nn.AdaptiveAvgPool2d
a = nn.AdaptiveAvgPool2d((5, 7))
output = a(input)
print('5x7:',output)
# 目标输出大小为 7x7（正方形）
m = nn.AdaptiveAvgPool2d(7)
output = m(input)
print('7x7:',output)
5x7: tensor([[[[ 0.3847, 0.6597, 0.0249, 0.3430, 0.2783, -0.3439, -1.1087],
[-0.0574, 0.7386, 0.6476, 0.3687, -0.1651, -0.3378, -0.3508],
[-0.3331, 0.3783, 0.4337, -0.2494, -0.6286, -0.0352, 0.0131],
[-0.5599, -0.1819, -0.3177, -0.5623, 0.0833, 0.1311, -0.4833],
[ 0.0169, 0.3534, -0.0434, -0.3368, 0.3731, -0.4099, -0.7534]]]])
7x7: tensor([[[[ 0.3847, 0.6597, 0.0249, 0.3430, 0.2783, -0.3439, -1.1087],
[-0.0040, 0.5401, 0.3237, 0.6172, 0.1927, -0.4535, -0.6810],
[-0.1919, 0.8235, 0.7367, 0.1908, -0.6345, -0.5306, -0.1679],
[-0.3331, 0.3783, 0.4337, -0.2494, -0.6286, -0.0352, 0.0131],
[-0.7913, -0.7301, -0.6630, -0.4539, 0.1089, 0.5917, -0.2622],
[-0.5888, -0.0834, -0.2626, -0.6581, 0.3132, 0.1788, -0.5833],
[ 0.0169, 0.3534, -0.0434, -0.3368, 0.3731, -0.4099, -0.7534]]]])
770
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
