mean_human_score
float32
1
5
binary_label
stringclasses
3 values
sentence_id
stringlengths
10
18
adjective
stringlengths
3
15
noun
stringlengths
3
12
premise
stringlengths
10
369
hypothesis
stringlengths
15
378
label
class label
0
1
3.33333
UNK
gutenberg_1252
wild
creature
My proud assurance was all wasted upon the creature .
My proud assurance was all wasted upon the wild creature .
0not-entailed
3.66667
YES
gutenberg_1253
wild
creature
The expectation of the creature waited for the manifestation.
The expectation of the wild creature waited for the manifestation.
1entailed
4.66667
YES
gutenberg_2804
human
character
The quarter-deck is perhaps the severest test of character known .
The quarter-deck is perhaps the severest test of human character known .
1entailed
3.66667
YES
gutenberg_2803
human
character
Can you control your character .
Can you control your human character .
1entailed
3
UNK
imagecaptions_439
dry
grass
Brown and white dog playing in the grass .
Brown and white dog playing in the dry grass .
0not-entailed
3
UNK
imagecaptions_440
dry
grass
A tan dog plays with a tennis ball while rolling around in the grass .
A tan dog plays with a tennis ball while rolling around in the dry grass .
0not-entailed
3
UNK
imagecaptions_1068
blue
tube
A young boy smiles while holding a tube of toothpaste.
A young boy smiles while holding a blue tube of toothpaste.
0not-entailed
3
UNK
imagecaptions_1069
blue
tube
The beautiful young girl goes through the tube .
The beautiful young girl goes through the blue tube .
0not-entailed
3
UNK
imagecaptions_1070
blue
tube
A rollerblader skating inside a tube .
A rollerblader skating inside a blue tube .
0not-entailed
3.33333
UNK
gutenberg_2996
long
talk
That talk had doneDiongood.
That long talk had doneDiongood.
0not-entailed
3
UNK
argumentation_83
american
society
However, it is almost necessary for a good job in today 's society .
However, it is almost necessary for a good job in today 's american society .
0not-entailed
3.66667
YES
argumentation_81
american
society
Lotteries are bad for a society .
Lotteries are bad for an american society .
1entailed
3.66667
YES
argumentation_82
american
society
Uniforms tell kids that society is a lot more important than their individual feelings.
Uniforms tell kids that american society is a lot more important than their individual feelings.
1entailed
2.33333
NO
gigaword_2116
international
experience
Japan 's experience also seems to argue for spending heavily to promote social development.
Japan 's international experience also seems to argue for spending heavily to promote social development.
0not-entailed
3
UNK
gigaword_2117
international
experience
So learn a few lessons from this fool 's experience .
So learn a few lessons from this fool 's international experience .
0not-entailed
2.66667
UNK
argumentation_1367
social
class
Parents who slag off the school and get private tutors actually undermine the learning their student does in class .
Parents who slag off the school and get private tutors actually undermine the learning their student does in social class .
0not-entailed
4.66667
YES
argumentation_1366
social
class
Most people die in the class to which they were born.
Most people die in the social class to which they were born.
1entailed
3
UNK
argumentation_1365
social
class
They brought in a retired teacher to take over the class .
They brought in a retired teacher to take over the social class .
0not-entailed
3.33333
UNK
argumentation_2861
true
statement
According to a statement President Bush made today, they're investigating the incident.
According to a true statement President Bush made today, they're investigating the incident.
0not-entailed
3.33333
UNK
argumentation_2860
true
statement
Ok you guys took my statement the wrong way.
Ok you guys took my true statement the wrong way.
0not-entailed
3
UNK
gutenberg_1042
quiet
talk
And the talk wandered away into a sort of paradoxical metaphysics.
And the quiet talk wandered away into a sort of paradoxical metaphysics.
0not-entailed
3
UNK
gutenberg_1041
quiet
talk
I tell you the truth, having heard all their talk .
I tell you the truth, having heard all their quiet talk .
0not-entailed
3
UNK
gutenberg_1043
quiet
talk
Walthamhad floated into a stream of talk .
Walthamhad floated into a stream of quiet talk .
0not-entailed
4.33333
YES
argumentation_2742
personal
definition
By my definition, by the objective definition; you are a racist.
By my personal definition, by the objective definition; you are a racist.
1entailed
2.33333
NO
argumentation_2743
personal
definition
