14_pytorch.where,pytorch.gather

我会带着你远行 2022-10-26 08:09 227阅读 0赞

## 1.14.高级OP ##

### 1.14.1.where ###

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70]  
给定一个条件**cond**，满足条件的**取x对应位置元素**，不满足的取**y对应元素**。

**代码示例：**

# -*- coding: UTF-8 -*-
    
    import torch
    
    a = torch.randn(3, 5)
    b = torch.ones(3, 5)
    
    print(a)
    """ 输出结果： tensor([[ 1.4446, 0.4373, 0.6730, 0.1549, -1.0664], [ 0.4106, 0.3604, 1.3719, -0.9672, 0.7382], [ 0.2602, -0.3008, 0.8865, -0.2711, 0.1028]]) """
    print(b)
    """ 输出结果： tensor([[1., 1., 1., 1., 1.], [1., 1., 1., 1., 1.], [1., 1., 1., 1., 1.]]) """
    
    print(torch.where(a > 0, a, b))
    """ 输出结果： tensor([[1.4446, 0.4373, 0.6730, 0.1549, 1.0000], [0.4106, 0.3604, 1.3719, 1.0000, 0.7382], [0.2602, 1.0000, 0.8865, 1.0000, 0.1028]]) """

### 1.14.2.gather ###

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70 1]**举例**  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70 2]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70]: /images/20221024/f55be578b4a94c72b92df4f8b41e2d9e.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70 1]: /images/20221024/6a338c351afa430e8a717d120439c93a.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3RvdG8xMjk3NDg4NTA0_size_16_color_FFFFFF_t_70 2]: /images/20221024/12af36aa7a9e409fa3359a944cce8804.png