Uselis, Arnas

6 publications

ICLR 2026 CLIP Behaves like a Bag-of-Words Model Cross-Modally but Not Uni-Modally Darina Koishigarina, Arnas Uselis, Seong Joon Oh

ICLRW 2025 Beyond Decodability: Linear Feature Spaces Enable Visual Compositional Generalization Arnas Uselis, Andrea Dittadi, Seong Joon Oh

NeurIPS 2025 Diffusion Classifiers Understand Compositionality, but Conditions Apply Yujin Jeong, Arnas Uselis, Seong Joon Oh, Anna Rohrbach

ICML 2025 Does Data Scaling Lead to Visual Compositional Generalization? Arnas Uselis, Andrea Dittadi, Seong Joon Oh

ICLR 2025 Intermediate Layer Classifiers for OOD Generalization Arnas Uselis, Seong Joon Oh

NeurIPS 2025 On the Rankability of Visual Embeddings Ankit Sonthalia, Arnas Uselis, Seong Joon Oh