The PULL? The NOSE was down by three off 0, and 5.5 off cruise. That's nearly six degrees to make up, and he was too swift, STALLSTALL. Also, the a/p was actively inputting NU to counter the PITCH down. So PITCH is iffy at the a/p drop, and PF needs to fly, and he did.
That can explain the initial nose up, if the aircraft was not Straight and Level at the point of "you have it, robot taking a timeout, old chap."
It does not explain subsequent nose up inputs that lack successful counter-corrections toward a more level attitude.