unlearning-cleanslate/generations-nemotron-nano-9b-v2-pre_val

该数据集是一个用于评估人工智能模型推理和思维链能力的集合，包含多个子集。主要子集包括：ARC挑战赛（ARC Challenge），这是一个多项选择题数据集，旨在测试AI的推理能力；以及多个BBH（Big-Bench Hard）思维链少样本任务子集，如布尔表达式、因果判断、日期理解、消歧问答、Dyck语言、形式谬误、几何形状、超序数、逻辑演绎（涉及三个、五个和七个对象）、电影推荐、多步算术、导航、对象计数、表格中的企鹅、关于彩色物体的推理和毁坏名称等。这些子集涵盖了各种复杂任务，要求模型进行逐步推理。数据集仅包含训练分割，每个子集有不同数量的示例，并包含输入、目标、生成参数、模型响应和评估指标等特征。

This dataset is a collection for evaluating AI model reasoning and chain-of-thought capabilities, comprising multiple subsets. Key subsets include: ARC Challenge, a multiple-choice question dataset designed to test AI reasoning; and several BBH (Big-Bench Hard) chain-of-thought few-shot task subsets, such as boolean expressions, causal judgement, date understanding, disambiguation QA, Dyck languages, formal fallacies, geometric shapes, hyperbaton, logical deduction (involving three, five, and seven objects), movie recommendation, multistep arithmetic, navigate, object counting, penguins in a table, reasoning about colored objects, and ruin names. These subsets cover a variety of complex tasks requiring step-by-step reasoning. The dataset includes only training splits, with varying numbers of examples per subset, and features such as input, target, generation arguments, model responses, and evaluation metrics.