random

Source code: tianshou/exploration/random.py

---

class BaseNoise

The action noise base class.

abstract reset() → None

Reset to the initial state.

class GaussianNoise(mu: float = 0.0, sigma: float = 1.0)

The vanilla Gaussian process, for exploration in DDPG by default.

reset() → None

Reset to the initial state.

class OUNoise(mu: float = 0.0, sigma: float = 0.3, theta: float = 0.15, dt: float = 0.01, x0: float | ndarray | None = None)

Class for Ornstein-Uhlenbeck process, as used for exploration in DDPG.

Usage:

# init
self.noise = OUNoise()
# generate noise
noise = self.noise(logits.shape, eps)

For required parameters, you can refer to the stackoverflow page. However, our experiment result shows that (similar to OpenAI SpinningUp) using vanilla Gaussian process has little difference from using the Ornstein-Uhlenbeck process.

reset() → None

Reset to the initial state.