ERUnderstand: A Benchmark for Evaluating Vision-Language Models on Entity-Relationship Diagrams
Date:
I had the pleasure of presenting my research paper, “ERUnderstand,” which is currently under review. The talk focused on evaluating and improving how Vision-Language Models (VLMs) and LLMs reason over structured Entity-Relationship diagrams.
It was a fantastic session with the UPenn Database Group, filled with insightful questions and great discussions on the future of multimodal AI in scientific contexts.