AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Large Multi-modal Models Can Interpret Features in Large Multi-modal Models

Kaichen Zhang, Yifei Shen, Bo Li, Ziwei Liu

ICCV 2025 · Vision-language and multimodal › Cross-modal learning · approach: Multimodal fusion with missing modalities · 13 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 SAE feature explanations from the V-Interp pipeline instead of random imagesSAE features · SAE explanations (total)IoU: 0.005 ± 2 × 10^-4 -> 0.20Improved
use SAE feature explanations from the V-Interp pipeline instead of random imagesSAE features · SAE explanations (total)CLIP score: 18.2 ± 1 × 10^-2 -> 23.6Improved
hook the SAE at a deeper LLaVA layer (8th to 32nd) (ablation)SAE on LLaVA-NeXT-LLaMA3-8B · Cross-layer SAEIoU: 0.30 -> 0.40Improved
hook the SAE at a deeper LLaVA layer (8th to 32nd) (ablation)SAE on LLaVA-NeXT-LLaMA3-8B · Cross-layer SAECLIP score: 22.82 -> 26.55Improved
hook the SAE at the 25th instead of the 8th layer (ablation)SAE on LLaVA-NeXT-LLaMA3-8B · Cross-layer SAEIoU: 0.30 -> 0.31No clear effect
apply the pipeline to InstructBLIP-7B instead of LLaVA (ablation)V-Interp · InstructBLIP-7B SAECLIP score: 17.71 -> 28.01Improved
apply the pipeline to Qwen2.5-VL (ablation)V-Interp · Qwen2.5-VL SAEIoU: 0.06 -> 26.67Improved
steer with identified features instead of random featuresLLaVA-NeXT · Steering effect evaluationGPT-4o score: 2.02 -> 6.36Improved
clamp a single OCR-related feature to a high valueLLaVA-NeXT · HallusionBench (100 examples)fraction of cases improved: 0.09 improved, 0.02 worse, 0.89 sameImproved
scale SAE from 2^17 to 2^18 features (ablation)SAE on LLaVA-NeXT · SAE training losslossNo clear effect

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