Empirical study shows bidirectional but sensitive relationship between compositionality and long-caption understanding in VLMs, promoted by high-quality grounded data and affected by architectural choices like frozen positional embeddings.
When are Lemons Purple? The Concept Association Bias of CLIP
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2representative citing papers
citing papers explorer
-
Long Story Short: Disentangling Compositionality and Long-Caption Understanding in Contrastive VLMs
Empirical study shows bidirectional but sensitive relationship between compositionality and long-caption understanding in VLMs, promoted by high-quality grounded data and affected by architectural choices like frozen positional embeddings.
- SRL-CLIP: Efficient CLIP Video Adaptation via Structured Semantic Role Labels