zoukankan      html  css  js  c++  java
  • gather函数

    gather(input, dim, index):根据  index,在  dim  维度上选取数据,输出的  size  与  index  一致

    # input (Tensor) – 源张量

    # dim (int) – 索引的轴

    # index (LongTensor) – 聚合元素的下标(index需要是torch.longTensor类型)

    # out (Tensor, optional) – 目标张量

    for 3D tensor:

                         out[i][j][k] = tensor[index[i][j][k]][j][k]   # dim=0

                         out[i][j][k] = tensor[i][index[i][j][k]][k]   # dim=1

                         out[i][j][k] = tensor[i][j][index[i][j][k]]   # dim=2

    for 2D tensor:

     

                            out[i][j] = input[index[i][j]][j]  # dim = 0

                            out[i][j] = input[i][index[i][j]]  # dim = 1

    import torch as t  # 导入torch模块
    c = t.arange(0, 60).view(3, 4, 5) # 定义tensor
    print(c)
    index = torch.LongTensor([[[0,1,2,0,2],
                    [0,0,0,0,0],
                    [1,1,1,1,1]],
                    [[1,2,2,2,2],
                     [0,0,0,0,0],
                    [2,2,2,2,2]]])
    b = t.gather(c, 0, index)
    print(b)

    输出:

    tensor([[[ 0, 1, 2, 3, 4],
                 [ 5, 6, 7, 8, 9],
                 [10, 11, 12, 13, 14],
                 [15, 16, 17, 18, 19]],

                 [[20, 21, 22, 23, 24],
                 [25, 26, 27, 28, 29],
                 [30, 31, 32, 33, 34],
                 [35, 36, 37, 38, 39]],

                 [[40, 41, 42, 43, 44],
                 [45, 46, 47, 48, 49],
                 [50, 51, 52, 53, 54],
                 [55, 56, 57, 58, 59]]])

    报错:

    Traceback (most recent call last):
    File "E:/Release02/my_torch.py", line 14, in <module>
    b = t.gather(c, 0, index)
    RuntimeError: Size does not match at dimension 1 get 4 vs 3

    (第1维尺寸不匹配)

    将index调整为:

    index = t.LongTensor([[[0, 1, 2, 0, 2], [0, 0, 0, 0, 0], [1, 1, 1, 1, 1], [1, 1, 1, 1, 1]],
    [[1, 2, 2, 2, 2], [0, 0, 0, 0, 0], [2, 2, 2, 2, 2], [1, 1, 1, 1, 1]],
    [[1, 2, 2, 2, 2], [0, 0, 0, 0, 0], [2, 2, 2, 2, 2], [1, 1, 1, 1, 1]]])
    则上文输出为:

    tensor([[[ 0, 21, 42, 3, 44],
                 [ 5, 6, 7, 8, 9],
                 [30, 31, 32, 33, 34],
                 [35, 36, 37, 38, 39]],

                [[20, 41, 42, 43, 44],
                 [ 5, 6, 7, 8, 9],
                 [50, 51, 52, 53, 54],
                 [35, 36, 37, 38, 39]],

                [[20, 41, 42, 43, 44],
                 [ 5, 6, 7, 8, 9],
                 [50, 51, 52, 53, 54],
                 [35, 36, 37, 38, 39]]])

    对于2D tensor 则无“index与tensor 的size一致”之要求,

    这个要求在官方文档和其他博文、日志中均无提到

    可能是个坑吧丨可能是个坑吧丨可能是个坑吧

    eg:

    代码(此部分来自https://www.yzlfxy.com/jiaocheng/python/337618.html):

    b = torch.Tensor([[1,2,3],[4,5,6]])
    print b
    index_1 = torch.LongTensor([[0,1],[2,0]])
    index_2 = torch.LongTensor([[0,1,1],[0,0,0]])
    print torch.gather(b, dim=1, index=index_1)
    print torch.gather(b, dim=0, index=index_2)

    输出:

     1 2 3
     4 5 6
    [torch.FloatTensor of size 2x3]
    
     1 2
     6 4
    [torch.FloatTensor of size 2x2]
    
     1 5 6
     1 2 3
    [torch.FloatTensor of size 2x3]



    官方文档:
    torch.gather(input, dim, index, out=None) → Tensor
    
     Gathers values along an axis specified by dim.
    
     For a 3-D tensor the output is specified by:
    
     out[i][j][k] = input[index[i][j][k]][j][k] # dim=0
     out[i][j][k] = input[i][index[i][j][k]][k] # dim=1
     out[i][j][k] = input[i][j][index[i][j][k]] # dim=2
    
     Parameters: 
    
      input (Tensor)-The source tensor
      dim (int)-The axis along which to index
      index (LongTensor)-The indices of elements to gather
      out (Tensor, optional)-Destination tensor
    
     Example:
    
     >>> t = torch.Tensor([[1,2],[3,4]])
     >>> torch.gather(t, 1, torch.LongTensor([[0,0],[1,0]]))
      1 1
      4 3
     [torch.FloatTensor of size 2x2]

    以上,学习中遇到的问题,记录方便回顾,亦示他人以之勉坑







  • 相关阅读:
    Java关键字:transient,strictfp和volatile简介
    freemarker 数字格式化函数
    使用 BeanCopier 复制对象
    扩展Smack Message
    JavaScript 中2个等号与3个等号的区别
    Eclipse 3.5使用dropins的插件安装方式
    常见的HTTP 状态代码
    使用python操作FTP上传和下载
    Python操作redis
    Ubuntu14.04安装redis和简单配置
  • 原文地址:https://www.cnblogs.com/zftoughe-9/p/12881711.html
Copyright © 2011-2022 走看看