README.md

metadata

license: mit
dataset_info:
  features:
    - name: id
      dtype: string
    - name: domain
      dtype: string
    - name: bucket
      dtype: float64
    - name: question_title
      dtype: string
    - name: question_text
      dtype: string
    - name: answer_title
      dtype: string
    - name: answer_text
      dtype: string
    - name: is_accepted
      dtype: bool
  splits:
    - name: train
      num_bytes: 234231847
      num_examples: 500000
  download_size: 57820994
  dataset_size: 234231847
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
task_categories:
  - question-answering
  - table-question-answering
language:
  - pt
tags:
  - code
pretty_name: Juan
size_categories:
  - 100K<n<1M

QA-PORTUGUESE-SMALL

Dataset Description

The qa-portuguese-small dataset is a collection of 500,000 question-answer pairs in Portuguese designed for Question Answering (QA) tasks. The dataset includes questions based on a wide variety of domains, such as news, general knowledge, and everyday facts, and provides corresponding answers in natural language.

The dataset is intended for training and evaluating machine learning models that can answer questions in Portuguese, making it suitable for research in multilingual NLP models and Portuguese language processing.

Dataset Details

• Task: Question Answering (QA)
• Languages: Portuguese
• Licensing: [MIT] (https://opensource.org/license/mit)
• Size: 500,000 question-answer pairs
• Number of Classes: N/A (QA task, answers are free-text)

Dataset Split

The dataset is split into three sections:

• Train: 400,000 question-answer pairs
• Validation: 50,000 question-answer pairs
• Test: 50,000 question-answer pairs

Features

The dataset contains the following fields:

• question: The question in Portuguese.
• context: The context or passage that contains the answer
• answer: The correct answer to the question in Portuguese.