citrinegui/countdown_n4t100_1-100

这是一个包含数学表达式推理步骤的数据集。数据集的特征字段包括目标值(target)，数字序列(nums)，表达式(expression)，推理步骤(reasoning_steps)，数据来源(data_source)，能力(ability)，奖励模型(reward_model)，以及额外信息(extra_info)。奖励模型包含地面真实值(ground_truth)和风格(style)，地面真实值又包括表达式(expression)，数字(numbers)和目标值(target)。额外信息包括索引(index)和数据集划分(split)。数据集分为训练集(train)和测试集(test)，其中训练集包含327680个样本，测试集包含1024个样本。

This is a dataset containing mathematical expression reasoning steps. The dataset features include target, nums, expression, reasoning_steps, data_source, ability, reward_model, and extra_info. The reward_model includes ground_truth and style, where ground_truth consists of expression, numbers, and target. Extra_info includes index and split. The dataset is split into training set (train) and test set (test), with the training set containing 327680 samples and the test set containing 1024 samples.