RewardClippingWrapper

class maze.core.wrappers.reward_clipping_wrapper.RewardClippingWrapper(env: MazeEnv, min_val: float, max_val: float)

Clips original step reward to range [min, max].

Parameters:

• env – The underlying environment.

• min_val – Minimum allowed reward value.

• max_val – Maximum allowed reward value.

clone_from(env: RewardClippingWrapper) → None

(overrides SimulatedEnvMixin)

implementation of SimulatedEnvMixin.

reward(reward: float) → float

(overrides RewardWrapper)

Clips the original reward.

param reward:

The original reward.

return:

The clipped reward.