| 161 | class TextEncoderBiGRUCo(nn.Module): |
| 162 | |
| 163 | def __init__(self, word_size, pos_size, hidden_size, output_size): |
| 164 | super(TextEncoderBiGRUCo, self).__init__() |
| 165 | |
| 166 | self.pos_emb = nn.Linear(pos_size, word_size) |
| 167 | self.input_emb = nn.Linear(word_size, hidden_size) |
| 168 | self.gru = nn.GRU(hidden_size, |
| 169 | hidden_size, |
| 170 | batch_first=True, |
| 171 | bidirectional=True) |
| 172 | self.output_net = nn.Sequential( |
| 173 | nn.Linear(hidden_size * 2, hidden_size), nn.LayerNorm(hidden_size), |
| 174 | nn.LeakyReLU(0.2, inplace=True), nn.Linear(hidden_size, |
| 175 | output_size)) |
| 176 | |
| 177 | self.input_emb.apply(init_weight) |
| 178 | self.pos_emb.apply(init_weight) |
| 179 | self.output_net.apply(init_weight) |
| 180 | # self.linear2.apply(init_weight) |
| 181 | # self.batch_size = batch_size |
| 182 | self.hidden_size = hidden_size |
| 183 | self.hidden = nn.Parameter( |
| 184 | torch.randn((2, 1, self.hidden_size), requires_grad=True)) |
| 185 | |
| 186 | # input(batch_size, seq_len, dim) |
| 187 | def forward(self, word_embs, pos_onehot, cap_lens): |