PyTorch构建GAN时出现'NoneType' object is not callable错误求助

你在构建GAN生成器时遇到的'NoneType' object is not callable错误，核心原因是**get_gen_block函数的缩进错误**，导致非final_block的场景下函数没有返回任何模块，最终self.block_1、self.block_2、self.block_3都为None，调用时自然会触发该错误。

问题代码中的错误点 #

你的get_gen_block函数里，第二个return语句被错误缩进在了if final_block == True:的代码块内，当final_block为False时，函数执行完if分支后没有任何返回，默认返回None：

def get_gen_block(in_channels, out_channels, kernel_size, stride, final_block = False):
  if final_block == True:
    return nn.Sequential(
        nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride),
        nn.Tanh()
    )

    return nn.Sequential(
        nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride),
        nn.BatchNorm2d(out_channels),
        nn.ReLU()
    )

修正后的代码 #

只需要把第二个return语句的缩进调整到和if同级即可：

def get_gen_block(in_channels, out_channels, kernel_size, stride, final_block = False):
  if final_block == True:
    return nn.Sequential(
        nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride),
        nn.Tanh()
    )

  return nn.Sequential(
      nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride),
      nn.BatchNorm2d(out_channels),
      nn.ReLU()
  )

验证修正效果 #

修正后，实例化Generator并测试前向传播：

import torch
import torch.nn as nn

# 实例化生成器
noise_dim = 100
gen = Generator(noise_dim)

# 生成测试噪声
test_noise = torch.randn(32, noise_dim)
output = gen(test_noise)
print(output.shape)  # 输出符合预期的张量形状，不再报错

内容的提问来源于stack exchange，提问作者Amit Pal