Rethinking Misalignment in Vision-Language Model Adaptation from a Causal Perspective

VLM
Re-assign community