Datasets:

HYUKI5
/

tasksource__cnli

premise stringlengths 9 285	hypothesis stringlengths 7 155	label stringclasses 3 values
A man is riding a red motorcycle with a small child sitting in front of him.	A man rides his motorcyle with his won.	neutral
A man is riding a motorcycle with a small child sitting in front of him.	A man rides his motorcyle with his won.	entailment
A man is riding a red motorcycle with a small child sitting in front of him.	A man rides his car with his son.	contradiction
A man is riding a blue truck with a small child sitting in front of him.	A man rides his motorcyle with his won.	contradiction
A man is riding a red motorcycle with a small child sitting in front of him.	A man rides his motorcycle with a child.	entailment
In a martial arts demonstration a man with two swords faces a woman with a Chinese spear.	The woman is watching.	neutral
In a martial arts demonstration a man with two swords faces a woman with a Chinese spear.	The woman is with a Chinese spear.	entailment
In a martial arts demonstration a man with two swords faces a woman with a Chinese spear while the men watch.	The woman is watching.	contradiction
In a martial arts demonstration a man with two swords faces a woman with a Chinese spear while another woman watches.	The woman is watching.	entailment
In a martial arts demonstration a man with two swords faces a woman with a Chinese spear.	The woman is watching the man with a Japanese spear.	contradiction
A woman wearing a green shirt stands on the railroad tracks.	The woman is wearing a green shirt.	entailment
A woman wearing a green shirt stands on the railroad tracks.	A railroad employee stands on the train tracks.	neutral
A railroad employee wearing a green shirt stands on the railroad tracks.	A railroad employee stands on the train tracks.	entailment
A woman wearing a green shirt stands on the railroad tracks and waves on the railroad employee in the booth.	A railroad employee stands on the train tracks.	contradiction
A woman wearing a green shirt stands on the railroad tracks.	The woman is not standing on the railroad tracks.	contradiction
A blond woman speaking to a brunette woman with her arms crossed.	A woman is talking to another woman.	entailment
A blond man speaking to a brunette woman with her arms crossed.	A woman is talking to another woman.	contradiction
A blond woman speaking to another person with their arms crossed.	A woman is talking to another woman.	neutral
A blond woman speaking to a brunette woman with her arms crossed.	A woman is talking to a family member.	neutral
A blond woman speaking to a brunette woman with her arms crossed.	A woman is talking to a man.	contradiction
Students are outside for a lecture hall.	Students are indoors.	contradiction
Students are inside of a lecture hall.	Students are on the soccer field.	contradiction
Students are inside of a lecture hall.	Students are inside of the science lecture hall.	neutral
Students are attending a lecture.	Students are indoors.	neutral
Students are inside of a lecture hall.	Students are indoors.	entailment
A woman in a red polka-dot dress sings into a mic.	A woman is wearing a dress.	entailment
A man in a pink polka-dot dress sings into a mic.	A woman is wearing a dress.	contradiction
A tall woman in a red polka-dot dress sings into a mic.	A woman is wearing a dress.	neutral
A woman in a red polka-dot dress sings into a mic.	A woman is wearing a silk dress.	neutral
A woman in a red polka-dot dress sings into a mic.	A woman is wearing pants.	contradiction
Men playing a game in the park.	Some men are playing a game.	entailment
Men playing a game in the park.	Some women are playing a game.	contradiction
Men playing a game at the park.	some men play for money	entailment
Men playing a game in the park.	some men play for money	neutral
All men playing a game at the park for free.	some men play for money	contradiction
Three children are playing together in some water.	Three smart children are playing in water.	neutral
Three children are playing together at the park.	Three children are playing in water.	contradiction
Three children are playing together in some water.	Four adults are playing in sand.	contradiction
Three children are playing together in some water.	Three children are playing in water.	entailment
Children are playing together in some water.	Three children are playing in water.	neutral
A woman and a young child are making sculptures out of clay.	A child and woman with clay.	entailment
A woman and her son are painting on a canvas	A man and woman painting on canvas.	entailment
A woman and a young child are making sculptures out of clay.	A boy and woman make something.	neutral
A woman and a young child are doing art.	A man and woman painting on canvas.	neutral
A woman and a young child are making sculptures out of clay.	A man and woman painting on canvas.	contradiction
A man not wearing a shirt is pouring something out a can while sitting on sand in front of a orange and red tent.	A man is pouring orange juice outside a tent.	neutral
A man not wearing a shirt is pouring something out a can in front of a tent.	A man is walking down the trail.	neutral
A man not wearing a shirt is pouring something out a can while sitting on sand in front of a tent.	A man is walking down the trail.	entailment
A man not wearing a shirt is pouring something out a can while sitting on sand in front of a orange and red tent.	A man is walking down the trail.	contradiction
A man not wearing a shirt is pouring something out a can while sitting on sand in front of a orange and red tent.	A man is pouring outside a tent.	entailment
A football player wearing a number ten jersey is catching a football with both hands.	A sports player is catching a baseball.	contradiction
A football player wearing a number ten jersey is catching a football with one hand.	A huge player	contradiction
A giant player wearing a number ten jersey is catching a football with one hand.	A huge player	entailment
