Get started
Author

Arnas Uselis

Observed name: Uselis, Arnas
OpenAlex ID: A5127837635
CLIP Behaves like a Bag-of-Words Model Cross-modally but not Uni-modally
2025 · Koishigarina, Darina, Uselis, Arnas, et al. · arXiv (Cornell University) · 22
PDF
Does Data Scaling Lead to Visual Compositional Generalization?
2025 · Uselis, Arnas, Dittadi, Andrea, et al. · arXiv (Cornell University) · 8
PDF
Intermediate Layer Classifiers for OOD generalization
2025 · Uselis, Arnas, Oh, Seong Joon · arXiv (Cornell University) · 8
PDF
Diffusion Classifiers Understand Compositionality, but Conditions Apply
2025 · Yujin Jeong, Arnas Uselis, et al. · neural information processing systems · 6
PDF
MEME: Multi-entity & Evolving Memory Evaluation
2026 · Jung, Seokwon, Rubinstein, Alexander, et al. · arXiv (Cornell University) · 5
PDF
Compositional Generalization Requires Linear, Orthogonal Representations in Vision Embedding Models
2026 · Uselis, Arnas, Dittadi, Andrea, et al. · arXiv (Cornell University) · 5
PDF
Sparse Autoencoders enable Robust and Interpretable Fine-tuning of CLIP models
2026 · Morelli, Fabian, Uselis, Arnas, et al. · arXiv (Cornell University) · 2
PDF
On the rankability of visual embeddings
2025 · Ankit Sonthalia, Arnas Uselis, et al. · neural information processing systems · 2
PDF
Half-Truths Break Similarity-Based Retrieval
2026 · Kargi, Bora, Uselis, Arnas, et al. · arXiv (Cornell University) · 1
PDF
Do vision models know that three is three?
2026 · Sonthalia, Ankit, Uselis, Arnas, et al. · arXiv (Cornell University)
PDF
Cross-Pollination Retrieval: Finding Transferable Ideas Across Fields in Scientific Literature
2026 · Jung, Seokwon, Kim, Yerang, et al. · arXiv (Cornell University)
PDF
How can embedding models bind concepts?
2026 · Uselis, Arnas, Koishigarina, Darina, et al. · arXiv (Cornell University)
PDF
When Do Diffusion Models learn to Generate Multiple Objects?
2026 · Jeong, Yujin, Uselis, Arnas, et al. · arXiv (Cornell University)
PDF
The Fragmented Linear Representation Hypothesis for Vision Embeddings
2026 · Sonthalia, Ankit, Uselis, Arnas, et al. · KAIST Institutional Repository (KAIST)
Necessary Conditions for Compositional Generalization of Embedding Models
2026 · Uselis, Arnas, Dittadi, Andrea, et al. · KAIST Institutional Repository (KAIST)
SAE-FT: Robust and Interpretable Fine-tuning of CLIP models via Sparse Autoencoders
2026 · Morelli, Fabian, Uselis, Arnas, et al. · KAIST Institutional Repository (KAIST)
Works count: 16
Citation count: 59
H-index: 5
I10-index: 1
Year
Topic
Type