gistfile1.txt

import time
import torch
import torch.nn as nn

class Model(nn.Module):
    def __init__(self):
        super(Model, self).__init__()

        self.conv1 = nn.Sequential(
            nn.Conv1d(1, 40, kernel_size=8, stride=1, padding=4),
            nn.BatchNorm1d(40)
        )
        self.conv2 = nn.Sequential(
            nn.Conv1d(40, 40, kernel_size=8, stride=1, padding=4),
            nn.BatchNorm1d(40)
        )
        self.pool2 = nn.MaxPool1d(kernel_size=160)

        self.conv3 = nn.Sequential(
            nn.Conv2d(1, 50, kernel_size=(8, 13), stride=(1, 1)),
            nn.BatchNorm2d(50)
        )
        self.pool3 = nn.MaxPool2d(kernel_size=(3, 3))

        self.conv4 = nn.Sequential(
            nn.Conv2d(50, 50, kernel_size=(1, 5), stride=(1, 1)),
            nn.BatchNorm2d(50)
        )
        self.pool4 = nn.MaxPool2d(kernel_size=(1, 3))

        self.fc5 = nn.Sequential(
            nn.Linear(50 * 11 * 14, 4096),
            nn.Dropout(0.5)
        )

        self.fc6 = nn.Sequential(
            nn.Linear(4096, 4096),
            nn.Dropout(0.5)
        )

        self.output = nn.Linear(4096, 50)

    def forward(self, x):
        x = self.conv1(x)
        x = self.conv2(x)
        x = self.pool2(x)
        x = x.unsqueeze(1)
        x = self.conv3(x)
        x = self.pool3(x)
        x = self.conv4(x)
        x = self.pool4(x)
        x = x.view(x.shape[0], -1)
        x = self.fc5(x)
        x = self.fc6(x)
        return self.output(x)

def main():
    with torch.no_grad():
        model = Model()

        sums = torch.zeros(1, 50)

        for _ in range(100):
            time.sleep(0.5)
            data = torch.rand(1, 1, 24000)
            output = model(data)
            print(output.requires_grad)
            sums += output

if __name__ == '__main__':
    main()