Tensorflow 有问题 tf.reduce_sum TypeError: Expected binary or unicode string, got 0.026435676566314795

Question

由于代码太长，无法粘贴整个代码。 X 有 9 列，rest 各有 1 列。 过去工作，更新了tf版本，现在不工作。

X = X.values
Y = Z.values
W = W.values
B = B.values

learning_rate = 0.01
epoch = 300 
samples = 150

x = tf.placeholder('float32',[None,9]) 
y = tf.placeholder('float32',[None,1])
w = tf.placeholder('float32',[None,1])
b = tf.placeholder('float32',[None,1])

prediction = W*X+B

cost = tf.reduce_sum((prediction - Y)**2)/(2*samples)

optimizer = tf.train.GradientDescentOptomizer(learning_rate).minimize(cost)
print('optimizer in:'+optimizer)
with tf.Session() as sess:
    for X,Y in zip(x_train,y_train):
        sess.run(optimizer,feed_dict={x:X,y:Y})
        weight = sess.run(W)
        print(weight)
        bias = sess.run(B)
        print(bias)

Following is the error==============
TypeError                                 Traceback (most recent call last)
<ipython-input-15-68f4f3d18cbe> in <module>()
---> 75 cost = tf.reduce_sum((prediction - Y)**2)/(2*samples)
     76 # define optimizer
     77 optimizer = tf.train.GradientDescentOptomizer(learning_rate).minimize(cost)``
TypeError: Expected binary or unicode string, got 0.026435676566314795