Adapting Vision-Language Models for Evaluating World Models

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Hendriksen, Mariya, Rashid, Tabish, Bignell, David, Georgescu, Raluca, Lemkhenter, Abdelhak, Hofmann, Katja, Devlin, Sam, Parisot, Sarah
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911284145946624
author Hendriksen, Mariya
Rashid, Tabish
Bignell, David
Georgescu, Raluca
Lemkhenter, Abdelhak
Hofmann, Katja
Devlin, Sam
Parisot, Sarah
author_facet Hendriksen, Mariya
Rashid, Tabish
Bignell, David
Georgescu, Raluca
Lemkhenter, Abdelhak
Hofmann, Katja
Devlin, Sam
Parisot, Sarah
contents World models - generative models that simulate environment dynamics conditioned on past observations and actions - are gaining prominence in planning, simulation, and embodied AI. However, evaluating their rollouts remains a fundamental challenge, requiring fine-grained, temporally grounded assessment of action alignment and semantic consistency - capabilities not captured by existing metrics. Vision-Language Models (VLMs) have shown promise as automatic evaluators of generative content due to their strong multimodal reasoning abilities. Yet, their use in fine-grained, temporally sensitive evaluation tasks remains limited and requires targeted adaptation. We introduce an evaluation protocol targeting two recognition tasks - action recognition and character recognition - each assessed across binary, multiple-choice, and open-ended formats. To support this, we present UNIVERSE (UNIfied Vision-language Evaluator for Rollouts in Simulated Environments), a VLM-based evaluator for video world model rollouts adapted under data and compute constraints. In our extensive experiments totaling over 5,154 GPU-days, we explore full, partial, and parameter-efficient adaptation methods across various task formats, context lengths, sampling methods, and data compositions. The resulting unified evaluator achieves parity with task-specific checkpoints. Human studies across seven diverse environments confirm strong alignment with human judgments, establishing UNIVERSE as a lightweight, adaptable, and semantics-aware evaluator for video world models.
format Preprint
id arxiv_https___arxiv_org_abs_2506_17967
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Adapting Vision-Language Models for Evaluating World Models
Hendriksen, Mariya
Rashid, Tabish
Bignell, David
Georgescu, Raluca
Lemkhenter, Abdelhak
Hofmann, Katja
Devlin, Sam
Parisot, Sarah
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
World models - generative models that simulate environment dynamics conditioned on past observations and actions - are gaining prominence in planning, simulation, and embodied AI. However, evaluating their rollouts remains a fundamental challenge, requiring fine-grained, temporally grounded assessment of action alignment and semantic consistency - capabilities not captured by existing metrics. Vision-Language Models (VLMs) have shown promise as automatic evaluators of generative content due to their strong multimodal reasoning abilities. Yet, their use in fine-grained, temporally sensitive evaluation tasks remains limited and requires targeted adaptation. We introduce an evaluation protocol targeting two recognition tasks - action recognition and character recognition - each assessed across binary, multiple-choice, and open-ended formats. To support this, we present UNIVERSE (UNIfied Vision-language Evaluator for Rollouts in Simulated Environments), a VLM-based evaluator for video world model rollouts adapted under data and compute constraints. In our extensive experiments totaling over 5,154 GPU-days, we explore full, partial, and parameter-efficient adaptation methods across various task formats, context lengths, sampling methods, and data compositions. The resulting unified evaluator achieves parity with task-specific checkpoints. Human studies across seven diverse environments confirm strong alignment with human judgments, establishing UNIVERSE as a lightweight, adaptable, and semantics-aware evaluator for video world models.
title Adapting Vision-Language Models for Evaluating World Models
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.17967