Dataset Viewer
Auto-converted to Parquet Duplicate
premise
stringlengths
14
81
hypothesis
stringlengths
12
49
label
stringclasses
3 values
config
stringclasses
1 value
The playwright is ambitious.
The playwright is ambitious.
entailment
aug_train
The playwright is ambitious.
The poet is not knowledgeable.
neutral
aug_train
The playwright is ambitious.
The playwright is not ambitious.
contradiction
aug_train
The walkman is expensive.
The walkman is expensive.
entailment
aug_train
The walkman is expensive.
The curtain is expensive.
neutral
aug_train
The walkman is expensive.
The walkman is not expensive.
contradiction
aug_train
The toothbrush is precious.
The toothbrush is precious.
entailment
aug_train
The toothbrush is precious.
The toothbrush is expensive.
neutral
aug_train
The toothbrush is precious.
The toothbrush is not precious.
contradiction
aug_train
Eric is dependable.
Eric is dependable.
entailment
aug_train
Eric is dependable.
Darcy is dependable.
neutral
aug_train
Eric is dependable.
Eric is not dependable.
contradiction
aug_train
The grapefruit is skinned.
The grapefruit is skinned.
entailment
aug_train
The grapefruit is skinned.
The pea is not canned.
neutral
aug_train
The grapefruit is skinned.
The grapefruit is not skinned.
contradiction
aug_train
The lady is patient.
The lady is patient.
entailment
aug_train
The lady is patient.
The lady is energetic.
neutral
aug_train
The lady is patient.
The lady is not patient.
contradiction
aug_train
Phoebe is mysterious.
Phoebe is mysterious.
entailment
aug_train
Phoebe is mysterious.
Brian is not gentle.
neutral
aug_train
Phoebe is mysterious.
Phoebe is not mysterious.
contradiction
aug_train
The chemist is blond.
The chemist is blond.
entailment
aug_train
The chemist is blond.
The kid is humorous.
neutral
aug_train
The chemist is blond.
The chemist is not blond.
contradiction
aug_train
The porter is patient.
The porter is patient.
entailment
aug_train
The porter is patient.
The philosopher is not patient.
neutral
aug_train
The porter is patient.
The porter is not patient.
contradiction
aug_train
Jessica is humpbacked.
Jessica is humpbacked.
entailment
aug_train
Jessica is humpbacked.
Jessica is not fit.
neutral
aug_train
Jessica is humpbacked.
Jessica is not humpbacked.
contradiction
aug_train
The celery is raw.
The celery is raw.
entailment
aug_train
The celery is raw.
The pomegranate is pulpy.
neutral
aug_train
The celery is raw.
The celery is not raw.
contradiction
aug_train
The peach is nutritious.
The peach is nutritious.
entailment
aug_train
The peach is nutritious.
The kumquat is not salable.
neutral
aug_train
The peach is nutritious.
The peach is not nutritious.
contradiction
aug_train
The melon is soft.
The melon is soft.
entailment
aug_train
The melon is soft.
The melon is withered.
neutral
aug_train
The melon is soft.
The melon is not soft.
contradiction
aug_train
The ball is classic.
The ball is classic.
entailment
aug_train
The ball is classic.
The kite is classic.
neutral
aug_train
The ball is classic.
The ball is not classic.
contradiction
aug_train
Chris is famous.
Chris is famous.
entailment
aug_train
Chris is famous.
Susan is famous.
neutral
aug_train
Chris is famous.
Chris is not famous.
contradiction
aug_train
The businessman is ambitious.
The businessman is ambitious.
entailment
aug_train
The businessman is ambitious.
The firefighter is not fashionable.
neutral
aug_train
The businessman is ambitious.
The businessman is not ambitious.
contradiction
aug_train
The shorts is durable.
The shorts is durable.
entailment
aug_train
The shorts is durable.
The shelf is not precious.
neutral
aug_train
The shorts is durable.
The shorts is not durable.
contradiction
aug_train
The coconut is plump.
The coconut is plump.
entailment
aug_train
The coconut is plump.
The pitaya is seedless.
neutral
aug_train
The coconut is plump.
The coconut is not plump.
contradiction
aug_train
The cabbage is fresh.
The cabbage is fresh.
entailment
aug_train
The cabbage is fresh.
The cabbage is not delicious.
neutral
aug_train
The cabbage is fresh.
The cabbage is not fresh.
contradiction
aug_train
The zoologist is blond.
The zoologist is blond.
entailment
aug_train
The zoologist is blond.
The zoologist is active.
neutral
aug_train
The zoologist is blond.
The zoologist is not blond.
contradiction
aug_train
The bat is clumsy.
The bat is clumsy.
entailment
aug_train
The bat is clumsy.
The penguin is clumsy.
neutral
aug_train
The bat is clumsy.
The bat is not clumsy.
contradiction
aug_train
Angelia is embarrassed.
Angelia is embarrassed.
entailment
aug_train
Angelia is embarrassed.
Francis is not embarrassed.
neutral
aug_train
Angelia is embarrassed.
Angelia is not embarrassed.
contradiction
aug_train
The firefly is migratory.
The firefly is migratory.
entailment
aug_train
The firefly is migratory.
The crane is migratory.
neutral
aug_train
The firefly is migratory.
The firefly is not migratory.
contradiction
aug_train
The mathematician is calculating.
The mathematician is calculating.
entailment
aug_train
The mathematician is calculating.
The professor is gentle.
neutral
aug_train
The mathematician is calculating.
The mathematician is not calculating.
contradiction
aug_train
The plum is delicious.
The plum is delicious.
entailment
aug_train
The plum is delicious.
The plum is not aromatic.
neutral
aug_train
The plum is delicious.
The plum is not delicious.
contradiction
aug_train
The photo is unique.
The photo is unique.
entailment
aug_train
The photo is unique.
The boat is weird.
neutral
aug_train
The photo is unique.
The photo is not unique.
contradiction
aug_train
The door is durable.
The door is durable.
entailment
aug_train
The door is durable.
The dish is oversized.
neutral
aug_train
The door is durable.
The door is not durable.
contradiction
aug_train
The persimmon is plump.
The persimmon is plump.
entailment
aug_train
The persimmon is plump.
The persimmon is organic.
neutral
aug_train
The persimmon is plump.
The persimmon is not plump.
contradiction
aug_train
The chemist is eloquent.
The chemist is eloquent.
entailment
aug_train
The chemist is eloquent.
The reporter is energetic.
neutral
aug_train
The chemist is eloquent.
The chemist is not eloquent.
contradiction
aug_train
The saleswoman is eccentric.
The saleswoman is eccentric.
entailment
aug_train
The saleswoman is eccentric.
The saleswoman is dependable.
neutral
aug_train
The saleswoman is eccentric.
The saleswoman is not eccentric.
contradiction
aug_train
The parrot is healthy.
The parrot is healthy.
entailment
aug_train
The parrot is healthy.
The sloth is healthy.
neutral
aug_train
The parrot is healthy.
The parrot is not healthy.
contradiction
aug_train
The jewelry is imported.
The jewelry is imported.
entailment
aug_train
The jewelry is imported.
The pot is not imported.
neutral
aug_train
The jewelry is imported.
The jewelry is not imported.
contradiction
aug_train
The pea is juicy.
The pea is juicy.
entailment
aug_train
The pea is juicy.
The radish is not sticky.
neutral
aug_train
The pea is juicy.
The pea is not juicy.
contradiction
aug_train
The boat is interesting.
The boat is interesting.
entailment
aug_train
End of preview. Expand in Data Studio

