I am more lazy, intermittently update the common function tf for reference:
一、tf.reduce_sum( )
reduce_sum () is a personal understanding summation function dimensionality reduction, in which tensorflow, Tensor are calculated, the summation may be controlled by adjusting the dimension of the axis dimension.
parameter:
- input_tensor:. To reduce the amount of digital photos should have a type.
- axis:. If you want to reduce the size of None (default), all of the reduced size must be in the range [-rank (input_tensor), rank (input_tensor)) inside.
- keep_dims: If true, the reserved length is reduced in size to 1.
- name: name of the operation (optional).
- reduction_indices: the name of the axis of the abandoned.
return:
This function returns reduce the amount of copies.
numpy compatibility
Equivalent np.sum;
Here is the order tensor axis using the function specified in order to eliminate tensor axis, while all of the lower order elements cumulatively summing operation.
A look at the official example:
x = tf.constant([[1, 1, 1], [1, 1, 1]]) tf.reduce_sum(x) # 6 tf.reduce_sum(x, 0) # [2, 2, 2] tf.reduce_sum(x, 1) # [3, 3] tf.reduce_sum(x, 1, keep_dims=True) # [[3], [3]] tf.reduce_sum(x, [0, 1]) # 6
This function calculates the sum of each dimension element of a tensor.
Input_tensor function in accordance with a given axis has reduced dimensions; unless keep_dims is true, otherwise it will decrease the rank tensor in an axis of each entry; keep_dims If true, the reduced dimension will remain length 1.
If there is no entry axis, all dimensions are reduced, and returns a single element having a tensor.
二、tf.ones_like | tf.zeros_like
tf.ones_like (tensor, dype = None, name = None)
tf.zeros_like (tensor, dype = None, name = None)
Create a consistent with the given tensor tensor type size, all of whose elements are 0 and 1, the exemplary as follows:
tensor=[[1, 2, 3], [4, 5, 6]] x = tf.ones_like(tensor) print(sess.run(x)) #[[1 1 1], # [1 1 1]]