Block-based diffusion generation of counterfactual image-text sets, combined with a set-aware loss, improves CLIP's compositional reasoning over several benchmarks, but the paper overstates one benchmark result and ships no code or data.
Vismin: Visual minimal-change understanding
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Visual Leap in CLIP Compositionality Reasoning through Generation of Counterfactual Sets
Block-based diffusion generation of counterfactual image-text sets, combined with a set-aware loss, improves CLIP's compositional reasoning over several benchmarks, but the paper overstates one benchmark result and ships no code or data.