opentau.configs.reward

Reward configuration module.

This module provides the RewardConfig class which contains configuration parameters for reward computation in reinforcement learning scenarios.

Classes

RewardConfig([number_of_bins, C_neg, ...])

Configuration for reward computation settings.

class opentau.configs.reward.RewardConfig(number_of_bins: int = 201, C_neg: float = -1000.0, reward_normalizer: int = 400, N_steps_look_ahead: int = 50)

Bases: object

Configuration for reward computation settings.

This configuration is used for reward modeling and computation in reinforcement learning scenarios.

Parameters:

• number_of_bins – Number of bins used for reward discretization or binning. Defaults to 201.

• C_neg – Negative constant used in reward computation. Defaults to -1000.0.

• reward_normalizer – Normalization factor for rewards. Defaults to 400.

• N_steps_look_ahead – Number of steps to look ahead when computing rewards. Defaults to 50.

C_neg: float = -1000.0

N_steps_look_ahead: int = 50

__init__(number_of_bins: int = 201, C_neg: float = -1000.0, reward_normalizer: int = 400, N_steps_look_ahead: int = 50) → None

number_of_bins: int = 201

reward_normalizer: int = 400