PuSH - Publikationsserver des Helmholtz Zentrums München

Schulze Buschoff, L.M. ; Voudouris, K. ; Akata, E. ; Bethge, M.* ; Tenenbaum, J.B.* ; Schulz, E.

Testing the Limits of Fine-Tuning for Improving Visual Cognition in Vision Language Models.

In: (42nd International Conference on Machine Learning, ICML 2025, 13-19 July 2025, Vancouver). 2025. 53645-53662 (Proceedings of Machine Learning Research ; 267)
Verlagsversion
Open Access Hybrid
Pre-trained vision language models still fall short of human visual cognition. In an effort to improve visual cognition and align models with human behavior, we introduce visual stimuli and human judgments on visual cognition tasks, allowing us to systematically evaluate performance across cognitive domains under a consistent environment. We fine-tune models on ground truth data for intuitive physics and causal reasoning and find that this improves model performance in the respective fine-tuning domain. Furthermore, it can improve model alignment with human behavior. However, we find that task-specific fine-tuning does not contribute to robust human-like generalization to data with other visual characteristics or to tasks in other cognitive domains.
Weitere Metriken?
[➜Einloggen]
Zusatzinfos bearbeiten [➜Einloggen]
Publikationstyp Artikel: Konferenzbeitrag
Konferenztitel 42nd International Conference on Machine Learning, ICML 2025
Konferzenzdatum 13-19 July 2025
Konferenzort Vancouver
Quellenangaben Band: 267, Heft: , Seiten: 53645-53662 Artikelnummer: , Supplement: ,
Institut(e) Institute of AI for Health (AIH)