Layer-wise Alignment: Examining Safety Alignment Across Image Encoder Layers in Vision Language Models

Re-assign community