Exploring the Limits of Transfer Learning with Unified Model in the Cybersecurity Domain