A direction associated with negation exists in CLIP embedding space and can be steered at test time via representation engineering to produce negation-aware outputs without fine-tuning.
Otter: A multi-modal model with in-context instruction tuning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 47:7543–7557,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
When Negation Is a Geometry Problem in Vision-Language Models
A direction associated with negation exists in CLIP embedding space and can be steered at test time via representation engineering to produce negation-aware outputs without fine-tuning.