如何在模型中引入可学习参数（Pytorch）

错误实例：

def init(self):
	self.w1 = torch.nn.Parameter(torch.FloatTensor(1),requires_grad=True).cuda()
	self.w2 = torch.nn.Parameter(torch.FloatTensor(1),requires_grad=True).cuda()
	self.w1.data.fill_(0.3)
	self.w2.data.fill_(0.3)
def forward(self, x):
	out = self.w1 * out1 + self.w2 * out2
	out = self.fc(out)

正确实例：

def init(self):
	self.w1 = torch.nn.Parameter(torch.FloatTensor(1),requires_grad=True)
	self.w2 = torch.nn.Parameter(torch.FloatTensor(1),requires_grad=True)
	self.w1.data.fill_(0.3)
	self.w2.data.fill_(0.3)
def forward(self, x):
	out = self.w1 * out1 + self.w2 * out2
	out = self.fc(out)