Python 构建第一个轴形状为“的条件图”；无”；在张量流中

在图形构建阶段，假设张量

x

，这是一个神经网络的完全连接层

因此假设

x

的形状是

（？，5）

。我想在python中设置最后一列，如下所示：

for i in range(x.shape[0]):
    if x[i,-1] < 0.5:
        x[i,-1] = 0.0
    else:
        x[i,-1] = 1.0

当

x

的第一个维度是

？

？

这样时，我如何检查上面的内容

import tensorflow as tf
import numpy as np

a = tf.placeholder(tf.int32, shape=[None, 5])

r, c = a.get_shape()
x_split = tf.split(1, c, a) # split a along axis 1
last_col = x_split[-1]


mask = tf.greater(last_col, tf.constant(6))
cond = tf.where(mask,
                tf.add(last_col, tf.constant(1)), # if true, add one
                tf.add(last_col, tf.constant(-1)))# if false, minus one

x_split = x_split[0:-1]
x_split.append(cond)
ans = tf.concat(1, x_split)


with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    arr = np.array([[1, 2, 3, 4, 5],
                    [6, 7, 8, 9, 10]])
    the_ans = sess.run(ans, feed_dict={a: arr})
'''
the_ans is 
[[ 1  2  3  4  4]
 [ 6  7  8  9 11]]
'''

像这样

import tensorflow as tf
import numpy as np

a = tf.placeholder(tf.int32, shape=[None, 5])

r, c = a.get_shape()
x_split = tf.split(1, c, a) # split a along axis 1
last_col = x_split[-1]


mask = tf.greater(last_col, tf.constant(6))
cond = tf.where(mask,
                tf.add(last_col, tf.constant(1)), # if true, add one
                tf.add(last_col, tf.constant(-1)))# if false, minus one

x_split = x_split[0:-1]
x_split.append(cond)
ans = tf.concat(1, x_split)


with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    arr = np.array([[1, 2, 3, 4, 5],
                    [6, 7, 8, 9, 10]])
    the_ans = sess.run(ans, feed_dict={a: arr})
'''
the_ans is 
[[ 1  2  3  4  4]
 [ 6  7  8  9 11]]
'''

---

非常感谢你的回答，这是我想要的：-）非常感谢你的回答，这是我想要的：-）