def embedding_layer(self):
# embedding layer
with tf.name_scope('embedding'):
glove_embedding = load_word_embedding(
word_index=self.word_index,
file='',
trimmed_filename=glove_embedding_save,
load=True,
dim=300
)
glove_w2v = tf.Variable(glove_embedding, dtype=tf.float32, name='glove_w2v')
sen_inputs_glove = tf.nn.embedding_lookup(glove_w2v, self.x_inputs)
return sen_inputs_glove
def embedding_layer(self):
# embedding layer
with tf.name_scope('embedding'):
glove_embedding = load_word_embedding(
word_index=self.word_index,
file='',
trimmed_filename=glove_embedding_save,
load=True,
dim=300)
glove_w2v = tf.Variable(glove_embedding,
dtype=tf.float32,
name='glove_w2v')
batch_embedding = tf.nn.embedding_lookup(glove_w2v, self.x_inputs)
# sen_inputs_glove = tf.expand_dims(embedded_chars, -1)
return batch_embedding # [b_s, max_len, 300]