What you have written is the definition of prejudice.
What you have written is the personal definition of prejudice.
0not-entailed
3
UNK
argumentation_2883
educational
system
The system has , as I think you are implying, broken down.
The educational system has , as I think you are implying, broken down.
0not-entailed
3
UNK
argumentation_2885
educational
system
Perhaps NY 's system is the exception instead of the rule.
Perhaps NY 's educational system is the exception instead of the rule.
0not-entailed
2.66667
UNK
argumentation_2884
educational
system
I think you are trapped in the system .
I think you are trapped in the educational system .
0not-entailed
4.33333
YES
gutenberg_1836
own
appearance
Jack 's appearance on the scene had setMrs.
Jack 's own appearance on the scene had setMrs.
1entailed
5
YES
gutenberg_1838
own
appearance
His appearance disarmed her.
His own appearance disarmed her.
1entailed
3
UNK
gigaword_371
annual
production
A production of the Educational Film Center.
A annual production of the Educational Film Center.
0not-entailed
2.33333
NO
gigaword_370
annual
production
In all, only 3,564 Scamps and Rampages were sold before production halted .
In all, only 3,564 Scamps and Rampages were sold before annual production halted .
0not-entailed
4.33333
YES
gigaword_2566
new
record
After setting the record, the train gradually slowed.
After setting the new record, the train gradually slowed.
1entailed
4
YES
argumentation_2731
significant
period
During period between 1975-2000 the annual GDP growth in South Africa was- 0,7%-LRB- that` s a negative-RRB-.
During significant period between 1975-2000 the annual GDP growth in South Africa was- 0,7%-LRB- that` s a negative-RRB-.
1entailed
4
YES
argumentation_2730
significant
period
Within this period of time the Ottoman and British Empires fell and the American Empire was born.
Within this significant period of time the Ottoman and British Empires fell and the American Empire was born.
1entailed
3.66667
YES
gigaword_1185
new
production
A movie about Larson starring Bill Murray begins production this December.
A movie about Larson starring Bill Murray begins new production this December.
1entailed
3
UNK
imagecaptions_865
wet
rock
A man in a blue jacket is climbing on a rock in the woods.
A man in a blue jacket is climbing on a wet rock in the woods.
0not-entailed
3.33333
UNK
imagecaptions_866
wet
rock
A woman wearing a red backpack is sitting on a rock overlooking the water.
A woman wearing a red backpack is sitting on a wet rock overlooking the water.
0not-entailed
2.33333
NO
gigaword_2891
controversial
plan
Do this by clicking on the name of a plan that interests you.
Do this by clicking on the name of a controversial plan that interests you.
0not-entailed
3.66667
YES
gigaword_2890
controversial
plan
But the Serbs refuse to discuss the plan unless the peacekeepers stay.
But the Serbs refuse to discuss the controversial plan unless the peacekeepers stay.
1entailed
1.66667
NO
gigaword_328
jewish
center
Southern Afghanistan is the center of the Taliban-led insurgency.
Southern Afghanistan is the jewish center of the Taliban-led insurgency.
0not-entailed
3
UNK
gigaword_327
jewish
center
His will be Manor House, right in the center of the village.
His will be Manor House, right in the jewish center of the village.
0not-entailed
3
UNK
imagecaptions_1578
rocky
beach
Dirty water near beach with large white cruiseship in background.
Dirty water near rocky beach with large white cruiseship in background.
0not-entailed
3
UNK
imagecaptions_1579
rocky
beach
Children playing on the beach .
Children playing on the rocky beach .
0not-entailed
2.66667
UNK
imagecaptions_1580
rocky
beach
A man is standing in front of 4 dogs on the beach .
A man is standing in front of 4 dogs on the rocky beach .
0not-entailed
4.66667
YES
gigaword_209
personal
history
He has a humble background and a history as an outspoken populist.
He has a humble background and a personal history as an outspoken populist.
1entailed
3
UNK
gigaword_2366
royal
family
He and his family became U.S. citizens.
He and his royal family became U.S. citizens.
0not-entailed
3
UNK
imagecaptions_573
green
chair
A young boy in a red jacker stands on a pier with a fishing pole and chair near him.
A young boy in a red jacker stands on a pier with a fishing pole and green chair near him.
0not-entailed
3
UNK
imagecaptions_575
green
chair
Man reading in a chair outside Man relaxing in a folding chair on the street.
Man reading in a green chair outside Man relaxing in a folding chair on the street.
0not-entailed
