Safeguarding Data in Multimodal AI: A Differentially Private Approach to CLIP Training

Open in new window