让大模型持续学习不遗忘，用门控机制控制更新幅度
STABLE: Gated Continual Learning for Large Language Models
- 用LoRA实现高效微调，通过门控判断每次更新是否安全
- 在Qwen-2.5-7B上测试，门控后准确率下降不超过5%且知识保留更好
- 适合需要长期更新知识的大模型应用，如智能客服、内容生成
大语言模型日益需要在不重新训练的情况下持续适应新信息。然而，顺序更新会导致灾难性遗忘，即新知识损害旧知识。本文提出STABLE，一种基于门控的持续自编辑框架，通过低秩适配（LoRA）实现参数高效微调，并利用三个指标评估每次候选更新的稳定性：(i) 精确匹配（EM）下降，衡量事实准确性损失；(ii) 位数增加，反映模型置信度降低；(iii) KL散度，量化基线模型与适配模型之间的分布偏移。若超过阈值，则对LoRA更新进行裁剪或拒绝。在Qwen-2.5-7B模型上的实验表明，门控有效缓解遗忘并保持可适应性。基于EM的门控在短序列中取得最高累积性能。不同门控策略虽可达到相近的分布偏移（以KL散度衡量），但准确率表现各异，凸显门控设计在持续适配中的重要性。该方法为持续模型编辑提供了原则性方案，使大模型能在保留可靠性的同时融合新知识。代码见：https://github.com/Bhoy1/STABLE
原文摘要 · Abstract (English)
Large language models (LLMs) increasingly require mechanisms for continual adaptation without full retraining. However, sequential updates can lead to catastrophic forgetting, where new edits degrade previously acquired knowledge. This work presents STABLE, a gated continual self editing framework that constrains forgetting during sequential updates using parameter efficient fine tuning via Low Rank Adaptation (LoRA; see arXiv:2106.09685). Each candidate edit is evaluated against a stability budget using one of three metrics: (i) Exact Match (EM) drop, capturing factual accuracy loss; (ii) bits increase, reflecting reduced model confidence; and (iii) KL divergence, quantifying distributional drift between the base and adapted models. If a threshold is exceeded, the LoRA update is rescaled through a clipping procedure or rejected. Experiments on the Qwen-2.5-7B model show that gating effectively mitigates forgetting while preserving adaptability. EM based gating achieved the highest cumulative performance in short continual learning sequences. Our results show that different gating strategies can achieve comparable distribution shift (measured by KL divergence) while producing different accuracy outcomes, highlighting the importance of gating design in continual adaptation. This approach offers a principled method for continual model editing, enabling LLMs to integrate new knowledge while maintaining reliability. Code: https://github.com/Bhoy1/STABLE
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。