AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Direct Preference Optimization for Perceptual Alignment via Vision-Language Consistency

Seungyeon Lee, Donggyu Lee

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Bias-aware multi-image robustness · 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 description-based responses (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 82.0 -> 87.2 F1; 82.6 -> 85.2 POPEImproved
add consistency filtering of response pairs (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 81.7 -> 87.2 F1; 81.8 -> 85.2 POPEImproved
add VLC-score difference penalty weight |wc| (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 77.9 -> 87.2 F1; 81.0 -> 85.2 POPEImproved
raise consistency threshold from 0.5 to 0.7 (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 82.9 -> 87.2 F1; 83.2 -> 85.2 POPEImproved
raise consistency threshold from 0.7 to 0.8 (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 87.2 -> 86.1 F1; 85.2 -> 84.7 POPEHurt
add mixture noise injection to rejected responses (vs noise-free) (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 81.4 -> 87.2 F1; 81.6 -> 85.2 POPEImproved
use order-shuffling noise only instead of noise-free (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 81.4 -> 79.0 F1; 81.6 -> 81.3 POPEHurt
use replacement noise only instead of mixture (ablation)VLC-DPO (LLaVA-1.5 7B) · AMBER (Disc.) / POPEF1 / accuracy: 87.2 -> 86.1 F1; 85.2 -> 83.9 POPEHurt
apply VLC-DPO to LLaVA-1.5 7BVLC-DPO (LLaVA-1.5 7B) · POPE (Adversarial)accuracy: 80.8 -> 85.2Improved
apply VLC-DPO to LLaVA-1.5 13BVLC-DPO (LLaVA-1.5 7B) · A-OKVQA (zero-shot)VQA score: 52.3 -> 66.5Improved
apply VLC-DPO to LLaVA-1.5 7BVLC-DPO (LLaVA-1.5 7B) · OK-VQA (zero-shot)VQA score: 46.4 -> 57.3Improved
apply VLC-DPO to LLaVA-1.5 7BVLC-DPO (LLaVA-1.5 7B) · MMHal-Benchscore: 1.86 -> 2.95Improved

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