Several simple examples showing the usage of scatter_nd_sub

TensorFlow中scatter_nd_sub用法示例

最新推荐文章于 2022-01-03 18:08:02 发布

原创最新推荐文章于 2022-01-03 18:08:02 发布 · 593 阅读

0 ·

CC 4.0 BY-SA版权

Tensorflow 专栏收录该内容

2 篇文章

订阅专栏

本文通过两个具体实例介绍了如何使用TensorFlow中的scatter_nd_sub操作进行张量元素的减法更新。第一个例子展示了如何在第二维度上进行向量的减法操作；第二个例子则演示了如何在第三维度上的特定位置进行单个元素的减法更新。

部署运行你感兴趣的模型镜像

scatter_nd_sub

Example 1: subtraction at the second dimension, [0,1], which corresponding to a vector.

>>> ref = tf.Variable(tf.ones([2,3,3],tf.int32))        
>>> indices = tf.constant([[0,1]])
>>> updates = tf.constant([[1,1,1]])    
>>> init = tf.global_variables_initializer()
>>> sess.run(init)
>>> print(ref.eval())
[[[1 1 1]
  [1 1 1]
  [1 1 1]]

 [[1 1 1]
  [1 1 1]
  [1 1 1]]]
>>> print(updates.eval())
[[1 1 1]]
>>> update = tf.scatter_nd_sub(ref,indices,updates)
>>> sess.run(update)
array([[[1, 1, 1],
    [0, 0, 0],
    [1, 1, 1]],

   [[1, 1, 1],
    [1, 1, 1],
    [1, 1, 1]]], dtype=int32)

Example 2: subtraction at the third dimension, [0,1,2], which points to a single element.

>>> ref = tf.Variable(tf.ones([2,3,3],tf.int32))
>>> indices = tf.constant([[0,1,2]])
>>> updates = tf.constant([1])
>>> init = tf.global_variables_initializer()
>>> sess.run(init)
>>> print(ref.eval())
[[[1 1 1]
  [1 1 1]
  [1 1 1]]

 [[1 1 1]
  [1 1 1]
  [1 1 1]]]
>>> print(updates.eval())
[1]
>>> update = tf.scatter_nd_sub(ref,indices,updates)
>>> sess.run(update)
array([[[1, 1, 1],
    [1, 1, 0],
    [1, 1, 1]],

   [[1, 1, 1],
    [1, 1, 1],
    [1, 1, 1]]], dtype=int32)

您可能感兴趣的与本文相关的镜像