用例对应的源代码，觉得有帮助可以Star
import tensorflow as tf
# tf.reduce_sum(
# input_tensor,
# axis=None,
# keepdims=None,
# name=None,
# reduction_indices=None,
# keep_dims=None
# )
#
# Computes the sum of elements across dimensions of a tensor. (deprecated arguments)
# 1 2 4
# 8 16 32
#
# x has a shape of (2, 3) (two rows and three columns):
#
# The 0 axis in tensorflow is the rows, 1 axis is the columns.
# By doing tf.reduce_sum(x, 0) the tensor is reduced along the first dimension
# (rows), so the result is [1, 2, 4] + [8, 16, 32] = [9, 18, 32].
#
# By doing tf.reduce_sum(x, 1) the tensor is reduced along the second dimension
# (columns), so the result is [1, 9] + [2, 16] + [4, 32] = [7, 56].
x = tf.constant([[1, 2, 4], [8, 16, 32]])
a = tf.reduce_sum(x, 0) # [ 9 18 36]
b = tf.reduce_sum(x, 1) # [ 7 56]
c = tf.reduce_sum(x, [0, 1]) # 63
d = tf.reduce_sum(x, 1, keepdims=True) # [[ 7]
# [56]]
with tf.Session() as sess:
print(sess.run(tf.rank(x)))
output_a = sess.run(a)
print(output_a)
output_b = sess.run(b)
print(output_b)
output_c = sess.run(c)
print(output_c)
output_d = sess.run(d)
print(output_d)
