pytorch-張量次元size和元素個數numel

1.tensor.size()， 檢視張量的次元資訊

import torch

a = torch.tensor([[ 0.0349, 0.0670, -0.0612, 0.0280, -0.0222, 0.0422],

[-1.6719, 0.1242, -0.6488, 0.3313, -1.3965, -0.0682],

[-1.3419, 0.4485, -0.6589, 0.1420, -0.3260, -0.4795],

[-0.0658, -0.1490, -0.1684, 0.7188, 0.3129, -0.1116],

[-0.2098, -0.2980, 0.1126, 0.9666, -0.0178, 0.1222],

[ 0.1179, -0.4622, -0.2112, 1.1151, 0.1846, 0.4283]])

# 檢視tensor的次元資訊：torch.Size([6, 6])

print(a.size())
           

2.numel()函數，檢視一個張量有多少元素

numel就是"number of elements"的簡寫。numel()可以直接傳回int類型的元素個數。

import torch


a = torch.tensor([[ 0.0349, 0.0670, -0.0612, 0.0280, -0.0222, 0.0422],

[-1.6719, 0.1242, -0.6488, 0.3313, -1.3965, -0.0682],

[-1.3419, 0.4485, -0.6589, 0.1420, -0.3260, -0.4795],

[-0.0658, -0.1490, -0.1684, 0.7188, 0.3129, -0.1116],

[-0.2098, -0.2980, 0.1126, 0.9666, -0.0178, 0.1222],

[ 0.1179, -0.4622, -0.2112, 1.1151, 0.1846, 0.4283]])

b = a.numel()

print(type(b)) # int

print(b) # 36