AI Paper Atlas

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

SiGMA: Sign-Guided Merging and Adaptation framework for Multimodal Continual Instruction Tuning

Keonhee Park, Gunhee Kim

ECCV 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Mixture-of-experts for multimodal continual tuning

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 sign-guided adaptive tuning (S-Tuning) to the expansion-merge baseline (ablation)expansion-merge LoRA baseline · UCITAvg. All: 62.42 -> 64.86Improved
add sign-guided merging (S-Merge) to the baseline (ablation)expansion-merge LoRA baseline · UCITAvg. All / Forgets: 62.42 -> 63.35 Avg. All; -5.98 -> -6.26 ForgetsMixed
add S-Merge on top of S-Tuning (ablation)SiGMA · UCITAvg. All: 64.86 -> 66.17Improved
add sign-based weight decoupling (SWD) when consolidating A and B (ablation)S-Tuning · UCITAvg. All: 62.42 -> 64.86Improved
add SWD when consolidating full LoRA deltas (ablation)S-Tuning · UCITAvg. All: 63.23 -> 64.15Improved
consolidate low-rank A and B separately instead of full deltas (with SWD) (ablation)S-Tuning · UCITAvg. All: 64.15 -> 64.86Improved
exclude the top layer from S-Tuning (k=1) instead of tuning all layers (ablation)S-Tuning · UCITAvg. All: 64.27 -> 66.17Improved
apply S-Tuning to more layers (k from 32 to 1) (ablation)S-Tuning · UCITAvg. All: 63.35 -> 66.17Improved
add cosine-distance amplification d_k in S-Merge (ablation)S-Merge · UCITAvg. All: 62.69 -> 66.17Improved
add magnitude threshold lambda=0.8 to S-Merge (ablation)S-Merge · UCITAvg. All / Forgets: 54.32 -> 66.17 Avg. All; -19.65 -> -4.81 ForgetsImproved
raise lambda from 0.8 to 1.0 (ablation)S-Merge · UCITAvg. All: 66.17 -> 65.89No clear effect
SiGMA versus strongest baseline DISCOSiGMA · UCITAvg. All: 62.91 -> 66.17Improved

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