mlsanityData to 14 Oct 2026

HANCLIP: A Family of Hyperbolic Angular Negation Vision Language Models

arXiv 2606.23843 ↗First posted 22 Jun 2026cs.CV cs.IRcited by 0

On the map: Multimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP

Abstract (arXiv)

Vision-language models (VLMs) achieve strong cross-modal alignment but remain brittle to negation, often relying on shallow word associations rather than compositional reasoning. Fine-tuning on negation-specific data can also compromise their general purpose capabilities through catastrophic forgetting. We introduce HANCLIP (Hyperbolic, Angular, and Negation), a geometry-aware framework that improves negation sensitivity while preserving the structure of the pretrained joint embedding space. HANCLIP combines a hyperbolic contrastive objective, which models hierarchical relations and semantic asymmetries, with an angular triplet loss that separates negated descriptions from their affirmative counterparts. Using only 20,000 image-text quadruplets, HANCLIP consistently improves performance across CLIP, LongCLIP, and SmartCLIP backbones on the NegBench benchmark, while maintaining or improving zero-shot classification and image-text retrieval performance. These results show that lightweight, geometry-guided objectives can enhance negation understanding without large-scale retraining.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 6 of its 55 references that are in the map. First posted on arXiv 22 Jun 2026.

Builds on

6 of 55 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Negation-Aware Test-Time Adaptation for Vision-Language Models25 Jul

Other references · 5

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2017Decoupled Weight Decay Regularizationcited by 1,831
  3. 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,188
  4. 2025PHyCLIP: ℓ1-Product of Hyperbolic Factors Unifies Hierarchy and Compositionality in Vision-Language Representation Learningcited by 7
  5. 2025On the Brittleness of CLIP Text Encoderscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026When Negation Is a Geometry Problem in Vision-Language Models
  2. 2025SpaceVLM: Sub-Space Modeling of Negation in Vision-Language Models
  3. 2024Learn "No" to Say "Yes" Better: Improving Vision-Language Models via Negations
  4. 2025Vision-Language Models Do Not Understand Negation
  5. 2026Not Just What's There: Enabling CLIP to Comprehend Negated Visual Descriptions Without Fine-tuning
  6. 2026Omni-NegCLIP: Enhancing CLIP with Front-Layer Contrastive Fine-Tuning for Comprehensive Negation Understanding
  7. 2025TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP
  8. 2025What "Not" to Detect: Negation-Aware VLMs via Structured Reasoning and Token Merging