r""" Borrowing the ``Policy`` class from the Reinforcement Learning example. Copying the code to make these two examples independent. See https://github.com/pytorch/examples/tree/main/reinforcement_learning
| 38 | |
| 39 | |
| 40 | class Policy(nn.Module): |
| 41 | r""" |
| 42 | Borrowing the ``Policy`` class from the Reinforcement Learning example. |
| 43 | Copying the code to make these two examples independent. |
| 44 | See https://github.com/pytorch/examples/tree/main/reinforcement_learning |
| 45 | """ |
| 46 | def __init__(self, batch=True): |
| 47 | super(Policy, self).__init__() |
| 48 | self.affine1 = nn.Linear(4, 128) |
| 49 | self.dropout = nn.Dropout(p=0.6) |
| 50 | self.affine2 = nn.Linear(128, 2) |
| 51 | self.dim = 2 if batch else 1 |
| 52 | |
| 53 | def forward(self, x): |
| 54 | x = self.affine1(x) |
| 55 | x = self.dropout(x) |
| 56 | x = F.relu(x) |
| 57 | action_scores = self.affine2(x) |
| 58 | return F.softmax(action_scores, dim=self.dim) |
| 59 | |
| 60 | |
| 61 | class Observer: |