PyTorch MLP 不训练？持续亏损

Question

我有以下网络来执行一些轨迹的二元分类：

建筑

class DCR(nn.Module):
    def __init__(self, kemb_size, nvar, points, device):
        super().__init__()
        self.phis = load_phis_dataset()
        self.kemb = get_kernel_embedding(self.phis, nvar, samples = kemb_size).to(device) # (concepts, kemb_size)
        _ = self.kemb.requires_grad_()
        
        self.fc1 = nn.Linear(kemb_size + (nvar*points), 64)
        self.fc2 = nn.Linear(64, 1)
        self.sigmoid = nn.Sigmoid()
                
    def forward(self, x):
        # concept truth degrees
        rhos = get_robustness(x, self.phis, time = False) # (trajectories, concepts)
        _ = rhos.requires_grad_()
        # embed trajectories in kernel space
        traj_emb = torch.matmul(rhos, self.kemb) # (trajectories, kemb_size)
        _ = traj_emb.requires_grad_()
        # combine info from traj_embed and x to predict class 
        x_new = x.view(x.size(0), -1) # flatten x
        combined_features = torch.cat((traj_emb, x_new), dim=1) # (trajectories, kemb_size + x.shape[0]*x.shape[1])

        output = self.fc1(combined_features)
        output = F.relu(output)
        output = self.fc2(output)
        output = self.sigmoid(output)
        
        return output.squeeze(1)

培训

model = DCR(kemb_size, nvar, points, device).to(device)
criterion = nn.BCELoss().to(device)
optimizer = optim.SGD(model.parameters(), lr=0.01)

model.train()
for epoch in range(10):
    epoch_loss = 0.0

    for batch, labels in train_loader:
        batch, labels = batch.to(device), labels.to(device)
                 
        y_preds = model(batch)
        loss = criterion(y_preds, labels.float())
        
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
        
        epoch_loss += y_preds.shape[0] * loss.item()
  
    print(f'Epoch: {epoch}, Loss: {epoch_loss/len(y_train):.5f}')

然而，训练损失在每个时期都保持完全恒定，并且权重不会更新。我可能做错了什么？

我尝试用一些

.requires_grad_()

解决问题，但没有成功。

一些解释

• phis = STL 公式列表
• kemb = 所述 STL 公式的内核嵌入
• rhos = STL 公式对输入轨迹的鲁棒性

所有形状均作为代码中的注释。

Answer 1

试试这个：我删除了损失计算中的

y_preds.shape[0] *

for epoch in range(10):
epoch_loss = 0.0

for batch, labels in train_loader:
    batch, labels = batch.to(device), labels.to(device)
    optimizer.zero_grad()         
    y_preds = model(batch)
    loss = criterion(y_preds, labels.float())
    
    loss.backward()
    optimizer.step()
    
    epoch_loss += loss.item()
  
print(f'Epoch: {epoch}, Loss: {epoch_loss/len(y_train):.5f}')