参考博客为:https://blog.youkuaiyun.com/xierhacker/article/details/53102355
在初步学习tensorflow的过程中,遇到了tensorflow实例.numpy() 报错
AttributeError: 'Tensor' object has no attribute 'numpy'
在国内网站上似乎没有找到解决办法,后来在StackOverflow中找到了答案:
由于该方法需要启用eager_execution(),也就是需要启用eager 链接在这
所以只需要在代码中加入
tf.enable_eager_execution()
即可。官方API 在这
HelloWord程序参考如下:
from __future__ import print_function,division
import tensorflow as tf
#define the graph
tf.enable_eager_execution()
info = tf.constant("hello world")
a=tf.constant([[1,2],[3,4]])
b=tf.constant([[1,1],[0,1]])
print("info:",info)
print("a:",a)
print("b:",b)
print("type of a:",type(a))
# mul option
c=tf.matmul(a,b)
print("c:",c)
print("c.numpy:\n",c.numpy())
print("type of c.numpy():",type(c.numpy()))
print("\n")
'''
output:
info: tf.Tensor(b'hello world', shape=(), dtype=string)
a: tf.Tensor(
[[1 2]
[3 4]], shape=(2, 2), dtype=int32)
b: tf.Tensor(
[[1 1]
[0 1]], shape=(2, 2), dtype=int32)
type of a: <class 'EagerTensor'>
c: tf.Tensor(
[[1 3]
[3 7]], shape=(2, 2), dtype=int32)
c.numpy:
[[1 3]
[3 7]]
type of c.numpy(): <class 'numpy.ndarray'>
'''
#show the dedvice
print("device:",c.device)
print("dtype:",c.dtype)
print("shape:",type(c.shape))
'''
output:
device: /job:localhost/replica:0/task:0/device:CPU:0
dtype: <dtype: 'int32'>
shape: <class 'tensorflow.python.framework.tensor_shape.TensorShape'>
'''