Question

我正在尝试将嵌入层与其他功能串联在一起。它不会给我任何错误，但也不会进行任何培训。这个模型定义有什么问题，如何调试？

注：X的最后一列（功能）是word2ix（单个单词）的功能。注意：如果没有嵌入功能/图层，网络工作正常

最初发布在pytorch forum上


class Net(torch.nn.Module):
    def __init__(self, n_features, h_sizes, num_words, embed_dim, out_size, dropout=None):
        super().__init__()


        self.num_layers = len(h_sizes)  # hidden + input


        self.embedding = torch.nn.Embedding(num_words, embed_dim)
        self.hidden = torch.nn.ModuleList()
        self.bnorm = torch.nn.ModuleList()
        if dropout is not None:
            self.dropout = torch.nn.ModuleList()
        else:
            self.dropout = None
        for k in range(len(h_sizes)):
            if k == 0:
                self.hidden.append(torch.nn.Linear(n_features, h_sizes[0]))
                self.bnorm.append(torch.nn.BatchNorm1d(h_sizes[0]))
                if self.dropout is not None:
                    self.dropout.append(torch.nn.Dropout(p=dropout))

            else:
                if k == 1:
                    input_dim = h_sizes[0] + embed_dim
                else:
                    input_dim = h_sizes[k-1]

                self.hidden.append(torch.nn.Linear(input_dim, h_sizes[k]))
                self.bnorm.append(torch.nn.BatchNorm1d(h_sizes[k]))
                if self.dropout is not None:
                    self.dropout.append(torch.nn.Dropout(p=dropout))

        # Output layer
        self.out = torch.nn.Linear(h_sizes[-1], out_size)

    def forward(self, inputs):

        # Feedforward

        for l in range(self.num_layers):
            if l == 0:
                x = self.hidden[l](inputs[:, :-1])
                x = self.bnorm[l](x)
                if self.dropout is not None:
                    x= self.dropout[l](x)

                embeds = self.embedding(inputs[:,-1])#.view((1, -1)
                x = torch.cat((embeds, x),dim=1)

            else:
                x = self.hidden[l](x)
                x = self.bnorm[l](x)
                if self.dropout is not None:
                    x = self.dropout[l](x)
            x = F.relu(x)
        output= self.out(x)

        return output

Answer 1

有一些问题。关键之一是数据类型。我混合了float特征和int索引。

修复前的样本数据和培训：

NUM_TARGETS = 4
NUM_FEATURES = 3
NUM_TEXT_FEATURES = 1

x = np.random.rand(5, NUM_FEATURES)
y = np.random.rand(5, NUM_TARGETS)

word_ix = np.arange(5).reshape(-1,1).astype(int)
x_train = np.append(x, word_ix, axis=1)

x_train = torch.from_numpy(x).float().to(device)
y_train = torch.from_numpy(y).float().to(device)

h_sizes = [2,2]

net = Net(x_train.shape[1] , h_sizes=h_sizes, num_words=5, embed_dim=2, out_size=y_train.shape[1],dropout=.01)     # define the network
print(net)  # net architecture
net = net.float()
net.to(device)

optimizer = torch.optim.Adam(net.parameters(), lr=0.0001, weight_decay=.01)
loss_func = torch.nn.MSELoss()  # this is for regression mean squared loss

# one training loop
prediction = net(x_train)     # input x and predict based on x

loss = loss_func(prediction, y_train)     # must be (1. nn output, 2. target)

optimizer.zero_grad()   # clear gradients for next train
loss.backward()         # backpropagation, compute gradients
optimizer.step()        # apply gradients       
# train_losses.append(loss.detach().to('cpu').numpy())

要解决此问题，我将单词索引功能与x分开，并删除了net.float()。

将dtypes转换更改为：

x_train = torch.from_numpy(x).float().to(device)
y_train = torch.from_numpy(y).float().to(device) 

# NOTE: word index needs to be long
word_ix = torch.from_numpy(word_ix).to(torch.long).to(device)

和forward的方法已更改为：


    def forward(self, inputs, word_ix):

        # Feedforward

        for l in range(self.num_layers):
            if l == 0:
                x = self.hidden[l](inputs)
                x = self.bnorm[l](x)
                if self.dropout is not None:
                    x = self.dropout[l](x)

                embeds = self.embedding(word_ix)
                # NOTE:
                # embeds has a shape of (batch_size, 1, embed_dim)
                # inorder to merge this change this with x, reshape this to
                # (batch_size, embed_dim)
                embeds = embeds.view(embeds.shape[0], embeds.shape[2])
                x = torch.cat((x, embeds.view(x.shape)),dim=1)

            else:
                x = self.hidden[l](x)
                x = self.bnorm[l](x)
                if self.dropout is not None:
                    x = self.dropout[l](x)
            x = F.relu(x)
        output= self.out(x)

        return output

如何在pytorch中串联嵌入层

1 个答案: