将 tensorflow 与 elu 一起使用时出现属性错误答案

【问题标题】：Attribute error when using tensorflow with elu将 tensorflow 与 elu 一起使用时出现属性错误
【发布时间】：2020-03-01 00:12:05
【问题描述】：

尝试使用 tensorflow 库时出现属性错误输出

TF 版本：

Name: tensorflow
Version: 2.0.0
Summary: TensorFlow is an open source machine learning framework for everyone.
Home-page: https://www.tensorflow.org/
Author: Google Inc.
Author-email: packages@tensorflow.org
License: Apache 2.0
Location: c:\users\vinhalivinhali\appdata\local\programs\python\python35\lib\site-packages
Requires: tensorboard, wrapt, keras-preprocessing, protobuf, wheel, numpy, gast, tensorflow-estimator, termcolor, google-pasta, grpcio, absl-py, keras-applications, astor, opt-einsum, six
Required-by:

Python 版本

Python 3.5.0

进口

import tensorflow as tf # for Deep Learning
import pandas as pd # for data manipulation
import numpy as np # for matrix manipulation
from matplotlib import pyplot as plt # for graphics

脚本

X0 = X[:, 0, :] # get the data in the first time period
Ht = tf.elu(tf.matmul(X0, Wx_h) + b) # uses X0 to initiate the hidden state
y = [] # list to be populated with outputs every time period

输出

Traceback (most recent call last):
  File "rnrs.py", line 40, in <module>
    Ht = tf.elu(tf.matmul(X0, Wx_h) + b) # uses X0 to initiate the hidden state
AttributeError: module 'tensorflow' has no attribute 'elu'

【问题讨论】：

改用tf.nn.elu
回溯（最近一次调用最后）：文件“rnrs.py”，第 40 行，在 Ht = tf.nn.elu(tf.matmul(X0, Wx_h) + b) NameError : 名称“Wx_h”未定义
有以下出口@MihailBurduja
你没有定义Wx_h。您应该将其声明为 Wx_h = tf.Variable(tf.random.normal((200, X0.shape[0]))) 和 b 变量。我建议你考虑使用 TF Keras，因为它更简单，层次更高

标签： tensorflow python-3.5

【解决方案1】：

elu 函数在tf.nn 模块中，参见documentation。

【讨论】：

我测试过但得到以下输出： Traceback (last recent call last): File "rnrs.py", line 40, in Ht = tf.nn.elu(tf.matmul( X0, Wx_h) + b) NameError: name 'Wx_h' is not defined