多通道求均值

最新推荐文章于 2024-10-18 00:00:00 发布

zhous_

最新推荐文章于 2024-10-18 00:00:00 发布

阅读量117

点赞数

文章标签：均值算法 python 深度学习

本文链接：https://blog.csdn.net/zhous_/article/details/130495031

版权

x=tensor([[[[ 1.1894,  0.3288,  1.2571],
          [ 0.0033, -0.2327, -0.6995],
          [-0.6776,  0.9034,  0.4211]],
         [[-1.9202,  0.9340, -1.3507],
          [ 0.7703,  0.9573,  1.0394],
          [ 0.8564,  0.2168,  0.4774]],
         [[-1.3796,  0.1197,  0.3392],
          [ 1.1925, -1.0463, -0.5019],
          [ 0.5268,  0.2200, -0.9236]]],
        [[[-1.2205,  0.9066,  0.4491],
          [-1.1902,  0.2367,  0.9661],
          [-0.9640,  1.0949,  1.9082]],
         [[-0.4913,  1.2142,  0.0562],
          [ 0.2626, -1.3402, -0.3504],
          [-0.4058,  0.5830,  0.2935]],
         [[ 1.2425, -0.0395,  1.3205],
          [ 0.2212,  0.4599, -0.2039],
          [-2.0330, -0.3581,  0.2684]]]])
 mean = torch.mean(x, dim=(0,2,3), keepdim=True)
mean: tensor([[[[ 0.2600]],
         [[ 0.1001]],
         [[-0.0319]]]])
         
sa = x.sum(dim=(2,3))
ensor([[ 2.4933,  1.9808, -1.4532],
        [ 2.1868, -0.1783,  0.8781]])

s6= x.sum(dim=(0,2,3))
tensor([ 4.6801,  1.8024, -0.5750])