LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Models | Read Paper on Bytez