Leveraging LLM-generated structural prior for causal inference with concurrent causes

In CALM @ NeurIPS, 2024