1 paper · 1 filter
William Hoy, Nurcin Celik
Large language models (LLMs) increasingly require mechanisms for continual adaptation without full retraining. However, sequential updates can lead to catastrophic forgetting, wher…