遇见数据集

nataliaElv/similarity-qa-with-vectors

收藏
Hugging Face2023-11-10 更新2024-03-04 收录
官方服务:

资源简介:

该数据集similarity-qa-with-vectors使用Argilla创建,并与HuggingFace的`datasets`库兼容。它包括Argilla的配置文件、数据集记录和注释指南。数据集结构包含字段、问题、建议、元数据、向量和指南。字段包括文本输入和输出,问题设计用于注释者评分和解释记录的质量,向量是可选的浮点数列,具有预定义的维度。根据其配置,该数据集可用于各种NLP任务。

The dataset similarity-qa-with-vectors is developed using Argilla and is compatible with HuggingFace's `datasets` library. It includes Argilla configuration files, dataset records, and annotation guidelines. The dataset structure consists of fields, prompts, suggestions, metadata, vectors, and guidelines. The fields include text inputs and outputs. The prompts are designed for annotators to score and explain the quality of records. Vectors are optional floating-point columns with predefined dimensions. Depending on its configuration, this dataset can be applied to various NLP tasks.

提供机构:
nataliaElv
原始信息汇总

数据集卡片 for similarity-qa-with-vectors

数据集描述

  • 数据集概述
    • 该数据集包含一个符合Argilla数据集格式的配置文件argilla.yaml,用于在使用FeedbackDataset.from_huggingface方法时配置数据集。
    • 数据集记录采用与HuggingFace datasets兼容的格式,可以通过datasets库的load_dataset方法独立加载。
    • 包含用于构建和整理数据集的标注指南(如果已在Argilla中定义)。

加载数据集

使用Argilla加载

python import argilla as rg

ds = rg.FeedbackDataset.from_huggingface("nataliaElv/similarity-qa-with-vectors")

使用datasets库加载

python from datasets import load_dataset

ds = load_dataset("nataliaElv/similarity-qa-with-vectors")

支持的任务和排行榜

  • 该数据集可以包含多个字段、问题和响应,因此可以用于不同的NLP任务,具体取决于配置。
  • 数据集结构在数据集结构部分中描述。
  • 没有与该数据集关联的排行榜。

数据集结构

数据在Argilla中的结构

  • 字段(Fields)

    • 数据集记录本身,目前仅支持文本字段。这些字段将用于提供对问题的响应。
    • 示例字段包括:
      • instruction(指令):文本类型,必需
      • input(输入):文本类型,非必需
      • output(输出):文本类型,必需
  • 问题(Questions)

    • 向标注者提出的问题。问题类型包括评分、文本、标签选择、多标签选择或排序。
    • 示例问题包括:
      • quality(记录质量评分):评分类型,必需,值为[1, 2, 3, 4, 5]
      • explanation(评分解释):文本类型,必需
  • 建议(Suggestions)

    • 人类或机器生成的推荐,用于辅助标注者在标注过程中的选择。
  • 元数据(Metadata)

    • 提供关于数据集记录的额外信息,如原始来源链接或记录的作者、日期和来源。
  • 向量(Vectors)

    • 包含浮点数的不同列,维度由数据集配置文件中的vectors_settings预定义。
    • 示例向量包括:
      • input(输入):维度为[1, 384]
      • instruction(指令):维度为[1, 384]
      • output(输出):维度为[1, 384]
      • testing(测试):维度为[1, 1]
  • 指南(Guidelines)

    • 提供给标注者的指令,可选。

数据实例

一个数据集实例在Argilla中的示例如下:

json { "external_id": null, "fields": { "input": "", "instruction": "Give three tips for staying healthy.", "output": "1. Eat a balanced diet and make sure to include plenty of fruits and vegetables. 2. Exercise regularly to keep your body active and strong. 3. Get enough sleep and maintain a consistent sleep schedule." }, "metadata": { "text_length": 241 }, "responses": [], "suggestions": [], "vectors": { "input": [ -0.025378959253430367, -0.005421411711722612, -0.005123426206409931, -0.015000881627202034, -0.010828345082700253, 0.011933867819607258, 0.019314972683787346, 0.040846794843673706, -0.009248972870409489, 0.015658004209399223, 0.0018413026118651032, -0.04884575679898262, 0.007001905702054501, 0.03489101678133011, 0.035010259598493576, 0.004000979475677013, 0.03179853782057762, 0.013713518157601357, -0.01575734093785286, 0.016500428318977356, 0.02162296697497368, -0.019962908700108528, 0.011788141913712025, -0.018135597929358482, 0.00479349447414279, 0.027265621349215508, -0.00592863280326128, -0.00819356832653284, -0.04846194013953209, -0.19176225364208221, -0.033277515321969986, -0.013714526779949665, 0.0032154761720448732, -0.009890320710837841, -0.010387021116912365, -0.009758984670042992, -0.01616772636771202, 0.013864913955330849, -0.010939724743366241, 0.04058735817670822, 0.021671248599886894, 0.01383791770786047, -0.01536033395677805, -0.010618588887155056, 0.005697894841432571, -0.02265983633697033, -0.016780417412519455, -0.006693877745419741, 0.05799293890595436, -0.006326382048428059, 0.002093177754431963, 0.010354680009186268, 0.0006329257157631218, 0.027090711519122124, 0.004488569684326649, 0.014552658423781395, 0.0180455781519413, 0.019452394917607307, 0.02411177195608616, 0.008954178541898727, 0.0015302742831408978, 0.029447568580508232, -0.16580072045326233, 0.02812054567039013, 0.009662247262895107, 0.009475956670939922, 0.013372445479035378, -0.016405431553721428, -0.001572685199789703, 0.051213230937719345, 0.003518211655318737, 0.015949634835124016, -0.0069265239872038364, 0.027317708358168602, 0.019327018409967422, -0.022707704454660416, 0.028689151629805565, -0.01890380308032036, -0.01167482603341341, 0.011035646311938763, 0.0040340544655919075, -0.012239952571690083, -0.006184910889714956, -0.005307812709361315, -0.03035779856145382, -0.041286271065473557, 0.010543900541961193, 0.014870839193463326, 0.00642419932410121, 0.01750650443136692, -0.024431902915239334, -0.0055658514611423016, 0.02791532501578331, 0.007770954631268978, -0.06280053406953812, -0.011230005882680416, 0.022709796205163002, 0.0036207374650985003, -0.032403528690338135, 0.7040055990219116, -0.018570110201835632, 0.00400574691593647, 0.03399886190891266, -0.049098845571279526, 0.0239898469299078, -0.01194965373724699, -0.018013538792729378, -0.012237226590514183, -0.008749520406126976, 0.0011163142044097185, 0.025379084050655365, -0.009777436032891273, 0.04108814150094986, -0.005716001149266958, 0.006996306125074625, 0.01101826224476099, 0.043749451637268066, 0.025922292843461037, -0.006995497737079859, -0.031284742057323456, -0.03961759805679321, 0.024092240259051323, -0.0037946782540529966, -0.016933923587203026, 0.009725619107484818, -0.09440258890390396, 0.008375165052711964, 0.04419294372200966, 0.01720806583762169, 0.025360679253935814, 0.024841418489813805, -0.037821535021066666, -0.002577421488240361, -0.008712586015462875, 0.007797832600772381, -0.0038116704672574997, 0.019269822165369987, -0.0267858728766441

二维码
社区交流群
二维码
科研交流群
商业服务