ruggsea/stanford-encyclopedia-of-philosophy_instruct

这是一个半合成的指令数据集，旨在监督微调大型语言模型，以正式方式回答哲学问题。数据集基于斯坦福哲学百科全书（SEP）。每篇文章被细分为多个部分，每个部分用于生成一个问题-答案对，通过提示模型编写可以由每个子部分回答的问题。子部分的标记数过高（>2000）或过低（<100）的都被丢弃。数据集总共包含12260个示例。

原始信息汇总

数据集概述

基本信息

• 语言: 英语
• 数据规模: 10K<n<100K
• 任务类别:
  • 文本到文本生成
  • 文本生成
  • 问答
• 数据集名称: Stanford Encyclopedia of Technology instruct

配置

• 配置名称: default
• 数据文件:
  • 分割: train
  • 路径: data/train-*

数据集信息

• 特征:
  • 名称: question
    • 数据类型: string
  • 名称: answer
    • 数据类型: string
• 分割:
  • 名称: train
    • 字节数: 30581169
    • 样本数: 11904
• 下载大小: 15654471
• 数据集大小: 30581169

描述

该数据集是一个半合成指令数据集，旨在用于监督大型语言模型在正式方式下回答哲学问题的微调任务。数据集基于斯坦福哲学百科全书（SEP）。每篇文章被细分为多个部分，每个部分用于生成一个问题-答案对，通过提示模型编写一个可以由每个子部分回答的问题。子部分中，令牌数过高（>2000）或过低（<100）的被丢弃。数据集总共包含12260个样本。

格式

数据集包含两列：question 和 answer。question 列包含模型生成的问题，answer 列包含问题所生成的子部分文本。

示例

json { "question": "What is the significance and role of the debate over adaptationism in biology and philosophy?", "answer": "5. Are constraint hypotheses genuine rivals to adaptive hypotheses?

Why should biologists and philosophers care about the debate over adaptationism? This debate has been and will continue to be important to biologists because it helps clarify how to do better evolutionary biology. Despite this, a minority of currently-active biologists contribute to the debate about adaptationism, and many ignore it or disparage it. For a variety of reasons, many of the important points that have arisen from the debate, such as how to structure empirical tests of adaptive hypotheses and optimality models, go unheeded. For example, claims for local optimality continue to appear that lack a specification of even a single alternative hypothesis to the adaptive hypothesis. This lack inevitably leaves ambiguous the status of the adaptive hypothesis. In addition, standards for rejection of the (only) adaptive hypothesis continue to be rarely delineated if at all.

The debate over adaptationism will continue to be important to philosophers of biology because much will be gained by a more informed understanding of the use of hypotheses about natural selection by biologists. In the 1970s and 1980s, there was a general tendency to misunderstand the conceptual commitments of users of optimality models. More recently, there has been a growing understanding of the variety of conceptual commitments held by users of such models, especially that the use and development of an optimality model need not bind the user to a blind acceptance of the claim that natural selection is all powerful." }