关于tensorflow中报错NotImplementedError的解决_set_input_embeddings notimplementederror

跟着tensorflow入门与实践学习，程序如下：

import tensorflow as tf
tf.compat.v1.disable_eager_execution() #此为V1与V2版本兼容增加，原程序没有这部分
 
weight1 = tf.compat.v1.Variable(0.001)
weight2 = tf.compat.v1.Variable(weight1.initial_value * 2) #weight2 = tf.Variable(weight1.initialized_value() * 2)报错NotImplementedError
init = tf.compat.v1.global_variables_initializer()
with tf.compat.v1.Session() as sess:
  sess.run(init)
  print("weight1 is:")
  print(sess.run(weight1))
  print("weight2 is:")
  print(sess.run(weight2))

需把原程序中的weight1.initialized_value()改为initial_value，就不会再报错了。