如何在pytorch中初始化不同风格的nn.Sequential块的不同层的权重？

import torch.nn as nn
import torch.nn.init as init

# 定义一个自定义的nn.Sequential块
class MySequential(nn.Sequential):
    def __init__(self, *args):
        super(MySequential, self).__init__(*args)

        # 遍历子模块并初始化权重
        for module in self.children():
            if isinstance(module, nn.Linear):
                # 对于Linear层，使用xavier_uniform_方法初始化权重
                init.xavier_uniform_(module.weight)
                if module.bias is not None:
                    init.constant_(module.bias, 0.0)
            elif isinstance(module, nn.Conv2d):
                # 对于Conv2d层，使用kaiming_normal_方法初始化权重
                init.kaiming_normal_(module.weight, mode='fan_out', nonlinearity='relu')
                if module.bias is not None:
                    init.constant_(module.bias, 0.0)
            elif isinstance(module, nn.BatchNorm2d):
                # 对于BatchNorm2d层，使用constant_方法初始化权重
                init.constant_(module.weight, 1)
                init.constant_(module.bias, 0.0)

# 创建一个示例MySequential块
model = MySequential(
    nn.Linear(10, 20),
    nn.ReLU(),
    nn.Conv2d(20, 64, kernel_size=3),
    nn.BatchNorm2d(64),
    nn.ReLU(),
    nn.Linear(64, 10)
)

# 打印模型权重
for name, param in model.named_parameters():
    print(name, param.data)