ValueError: 目标尺寸 (torch.Size([128])) 必须与输入尺寸 (torch.Size([112])) 相同

Posted 2023-03-12

【中文标题】ValueError: 目标尺寸 (torch.Size([128])) 必须与输入尺寸 (torch.Size([112])) 相同

【英文标题】：ValueError: Target size (torch.Size([128])) must be the same as input size (torch.Size([112]))

【发布时间】：2021-11-12 22:30:32

【问题描述】：

我有一个训练函数，里面有两个向量：

d_labels_a = torch.zeros(128)
d_labels_b = torch.ones(128)

那我有这些功能：

# Compute output
features_a = nets[0](input_a)
features_b = nets[1](input_b)
features_c = nets[2](inputs)

然后域分类器（nets[4]）进行预测：

d_pred_a = torch.squeeze(nets[4](features_a))
d_pred_b = torch.squeeze(nets[4](features_b))
d_pred_a = d_pred_a.float()
d_pred_b = d_pred_b.float()
print(d_pred_a.shape)

损失函数中出现错误：` pred_a = torch.squeeze(nets3) pred_b = torch.squeeze(nets3) pred_c = torch.squeeze(nets3)

        loss = criterion(pred_a, labels_a) + criterion(pred_b, labels_b) + criterion(pred_c, labels) + d_criterion(d_pred_a, d_labels_a) + d_criterion(d_pred_b, d_labels_b)

The problem is that d_pred_a/b is different from d_labels_a/b, but only after a certain point. Indeed, when I print the shape of d_pred_a/b it istorch.Size([128])but then it changes totorch.Size([112])` 独立。

它来自这里：

# Compute output
features_a = nets[0](input_a)
features_b = nets[1](input_b)
features_c = nets[2](inputs)

因为如果我打印 features_a 的形状是 torch.Size([128, 2048]) 但它会变成 torch.Size([112, 2048]) nets[0] 是一个 VGG，像这样：

class VGG16(nn.Module):

  def __init__(self, input_size, batch_norm=False):
    super(VGG16, self).__init__()

    self.in_channels,self.in_width,self.in_height = input_size

    self.block_1 = VGGBlock(self.in_channels,64,batch_norm=batch_norm)
    self.block_2 = VGGBlock(64, 128,batch_norm=batch_norm)
    self.block_3 = VGGBlock(128, 256,batch_norm=batch_norm)
    self.block_4 = VGGBlock(256,512,batch_norm=batch_norm)


  @property
  def input_size(self):
      return self.in_channels,self.in_width,self.in_height

  def forward(self, x):

    x = self.block_1(x)
    x = self.block_2(x)
    x = self.block_3(x)
    x = self.block_4(x)
    # x = self.avgpool(x)
    x = torch.flatten(x,1)

    return x

【参考方案1】：

我解决了。问题是最后一批。我在数据加载器中使用了drop_last=True，它成功了。