3
UNK
gigaword_40
american
music
And my music keeps me focused.
And my american music keeps me focused.
0not-entailed
3
UNK
gigaword_41
american
music
And the music is enchanting.
And the american music is enchanting.
0not-entailed
2.66667
UNK
argumentation_2352
illegal
act
Pulling the troops would be just the opportunity they need; an act that they interpret as weakness.
Pulling the troops would be just the opportunity they need; an illegal act that they interpret as weakness.
0not-entailed
3
UNK
argumentation_2354
illegal
act
Evidenced by the act of sending unarmed teens into the fury of machine guns of Hussein's army.
Evidenced by the illegal act of sending unarmed teens into the fury of machine guns of Hussein's army.
0not-entailed
3.33333
UNK
gutenberg_1941
peculiar
expression
And there was an expression about his beak which I distrusted from the first.
And there was a peculiar expression about his beak which I distrusted from the first.
0not-entailed
2.66667
UNK
gutenberg_1943
peculiar
expression
There was added dignity as well as appeal in his attitude and expression .
There was added dignity as well as appeal in his attitude and peculiar expression .
0not-entailed
3
UNK
imagecaptions_929
personal
tent
Three men posing in a tent .
Three men posing in a personal tent .
0not-entailed
3.66667
YES
imagecaptions_928
personal
tent
A hiker in a tent on a mountain.
A hiker in a personal tent on a mountain.
1entailed
3
UNK
imagecaptions_1794
blue
fence
The dog is on the fence The dog is playing in the water.
The dog is on the blue fence The dog is playing in the water.
0not-entailed
3
UNK
imagecaptions_1795
blue
fence
Two children look down squeezing through a fence .
Two children look down squeezing through a blue fence .
0not-entailed
2.66667
UNK
gigaword_45
special
meeting
The participants of the meeting have made a decision`` to restore normalcy.
The participants of the special meeting have made a decision`` to restore normalcy.
0not-entailed
3
UNK
gigaword_47
special
meeting
The meeting is still going on at press time.
The special meeting is still going on at press time.
0not-entailed
3.66667
YES
argumentation_523
formal
education
People come to their opinions through upbringing, religion , education and life experience.
People come to their opinions through upbringing, religion , formal education and life experience.
1entailed
3
UNK
argumentation_522
formal
education
The Taliban ended, for all practical purposes , education for girls.
The Taliban ended, for all practical purposes , formal education for girls.
0not-entailed
4.66667
YES
gutenberg_1061
single
word
Time and again he repeated the word, going closer and closer.
Time and again he repeated the single word, going closer and closer.
1entailed
5
YES
gutenberg_1060
single
word
At that word he rebelled.
At that single word he rebelled.
1entailed
3.33333
UNK
imagecaptions_1668
sandy
track
A dirt bike rider riding around a track .
A dirt bike rider riding around a sandy track .
0not-entailed
3
UNK
imagecaptions_1669
sandy
track
A man in red shorts runs on a track .
A man in red shorts runs on a sandy track .
0not-entailed
3
UNK
imagecaptions_1670
sandy
track
Jockeys on horses race around a track .
Jockeys on horses race around a sandy track .
0not-entailed
3.33333
UNK
imagecaptions_140
snowy
bench
Woman sitting alone with drink on a bench as people walk by.
Woman sitting alone with drink on a snowy bench as people walk by.
0not-entailed
2.66667
UNK
imagecaptions_138
snowy
bench
A family sits on a bench near a beach.
A family sits on a snowy bench near a beach.
0not-entailed
3
UNK
imagecaptions_139
snowy
bench
The man and woman sitting on a bench are kissing.
The man and woman sitting on a snowy bench are kissing.
0not-entailed
2.66667
UNK
gutenberg_122
little
thing
Now that the thing had come, he was glad to face it.
Now that the little thing had come, he was glad to face it.
0not-entailed
3.33333
UNK
gutenberg_2764
small
boy
It is seldom a boy of your age exhibits such good judgment and discretion.
It is seldom a small boy of your age exhibits such good judgment and discretion.
0not-entailed
3.33333
UNK
gutenberg_2763
small
boy
The boy stood first on one foot and then on the other.
The small boy stood first on one foot and then on the other.
0not-entailed
3.66667
YES
imagecaptions_1514
small
boy
A boy leaps off his skateboard over some steps near a house.
A small boy leaps off his skateboard over some steps near a house.
1entailed
3.66667
YES
gutenberg_1965
young
child
I hope that child will .
I hope that young child will .
1entailed
