← all datasets

B-XAIC

Emerging
1papers using it
2025first seen

B-XAIC is a benchmark constructed from real-world molecular data and diverse tasks that includes known ground-truth rationales for assigned labels, used to evaluate the faithfulness of Explainable AI methods for Graph Neural Networks in cheminformatics and drug discovery.

🔎 Find this dataset

Papers using B-XAIC (1)

B-XAIC — datasets — graph-learning