| 195 | end_index = min(start_index + self.batch_size, data_size) |
| 196 | |
| 197 | def get_embeddings(self, ): |
| 198 | self._embeddings = {} |
| 199 | if self.order == 'first': |
| 200 | embeddings = self.embedding_dict['first'].get_weights()[0] |
| 201 | elif self.order == 'second': |
| 202 | embeddings = self.embedding_dict['second'].get_weights()[0] |
| 203 | else: |
| 204 | embeddings = np.hstack((self.embedding_dict['first'].get_weights()[ |
| 205 | 0], self.embedding_dict['second'].get_weights()[0])) |
| 206 | idx2node = self.idx2node |
| 207 | for i, embedding in enumerate(embeddings): |
| 208 | self._embeddings[idx2node[i]] = embedding |
| 209 | |
| 210 | return self._embeddings |
| 211 | |
| 212 | def train(self, batch_size=1024, epochs=1, initial_epoch=0, verbose=1, times=1): |
| 213 | self.reset_training_config(batch_size, times) |