README.md

metadata

annotations_creators:
  - expert-generated
  - found
language_creators:
  - found
  - expert-generated
license: []
multilinguality:
  - translation
pretty_name: opus
size_categories: []
source_datasets: []
tags:
  - parallel-corpus
task_categories:
  - translation
task_ids: []

Dataset Card for [opus]

Table of Contents

• Table of Contents
• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Source Data
  • Annotations
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
  • Other Known Limitations
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information
  • Contributions

Dataset Description

Disclaimer. Loading of dataset is slow, thus it may not be feasible when loading at scale. I'd suggest to use the other OPUS datasets on Huggingface which loads a specific corpus.

Loads OPUS as HuggingFace dataset. OPUS is an open parallel corpus covering 700+ languages and 1100+ datasets.

Given a src and tgt language, this repository can load all available parallel corpus. To my knowledge, other OPUS datasets on Huggingface loads a specific corpus

Requirements.

pip install pandas

# pip install my fork of `opustools`
git clone https://github.com/larrylawl/OpusTools.git
pip install -e OpusTools/opustools_pkg

Example Usage.

# args follows `opustools`: https://pypi.org/project/opustools/

src="en"
tgt="id"
download_dir="data"  # dir to save downloaded files
corpus="bible-uedin" # corpus name. Leave as `None` to download all available corpus for the src-tgt pair.


dataset = load_dataset("larrylawl/opus", 
                       src=src, 
                       tgt=tgt,
                       download_dir=download_dir,
                       corpus=corpus)
                       )

Disclaimer.

This repository is still in active development. Do make a PR if there're any issues!

Dataset Summary

[More Information Needed]

Supported Tasks and Leaderboards

[More Information Needed]

Languages

Available languages can be viewed on the OPUS API

Dataset Structure

Data Instances

{'src': 'In the beginning God created the heavens and the earth .',
 'tgt': 'Pada mulanya , waktu Allah mulai menciptakan alam semesta'}

Data Fields

features = {
            "src": datasets.Value("string"),
            "tgt": datasets.Value("string"),
        }

Data Splits

Merged all data into train split.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

[More Information Needed]

Contributions

Thanks to @larrylawl for adding this dataset.