A hybrid cross-modal attention model using CLIP and BGE-M3 improves hate detection F1-macro by 5.9% over text-only baselines on Nepali memes while revealing failures of English-centric vision models and ensembles on small Devanagari data.
In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1994-2003
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MEME-Fusion@CHiPSAL 2026: Multimodal Ablation Study of Hate Detection and Sentiment Analysis on Nepali Memes
A hybrid cross-modal attention model using CLIP and BGE-M3 improves hate detection F1-macro by 5.9% over text-only baselines on Nepali memes while revealing failures of English-centric vision models and ensembles on small Devanagari data.