【Tensorflow】报错：Cannot interpret feed_dict key as Tensor: The name 'x' refers to an operation, # > no

问题描述：

我尝试给一个tensor输入值的时候报错：

 Cannot interpret feed_dict key as Tensor: The name 'x' refers to an operation,
not a Tensor. Tensor names must be of the form "<op_name>:<output_index>".

代码如下：

import tensorflow as tf
x = tf.placeholder(tf.float32, (None,), 'x')
y = tf.reduce_sum(x)
sess = tf.Session()

sess.run(y, {x: [1, 2, 3]}
# > 6.0

sess.run(y, {'x': [1, 2, 3]}
# > Cannot interpret feed_dict key as Tensor: The name 'x' refers to an operation,
# > not a Tensor. Tensor names must be of the form "<op_name>:<output_index>".

sess.run(y, {tf.get_default_graph().get_operation_by_name('x').outputs[0]: [1, 2, 3]})
# > 6.0

有没有可能feed placeholders以他们的变量名呢？如果不能，为什么？这对随后从磁盘恢复网络后进行feed很有用。

问题解决：

你需要添加 ":0"，例如：

print sess.run(y, {'x:0': [1, 2, 3]})

这里是为什么 ":0"要被加上的原因：http://stackoverflow.com/a/37870634/419116
.