metadata

license: cc-by-4.0
language:
  - gl
task_categories:
  - text-classification
pretty_name: paws-ca
tags:
  - paraphrase-identification
dataset_info:
  features:
    - name: id
      dtype: string
    - name: sentence1
      dtype: string
    - name: sentence2
      dtype: string
    - name: label
      dtype:
        class_label:
          names:
            '0': '0'
            '1': '1'

Dataset Card for PAWS-gl

The PAWS-gl dataset (Paraphrase Adversaries from Word Scrambling in Galician) is a translation of the English PAWS dataset.

Dataset Details

Curated by: Proxecto Nós
Language(s) (NLP): Galician
License: CC-BY 4.0

Dataset Sources

Repository: Proxecto NÓS at HuggingFace

Supported Tasks and Leaderboards

Paraphrase Identification, Language Model

Dataset Structure

Data Instances

This dataset contains more than 2000 entries. Two examples are provided below:

Id	Sentence 1	Sentence 2	Label
10	A excepción foi entre finais de 2005 e 2009 cando xogou en Suecia co Carlstad United BK, Serbia co FK Borac Čačak e o FC ruso Terek Grozny.	A excepción deuse entre finais de 2005 e 2009, cando xogou en Suecia co Carlstad United BK, Serbia co FK Borac Čačak e o FC ruso Terek Grozny.	1
12	O Tabaci é un afluente do río Leurda en Romanía.	O Leurda é un afluente do río Tabaci en Romanía.	0

Each label has two possible values: 0 indicates the pair has different meaning, while 1 indicates the pair is a paraphrase.

Data Fields

id: a string identifier
sentence1: a string containing the first sentence
sentence2: a string containing the second sentence
label: a binary label indicating whether the two sentences are paraphrases of each other