PyTorch 1.x Reinforcement Learning Cookbook

How to do it...

We develop the actor-critic algorithm in order to solve the CartPole environment as follows:

1. Import all the necessary packages and create a CartPole instance:

>>> import gym>>> import torch>>> import torch.nn as nn>>> import torch.nn.functional as F>>> env = gym.make('CartPole-v0')

2. Let's start with the actor-critic neural network model:

>>> class ActorCriticModel(nn.Module): ...     def __init__(self, n_input, n_output, n_hidden): ...         super(ActorCriticModel, self).__init__() ...         self.fc = nn.Linear(n_input, n_hidden) ...         self.action = nn.Linear(n_hidden, n_output) ...         self.value = nn.Linear(n_hidden, 1) ... ...     def forward(self, x): ...         x = torch.Tensor(x) ...         x = F.relu(self.fc(x)) ... action_probs = F.softmax(self.action(x), ...