python - 如何正确转发dropout层

Question

我创建了以下带有 dropout 层的深度网络，如下所示:

class QNet_dropout(nn.Module):

    """
        A MLP with 2 hidden layer and dropout

        observation_dim (int): number of observation features
        action_dim (int): Dimension of each action
        seed (int): Random seed
    """

    def __init__(self, observation_dim, action_dim, seed):
        super(QNet_dropout, self).__init__()
        self.seed = torch.manual_seed(seed)
        self.fc1 = nn.Linear(observation_dim, 128)
        self.fc2 = nn.Dropout(0.5)
        self.fc3 = nn.Linear(128, 64)
        self.fc4 = nn.Dropout(0.5)
        self.fc5 = nn.Linear(64, action_dim)

    def forward(self, observations):
        """
           Forward propagation of neural network

        """

        x = F.relu(self.fc1(observations))
        x = F.linear(self.fc2(x))
        x = F.relu(self.fc3(x))
        x = F.linear(self.fc4(x))
        x = self.fc5(x)
        return x

但是，当我尝试运行代码时，出现以下错误:

/home/workspace/QNetworks.py in forward(self, observations)
     90 
     91         x = F.relu(self.fc1(observations))
---> 92         x = F.linear(self.fc2(x))
     93         x = F.relu(self.fc3(x))
     94         x = F.linear(self.fc4(x))

TypeError: linear() missing 1 required positional argument: 'weight'

看来我没有正确使用/转发 dropout 层。对 dropout 层进行 Forward 的正确方法应该是什么？谢谢!

Answer 1

F.linear() 函数使用不正确。您应该使用您指定的线性函数而不是 torch.nn.function。 Dropout层应该在Relu之后。您可以从 torch.nn.function 调用 Relu 函数。

import torch
import torch.nn.functional as F

class QNet_dropout(nn.Module):

    """
        A MLP with 2 hidden layer and dropout

        observation_dim (int): number of observation features
        action_dim (int): Dimension of each action
        seed (int): Random seed
    """

    def __init__(self, observation_dim, action_dim, seed):
        super(QNet_dropout, self).__init__()
        self.seed = torch.manual_seed(seed)
        self.fc1 = nn.Linear(observation_dim, 128)
        self.fc2 = nn.Dropout(0.5)
        self.fc3 = nn.Linear(128, 64)
        self.fc4 = nn.Dropout(0.5)
        self.fc5 = nn.Linear(64, action_dim)

    def forward(self, observations):
        """
           Forward propagation of neural network

        """
        x = self.fc2(F.relu(self.fc1(observations)))
        x = self.fc4(F.relu(self.fc3(x)))
        x = self.fc5(x)
        return x

observation_dim = 512
model = QNet_dropout(observation_dim, 10, 512)
batch_size = 8
inpt  = torch.rand(batch_size, observation_dim)
output = model(inpt)
print ("output shape: ", output.shape)