Deep learning DCgAN只产生噪声

我是GANs的新手，在mnist上训练DCGAN时遇到了麻烦，当我使用GAN线性层时，一切都很好，生成器生成了非常好的图像。但当我开始研究卷积GAN时，发电机只产生噪音。有人知道如何解决这个问题吗

这是我的神经网络和训练循环

神经网络：

鉴别器 发电机 训练循环：

loss_G, loss_D = [],[]
for i in range(epochs):
    D.train()
    G.train()
    st = time.time()
    for idx, (img, _ ) in enumerate(mnist):
        img = img.to(device)
        ## Discriminator ##
        D.zero_grad(set_to_none=True)
        lable = torch.ones(bs,device=device)*0.9
        pred = D(img).reshape(-1)
        loss_d_real = criterion(pred,lable)

        z = torch.randn(img.shape[0],ch_z, 1,1,device=device)
        fake_img = G(z)
        lable = torch.ones(bs,device=device)*0.1
        pred = D(fake_img.detach()).reshape(-1)
        loss_d_fake = criterion(pred,lable)
        D_loss = loss_d_real + loss_d_fake
        D_loss.backward()
        optim_d.step()
        ## Generator ##
        G.zero_grad(True)
        lable = torch.randn(bs,device=device)
        pred = D(fake_img).reshape(-1)
        G_loss = criterion(pred,lable)
        G_loss.backward()
        optim_g.step()
        ## printing on terminal
        if idx % 100 == 0:
            print(f'\nBatches done : {idx}/{len(mnist)}')
            print(f'Loss_D : {D_loss.item():.4f}\tLoss_G : {G_loss.item():.4f}')
    et = time.time()
    print(f'\nEpoch : {i+1}\n{time_cal(st,et)}')
    G.eval()
    with torch.no_grad():
        fake_image = G(fixed_noise)
        save_image(fake_image[:25],fp=f'{path_to_img}/{i+1}_fake.png',nrow=5,normalize=True)

    loss_G.append(G_loss.item())
    loss_D.append(D_loss.item())

class Generator(nn.Module):
    def __init__(self, ch_noise, ch_img, features_g):
        super(Generator, self).__init__()

        def block(in_ch,out_ch,k,s,p,final=False):
            block = []
            block.append(nn.ConvTranspose2d(in_ch,out_ch,kernel_size=k,stride=s,padding=p))
            if not final:
                block.append(nn.BatchNorm2d(out_ch))
                block.append(nn.ReLU())
            if final:
                block.append(nn.Tanh())
            return block

        self.G = nn.Sequential(
            *block(ch_noise, features_g*16, k=4,s=1,p=0),
            *block(features_g*16, features_g*8, k=4, s=2, p=1),
            *block(features_g*8, features_g*4, k=4, s=2, p=1),
            *block(features_g*4, features_g*2, k=4, s=2, p=1),
            *block(features_g*2, 1, k=4, s=2, p=1,final=True)
        )
    def forward(self,z): return self.G(z)

loss_G, loss_D = [],[]
for i in range(epochs):
    D.train()
    G.train()
    st = time.time()
    for idx, (img, _ ) in enumerate(mnist):
        img = img.to(device)
        ## Discriminator ##
        D.zero_grad(set_to_none=True)
        lable = torch.ones(bs,device=device)*0.9
        pred = D(img).reshape(-1)
        loss_d_real = criterion(pred,lable)

        z = torch.randn(img.shape[0],ch_z, 1,1,device=device)
        fake_img = G(z)
        lable = torch.ones(bs,device=device)*0.1
        pred = D(fake_img.detach()).reshape(-1)
        loss_d_fake = criterion(pred,lable)
        D_loss = loss_d_real + loss_d_fake
        D_loss.backward()
        optim_d.step()
        ## Generator ##
        G.zero_grad(True)
        lable = torch.randn(bs,device=device)
        pred = D(fake_img).reshape(-1)
        G_loss = criterion(pred,lable)
        G_loss.backward()
        optim_g.step()
        ## printing on terminal
        if idx % 100 == 0:
            print(f'\nBatches done : {idx}/{len(mnist)}')
            print(f'Loss_D : {D_loss.item():.4f}\tLoss_G : {G_loss.item():.4f}')
    et = time.time()
    print(f'\nEpoch : {i+1}\n{time_cal(st,et)}')
    G.eval()
    with torch.no_grad():
        fake_image = G(fixed_noise)
        save_image(fake_image[:25],fp=f'{path_to_img}/{i+1}_fake.png',nrow=5,normalize=True)

    loss_G.append(G_loss.item())
    loss_D.append(D_loss.item())