OakEmu-RL v0.1.0

Appearance

DQNModelConfiguration ​

Configuration for the DQN model architecture.

Source

Values ​

categorical ​

python
categorical: CategoricalConfiguration | None = None

Configuration for Categorical DQN (C51). If None, categorical DQN is disabled.

Source

conv_activation ​

python
conv_activation: str = 'relu'

Activation function for convolutional layers.

Source

conv_filters ​

python
conv_filters: list[tuple[int, int | tuple[int, int], int | tuple[int, int]]] = [(32, 8, 4), (64, 4, 2), (64, 3, 1)]

Convolutional filters: (out_channels, kernel_size, stride).

Source

dueling ​

python
dueling: bool = False

Whether to use Duelling architecture.

Source

fcnet_activation ​

python
fcnet_activation: str = 'relu'

Activation function for initial fully connected layers.

Source

fcnet_hiddens ​

python
fcnet_hiddens: list[Annotated[int, Field(gt=0)]] = [512]

Hidden layer sizes for the initial fully connected network.

Source

head_fcnet_activation ​

python
head_fcnet_activation: str = 'relu'

Activation function for head fully connected layers.

Source

head_fcnet_hiddens ​

python
head_fcnet_hiddens: list[Annotated[int, Field(gt=0)]] = []

Hidden layer sizes for the head fully connected network.

Source

noisy_nets ​

python
noisy_nets: NoisyNetsConfiguration | None = None

Configuration for Noisy Networks. If None, Noisy Networks are disabled.

Source