Pytorch中高维度张量理解
创建一个tensor
python
tensor = torch.rand(3,5,3,2)结果如下：
python
```python
tensor([[[[0.3844, 0.9532],
[0.0787, 0.4187],
[0.4144, 0.9552]],
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]],
[[0.0040, 0.1001],
[0.3837, 0.6088],
[0.1752, 0.3184]],
[[0.2762, 0.8417],
[0.5438, 0.4406],
[0.0529, 0.5175]],
[[0.1038, 0.7948],
[0.4991, 0.5155],
[0.4651, 0.8095]]],
[[[0.0377, 0.0249],
[0.2440, 0.8501],
[0.1176, 0.7303]],
[[0.9979, 0.6738],
[0.2486, 0.4152],
[0.5896, 0.8879]],
[[0.3499, 0.6918],
[0.4399, 0.5192],
[0.1783, 0.5962]],
[[0.3021, 0.4297],
[0.9558, 0.0046],
[0.9994, 0.1249]],
[[0.8348, 0.7249],
[0.1525, 0.3867],
[0.8992, 0.6996]]],
[[[0.5918, 0.9135],
[0.8205, 0.5719],
[0.8127, 0.3856]],
[[0.1870, 0.6190],
[0.2991, 0.9424],
[0.5405, 0.4200]],
[[0.9396, 0.8072],
[0.0319, 0.6586],
[0.4849, 0.6193]],
[[0.5268, 0.2794],
[0.7877, 0.9502],
[0.6553, 0.9574]],
[[0.4079, 0.4648],
[0.6375, 0.8829],
[0.6280, 0.1463]]]])现在我想获取
python
tensor[0,0,0,0]获取第一个维度的第0个元素：
python
[[[0.3844, 0.9532],
[0.0787, 0.4187],
[0.4144, 0.9552]],
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]],
[[0.0040, 0.1001],
[0.3837, 0.6088],
[0.1752, 0.3184]],
[[0.2762, 0.8417],
[0.5438, 0.4406],
[0.0529, 0.5175]],
[[0.1038, 0.7948],
[0.4991, 0.5155],
[0.4651, 0.8095]]]获取第二个维度的第0个元素：
python
[[0.3844, 0.9532],
[0.0787, 0.4187],
[0.4144, 0.9552]]获取第三个维度的第0个元素：
python
[0.3844, 0.9532]获取第四个维度的第0个元素：
python
0.3844其他情况
tensor-1
获取第1个维度的最后一个元素：
python
[[[0.5918, 0.9135],
[0.8205, 0.5719],
[0.8127, 0.3856]],
[[0.1870, 0.6190],
[0.2991, 0.9424],
[0.5405, 0.4200]],
[[0.9396, 0.8072],
[0.0319, 0.6586],
[0.4849, 0.6193]],
[[0.5268, 0.2794],
[0.7877, 0.9502],
[0.6553, 0.9574]],
[[0.4079, 0.4648],
[0.6375, 0.8829],
[0.6280, 0.1463]]]tensor0,1
获取第1个维度的第0个元素 ：
python
[[[0.3844, 0.9532],
[0.0787, 0.4187],
[0.4144, 0.9552]],
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]],
[[0.0040, 0.1001],
[0.3837, 0.6088],
[0.1752, 0.3184]],
[[0.2762, 0.8417],
[0.5438, 0.4406],
[0.0529, 0.5175]],
[[0.1038, 0.7948],
[0.4991, 0.5155],
[0.4651, 0.8095]]]第2个维度的第1个元素：
python
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]]tensor:,1,0,1
获取第1个维度的所有元素：
python
[[[0.3844, 0.9532],
[0.0787, 0.4187],
[0.4144, 0.9552]],
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]],
[[0.0040, 0.1001],
[0.3837, 0.6088],
[0.1752, 0.3184]],
[[0.2762, 0.8417],
[0.5438, 0.4406],
[0.0529, 0.5175]],
[[0.1038, 0.7948],
[0.4991, 0.5155],
[0.4651, 0.8095]]],
[[[0.0377, 0.0249],
[0.2440, 0.8501],
[0.1176, 0.7303]],
[[0.9979, 0.6738],
[0.2486, 0.4152],
[0.5896, 0.8879]],
[[0.3499, 0.6918],
[0.4399, 0.5192],
[0.1783, 0.5962]],
[[0.3021, 0.4297],
[0.9558, 0.0046],
[0.9994, 0.1249]],
[[0.8348, 0.7249],
[0.1525, 0.3867],
[0.8992, 0.6996]]],
[[[0.5918, 0.9135],
[0.8205, 0.5719],
[0.8127, 0.3856]],
[[0.1870, 0.6190],
[0.2991, 0.9424],
[0.5405, 0.4200]],
[[0.9396, 0.8072],
[0.0319, 0.6586],
[0.4849, 0.6193]],
[[0.5268, 0.2794],
[0.7877, 0.9502],
[0.6553, 0.9574]],
[[0.4079, 0.4648],
[0.6375, 0.8829],
[0.6280, 0.1463]]]第2个维度的第1个元素：
python
[[0.0713, 0.5281],
[0.0230, 0.8433],
[0.1113, 0.5927]]
[[0.9979, 0.6738],
[0.2486, 0.4152],
[0.5896, 0.8879]]
[[0.1870, 0.6190],
[0.2991, 0.9424],
[0.5405, 0.4200]]第3个维度的第0个元素：
python
[0.0713, 0.5281]
[0.9979, 0.6738]
[0.1870, 0.6190]第4个维度的第1个元素：
python
0.5281
0.6738
0.6190最终结果：
python
tensor([0.5281, 0.6738, 0.6190])