VIPER exposes Functional Fusion in dynamic prompt architectures, enabling a backdoor that resists pruning by tightly integrating attack and utility parameters in the same high-magnitude core.
Describing textures in the wild
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
representative citing papers
CLIP-Inspector reconstructs OOD triggers to detect backdoors in prompt-tuned CLIP models with 94% accuracy and higher AUROC than baselines, plus a repair step via fine-tuning.
Dual-modality anchors from text descriptions and test-time image statistics filter views and ensemble predictions to improve test-time prompt tuning, achieving SOTA on 15 datasets.
SigLino distills SigLIP2 and DINOv3 into efficient vision models via asymmetric relation-knowledge distillation, token-balanced batching, and hierarchical data sampling on a new 200M-image corpus, yielding better transfer to grounding VLMs than training from scratch.
citing papers explorer
-
Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures
VIPER exposes Functional Fusion in dynamic prompt architectures, enabling a backdoor that resists pruning by tightly integrating attack and utility parameters in the same high-magnitude core.
-
CLIP-Inspector: Model-Level Backdoor Detection for Prompt-Tuned CLIP via OOD Trigger Inversion
CLIP-Inspector reconstructs OOD triggers to detect backdoors in prompt-tuned CLIP models with 94% accuracy and higher AUROC than baselines, plus a repair step via fine-tuning.
-
Dual-Modality Anchor-Guided Filtering for Test-time Prompt Tuning
Dual-modality anchors from text descriptions and test-time image statistics filter views and ensemble predictions to improve test-time prompt tuning, achieving SOTA on 15 datasets.
-
SigLino: Efficient Multi-Teacher Distillation for Agglomerative Vision Foundation Models
SigLino distills SigLIP2 and DINOv3 into efficient vision models via asymmetric relation-knowledge distillation, token-balanced batching, and hierarchical data sampling on a new 200M-image corpus, yielding better transfer to grounding VLMs than training from scratch.