3.33333
UNK
imagecaptions_1213
young
child
A girl rides a unicycle with a child who rides a scooter.
A girl rides a unicycle with a young child who rides a scooter.
0not-entailed
3.33333
UNK
gigaword_1251
new
building
More than 2,000 employees work in the building and will return in phases.
More than 2,000 employees work in the new building and will return in phases.
0not-entailed
3
UNK
gigaword_1252
new
building
Devananda himself was slightly injured after escaping from the building .
Devananda himself was slightly injured after escaping from the new building .
0not-entailed
3
UNK
gigaword_1253
new
building
LeFevre was in his upstairs office when the building started to shake.
LeFevre was in his upstairs office when the new building started to shake.
0not-entailed
3
UNK
imagecaptions_1057
orange
vest
Three dogs wearing vest are running on the green grass.
Three dogs wearing orange vest are running on the green grass.
0not-entailed
3
UNK
imagecaptions_1056
orange
vest
A shirtless man wearing a vest walks on a stage with his arms up.
A shirtless man wearing an orange vest walks on a stage with his arms up.
0not-entailed
2.33333
NO
argumentation_1512
hostile
environment
What we in the west are is solid evidence that responsible anti-pollution policies will allow us to live lives of luxury without trashing our environment or destroying our health.
What we in the west are is solid evidence that responsible anti-pollution policies will allow us to live lives of luxury without trashing our hostile environment or destroying our health.
0not-entailed
2.33333
NO
argumentation_1513
hostile
environment
Every single one of us grew up in an environment that taught us that cows, pigs, and chickens are food.
Every single one of us grew up in a hostile environment that taught us that cows, pigs, and chickens are food.
0not-entailed
4
YES
gutenberg_1376
last
word
She turned without a word and went up the steep incline.
She turned without a last word and went up the steep incline.
1entailed
5
YES
gigaword_2069
own
record
Since then, their record is 29-25.
Since then, their own record is 29-25.
1entailed
3.33333
UNK
imagecaptions_2247
rocky
trail
The hiker is following a trail uphill .
The hiker is following a rocky trail uphill .
0not-entailed
3.33333
UNK
imagecaptions_2249
rocky
trail
A woman rides a bike on a trail through a field.
A woman rides a bike on a rocky trail through a field.
0not-entailed
3.33333
UNK
imagecaptions_2248
rocky
trail
A man in a black helmet riding his bike on a trail .
A man in a black helmet riding his bike on a rocky trail .
0not-entailed
3
UNK
imagecaptions_600
brown
rock
A rock climber practices on a rock climbing wall.
A rock climber practices on a brown rock climbing wall.
0not-entailed
3.33333
UNK
imagecaptions_601
brown
rock
A man wearing a wet suit is climbing a rock along a beach.
A man wearing a wet suit is climbing a brown rock along a beach.
0not-entailed
3.33333
UNK
gutenberg_1715
choppy
sea
We have done it at sea and on land.
We have done it at choppy sea and on land.
0not-entailed
3.66667
YES
gutenberg_1714
choppy
sea
TheFrenchseldom ever ventured in the sea or the stream or to the reef.
TheFrenchseldom ever ventured in the choppy sea or the stream or to the reef.
1entailed
3
UNK
gutenberg_1713
choppy
sea
At the bottom of the sea-- with all the other navies.
At the bottom of the choppy sea-- with all the other navies.
0not-entailed
3
UNK
gigaword_753
easy
victory
Golkar has claimed victory in the April 5 parliamentary election.
Golkar has claimed easy victory in the April 5 parliamentary election.
0not-entailed
3
UNK
gigaword_621
black
woman
The woman, 67, later hanged herself in the barn.
The black woman, 67, later hanged herself in the barn.
0not-entailed
3
UNK
gigaword_622
black
woman
A woman was among the dead and three women were also wounded.
A black woman was among the dead and three women were also wounded.
0not-entailed
2.66667
UNK
gigaword_2155
federal
building
They're the real neighbors who live in and around the building .
They're the real neighbors who live in and around the federal building .
0not-entailed
3.33333
UNK
gigaword_1994
special
unit
The unit has suffered some of the highest casualties of the Iraq conflict.
The special unit has suffered some of the highest casualties of the Iraq conflict.
0not-entailed
3
UNK
gigaword_1992
special
unit
Sales in the unit fell 18 percent to 73.9 million from 90.5 million.
Sales in the special unit fell 18 percent to 73.9 million from 90.5 million.
0not-entailed
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

