analiza aiqbench
This study demonstrates how inference engineering techniques—such as direct candidate scoring from VLM logits, deterministic decoding, and targeted post-processing—can significantly boost the reliability of vision-language models without requiring task-specific training. By prioritizing robust output control over model selection alone, this approach achieves top-tier performance in multilingual visual reasoning benchmarks while maintaining strict formatting constraints essential for real-world applications.
FAU at ImageCLEF 2026 Task on Multimodal Reasoning Robust Candidate Scoring and Concise Multilingual Visual Answering
arXiv:2608.01664v1 Announce Type: cross Abstract: We present our ImageCLEF 2026 Multimodal Reasoning system for the Visual Multiple Choice Question Answering (Visual MCQ) and Visual Open Question Answering (Visual OpenQA) subtasks. The challenge requires reliable reasoning over multilingual educational and scientific images with dense text, diagrams, charts, tables, formulas, and units, while enforcing strict answer formats. Our central finding is that robust output control is as important as model choice. For Visual MCQ, we replace fragile free-form generation with direct candidate label scoring from vision-language model logits, then combine complementary runs through score fusion and voting. For Visual OpenQA, we use image enhancement, concise final answer prompting, deterministic decoding, and targeted post-processing to remove reasoning traces and formatting artifacts. Without task-specific model training, our official submissions achieved third place in Visual MCQ with 0.7108 accuracy and first place in Visual OpenQA with 0.6488 COMET, 0.1391 BLEU, 0.2762 ROUGE L, and 0.2383 METEOR. The results highlight the practical value of inference engineering: careful scoring, ensembling, prompting, and cleanup can turn strong VLMs into reliable competition systems.
Abstract:We present our ImageCLEF 2026 Multimodal Reasoning system for the Visual Multiple Choice Question Answering (Visual MCQ) and Visual Open Question Answering (Visual OpenQA) subtasks. The challenge requires reliable reasoning over multilingual educational and scientific images with dense text, diagrams, charts, tables, formulas, and units, while enforcing strict answer formats. Our central finding is that robust output control is as important as model choice. For Visual MCQ, we replace fragile free-form generation with direct candidate label scoring from vision-language model logits, then combine complementary runs through score fusion and voting. For Visual OpenQA, we use image enhancement, concise final answer prompting, deterministic decoding, and targeted post-processing to remove reasoning traces and formatting artifacts. Without task-specific model training, our official submissions achieved third place in Visual MCQ with 0.7108 accuracy and first place in Visual OpenQA with 0.6488 COMET, 0.1391 BLEU, 0.2762 ROUGE L, and 0.2383 METEOR. The results highlight the practical value of inference engineering: careful scoring, ensembling, prompting, and cleanup can turn strong VLMs into reliable competition systems.
16 pages, 3 figures, 7 tables. CLEF 2026 Working Notes, ImageCLEF 2026 Multimodal Reasoning Task
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2608.01664 [cs.CV]
(or arXiv:2608.01664v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.01664 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 284 cuvinte
Subiecte conexe
Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering
arXiv · 22 sept. 2026
2026 BAIR Graduate Showcase
BAIR Blog · 1 iul. 2026
[AINews] High Taste Testers Ep. 1
Latent Space · 17 oct. 2026
[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model
Latent Space · 13 oct. 2026