基于LoRA的GEM风格约束与双梯度投影的持续学习
GEM-Style Constraints for PEFT with Dual Gradient Projection in LoRA
AI总结：
I-GEM通过在LoRA子空间内应用GEM约束，实现低开销的持续学习，有效提升模型性能并减少计算成本。
摘要：
大型语言模型（LLMs）的全量微调计算成本高昂，推动了利用参数高效适配器的持续学习（CL）方法。我们重新审视了在低秩适配器（LoRA）子空间内的梯度周期记忆（GEM），并引入I-GEM：一种固定预算、驻留GPU的双投影梯度近似方法，用于近似GEM的二次投影。通过仅在适配器参数内限制非干扰，I-GEM在数量级更低的平均投影开销下保持GEM-like的稳定性。在具有诱导领域漂移的3任务AG新闻拆分上，使用GPT-2（355M）和LoRA（r=8），I-GEM在平均精度上与GEM相当（在约0.04点内），并且比A-GEM高出约1.4点。关键的是，它将投影时间与GEM相比减少了约1000倍。这些结果表明，在LoRA子空间中应用GEM约束是实现大规模LLM持续学习的一种可行路径。
英文摘要：
Full fine-tuning of Large Language Models (LLMs) is computationally costly, motivating Continual Learning (CL) approaches that utilize parameter-efficient adapters. We revisit Gradient Episodic Memory (GEM) within the Low-Rank Adapter (LoRA) subspace and introduce I-GEM: a fixed-budget, GPU-resident dual projected-gradient approximation to GEM's quadratic projection. By constraining non-interference solely within the adapter parameters, I-GEM preserves GEM-like stability with orders-of-magnitude lower mean projection overhead. On a 3-task AG News split with induced domain drift, using GPT-2 (355M) and LoRA ($r=8$), I-GEM matches GEM's average accuracy (within $\sim\!0.04$ pts) and outperforms A-GEM by $\sim\!1.4$ pts. Crucially, it reduces projection time vs.\ GEM by a factor of $\sim\!10^3$. These results suggest that applying GEM constraints in the LoRA subspace is a practical pathway for continual learning at the LLM scale.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
