timesteps option.

Returns:

The altered (or newly created) parser object.

Return type:

argparse.ArgumentParser