Com$^2$

该数据集是一个专注于利用因果事件图和多种大型语言模型进行复杂常识推理的基准测试。它包含了不同复杂度的决策、转换、反事实和干预等任务，以评估复杂常识推理的能力。

This dataset is a benchmark focused on leveraging causal event graphs and multiple large language models (LLMs) for complex commonsense reasoning. It includes tasks of varying complexities, such as decision-making, transformation, counterfactual, and intervention tasks, to evaluate complex commonsense reasoning capabilities.