Calibrating MLLM-as-a-judge via Multimodal Bayesian Prompt Ensembles

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Slyman, Eric, Tanjim, Mehrab, Kafle, Kushal, Lee, Stefan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911148135153664
author Slyman, Eric
Tanjim, Mehrab
Kafle, Kushal
Lee, Stefan
author_facet Slyman, Eric
Tanjim, Mehrab
Kafle, Kushal
Lee, Stefan
contents Multimodal large language models (MLLMs) are increasingly used to evaluate text-to-image (TTI) generation systems, providing automated judgments based on visual and textual context. However, these "judge" models often suffer from biases, overconfidence, and inconsistent performance across diverse image domains. While prompt ensembling has shown promise for mitigating these issues in unimodal, text-only settings, our experiments reveal that standard ensembling methods fail to generalize effectively for TTI tasks. To address these limitations, we propose a new multimodal-aware method called Multimodal Mixture-of-Bayesian Prompt Ensembles (MMB). Our method uses a Bayesian prompt ensemble approach augmented by image clustering, allowing the judge to dynamically assign prompt weights based on the visual characteristics of each sample. We show that MMB improves accuracy in pairwise preference judgments and greatly enhances calibration, making it easier to gauge the judge's true uncertainty. In evaluations on two TTI benchmarks, HPSv2 and MJBench, MMB outperforms existing baselines in alignment with human annotations and calibration across varied image content. Our findings highlight the importance of multimodal-specific strategies for judge calibration and suggest a promising path forward for reliable large-scale TTI evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2509_08777
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Calibrating MLLM-as-a-judge via Multimodal Bayesian Prompt Ensembles
Slyman, Eric
Tanjim, Mehrab
Kafle, Kushal
Lee, Stefan
Computer Vision and Pattern Recognition
Computation and Language
Multimodal large language models (MLLMs) are increasingly used to evaluate text-to-image (TTI) generation systems, providing automated judgments based on visual and textual context. However, these "judge" models often suffer from biases, overconfidence, and inconsistent performance across diverse image domains. While prompt ensembling has shown promise for mitigating these issues in unimodal, text-only settings, our experiments reveal that standard ensembling methods fail to generalize effectively for TTI tasks. To address these limitations, we propose a new multimodal-aware method called Multimodal Mixture-of-Bayesian Prompt Ensembles (MMB). Our method uses a Bayesian prompt ensemble approach augmented by image clustering, allowing the judge to dynamically assign prompt weights based on the visual characteristics of each sample. We show that MMB improves accuracy in pairwise preference judgments and greatly enhances calibration, making it easier to gauge the judge's true uncertainty. In evaluations on two TTI benchmarks, HPSv2 and MJBench, MMB outperforms existing baselines in alignment with human annotations and calibration across varied image content. Our findings highlight the importance of multimodal-specific strategies for judge calibration and suggest a promising path forward for reliable large-scale TTI evaluation.
title Calibrating MLLM-as-a-judge via Multimodal Bayesian Prompt Ensembles
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2509.08777