AI Paper Atlas

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

Sparse Spectral LoRA: Routed Experts for Medical VLMs

Omid Nejatimanzari, Hojat Asgariandehkordi, Taha Koleilat, Yiming Xiao, Hassan Rivaz

CVPR 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
use SVD-segment experts in MoE vs single LoRA with SVD fragments (ablation)MedQwen · medical VQA averageavg accuracy: 60.1 -> 68.2Improved
add SVD-segment initialization to MoE with scaling (ablation)MedQwen · medical VQA averageavg accuracy: 67.3 -> 68.2Improved
add MoE scaling to SVD-initialized MoE (ablation)MedQwen · medical VQA averageavg accuracy: 67.3 -> 68.2Improved
replace MoELoRA with SVD-structured MoE (ablation)MedQwen · PathVQA close-endedaccuracy: 78.4 -> 84.2Improved
replace standard LoRA with MedQwen in sequential training (Harvard-FairVLMed to PathVQA) (ablation)Qwen2.5-VL 7B · Harvard-FairVLMed after PathVQAaccuracy retention: over 50% drop -> 5% dropImproved
increase LoRA rank from 64 to 128 (ablation)MedQwen · radiology zero-shot classificationaccuracy: +0.3%No clear effect
activate more experts than 2-of-8 (ablation)MedQwen · medical VQAaccuracyHurt
increase total experts (2-of-8 to 4-of-16 / 8-of-32) (ablation)MedQwen · medical VQAaccuracyMixed
use MedQwen instead of single rank-32 LoRABiomedCLIP ViT-B/16 · zero-shot radiology classificationavg accuracy: 55.45 -> 58.83Improved
use MedQwen instead of PiSSABiomedCLIP ViT-B/16 · zero-shot radiology classificationavg accuracy: 50.08 -> 58.83Improved
use MedQwen instead of general Qwen-2.5-VLQwen2.5-VL 7B · medical VQA averageavg accuracy: 49.3 -> 68.2Improved

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