HYUKI5's picture
カードを組み立て直す(訳文と書式の修正を反映)
d4f7f54 verified
|
Raw History Blame Contribute Delete
6.16 kB
metadata
license: other
pretty_name: tasksource/ScienceQA_text_only (mirror)
source_datasets:
  - tasksource/ScienceQA_text_only
tags:
  - laya
  - laya-finetune
  - mirror
  - domain-science
  - domain-education
  - use-unclear
  - genre-exam
  - label-choice
  - annot-natural
  - lang-en
language: en
dataset_info:
  features:
    - name: question
      dtype: string
    - name: choices
      sequence: string
    - name: answer
      dtype: int8
    - name: hint
      dtype: string
    - name: task
      dtype: string
    - name: grade
      dtype: string
    - name: subject
      dtype: string
    - name: topic
      dtype: string
    - name: category
      dtype: string
    - name: skill
      dtype: string
    - name: lecture
      dtype: string
    - name: solution
      dtype: string
  splits:
    - name: train
      num_bytes: 8105771.787521609
      num_examples: 6508
    - name: validation
      num_bytes: 2638142.7097382694
      num_examples: 2144
    - name: test
      num_bytes: 2757852.295213393
      num_examples: 2224
  download_size: 2925662
  dataset_size: 13501766.792473271

tasksource/ScienceQA_text_only(複製)

データの内容

ScienceQA の問題のうち、もとから画像を使わない例だけを集めたデータ。テキストだけのモデルでも解けるとされる。1 行が 1 問で、question に問い、choices に選択肢の一覧、answer に正解の番号、hint に補足が入る。台帳では元の注釈を決まった手順で変換して作られたとされる。

データの概要

項目 値
行数 10,876(1 構成・3 分割の合計)
列数 12
容量 2.93 MB(ファイル 5 本)
ファイルの形式 parquet 3
言語 en
公開 public

構成と分割

構成(config) 分割(split)ごとの行数 列数
default train 6,508、validation 2,144、test 2,224 12

列(構成 default・分割 train)

列 型 先頭の行の値
question string Which tense does the sentence use? Mona will pri…
choices list<string> [3 件]
answer int8 1
hint string
task string closed choice
grade string grade2
subject string language science
topic string verbs
category string Verb tense
skill string Is the sentence in the past, present, or future …
lecture string Present tense verbs tell you about something tha…
solution string The sentence is in future tense. You can tell be…

行数と列は HF のデータセットビューアの集計(元のデータセットの現在の版で数えたもの)。全データセットの一覧は HYUKI5/dataset-catalog にある。

複製について

tasksource/ScienceQA_text_only をリビジョン 37ca6a0e9f35a7002e46afe5c2bbc047c7c3598f の時点でそのまま複製したもの。中身は変えていない。判断モデル Laya(convaiinnovations/laya)の微調整に使う候補として、調査メモ(Laya 学習用データセット調査)に載せたデータを集めている。

出典

論文

  • 元のカードに論文へのリンクは見つからなかった

アノテーションの方法

元のカードにアノテーションの節が無かった。下の「調査メモの記述」と「元のカード」を参照。

微調整に使うときの注意

  • 調査メモの台帳での区分: D(記載なし・other・unknown・資料間の食い違い)。区分は記載の文言で機械的に付けたもので、法的な判定ではない。学習に入れる前に元のカードとライセンスの原文で確かめる。
  • Laya の入力は 2,048 token まで(ModernBERT-large / mmBERT)。長い本文は先頭から黙って切られるので、切れる行は事前に除くか、要約・切り出しをしてから入れる。
  • 分割は記事・文書単位で行い、同じ本文から作った別の問いを学習側と較正側に分けない。温度と閾値は学習に使っていない分割で別々に当てる。

調査メモの記述

メモ 節 ライセンス(記載) 確認 注記
jev_sources_tasksource_bekko.csv CSV(tasksource-jev-typed-decisions) unspecified / license_use=unspecified 確認済 license_use=unspecified / [Jev 元データ source] source_annotation_converted(deterministic recast; no model r

元のカード(日本語訳)

以下は元のデータセットのカードの本文を日本語に訳したもの(機械翻訳)。訳に疑問があるときは、末尾の原文を確かめる。


"scienceQA_text_only" のデータセットカード

ScienceQA のテキストのみの例(もとは画像がなかった例で、テキストだけのモデルでも解けるはずのもの)。

@article{10.1007/s00799-022-00329-y,
author = {Saikh, Tanik and Ghosal, Tirthankar and Mittal, Amish and Ekbal, Asif and Bhattacharyya, Pushpak},
title = {ScienceQA: A Novel Resource for Question Answering on Scholarly Articles},
year = {2022},
journal = {Int. J. Digit. Libr.},
month = {sep}
}
原文(英語)

Dataset Card for "scienceQA_text_only"

ScienceQA text-only examples (examples where no image was initially present, which means they should be doable with text-only models.)

@article{10.1007/s00799-022-00329-y,
author = {Saikh, Tanik and Ghosal, Tirthankar and Mittal, Amish and Ekbal, Asif and Bhattacharyya, Pushpak},
title = {ScienceQA: A Novel Resource for Question Answering on Scholarly Articles},
year = {2022},
journal = {Int. J. Digit. Libr.},
month = {sep}
}