How to concatenate sentence level features with word level features

How can I concatenate sentence level embeddings with word level embeddings for padded sequences?
embedded_seqs has to be a PackedSequence before I feed it to the RNN.

class Model(nn.Module):
    def __init__(self, vocab_size, word_embed_dim,
                 num_labels, labels_embed_dim,
                 hidd_dim, padding_idx=0):
        super(Model, self).__init__()
        self.w_embedding = nn.Embedding(vocab_size, word_embed_dim, padding_idx=padding_idx)
        self.l_embedding = nn.Embedding(num_labels, labels_embed_dim)
        self.rnn = nn.GRU(word_embed_dim + labels_embed_dim, 
                         hidd_dim, batch_first=True, bidirectional=True)
    def forward(self, input_seqs, input_lengths, input_labels):

        embedded_seqs = self.w_embedding(input_seqs)
        embedded_labels = self.l_embedding(input_labels)

        # How can I concatenate embedded_seqs with embedded_labels?

        packed_seqs = pack_padded_sequence(embedded_seqs, input_lengths, 
        output, h_t = self.rnn(packed_seqs)