tf.nn.embedding_lookup函数的介绍

tf.nn.embedding_lookup函数的用法主要是选取一个张量里面索引对应的元素。tf.nn.embedding_lookup(tensor, id):tensor就是输入张量,id就是张量对应的索引,其他的参数不介绍。id可以是一个数字,或者一维数组,也可以是矩阵

例如:import tensorflow as tf;
import numpy as np;
 
c = np.random.random([10,1])
b = tf.nn.embedding_lookup(c,[[1, 3, 6],
                                                    [2, 4, 7]])
 
with tf.Session() as sess:
    sess.run(tf.initialize_all_variables())
    print(sess.run(b))
    print (c)

c:[[[0.01653263]
  [0.23159859]
  [0.47069613]]

 [[0.53892802]
  [0.84475463]
  [0.26659862]]]

b:[[0.1707615 ]
 [0.01653263]
 [0.53892802]
 [0.23159859]
 [0.84475463]
 [0.71732015]
 [0.47069613]
 [0.26659862]
 [0.59343494]
 [0.83624844]]

猜你喜欢

转载自blog.csdn.net/qq_41424519/article/details/81778000