The benchmark proposed in paper: GraphInstruct: Empowering Large Language Models with Graph Understanding and Reasoning Capability - View it on GitHub