gyung/RevThink-financial-mmlu-ko

RevThink-financial-mmlu-ko数据集是基于allganize/financial-mmlu-ko数据集，通过ChatGPT和GPT-4o生成的，采用了Reverse Thinking Makes LLMs Stronger Reasoners论文中的方法进行增强。该数据集包含原始问题、原始答案、正向推理、反向问题和反向推理等特征，旨在通过逆向思维增强大语言模型的推理能力。数据集的预处理部分提供了如何加载和格式化数据集的代码示例。

This dataset is an enhanced version of the [allganize/financial-mmlu-ko](https://huggingface.co/datasets/allganize/financial-mmlu-ko) dataset, augmented using the method described in the paper Reverse Thinking Makes LLMs Stronger Reasoners. The dataset includes features such as original question, original answer, forward reasoning, backward question, and backward reasoning. The purpose of the dataset is to enhance reasoning capabilities in the financial domain by generating prompts for forward and backward reasoning. The preprocessing steps include loading the dataset, defining a format template, and applying a formatting function.