Nam, Y.; Kim, J.; Jeong, J. Learning from the Undesirable: Robust Adaptation of Language Models Without Forgetting. AAAI 2026, 40, 32537-32545.