Encyclo-K is a statement-based benchmark that rethinks benchmark construction from the ground up. Our key observation is that the question itself need not be the atomic unit of curation—individual knowledge statements can be.
Dynamic Evaluation: We extract standalone knowledge statements from authoritative textbooks and dynamically compose them into evaluation… See the full description on the dataset page:
https://huggingface.co/datasets/yimingliang/Encyclo-K.
