http://www.atsb.gov.au/newsroom/2008...e/2008_43.aspx
In the Q&A that exists in the audio recording, it was stated that the aircraft has returned to Sydney this evening. They also clarified that the initial gradual increase in altitude occured after the autopilot first disengaged (and therefore while the aircraft was being flown manually). The autopilot was re-engaged and the aircraft returned to the selected altitude. The autopilot was then turned off, before the first of the pitch down incidents.
From my reading of it, the dodgy data coming out of the Air Data Inertial Reference Unit and into Flight Control Primary Computers caused them to believe the aircraft was operating with an angle of attack well outside of its flight envelope (and in a stall) and hence, being a Fly-By-Wire aircraft, took swift "corrective" action to bring the nose down. Except of course this wasn't corrective at all and took the aircraft out of stable level flight.