Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models

DSpace Repository

Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models

Author: Hein, Matthias; Croce, Francesco; Singh, Naman Deep; Schlarmann, Christian
Tübinger Autor(en):
Hein, Matthias
Croce, Francesco
Singh, Naman D.
Issue year: 2024-02-19
Verlagsangabe: arXiv
Language: English
Full text: https://doi.org/10.48550/arXiv.2402.12336
DDC Classifikation: 004 - Data processing and computer science
Dokumentart: Preprint
Show full item record

This item appears in the following Collection(s)