AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Controlling Multimodal LLMs via Reward-guided Decoding

Oscar Mañas, Pierluca D'Oro, Koustuv Sinha, Adriana Romero-Soriano, Michal Drozdzal, Aishwarya Agrawal

ICCV 2025 · 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 reward-guided decoding with hallucination reward (w=1.0) over greedyLLaVA-1.5-7B · COCO captionsCHAIR_i: 15.05 -> 4.53Improved
use hallucination reward only (w=1.0) instead of greedy, recall costLLaVA-1.5-7B · COCO captionsobject recall: 81.30 -> 76.04Hurt
mix recall reward (w=0.5) instead of w=1.0 (ablation)MRGD LLaVA-1.5-7B · AMBERcoverage: 52.4 -> 60.8Improved
use recall reward only (w=0.0) instead of greedyLLaVA-1.5-7B · COCO captionsrecall and CHAIR_i: recall 81.30 -> 85.23; CHAIR_i 15.05 -> 24.20Mixed
use MRGD instead of CGDLLaVA-1.5-7B · COCO captionsCHAIR_i: 9.48 -> 4.53Improved
fine-tune PaliGemma on preference data instead of off-the-shelf SigLIP for r_hal (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: 7.19 -> 4.53Improved
add RLAIF-V to the r_hal preference data (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: 5.34 -> 7.83Hurt
add RLAIF-V and remove POVID from r_hal data (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: 5.34 -> 8.17Hurt
use PaliGemma-2 backbone instead of PaliGemma for r_hal (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: 5.34 -> 5.88No clear effect
use DETR instead of OWLv2 object detector for r_rec (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: 5.34 -> 5.37No clear effect
evaluate rewards every sentence (T=1) with k=5 instead of rejection sampling with k=30 (ablation)MRGD LLaVA-1.5-7B · COCO captionsCHAIR_i: k=5 beats rejection sampling k=30; over 6x more sample-efficientImproved
add prompting for accuracyLlama-3.2-Vision-11B · COCO captionsCHAIR_s: 20.52 -> 25.24Hurt

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