TensorFlow InvalidArgumentError when feeding a placeholder

Question

I have this placeholder:

__y = tf.placeholder(tf.int32)

And then I use it in the following code:

self.session.run(tf.global_variables_initializer())
self.cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=self.__y, logits=self.super_source))
opt =  tf.train.GradientDescentOptimizer(0.1).minimize(self.cost)
not_important, c = self.session.run(opt, feed_dict={labels: label})

And I get this error:

InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'Placeholder' with dtype int32
     [[Node: Placeholder = Placeholder[dtype=DT_INT32, shape=<unknown>, _device="/job:localhost/replica:0/task:0/gpu:0"]()]]

I don't understand the error that I get and so I cant solve my problem. Can someone explain me at least what's happening?

Answer 1

If I rewrite your example as:

import tensorflow as tf

__y = tf.placeholder(tf.int32)

with tf.Session() as session:
    session.run(__y)

I achieve the same error:

InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'Placeholder' with dtype int32
     [[Node: Placeholder = Placeholder[dtype=DT_INT32, shape=<unknown>, _device="/job:localhost/replica:0/task:0/cpu:0"]()]]

The issue here is that as you did not provide an explicit name for '__y' is has gotten the default name 'Placeholder', so it should be fed as follows:

with tf.Session() as session:
   print(session.run(__y, feed_dict={__y: 123}))

Answer 2

问题很愚蠢，我没喂__y：

not_important, c = self.session.run([optimizer, self.cost], feed_dict={self.__labels: label, self.__y: 3.0})