st04.py

def content_loss(c_features, t_features):
    """
    Compute mean squared content loss of all feature maps. 
    """
    loss = 0.5 * (t_features['conv4_2'] - c_features['conv4_2']) ** 2
    return torch.mean(loss)