K.one_hot & K.sum

one-hot

one_hot(indices, nb_classes)

输入为n维的整数张量，形如(batch_size, dim1, dim2, ... dim(n-1))，输出为(n+1)维的one-hot编码，形如(batch_size, dim1, dim2, ... dim(n-1), nb_classes)

---

sum

sum(x, axis=None, keepdims=False)

在给定轴上计算张量中元素之和

K.sum 的源码

def sum(x, axis=None, keepdims=False):
    """Sum of the values in a tensor, alongside the specified axis.

    # Arguments
        x: A tensor or variable.
        axis: An integer or list of integers in [-rank(x), rank(x)),
            the axes to sum over. If `None` (default), sums over all
            dimensions.
        keepdims: A boolean, whether to keep the dimensions or not.
            If `keepdims` is `False`, the rank of the tensor is reduced
            by 1. If `keepdims` is `True`,
            the reduced dimension is retained with length 1.

    # Returns
        A tensor with sum of `x`.
    """
    return tf.reduce_sum(x, axis, keepdims)

返回的是： tf.reduce_sum(x, axis, keepdims)

---

接下来给出  tf.reduce_sum(x, axis, keepdims) 的实例，了解sum 的用法

例子：

import tensorflow as tf                                                   
                                                                          
x = tf.constant([[1, 1, 1], [1, 1, 1]])                                   
y1=tf.reduce_sum(x)  # 6                                                  
y2=tf.reduce_sum(x, 0)  # [2, 2, 2]                                       
y3=tf.reduce_sum(x, 1)  # [3, 3]                                          
y4=tf.reduce_sum(x, 1, keepdims=True)  # [[3], [3]]                       
y5=tf.reduce_sum(x, [0, 1])  # 6                                          
                                                                          
print("y1 value:",tf.Session().run(y1)," --- y1 shape:",y1.shape)         
print("y2 value:",tf.Session().run(y2)," --- y2 shape:",y2.shape)         
print("y3 value:",tf.Session().run(y3)," --- y3 shape:",y3.shape)         
print("y4 value:",tf.Session().run(y4)," --- y4 shape:",y4.shape)         
print("y5 value:",tf.Session().run(y5)," --- y5 shape:",y5.shape)