tasksource/simple_pair(複製)

データの内容

単純な文に対する NLI モデルの失敗を調べた Luo ら(2022)の研究のデータ。1 行は前提(premise)と仮説(hypothesis)の組で、正解が label に入り、config は設定名の列である。元データの注釈を決定的な変換で作り直したものである。

データの概要

項目 値
行数 230,000(1 構成・2 分割の合計)
列数 4
容量 4.75 MB(ファイル 4 本)
ファイルの形式 parquet 2
公開 public

構成と分割

構成(config) 分割(split)ごとの行数 列数
default train 6,000、test 224,000 4

列(構成 default・分割 train)

列 型 先頭の行の値
premise string The playwright is ambitious.
hypothesis string The playwright is ambitious.
label string entailment
config string aug_train

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/simple_pair をリビジョン 3c968fc6a34cb204f827a86504337b95a7a99e15 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

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。


データセットカード: "simple_pair"

@inproceedings{luo-etal-2022-simple-challenging,
    title = "Simple but Challenging: Natural Language Inference Models Fail on Simple Sentences",
    author = "Luo, Cheng  and
      Liu, Wei  and
      Lin, Jieyu  and
      Zou, Jiajie  and
      Xiang, Ming  and
      Ding, Nai",
    editor = "Goldberg, Yoav  and
      Kozareva, Zornitsa  and
      Zhang, Yue",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.findings-emnlp.252",
    doi = "10.18653/v1/2022.findings-emnlp.252",
    pages = "3449--3462",
}
原文(英語)

Dataset Card for "simple_pair"

@inproceedings{luo-etal-2022-simple-challenging,
    title = "Simple but Challenging: Natural Language Inference Models Fail on Simple Sentences",
    author = "Luo, Cheng  and
      Liu, Wei  and
      Lin, Jieyu  and
      Zou, Jiajie  and
      Xiang, Ming  and
      Ding, Nai",
    editor = "Goldberg, Yoav  and
      Kozareva, Zornitsa  and
      Zhang, Yue",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.findings-emnlp.252",
    doi = "10.18653/v1/2022.findings-emnlp.252",
    pages = "3449--3462",
}
Downloads last month
29