IndexError: Target 4 is out of bounds

After I started training the model, jumped out of the error.
I didn’t find any way to deal with such an error on Google. Could someone kindly help?
Thank you very much!

The next part is training the model.

def train(net, epochs=10, batch_size=100, lr=0.01):
    
    opt = torch.optim.SGD(net.parameters(), lr=lr, momentum=0.9, weight_decay=1e-4)
    criterion = nn.CrossEntropyLoss()
    
    if(train_on_gpu):
        net.cuda()
     
    for e in range(epochs):
        
        # initialize hidden state
        h = net.init_hidden(batch_size)         
        train_losses = []    
        net.train()
        for batch in iterate_minibatches(X_train, y_train, batch_size):
            x, y = batch

            inputs, targets = torch.from_numpy(x), torch.from_numpy(y)

            if(train_on_gpu):
                    inputs, targets = inputs.cuda(), targets.cuda()

            # Creating new variables for the hidden state, otherwise
            # we'd backprop through the entire training history
            h = tuple([each.data for each in h])
            
            # zero accumulated gradients
            opt.zero_grad()   
            
            # get the output from the model
            output, h = net(inputs, h, batch_size)
            
            loss = criterion(output, targets.long())
            train_losses.append(loss.item())
            loss.backward()
            opt.step()
            
        val_h = net.init_hidden(batch_size)
        val_losses = []
        accuracy=0
        f1score=0
        net.eval()
        with torch.no_grad():
            for batch in iterate_minibatches(X_test, y_test, batch_size):
                x, y = batch     

                inputs, targets = torch.from_numpy(x), torch.from_numpy(y)

                val_h = tuple([each.data for each in val_h])

                if(train_on_gpu):
                    inputs, targets = inputs.cuda(), targets.cuda()
                    
                output, val_h= net(inputs, val_h, batch_size)

                val_loss = criterion(output, targets.long())
                val_losses.append(val_loss.item())

                top_p, top_class = output.topk(1, dim=1)
                equals = top_class == targets.view(*top_class.shape).long()
                accuracy += torch.mean(equals.type(torch.FloatTensor))
                f1score += metrics.f1_score(top_class.cpu(), targets.view(*top_class.shape).long().cpu(), average='weighted')
            
        net.train() # reset to train mode after iterationg through validation data
                
        print("Epoch: {}/{}...".format(e+1, epochs),
        "Train Loss: {:.4f}...".format(np.mean(train_losses)),
        "Val Loss: {:.4f}...".format(np.mean(val_losses)),
        "Val Acc: {:.4f}...".format(accuracy/(len(X_test)//batch_size)),
        "F1-Score: {:.4f}...".format(f1score/(len(X_test)//batch_size)))

train(net)

When I run the code above, the error is shown below

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-16-9c574dc76081> in <module>()
     70         "F1-Score: {:.4f}...".format(f1score/(len(X_test)//batch_size)))
     71 
---> 72 train(net)

4 frames
/usr/local/lib/python3.6/dist-packages/torch/nn/functional.py in nll_loss(input, target, weight, size_average, ignore_index, reduce, reduction)
   2216                          .format(input.size(0), target.size(0)))
   2217     if dim == 2:
-> 2218         ret = torch._C._nn.nll_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
   2219     elif dim == 4:
   2220         ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index)

IndexError: Target 4 is out of bounds.

This is my whole model

NB_SENSOR_CHANNELS = 113
SLIDING_WINDOW_LENGTH = 24
SLIDING_WINDOW_STEP = 12
class HARModel(nn.Module):
    
    def __init__(self, n_hidden=128, n_layers=1, n_filters=64, 
                 n_classes=4, filter_size=5, drop_prob=0.5):
        super(HARModel, self).__init__()
        self.drop_prob = drop_prob
        self.n_layers = n_layers
        self.n_hidden = n_hidden
        self.n_filters = n_filters
        self.n_classes = n_classes
        self.filter_size = filter_size
             
        self.conv1 = nn.Conv1d(NB_SENSOR_CHANNELS, n_filters, filter_size)
        self.conv2 = nn.Conv1d(n_filters, n_filters, filter_size)
        self.conv3 = nn.Conv1d(n_filters, n_filters, filter_size)
        self.conv4 = nn.Conv1d(n_filters, n_filters, filter_size)
        
        self.lstm1  = nn.LSTM(n_filters, n_hidden, n_layers)
        self.lstm2  = nn.LSTM(n_hidden, n_hidden, n_layers)
        
        self.fc = nn.Linear(n_hidden, n_classes)

        self.dropout = nn.Dropout(drop_prob)
    
    def forward(self, x, hidden, batch_size):
        
        x = x.view(-1, NB_SENSOR_CHANNELS, SLIDING_WINDOW_LENGTH)
        x = F.relu(self.conv1(x))
        x = F.relu(self.conv2(x))
        x = F.relu(self.conv3(x))
        x = F.relu(self.conv4(x))
        
        x = x.view(8, -1, self.n_filters)
        x, hidden = self.lstm1(x, hidden)
        x, hidden = self.lstm2(x, hidden)
        
        x = x.contiguous().view(-1, self.n_hidden)
        x = self.dropout(x)
        x = self.fc(x)
        
        out = x.view(batch_size, -1, self.n_classes)[:,-1,:]
        
        return out, hidden
    
    def init_hidden(self, batch_size):
        ''' Initializes hidden state '''
        # Create two new tensors with sizes n_layers x batch_size x n_hidden,
        # initialized to zero, for hidden state and cell state of LSTM
        weight = next(self.parameters()).data
        
        if (train_on_gpu):
            hidden = (weight.new(self.n_layers, batch_size, self.n_hidden).zero_().cuda(),
                  weight.new(self.n_layers, batch_size, self.n_hidden).zero_().cuda())
        else:
            hidden = (weight.new(self.n_layers, batch_size, self.n_hidden).zero_(),
                      weight.new(self.n_layers, batch_size, self.n_hidden).zero_())
        
        return hidden
    
net = HARModel()

def init_weights(m):
    if type(m) == nn.LSTM:
        for name, param in m.named_parameters():
            if 'weight_ih' in name:
                torch.nn.init.orthogonal_(param.data)
            elif 'weight_hh' in name:
                torch.nn.init.orthogonal_(param.data)
            elif 'bias' in name:
                param.data.fill_(0)
    elif type(m) == nn.Conv1d or type(m) == nn.Linear:
        torch.nn.init.orthogonal_(m.weight)
        m.bias.data.fill_(0)
net.apply(init_weights)

def iterate_minibatches(inputs, targets, batchsize, shuffle=True):
    assert len(inputs) == len(targets)
    if shuffle:
        indices = np.arange(len(inputs))
        np.random.shuffle(indices)
    for start_idx in range(0, len(inputs) - batchsize + 1, batchsize):
        if shuffle:
            excerpt = indices[start_idx:start_idx + batchsize]
        else:
            excerpt = slice(start_idx, start_idx + batchsize)
        yield inputs[excerpt], targets[excerpt]

Hi,

if your model outputs 4 classes, CrossEntropyLoss expects your class labels to be in the range [0,C−1], so in this case [0, 3]. Can you make sure your labels are in this range?

3 Likes

Hi, Caruso

I am very appreciative of your suggestion. You are right. I forget one label. Now the model is working.

Again, Thank you for your help!

1 Like