Overview

Original data available here.

Dataset curation

premise and hypothesis columns have been cleaned following common practices (1, 2), that is

  • remove HTML tags <b>, <u>, </b>, </u>
  • normalize repeated white spaces
  • strip

mean_human_score has been transformed into class labels following common practices (1, 2), that is

  • for test set: mean_human_score <= 3 -> "not-entailed" and mean_human_score >= 4 -> "entailed" (anything between 3 and 4 has been removed)
  • for all other splits: mean_human_score < 3.5 -> "not-entailed" else "entailed"

more details below.

Code to generate the dataset

import pandas as pd
from datasets import Features, Value, ClassLabel, Dataset, DatasetDict


def convert_label(score, is_test):
  if is_test:
    if score <= 3:
      return "not-entailed"
    elif score >= 4:
      return "entailed"
    return "REMOVE"

  if score < 3.5:
      return "not-entailed"
  return "entailed"


ds = {}
for split in ("dev", "test", "train"):

    # read data
    df = pd.read_csv(f"<path to folder>/AN-composition/addone-entailment/splits/data.{split}", sep="\t", header=None)
    df.columns = ["mean_human_score", "binary_label", "sentence_id", "adjective", "noun", "premise", "hypothesis"]

    # clean text from html tags and useless spaces
    for col in ("premise", "hypothesis"):
        df[col] = (
            df[col]
            .str.replace("(<b>)|(<u>)|(</b>)|(</u>)", " ", regex=True)
            .str.replace(" {2,}", " ", regex=True)
            .str.strip()
        )

    # encode labels
    if split == "test":
        df["label"] = df["mean_human_score"].map(lambda x: convert_label(x, True))
        df = df.loc[df["label"] != "REMOVE"]
    else:
        df["label"] = df["mean_human_score"].map(lambda x: convert_label(x, False))
    assert df["label"].isna().sum() == 0

    df["label"] = df["label"].map({"not-entailed": 0, "entailed": 1})

    # cast to dataset
    features = Features({
        "mean_human_score": Value(dtype="float32"), 
        "binary_label": Value(dtype="string"), 
        "sentence_id": Value(dtype="string"), 
        "adjective": Value(dtype="string"), 
        "noun": Value(dtype="string"), 
        "premise": Value(dtype="string"), 
        "hypothesis": Value(dtype="string"),
        "label": ClassLabel(num_classes=2, names=["not-entailed", "entailed"]),
    })    
    ds[split] = Dataset.from_pandas(df, features=features)

ds = DatasetDict(ds)
ds.push_to_hub("add_one_rte", token="<token>")


# check overlap between splits
from itertools import combinations
for i, j in combinations(ds.keys(), 2):
    print(
        f"{i} - {j}: ",
        pd.merge(
            ds[i].to_pandas(), 
            ds[j].to_pandas(), 
            on=["premise", "hypothesis", "label"], 
            how="inner",
        ).shape[0],
    )
#> dev - test:  0
#> dev - train:  0
#> test - train:  0
Downloads last month
24
Edit dataset card

Models trained or fine-tuned on pietrolesci/add_one_rte

Collection including pietrolesci/add_one_rte