opus-mt-en-alv Open-source Translation Model - Freely Realize Mutual Translation between English and 18 Atlantic-Congo Languages

Opus Mt En Alv

Developed by Helsinki-NLP

This is a multilingual translation model based on the Transformer architecture, supporting translation tasks from English to 18 Atlantic-Congo languages.

Machine Translation

Transformers

Supports Multiple LanguagesOpen Source License:Apache-2.0 #Multilingual Translation #African Language Support #Atlantic-Congo Languages

Downloads 42

Release Time : 3/2/2022

Model Overview

This model is specifically designed for translation tasks between English and Atlantic-Congo languages, utilizing SentencePiece tokenization technology to support mutual translation among various African languages.

Model Features

Multilingual Support

Supports translation for 18 Atlantic-Congo languages, covering linguistic needs across multiple African regions.

Standardized Preprocessing

Employs standardized text preprocessing with SentencePiece tokenization (spm32k) to enhance translation quality.

Target Language Identification

Achieves multilingual target translation by adding target language identifiers (>>id<<) at the beginning of sentences.

Model Capabilities

English-to-African language translation

Multilingual parallel translation

Cross-language text conversion

Use Cases

Language Services

Multilingual Content Localization

Translates English content into local African languages for disseminating educational, medical, and other materials.

On the Tatoeba test set, BLEU scores reached 46.7 for some language pairs (English-Tsonga).

Cross-Language Communication Aid

Assists English speakers in basic communication with speakers of African languages.

Supports daily phrase translation for 18 African languages.

🚀 English - Atlantic-Congo Languages Translation Model

This project focuses on translating English to various Atlantic-Congo languages, providing a practical solution for cross - language communication.

✨ Features

Translation Scope: Translates from English to multiple Atlantic - Congo languages, including Ewe, Fula, Igbo, Kinyarwanda, etc.
Model Type: Utilizes a Transformer model for translation tasks.
Pre - processing: Applies normalization and SentencePiece (spm32k, spm32k) for text pre - processing.

📦 Installation

No specific installation steps are provided in the original document.

📚 Documentation

eng - alv

Source Group: English
Target Group: Atlantic - Congo languages
OPUS Readme: [eng - alv](https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/eng - alv/README.md)
Model: Transformer
Source Language(s): eng
Target Language(s): ewe fuc fuv ibo kin lin lug nya run sag sna swh toi_Latn tso umb wol xho yor zul
Pre - processing: normalization + SentencePiece (spm32k, spm32k)
Requirement: A sentence initial language token is required in the form of >>id<< (id = valid target language ID)
Download Original Weights: [opus2m - 2020 - 08 - 01.zip](https://object.pouta.csc.fi/Tatoeba - MT - models/eng - alv/opus2m - 2020 - 08 - 01.zip)
Test Set Translations: [opus2m - 2020 - 08 - 01.test.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/eng - alv/opus2m - 2020 - 08 - 01.test.txt)
Test Set Scores: [opus2m - 2020 - 08 - 01.eval.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/eng - alv/opus2m - 2020 - 08 - 01.eval.txt)

Benchmarks

Testset	BLEU	chr - F
Tatoeba - test.eng - ewe.eng.ewe	4.9	0.212
Tatoeba - test.eng - ful.eng.ful	0.6	0.079
Tatoeba - test.eng - ibo.eng.ibo	3.5	0.255
Tatoeba - test.eng - kin.eng.kin	10.5	0.510
Tatoeba - test.eng - lin.eng.lin	1.1	0.273
Tatoeba - test.eng - lug.eng.lug	5.3	0.340
Tatoeba - test.eng.multi	11.4	0.429
Tatoeba - test.eng - nya.eng.nya	18.1	0.595
Tatoeba - test.eng - run.eng.run	13.9	0.484
Tatoeba - test.eng - sag.eng.sag	5.3	0.194
Tatoeba - test.eng - sna.eng.sna	26.2	0.623
Tatoeba - test.eng - swa.eng.swa	1.0	0.141
Tatoeba - test.eng - toi.eng.toi	7.0	0.224
Tatoeba - test.eng - tso.eng.tso	46.7	0.643
Tatoeba - test.eng - umb.eng.umb	7.8	0.359
Tatoeba - test.eng - wol.eng.wol	6.8	0.191
Tatoeba - test.eng - xho.eng.xho	27.1	0.629
Tatoeba - test.eng - yor.eng.yor	17.4	0.356
Tatoeba - test.eng - zul.eng.zul	34.1	0.729

System Info

Property	Details
hf_name	eng - alv
source_languages	eng
target_languages	alv
opus_readme_url	https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/eng - alv/README.md
original_repo	Tatoeba - Challenge
tags	['translation']
languages	['en', 'sn', 'rw', 'wo', 'ig', 'sg', 'ee', 'zu', 'lg', 'ts', 'ln', 'ny', 'yo', 'rn', 'xh', 'alv']
src_constituents	{'eng'}
tgt_constituents	{'sna', 'kin', 'wol', 'ibo', 'swh', 'sag', 'ewe', 'zul', 'fuc', 'lug', 'tso', 'lin', 'nya', 'yor', 'run', 'xho', 'fuv', 'toi_Latn', 'umb'}
src_multilingual	False
tgt_multilingual	True
prepro	normalization + SentencePiece (spm32k, spm32k)
url_model	https://object.pouta.csc.fi/Tatoeba - MT - models/eng - alv/opus2m - 2020 - 08 - 01.zip
url_test_set	https://object.pouta.csc.fi/Tatoeba - MT - models/eng - alv/opus2m - 2020 - 08 - 01.test.txt
src_alpha3	eng
tgt_alpha3	alv
short_pair	en - alv
chrF2_score	0.429
bleu	11.4
brevity_penalty	1.0
ref_len	10603.0
src_name	English
tgt_name	Atlantic - Congo languages
train_date	2020 - 08 - 01
src_alpha2	en
tgt_alpha2	alv
prefer_old	False
long_pair	eng - alv
helsinki_git_sha	480fcbe0ee1bf4774bcbe6226ad9f58e63f6c535
transformers_git_sha	2207e5d8cb224e954a7cba69fa4ac2309e9ff30b
port_machine	brutasse
port_time	2020 - 08 - 21 - 14:41

📄 License

This project is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご