tensorflow中shape,get_shape,reshape,set_shape的对比

shape和get_shape

import tensorflow as tfx1 &＃61; tf.placeholder(tf.int32,shape&＃61;[2,2])
print(tf.shape(x1))
print(x1.get_shape())

输出结果&＃xff1a;

"C:\Program Files\Anaconda3\python.exe" D:/pycharmprogram/tensorflow_learn/shape_learn/shape_get_shape.py
Tensor("Shape:0", shape&＃61;(2,), dtype&＃61;int32)
(2, 2)Process finished with exit code 0

shape返回值是一个tensor,而get_shape返回的是一个tuple.

reshape和set_shape

import tensorflow as tfx1 &＃61; tf.placeholder(tf.int32)
x2&＃61;tf.reshape(x1,[2,2])
print(tf.shape(x1))sess &＃61; tf.Session()
print(sess.run(tf.shape(x2), feed_dict&＃61;{x1:[0,1,2,3]}))
print(sess.run(tf.shape(x2), feed_dict&＃61;{x1: [[0, 1], [2, 3]]}))

reshape生成一个新的shape,x2跟x1是一个不同的tensor

import tensorflow as tfx1 &＃61; tf.placeholder(tf.int32)
x1.set_shape([2,2])
print(tf.shape(x1))sess &＃61; tf.Session()
print(sess.run(tf.shape(x1), feed_dict&＃61;{x1:[0,1,2,3]}))
print(sess.run(tf.shape(x1), feed_dict&＃61;{x1: [[0, 1], [2, 3]]}))

set_shape只是设置placeholder的shape

程序员有偿接单和程序指导加QQ:734564390