kenhktsui/minimath

`minimath`数据集的目的是评估语言模型在数学能力方面的表现。该数据集通过从多个数学相关数据集中采样构建，包括math_dataset、math_qa、competition_math、stackexchange_math_jsonl和grade-school-math-instructions。数据集包含多个特征，如问题、答案、来源、推理、注释公式和线性公式等。数据集分为训练集，包含2880个样本，总大小为1114848字节。

The `minimath` dataset is designed to evaluate the mathematical capabilities of language models. It is constructed by sampling from multiple math-related datasets, including math_dataset, math_qa, competition_math, stackexchange_math_jsonl, and grade-school-math-instructions. The dataset includes various features such as question, answer, source, reasoning, annotated formula, linear formula, among others. It is divided into a training set containing 2880 samples with a total size of 1,114,848 bytes.

原始信息汇总

数据集概述

数据集名称

• 名称: minimath

数据集特征

• 特征列表:
  • question: 数据类型为字符串
  • answer: 数据类型为字符串
  • source: 数据类型为字符串
  • Rationale: 数据类型为字符串
  • annotated_formula: 数据类型为字符串
  • linear_formula: 数据类型为字符串

数据集分割

• 训练集:
  • num_examples: 2880
  • num_bytes: 1114848

数据集大小

• 下载大小: 543796
• 数据集大小: 1114848