多层感知机 (MLP) 网络设计与实现

import torch.nn as nn

class MLP(nn.Module): def init(self, input_size, hidden_size_1, hidden_size_2, output_size): super(MLP, self).init() self.flatten = nn.Flatten() self.fc1 = nn.Linear(input_size, hidden_size_1) self.sigmoid = nn.Sigmoid() self.fc2 = nn.Linear(hidden_size_1, hidden_size_2) self.relu = nn.ReLU() self.fc3 = nn.Linear(hidden_size_2, output_size)

def forward(self, x):
    x = self.flatten(x)
    print('Flatten output shape: 	', x.shape)
    x = self.fc1(x)
    print('Linear output shape: 	', x.shape)
    print('	 Linear weight's mean: 	', x.mean().item())
    print('	 Linear bias's mean: 	', self.fc1.bias.mean().item())
    x = self.sigmoid(x)
    print('Sigmoid output shape: 	', x.shape)
    x = self.fc2(x)
    print('Linear output shape: 	', x.shape)
    print('	 Linear weight's mean: 	', x.mean().item())
    print('	 Linear bias's mean: 	', self.fc2.bias.mean().item())
    x = self.relu(x)
    print('ReLU output shape: 	', x.shape)
    x = self.fc3(x)
    print('Linear output shape: 	', x.shape)
    print('	 Linear weight's mean: 	', x.mean().item())
    print('	 Linear bias's mean: 	', self.fc3.bias.mean().item())
    return x

model = MLP(input_size=32323, hidden_size_1=2048, hidden_size_2=512, output_size=20) print(model)