torch.nn.functional.pad函数

函数原型位于torch.nn.functional.py

def pad(input, pad, mode='constant', value=0): 
    
    Args:
        input (Tensor):4维或者5维的tensor 
        pad (tuple): 不同tensor的填充方式
           1.四维tensor,传入4元素tuple(pad_l,pad_r,pad_t,pad_b),指的是（左填充，
             右填充，上填充，下填充），其数值代表填充次数
           2.六维tensor,传入6元素tuple(pleft,pright,ptop,pbottom,pfront,pback)
        mode:'constant常量', 'reflect反射`, `replicate复制` or ``'circular'``.
            Default: ``'constant'``
        value: 填充的数值，在'constant'模式下，默认填充`0``   

例子

import torch
import torch.nn.functional as F
original_values = torch.randn([1,1, 3, 2])#生成一个[1,1,3,2]的tensor
print("original_values: ",original_values,"\n")
#tensor([[[[-0.6208, -1.4084],
          # [-0.2202,  0.1708],
          # [ 0.9050, -1.2441]]]])
print("original_values的shape: ",original_values.shape)#torch.Size([1, 1, 3, 2])
padding_values = F.pad(original_values, pad=(1,0,0,0))
print("padding_values: ",padding_values,"\n")
#tensor([[[[ 0.0000, -0.6208, -1.4084],
          # [ 0.0000, -0.2202,  0.1708],
          # [ 0.0000,  0.9050, -1.2441]]]])
print("padding_values的shape: ",padding_values.shape)#torch.Size([1, 1, 3, 3])