language:
- gl
task_categories:
- question-answering
- multiple-choice
- text-generation
pretty_name: xstorycloze_gl
dataset_info:
config_name: gl
features:
- name: InputStoryid
dtype: string
- name: InputSentence1
dtype: string
- name: InputSentence2
dtype: string
- name: InputSentence3
dtype: string
- name: InputSentence4
dtype: string
- name: RandomFifthSentenceQuiz1
dtype: string
- name: RandomFifthSentenceQuiz2
dtype: string
- name: AnswerRightEnding
dtype: int32
splits:
- name: train
num_examples: 360
- name: test
num_examples: 1511
configs:
- config_name: gl
data_files:
- split: train
path: XStoryCloze_train_gl.tsv
- split: test
path: XStoryCloze_test_gl.tsv
default: true
license: cc-by-4.0
size_categories:
- 1K<n<10K
Dataset Card for xstorycloze_gl
xstorycloze_gl is a question answering dataset in Galician, translated from the English StoryCloze dataset.
Dataset Details
Dataset Description
xstorycloze_gl is based on multiple-choice narrative completions. The dataset consists of 360 instances in the train split and 1511 instances in the test split. Each instance contains a story stem, divided in 4 sentences, 2 possible completions, and the number indicating the correct answer.
- Curated by: Proxecto Nós
- Language(s) (NLP): Galician
- License: CC-BY 4.0
Dataset Sources
- Repository: Proxecto NÓS at HuggingFace
Uses
xstorycloze_gl is intended to evaluate reading comprehension of language models. Some suitable use cases for the dataset are:
- Common sense reasoning: xstorycloze_ca contains stories that require basic background knowledge, such as the cost of things or the consequences of social interactions.
- Casuality understanding: The stories in xstorycloze_ca are full of temporal and causal relationships between events, which requires a coherent way of understanding causality in narratives.
- Multiple choice test: For each story, xstorycloze_ca has 2 different completions which require reasoning between different options.
- Reading comprehension: Problems and answers in xstorycloze_ca are formulated in natural language.
Dataset Structure
The dataset is provided in CSV format where each row corresponds to a four-sentence story and contains an instance identifier, the story divided in four fields for each sentence, 2 possible completions for the story, and the number - either 1 or 2 - corresponding to the correct completion. Each line contains the following fields:
InputStoryid
: text string containing the identifier of the story.InputSentence1
: The first statement in the story.InputSentence2
: The second statement in the story.InputSentence3
: The third statement in the story.InputSentence4
: The forth statement in the story.RandomFifthSentenceQuiz1
: first possible continuation of the story.RandomFifthSentenceQuiz2
: second possible continuation of the story.AnswerRightEnding
: correct possible ending; either 1 or 2.