Newest papers

RACE: Learning Multimodal Search through Referential Abstraction and Compositional Evidence

Eric Tao, Minhao Sun, Liyan Liu, Zhijing Wu, Xiaohan Yu, Lang Mei, Li Huang, Chao Feng, Jinchao Ma, Sichen Kang, Siyu Mo, Chong Chen · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28382 · cited by 0 works (OpenAlex) · open-access PDF

French

Les agents de recherche multimodales de formation nécessitent des tâches qui rendent la recherche visuellement fondée nécessaire pour déterminer la réponse. Les indices d'identité textuels et les preuves cibles inutilisées permettent des raccourcis dans les questions synthétiques, affaiblissant la surveillance pour la recherche visuelle et l'intégration des preuves. Nous proposons RACE, un cadre de synthèse des tâches conditionnée par l'image qui construit des dépendances d'images à des cibles et de preuves spécifiques à des réponses. L'abstraction de référence (RA) remplace le texte révélateur d'identité par des références visuelles spécifiques ou des références relationnelles ambiguës résolues par la recherche parmi les candidats représentés. Evidence compos

Machine translation (Meta M2M100); the English original follows.

Abstract

Training multimodal search agents requires tasks that make visually grounded search necessary for determining the answer. Textual identity clues and unused target evidence allow shortcuts in synthetic questions, weakening supervision for visual search and evidence integration. We propose RACE, an image-conditioned task synthesis framework that constructs dependencies from images to targets and from target-specific evidence to answers. Referential Abstraction (RA) replaces identity-revealing text with specific visual references or ambiguous relational references resolved by searching among depicted candidates. Compositional Evidence (CE) links retrieved facts about multiple entities in one image or entities from different images into joint questions whose answers require every designated contribution. Both operations also apply to images found during search, allowing new visual observations to determine subsequent search targets or supply answer evidence. Fine-tuning Qwen3.5-27B on only 12K teacher-generated trajectories from RACE raises average accuracy across ten multimodal benchmarks from 53.5 to 56.0, a relative improvement of approximately 4.7% over the matched base agent. Our code, data, and model will be publicly available.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Abstraction, Computer science, Task (project management), Artificial intelligence, Visual search, Natural language processing

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange