Catastrophic forgetting is a central obstacle in single-stage fine-tuning of large language models (LLMs) on a new task, especially when the original pretrai…
机构:中科院
来源:arXiv 2610.25751 | AI4Papers 论文推荐平台