pltops/evosynth-gsm8k-smoke-cerebras-gsm8k

EvoSynth是一个基于gsm8k数据集通过进化提示生成的合成数学应用题数据集。该数据集采用遗传算法循环生成，包括初始种群、多代适应性选择、交叉和变异等步骤。数据集包含70行数据，每行数据包含问题文本、完整解答和生成代数等信息。数据集的语言为英语，适用于问答任务，标签包括synthetic、evolutionary和mathematical-reasoning。

EvoSynth is a synthetic math word problem dataset generated from gsm8k using evolutionary prompting. The dataset is produced through a genetic algorithm loop involving initial population, fitness-guided selection, crossover, and mutation over multiple generations. It contains 70 rows of data, each including the question text, full worked solution, and generation number. The dataset is in English, suitable for question-answering tasks, and tagged as synthetic, evolutionary, and mathematical-reasoning.