Tensorflow：Tensor Tensor("Placeholder:0", shape=(?, 3), dtype=float32) 不是此图的元素

Question

我正在尝试使用 Tensorflow 编写线性回归代码。但我收到一个错误：

> TypeError: Cannot interpret feed_dict key as Tensor: Tensor
> Tensor("Placeholder:0", shape=(?, 3), dtype=float32) is not an element
> of this graph.

也许我把 tf.Graph() 搞砸了，但我试图删除那些部分并尝试重新运行但得到同样的错误。

我的代码：

import numpy as np
import tensorflow as tf

x_data = np.random.randn(2000, 3)
w_real = [0.3, 0.5, 0.1]
b_real = -0.2
noise = np.random.randn(1,2000)*0.1
y_data = np.matmul(w_real, x_data.T) + b_real + noise

NUM_STEPS = 100
g = tf.Graph()

wb = []

with g.as_default():
    x = tf.placeholder(tf.float32, shape=[None,3])
    y_true = tf.placeholder(tf.float32, shape=None)

    with tf.name_scope('inference') as scope:
        w = tf.Variable([[0,0,0]], dtype=tf.float32,name='weights')
        b = tf.Variable(0, dtype=tf.float32, name='bias')
        y_pred = tf.matmul(w,tf.transpose(x)) + b

    with tf.name_scope('loss_function') as scope:
        loss = tf.reduce_mean(tf.square(y_true-y_pred))

    with tf.name_scope('train') as scope:
        learning_rate = 0.5
        optimizer = tf.train.GradientDescentOptimizer(learning_rate)
        train = optimizer.minimize(loss)

init = tf.global_variables_initializer()
with tf.Session() as sess:
    sess.run(init)
    for step in range(NUM_STEPS):
        sess.run(train, {x:x_data, y_true:y_data})
        if step%10 == 0:
            print(step, sess.run([w,b]))
        wb.append(sess.run([w,b]))

    print("Final Weights: ", wb)

我已经查看了Similar Question，但遇到了同样的错误。

Answer 1

每个会话都与一个图表相关联，您正在创建的会话：

with tf.Session() as sess:

适用于默认图表，不适用于g。要修复它，您只需执行以下操作：

with g.as_default(), tf.Session() as sess:

请注意，init 的定义也存在问题，它将是默认图中变量的初始化器 - 为空。改为：

with g.as_default(), tf.Session() as sess:
    init = tf.global_variables_initializer()