3 ms·

I think this kind of overthinking is an extremely common pattern in the Chinese models. GLM's models are also very much like this.