面向持续学习的基于LoRA的任务算术方法
Task Arithmetic with LoRA for Continual Learning
AI总结：
针对持续学习中的灾难性遗忘和计算成本高的问题，提出结合LoRA与任务算术的Transformer视觉模型持续训练方法，可规避遗忘、降低计算量，小内存辅助下性能接近全量微调。
摘要：
持续学习指的是训练数据以被称为“任务”的连续数据块形式可用的问题。持续学习领域的大部分进展因灾难性遗忘问题而受阻，该问题由模型在数据流上进行连续训练导致。此外，多次连续训练大模型的计算成本很高。为同时缓解这两个问题，我们提出一种新方法，结合低秩适配（LoRA）和任务算术来持续训练基于Transformer的视觉模型。我们的方法完全规避了灾难性遗忘问题，同时降低了在每个任务上训练模型的计算需求。当借助每类10个样本的小型内存时，我们的方法取得了接近全量微调的性能。我们开展了严格的消融实验以验证该方法的优势。
英文摘要：
Continual learning refers to the problem where the training data is available in sequential chunks, termed "tasks". The majority of progress in continual learning has been stunted by the problem of catastrophic forgetting, which is caused by sequential training of the model on streams of data. Moreover, it becomes computationally expensive to sequentially train large models multiple times. To mitigate both of these problems at once, we propose a novel method to continually train transformer-based vision models using low-rank adaptation and task arithmetic. Our method completely bypasses the problem of catastrophic forgetting, as well as reducing the computational requirement for training models on each task. When aided with a small memory of 10 samples per class, our method achieves performance close to full-set finetuning. We present rigorous ablations to support the prowess of our method.
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。
