TLDR AI · RESEARCH & INNOVATION
A WATERMARKING METHOD FOR PROTECTING VISION-LANGUAGE MODELS
Summary from the TLDR AI newsletter
VLPMarker, a new watermarking method, protects Embedding as a Service Vision-Language Pre-Trained Models from extraction attacks.