Closed lizaitang closed 1 year ago
Hi,
We have recently extended our attacks to Instruct-BLIP, LLaVA (vicuna based) and LLaVA (llama2-based). Similar results are also observed. We would update our paper as well as the github repository in our next iteration. Please stay tuned!
We will also present this paper in ICML AdvML-Frontiers workshop on July 28. If you are attending, would be happy to chat more as well.
Best
Dear Author, Your paper really helps, have you ever considering the difficulty of extending the work to other VLM model such as LLava?