InductionBench

该数据集是一个旨在评估大型语言模型（LLM）归纳推理能力的基准测试。它包含了在多个LLM和无-shot思维链提示上的评估，同时采用了精确度、召回率和兼容性等指标。为了实验设置，生成了10个函数，以通过生成函数和性能指标来评估LLM的归纳推理能力。

This dataset is a benchmark designed to evaluate the inductive reasoning capabilities of Large Language Models (LLMs). It includes evaluations across multiple LLMs and zero-shot Chain-of-Thought (CoT) prompting settings, adopting metrics such as precision, recall, and compatibility. For the experimental setup, 10 functions were generated to assess the inductive reasoning abilities of LLMs using these generated functions and the aforementioned performance metrics.