A football player wearing a number ten jersey is catching a football with both hands.	A sports player is catching a football.	entailment
A football player wearing a number ten jersey is catching a football with both hands.	A huge player	neutral
The back of a woman wearing blue jeans, black boots, and a blue and black shirt.	The back of someone in boots.	entailment
The back of a woman wearing blue jeans, and a shirt.	Someone in sandals walks away.	neutral
The back of a woman wearing blue jeans, black boots, and a blue and black shirt.	Someone walks away.	neutral
The back of a woman wearing blue jeans, sandals, and a shirt.	Someone in sandals walks away.	entailment
The back of a woman wearing blue jeans, black boots, and a blue and black shirt.	Someone in sandals walks away.	contradiction
A few kids wearing white and red are playing soccer.	Children are wearing their soccer uniform	entailment
Some kids wearing white and red are playing soccer.	Children are wearing their soccer uniform	neutral
Some kids wearing white and red are playing soccer.	Children are wearing their white and red clothes	entailment
Some kids wearing white and red are playing soccer.	Children are wearing their black uniform	contradiction
A few kids wearing white and red are playing rugby.	Children are wearing their soccer uniform	contradiction
A young lady is dressed in a blue top with black leotards and is in a dance pose.	The young lady is taking a break from exercising	contradiction
A young lady is dressed in a blue top with black leotards and is in a dance pose.	The young lady is about to dance.	entailment
A young lady is dressed in a top with black leotards and is in a dance pose after being exhausted from weightlifting.	The young lady is taking a break from exercising	entailment
A young lady is dressed in a top with black leotards.	The young lady is taking a break from exercising	neutral
A young lady is dressed in a blue top with black leotards and is in a dance pose.	The young lady is a professional dancer.	neutral
Four people trudge through the snow wearing snowshoes and using walking sticks.	Three crash survivors trying to reach civilization	contradiction
Three people trudge through the snow wearing snowshoes and using walking sticks.	Three people are in snow	entailment
Three people trudge through the snow wearing snowshoes and using walking sticks.	Six people are in mud	contradiction
Three people trudge through the snow to reach the city after emerging from a crashed airplane.	Three crash survivors trying to reach civilization	entailment
Three people trudge through the snow wearing snowshoes and using walking sticks.	Three crash survivors trying to reach civilization	neutral
Recked vehicle being towed by a large tow truck.	The firetruck was towed.	neutral
Recked ambulance being towed by a large tow truck.	The firetruck was towed.	contradiction
Recked ambulance being towed by a large tow truck.	The white ambulance was towed.	neutral
Recked ambulance being towed by a large tow truck.	The ambulance was towed.	entailment
Recked firetruck being towed by a large tow truck.	The firetruck was towed.	entailment
A man in a blue shirt is building a sand castle at the beach.	A man in a blue shirt is building a sand castle at the beach	entailment
A man in a blue shirt is at the beach.	A man in a blue shirt is building a sand castle at the beach	neutral
A man in a blue shirt is building a bank at the beach.	A man in a blue shirt is building a sand castle at the beach	contradiction
A man in a blue shirt is building a sand castle at the beach.	A man in a blue shirt is building nothing at the beach	contradiction
A man in a blue shirt is building a sand castle at the beach.	A man in a blue shirt is being forced to build a sand castle at the beach	neutral
An older man and woman sitting on a cement bench with frowns on their faces.	The man and woman are frowning.	entailment
An older man and woman sitting on a cement bench with frowns on their faces.	The man and woman are smiling.	contradiction
An older man and woman sitting on a cement bench with frowns on their faces.	The man and woman are arguing about money	neutral
An older man and woman sitting on a cement bench with frowns on their faces are arguing about their kids.	The man and woman are arguing about money	contradiction
An older man and woman sitting on a cement bench with frowns on their faces are arguing about money.	The man and woman are arguing about money	entailment
Two horse racing jockeys, one in checkered blue and red and the other in orange and brown, are racing against a blurry background.	Two horse racing jockeys are racing against a blurry background.	entailment
Two horse racing jockeys, one in checkered blue and red and the other in orange and brown, are racing against a blurry background, while two rodeo clowns are riding old sad-looking ponies.	Two rodeo clowns riding old sad-looking ponies	entailment
Two horse racing jockeys, one in checkered blue and red and the other in orange and brown, are racing against a blurry background, while two people were riding ponies near them.	Two rodeo clowns riding old sad-looking ponies	neutral
Two horse racing jockeys, one in checkered blue and red and the other in orange and brown, are racing against a blurry background.	Several horse racing jockeys are racing against a blurry background	neutral
Two horse racing jockeys, one in checkered blue and red and the other in orange and brown, are racing against a blurry background.	Two rodeo clowns riding old sad-looking ponies	contradiction
Several man in black tops sit silently.	People are playing music.	contradiction
Several man in black tops play the guitar.	People are playing music.	entailment
Several man in black tops play the guitar.	People are playing rock music.	neutral
Several man in black tops play the guitar.	People are playing baseball.	contradiction
Several man in black tops have guitars.	People are playing music.	neutral

