AI Paper Atlas

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

Beyond Disjoint Tasks: Towards More Natural Continual Learning for Vision-Language Models

Xiang Xu, Yiyang Su, Tianchen Zhao, Zheng Zhang, Zhuowen Tu, Anil Jain, Jonathan Wu

ECCV 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Mixture-of-experts for multimodal continual tuning · 0 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 interference-free delta-space merging instead of factor-level linear merging (ablation)SOCiaL merge · UCIT-O Protocol I (Tasks 1 and 7)average accuracy: 65.47 -> 69.85Improved
add DARE sparsification to delta merge (ablation)SOCiaL merge · UCIT-O Protocol I (Tasks 1 and 7)average accuracy: 69.85 -> 69.91No clear effect
add TIES trimming to delta merge (ablation)SOCiaL merge · UCIT-O Protocol I (Tasks 1 and 7)average accuracy: 69.85 -> 68.04Hurt
add regularized fine-tuning after merge (ablation)SOCiaL merge · UCIT-O Protocol I (Tasks 1 and 7)average accuracy: 69.91 -> 86.80Improved
replace MLP router with bandit router (ablation)SOCiaL · CoINlast average task accuracy: 32.74 -> 70.17Improved
replace mean prototype router with GMM and bandit router (ablation)SOCiaL · UCITlast average task accuracy: 69.45 -> 72.23Improved
raise containment threshold to 0.7 (ablation)SOCiaL · UCIT-O Protocol Iaverage accuracy: 64.50 -> 41.79Hurt
increase GMM components from 1 to 3 (ablation)SOCiaL · UCIT-O Protocol Iaverage accuracy: 50.62 -> 64.50Improved
increase GMM components from 3 to 8 (ablation)SOCiaL · UCIT-O Protocol Iaverage accuracy: 64.50 -> 69.41Improved
increase semantic overlap across protocolsHiDe-LLaVA · UCIT-Oaverage accuracy: 42.65 -> 34.03 -> 38.76Hurt
use SOCiaL instead of strongest baseline as overlap growsSOCiaL · UCIT-Oaverage accuracy gain: +3.07 on Protocol I to +9.77 on Protocol IIIImproved
merge overlapping adaptersSOCiaL · UCIT-O Protocol Inumber of deployed adapters: 10 -> 6Improved

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