It is enough to use a different seed to get a different result; I do not think it is important to recreate a copy of the Stanford model weights, as long as the

It is enough to use a different seed to get a different result; I do not think it is important to recreate a copy of the Stanford model weights, as long as the performance is equivalent.