Consider the following Module or class :123456789101112class Net(nn.Module):    def __init__(self, in_size, n_hidden, out_size, p)        super(Net, self).__init__()        self.drop=nn.Dropout(p=p)        self.linear1=nn.Linear(in_size, n_hidden)        self.linear2=nn.Linear(n_hidden, out_size)    def forward(self, x):        x=torch.relu(self.linear1(x))        x=self.drop(x)        x=self.linear2(x)        return x how would you create a neural network with a dropout parameter of 0.9 1 pointmodel =Net( in_size=10, n_hidden=100, out_size=10, p=0.9)model =Net( in_size=0.9, n_hidden=100, out_size=10, p=10)model =Net( in_size=0.9, n_hidden=0.9, out_size=10, p=10)

Question 2What's wrong with the following function :123456789101112 ]:class Net(nn.Module):    def __init__(self,D_in,H,D_out):        super(Net,self).__init__()        self.linear1=nn.Linear(D_in,H)        self.linear2=nn.Linear(H,D_out)             def forward(self,x):        x=torch.sigmoid(linear1(x))          x=torch.sigmoid(linear2(x))        return x1 pointyou did not call self.linear1(x) and self .linear2(x)nothing

Consider the following neural network model or class:1234567891011class Net(nn.Module):    def __init__(self,D_in,H,D_out):        super(Net,self).__init__()        self.linear1=nn.Linear(D_in,H)        self.linear2=nn.Linear(H,D_out)             def forward(self,x):        x=torch.sigmoid(self.linear1(x))          x=torch.sigmoid(self.linear2(x))        return xHow many hidden neurons does the following neural network object have?1model=Net(1,6,1)

import randomimport torchclass DynamicNet(torch.nn.Module): def __init__(self, D_in, H, D_out): """ 在构造函数中，我们构造了三个nn.Linear实例，它们将在前向传播时被使用。 """ super(DynamicNet, self).__init__() self.input_linear = torch.nn.Linear(D_in, H) self.middle_linear = torch.nn.Linear(H, H) self.output_linear = torch.nn.Linear(H, D_out) def forward(self, x): """ 对于模型的前向传播，我们随机选择0、1、2、3， 并重用了多次计算隐藏层的middle_linear模块。 由于每个前向传播构建一个动态计算图， 我们可以在定义模型的前向传播时使用常规Python控制流运算符，如循环或条件语句。 在这里，我们还看到，在定义计算图形时多次重用同一个模块是完全安全的。 这是Lua Torch的一大改进，因为Lua Torch中每个模块只能使用一次。 """ h_relu = self.input_linear(x).clamp(min=0) for _ in range(random.randint(0, 3)): h_relu = self.middle_linear(h_relu).clamp(min=0) y_pred = self.output_linear(h_relu) return y_pred# N是批大小；D是输入维度# H是隐藏层维度；D_out是输出维度N, D_in, H, D_out = 64, 1000, 100, 10# 产生输入和输出随机张量x = torch.randn(N, D_in)y = torch.randn(N, D_out)# 实例化上面定义的类来构造我们的模型model = DynamicNet(D_in, H, D_out)# 构造我们的损失函数（loss function）和优化器（Optimizer）。# 用平凡的随机梯度下降训练这个奇怪的模型是困难的，所以我们使用了momentum方法。criterion = torch.nn.MSELoss(reduction='sum')optimizer = torch.optim.SGD(model.parameters(), lr=1e-4, momentum=0.9)for t in range(500): # 前向传播：通过向模型传入x计算预测的y。 y_pred = model(x) # 计算并打印损失 loss = criterion(y_pred, y) print(t, loss.item()) # 清零梯度，反向传播，更新权重 optimizer.zero_grad() loss.backward() optimizer.step()

What activation function is used in the following class 123456789class NetRelu(nn.Module):        def __init__(self,D_in,H,D_out):                                     super(NetRelu,self).__init__()                    self.linear1=nn.Linear(D_in,H)                     self.linear2=nn.Linear(H,D_out)                      def forward(self,x):                    x=torch.relu(self.linear1(x)))                      x=self.linear2(x)                 return x 1 pointrelutanh Sigmoid

Consider the forward function , fill out the value for the if statement marked BLANK . 456789321        for (l, linear_transform) in zip(range(L), self.hidden):            if #BLANK                  activation = torch.relu(linear_transform(activation))            else:                activation = linear_transform(activation)        return activation        L=len(self.hidden)    def forward(self, activation):# Section 2: 1 pointl>Ll > L-1l<L-1