A training-time pipeline that synthesizes diverse negation captions from batch neighbors improves CLIP's negation accuracy on matching and generation benchmarks, and a new NEG-TTOI benchmark measures negation handling in text-to-image generation.
Title resolution pending
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
-
TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP
A training-time pipeline that synthesizes diverse negation captions from batch neighbors improves CLIP's negation accuracy on matching and generation benchmarks, and a new NEG-TTOI benchmark measures negation handling in text-to-image generation.