From Weights to Concepts: Data-Free Interpretability of CLIP via Singular Vector Decomposition

Documenti analoghi

• AL-GTD: Deep Active Learning for Gaze Target Detection
  di: Tonini, Francesco, et al.
  Pubblicazione: (2024)
• Retrieval-enriched zero-shot image classification in low-resource domains
  di: Dall'Asen, Nicola, et al.
  Pubblicazione: (2024)
• Training-Free Semantic Multi-Object Tracking with Vision-Language Models
  di: Bonat, Laurence, et al.
  Pubblicazione: (2026)
• Dynamic Scoring with Enhanced Semantics for Training-Free Human-Object Interaction Detection
  di: Tonini, Francesco, et al.
  Pubblicazione: (2025)
• Adv-KD: Adversarial Knowledge Distillation for Faster Diffusion Sampling
  di: Mekonnen, Kidist Amde, et al.
  Pubblicazione: (2024)