Detail usage of assign assignment for TensorFlow
- 2020-11-30 08:26:15
- OfStack
After TensorFlow changes the value of the variable, it needs to be re-assigned, and assign is a little tricky to use, but you need to get an operand, run 1.
You can't use it this way
import tensorflow as tf
import numpy as np
x = tf.Variable(0)
init = tf.initialize_all_variables()
sess = tf.InteractiveSession()
sess.run(init)
print(x.eval())
x.assign(1)
print(x.eval())
Proper use
1.
import tensorflow as tf
x = tf.Variable(0)
y = tf.assign(x, 1)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
print sess.run(x)
print sess.run(y)
print sess.run(x)
2.
In [212]: w = tf.Variable(12)
In [213]: w_new = w.assign(34)
In [214]: with tf.Session() as sess:
...: sess.run(w_new)
...: print(w_new.eval())
# output
34
3.
import tensorflow as tf
x = tf.Variable(0)
sess = tf.Session()
sess.run(tf.global_variables_initializer())
print(sess.run(x)) # Prints 0.
x.load(1, sess)
print(sess.run(x)) # Prints 1.
My method
import numpy as np # This is a Python the 1 A very powerful open source extension of numerical computing
import tensorflow as tf # The import tensorflow
## Structural data ##
x_data=np.random.rand(100).astype(np.float32) # Randomly generated 100 A type of float32 The value of the
y_data=x_data*0.1+0.3 # Defining equation y=x_data*A+B
##-------##
## To establish TensorFlow Neural computational structure ##
weight=tf.Variable(tf.random_uniform([1],-1.0,1.0))
biases=tf.Variable(tf.zeros([1]))
y=weight*x_data+biases
w1=weight*2
loss=tf.reduce_mean(tf.square(y-y_data)) # The difference between the judgment and the correct value
optimizer=tf.train.GradientDescentOptimizer(0.5) # The parameters are corrected by back propagation according to the gap
train=optimizer.minimize(loss) # Set up a trainer
init=tf.global_variables_initializer() # Initialize the TensorFlow Training structure
#sess=tf.Session() # To establish TensorFlow The training session
sess = tf.InteractiveSession()
sess.run(init) # Load the training structure into the session
print('weight',weight.eval())
for step in range(400): # Circuit training 400 time
sess.run(train) # Use a trainer to train according to the training structure
if step%20==0: # every 20 Time to print 1 Secondary training results
print(step,sess.run(weight),sess.run(biases)) # Training times, A Value, B value
print(sess.run(loss))
print('weight new',weight.eval())
#wop=weight.assign([3])
#wop.eval()
weight.load([1],sess)
print('w1',w1.eval())