arlbench.core.algorithms.dqn.models

Q-Networks for DQN.

Classes

CNNQ(action_dim[, activation, hidden_size, ...])

A CNN-based Q-Network for DQN.

MLPQ(action_dim[, activation, hidden_size, ...])

An MLP-based Q-Network for DQN.

class arlbench.core.algorithms.dqn.models.CNNQ(action_dim, activation='tanh', hidden_size=512, discrete=True, parent=<flax.linen.module._Sentinel object>, name=None)[source]

Bases: Module

A CNN-based Q-Network for DQN.

__call__(x)[source]

Applies the CNN to the input.

setup()[source]

Initializes the CNN Q-Network.

class arlbench.core.algorithms.dqn.models.MLPQ(action_dim, activation='tanh', hidden_size=64, discrete=True, parent=<flax.linen.module._Sentinel object>, name=None)[source]

Bases: Module

An MLP-based Q-Network for DQN.

__call__(x)[source]

Applies the MLP to the input.

setup()[source]

Initializes the MLP Q-Network.