CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning

TL;DR

CL-LoRA introduces a dual-adapter architecture with shared and task-specific modules, enabling efficient rehearsal-free class-incremental learning by leveraging shared knowledge and reducing parameter redundancy.

Contribution

The paper proposes CL-LoRA, a novel dual-adapter approach that combines shared and task-specific adapters for improved continual learning with pre-trained models.

Findings

Achieves competitive performance on multiple benchmarks.

Reduces training and inference computation.

Effectively balances shared knowledge and task-specific features.

Abstract

Class-Incremental Learning (CIL) aims to learn new classes sequentially while retaining the knowledge of previously learned classes. Recently, pre-trained models (PTMs) combined with parameter-efficient fine-tuning (PEFT) have shown remarkable performance in rehearsal-free CIL without requiring exemplars from previous tasks. However, existing adapter-based methods, which incorporate lightweight learnable modules into PTMs for CIL, create new adapters for each new task, leading to both parameter redundancy and failure to leverage shared knowledge across tasks. In this work, we propose ContinuaL Low-Rank Adaptation (CL-LoRA), which introduces a novel dual-adapter architecture combining \textbf{task-shared adapters} to learn cross-task knowledge and \textbf{task-specific adapters} to capture unique features of each new task. Specifically, the shared adapters utilize random orthogonal matrices and leverage knowledge distillation with gradient reassignment to preserve essential shared knowledge. In addition, we introduce learnable block-wise weights for task-specific adapters, which mitigate inter-task interference while maintaining the model's plasticity. We demonstrate CL-LoRA consistently achieves promising performance under multiple benchmarks with reduced training and inference computation, establishing a more efficient and scalable paradigm for continual learning with pre-trained models.