AI Paper Atlas

Research map / Continual learning / Continual fine-tuning of LLMs

Decomposing and Composing: Towards Efficient Vision-Language Continual Learning via Rank-1 Expert Pool in a Single LoRA

Zhan Fa, Yue Duan, Jian Zhang, Lei Qi, Wanqi Yang, Yinghuan Shi

AAAI 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add rank-1 expert pool (dynamic composition) to vanilla LoRA (ablation)Vanilla LoRA · MTIL Order 1Transfer: 65.0 -> 69.7Improved
Add rank-1 expert pool to vanilla LoRA (ablation)Vanilla LoRA · MTIL Order 1Last: 73.8 -> 77.1Improved
Add AGO loss to vanilla LoRA (ablation)Vanilla LoRA · MTIL Order 1Transfer: 65.0 -> 65.3No clear effect
Add AGO loss to vanilla LoRA (ablation)Vanilla LoRA · MTIL Order 1Average: 68.5 -> 69.2Improved
Add AGO loss on top of expert pool (ablation)Expert pool LoRA · MTIL Order 1Last: 77.1 -> 77.9Improved
Merge Top-8 experts instead of Top-4 (ablation)Ours · MTIL Order 1Last: 77.9 -> 73.7Hurt
Merge Top-2 experts instead of Top-4 (ablation)Ours · MTIL Order 1Transfer: 70.3 -> 70.5No clear effect
Merge low-frequency experts (Down-4) instead of high-frequency (Top-4) (ablation)Ours · MTIL Order 1Last: 77.9 -> 62.9Hurt
Merge Mid-4 instead of Top-4 experts (ablation)Ours · MTIL Order 1Last: 77.9 -> 77.4Hurt
Increase AGO loss weight lambda (ablation)Ours · MTILTransfer vs downstream accuracyMixed
Rank-1 expert pool single LoRA vs ZSCLZSCL · MTIL Order 1 task-ID freeAverage: 64.4 -> 73.0Improved
Rank-1 expert pool vs ZSCL training costZSCL · Training paramsMB: 570.76 -> 18.99Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy