JumpLoRA: Sparse Adapters for Continual Learning in Large Language Models

arXiv:2604.16171v1 Announce Type: cross Abstract: Adapter-based methods have become a cost-effective approach to continual learning (CL) for Large Language Models (LLMs), by sequentially learning a low-rank update matrix for each task. To mitigate catastrophic forgetting, state-of-the-art...