End of preview. Expand in Data Studio

tasksource/cnli（複製）

データの内容

SNLI を反事実的に拡張した自然言語推論（NLI）のデータ。1 行が前提（premise）と仮説（hypothesis）の組で、label に推論の関係が入る。元の論文は Huang ら（2020）で、反事実的に拡張した学習データは、拡張していないデータより汎化が良くなるわけではないことを扱う。

データの概要

項目	値
行数	8,330（1 構成・1 分割の合計）
列数	3
容量	1.36 MB（ファイル 3 本）
ファイルの形式	jsonl 1
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` 8,330	3

列（構成 `default`・分割 `train`）

列	型	先頭の行の値
`premise`	`string`	`A man is riding a red motorcycle with a small ch…`
`hypothesis`	`string`	`A man rides his motorcyle with his won.`
`label`	`string`	`neutral`

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/cnli をリビジョン 4cee00e6022028f7fb3a903ffdeac426d3c221dd 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

元のデータセット: https://huggingface.co/datasets/tasksource/cnli
複製したリビジョン: 4cee00e6022028f7fb3a903ffdeac426d3c221dd（2023-04-07 09:09:37+00:00）
元のライセンス表記: other

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。

Counterfactually-Augmented NLI データの一般化

@inproceedings{huang2020cnligeneralization,
 title={Counterfactually-Augmented {SNLI} Training Data Does Not Yield Better Generalization Than Unaugmented Data},
 author={William Huang and Haokun Liu and Samuel R. Bowman},
 booktitle = {Proceedings of the 2020 EMNLP Workshop on Insights from Negative Results in NLP},
 year={2020},
 publisher = {The Association for Computational Linguistics}
}

原文（英語）

Generalization of Counterfactually-Augmented NLI Data

@inproceedings{huang2020cnligeneralization,
 title={Counterfactually-Augmented {SNLI} Training Data Does Not Yield Better Generalization Than Unaugmented Data},
 author={William Huang and Haokun Liu and Samuel R. Bowman},
 booktitle = {Proceedings of the 2020 EMNLP Workshop on Insights from Negative Results in NLP},
 year={2020},
 publisher = {The Association for Computational Linguistics}
}