深度学习python代码提问

Question

cnn代码报错提问
报错信息如下：

/root/autodl-tmp/project/twst2.py:108: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
  train_data = torch.tensor(train_data, device=device, dtype=torch.float32)
Traceback (most recent call last):
  File "/root/autodl-tmp/project/twst2.py", line 279, in <module>
    loss, r2, MAE = train(model, optimizer, criterion, batch_data, batch_label)
  File "/root/autodl-tmp/project/twst2.py", line 121, in train
    SSR = ((output - train_label) ** 2).sum().clone().detach().numpy()
  File "/root/miniconda3/lib/python3.8/site-packages/torch/tensor.py", line 630, in __array__
    return self.numpy()
RuntimeError: Can't call numpy() on Tensor that requires grad. Use tensor.detach().numpy() instead.

代码如下：

def train(model, optimizer, criterion, train_data, train_label):
    device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
    train_data = torch.tensor(train_data).to(device)
    train_label = train_label.astype(np.float32)
    train_label = torch.tensor(train_label).to(device)
    train_label,train_data = train_label.to(device), train_data.to(device)
    model.train()
    optimizer.zero_grad()
    output = model(train_data)
    output = output.squeeze()
    train_label = train_label.cpu().numpy().astype(float)
    train_label_tensor = torch.tensor(train_label, dtype=torch.float32).to(device)
    #train_label = train_label.float()xiugai
    mean_train_label = torch.mean(train_label_tensor)
    SST = ((train_label_tensor - mean_train_label) ** 2).sum()
    SSR = ((output - train_label) ** 2).sum()..clone().detach().numpy()
    SSR = SSR.detach().numpy()
    r2 = 1 - SSR / SST
    MAE = torch.mean(torch.abs(train_label - output))
    loss = criterion(output, train_label)
   # loss = sum(loss_num)/len(loss_num)
    loss.backward()
    optimizer.step()
    loss_num = torch.sqrt(loss).detach().numpy
    # r2 = torch.tensor(r2)
    return loss_num, r2, MAE

Answer 1

【以下回答由 GPT 生成】

问题已解决，需要将以下行代码：
SSR = ((output - train_label) ** 2).sum().clone().detach().numpy()
修改为：
SSR = ((output - train_label) ** 2).sum().clone().detach().numpy()

【相关推荐】

• 这有个类似的问题, 你可以参考下: https://ask.csdn.net/questions/7427258
• 我还给你找了一篇非常好的博客，你可以看看是否有帮助，链接：【深度学习】Python实现CNN操作（附代码）
• 您还可以看一下 王而川老师的深度学习之卷积神经网络CNN以及python案例图片中文本的识别课程中的 手写文本识别案例一小节, 巩固相关知识点
  
• 除此之外, 这篇博客: CNN的Python实现——第二章：线性分类器中的 2.4.4 梯度下降法代码实现 部分也许能够解决你的问题。

如果你已经解决了该问题, 非常希望你能够分享一下解决方案, 写成博客, 将相关链接放在评论区, 以帮助更多的人 ^-^