日志：2021_3_4 tf.nn.embedding_lookup

tf.nn.embedding_lookup

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。

example:

import tensorflow as tf

a=[1,2,3,4,5,6,7,8,9,10]
# 用数组a生成张量b
b = tf.Variable(a)  
#查找张量b中的第二个和第四个元素
c = tf.nn.embedding_lookup(b, [1, 3])  

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(b))
    print(sess.run(c))

output:

[ 1  2  3  4  5  6  7  8  9 10]
[2 4]

如果张量不是一维的,如下，创建一个二维数组，用tf.nn.embedding_lookup查找。

a = [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10, 11, 12], [13, 14, 15]]
a = np.asarray(a) #将列表转换为数组，这里a变成一个二维数组
a= tf.Variable(a)
array1 = tf.Variable([0, 2, 3, 1], tf.int32)
array2 = tf.Variable([[0, 2, 3, 1], [4, 0, 2, 2]], tf.int32)
out1 = tf.nn.embedding_lookup(a, array1)
out2 = tf.nn.embedding_lookup(a, array2)
init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    print (sess.run(out1))
    print (out1.shape)
    print ('==================')
    print (sess.run(out2))
    print (out2.shape)

输出如下：out1查找的是张量a中的索引为0，2，3，1（array1）的元素。out1是一个4×3的张量矩阵。out查找的是张量a中的索引为[0，2，3，1]（array1）,[4, 0, 2, 2]（array2）的元素。out2是一个2×4×3的三维张量矩阵。

[[ 1  2  3]
 [ 7  8  9]
 [10 11 12]
 [ 4  5  6]]
(4, 3)
==================
[[[ 1  2  3]
  [ 7  8  9]
  [10 11 12]
  [ 4  5  6]]

 [[13 14 15]
  [ 1  2  3]
  [ 7  8  9]
  [ 7  8  9]]]
(2, 4, 3)

Process finished with exit code 0