אנלי

  • תיאור :

אדוורסריאל NLI (ANLI) הוא מערך נתונים בקנה מידה גדול של NLI, שנאסף באמצעות פרוצדורה איטרטיבית של אדם ומודל בלולאה.

FeaturesDict({
    'context': Text(shape=(), dtype=string),
    'hypothesis': Text(shape=(), dtype=string),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=3),
    'uid': Text(shape=(), dtype=string),
})
  • תיעוד תכונה :
תכונה מעמד צוּרָה Dtype תיאור
FeaturesDict
הֶקשֵׁר טֶקסט חוּט
הַשׁעָרָה טֶקסט חוּט
תווית ClassLabel int64
uid טֶקסט חוּט
@inproceedings{Nie2019AdversarialNA,
    title = "Adversarial NLI: A New Benchmark for Natural Language Understanding",
    author = "Nie, Yixin and
      Williams, Adina and
      Dinan, Emily  and
      Bansal, Mohit and
      Weston, Jason and
      Kiela, Douwe",
      year="2019",
    url ="https://arxiv.org/abs/1910.14599"
}

anli/r1 (תצורת ברירת המחדל)

  • תיאור תצורה : סיבוב ראשון

  • גודל מערך נתונים : 9.04 MiB

  • פיצולים :

לְפַצֵל דוגמאות
'test' 1,000
'train' 16,946
'validation' 1,000

anli/r2

  • תיאור תצורה : סיבוב שני

  • גודל ערכת נתונים : 22.39 MiB

  • פיצולים :

לְפַצֵל דוגמאות
'test' 1,000
'train' 45,460
'validation' 1,000

anli/r3

  • תיאור תצורה : סיבוב שלישי

  • גודל ערכת נתונים: 47.03 MiB

  • פיצולים :

לְפַצֵל דוגמאות
'test' 1,200
'train' 100,459
'validation' 1,200