Chain Necklace and Controlled Threat: A Pilot Similarity Test in CLIP (ViT-B/32)
Abstract
This project reports a small pilot similarity study using CLIP (ViT-B/32). Using 12 images and 8 text prompts, image-text and image-image cosine similarities were computed in the CLIP embedding space. The project shares stimuli links, similarity outputs, and figures. No human data or brain data were used.
Fields of Study & Subjects
Multimodal Machine Learning Applications
Image Retrieval and Classification Techniques
Topic Modeling
Publication details
- Publication date
- 1 January 2026
- Document type
- Preprint