additionally the training part

gistfile1.txt

def trainshit(model:NeuralNetwork, samples:list[tuple[torch.Tensor, torch.Tensor]], epochs=100):
    criterion = nn.BCELoss().double()
    optimizer = torch.optim.SGD(model.parameters(),lr=0.01, momentum=0.9)
    print()

    for epoch in range(epochs):

        i = 0

        while i < len(samples):
            print(f'\repoch: {epoch} input: {i}       ',end=' ')
            input = samples[i][0]
            target = samples[i][1]
            # enumerate mini batches   
            # clear the gradients
            optimizer.zero_grad()
            # compute the model output
            model.double()
            yhat = model(input)
            # calculate loss
            loss = criterion(yhat, target)
            # credit assignment
            loss.backward()
            # update model weights
            optimizer.step()

            i+=1