3 ms·

The paper wasn't about deep learning