src.gridmind.policies.parameterized.discrete_action_cnn_policy

Classes

DiscreteActionCNNPolicy

Helper class that provides a standard way to create an ABC using

Module Contents

class src.gridmind.policies.parameterized.discrete_action_cnn_policy.DiscreteActionCNNPolicy(observation_shape: tuple, num_actions: int)

Bases: gridmind.policies.parameterized.base_parameterized_policy.BaseParameterizedPolicy

Helper class that provides a standard way to create an ABC using inheritance.

conv1

conv2

conv3

fc1

fc2

_get_conv_output_size(x)

Helper function to compute the size of the flattened output after convolutions.

forward(x)

get_action(state)

get_action_prob(state, action)

update(state, action, value)