Opus-mt-ja-pt Open-source Translation Model - Free Implementation of Multivariate Japanese-to-Portuguese Translation

Opus Mt Ja Pt

Developed by Helsinki-NLP

Japanese to Portuguese neural machine translation model based on transformer-align architecture, supporting multiple text variant processing

Machine Translation

Transformers

Supports Multiple LanguagesOpen Source License:Apache-2.0 #Multilingual mixed translation #SentencePiece tokenization #Low-resource language pairs

Downloads 414

Release Time : 3/2/2022

Model Overview

This model is specifically designed for Japanese to Portuguese text translation tasks, employing standardized preprocessing and SentencePiece tokenization techniques, capable of translating mixed texts containing kanji, hiragana, katakana, etc.

Model Features

Mixed script processing capability

Supports processing complex Japanese texts containing kanji, hiragana, katakana, Latin letters, and Yi script variants

Standardized preprocessing

Adopts a preprocessing scheme combining standardization and SentencePiece tokenization (spm32k) to improve translation quality

Multilingual identifier support

Requires adding target language identifiers (e.g., >>por<<) at the beginning of sentences for accurate translation

Model Capabilities

Japanese to Portuguese text translation

Mixed script processing

Long sentence translation

Use Cases

Cross-language communication

Japanese-Portuguese daily conversation translation

Used for instant translation in daily conversation scenarios

BLEU 22.2/chr-F 0.444

Mixed script content translation

Processes mixed texts containing kanji, kana, and Latin letters

🚀 JPN - POR Translation Model

This project focuses on the translation from Japanese to Portuguese. It provides a transformer - based model with detailed performance benchmarks and system information.

✨ Features

Language Pair: Translates from Japanese to Portuguese.
Model Type: Uses the transformer - align model.
Pre - processing: Applies normalization and SentencePiece (spm32k, spm32k).
Language Token Requirement: Requires a sentence - initial language token in the form of >>id<< (id = valid target language ID).

📦 Installation

There is no specific installation command provided in the original document. However, you can download the original weights and test set from the following links:

Download original weights: [opus - 2020 - 06 - 17.zip](https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.zip)
Test set translations: [opus - 2020 - 06 - 17.test.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.test.txt)
Test set scores: [opus - 2020 - 06 - 17.eval.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.eval.txt)

📚 Documentation

Model Details

Source group: Japanese
Target group: Portuguese
OPUS readme: [jpn - por](https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/jpn - por/README.md)
Model: transformer - align
Source language(s): jpn jpn_Hani jpn_Hira jpn_Kana jpn_Latn jpn_Yiii
Target language(s): por por_Hira

Benchmarks

testset	BLEU	chr - F
Tatoeba - test.jpn.por	22.2	0.444

System Info

Property	Details
hf_name	jpn - por
source_languages	jpn
target_languages	por
opus_readme_url	[https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/jpn - por/README.md](https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/jpn - por/README.md)
original_repo	Tatoeba - Challenge
tags	['translation']
languages	['ja', 'pt']
src_constituents	{'jpn_Hang', 'jpn', 'jpn_Yiii', 'jpn_Kana', 'jpn_Hani', 'jpn_Bopo', 'jpn_Latn', 'jpn_Hira'}
tgt_constituents	{'por'}
src_multilingual	False
tgt_multilingual	False
prepro	normalization + SentencePiece (spm32k, spm32k)
url_model	[https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.zip](https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.zip)
url_test_set	[https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.test.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/jpn - por/opus - 2020 - 06 - 17.test.txt)
src_alpha3	jpn
tgt_alpha3	por
short_pair	ja - pt
chrF2_score	0.444
bleu	22.2
brevity_penalty	0.922
ref_len	15570.0
src_name	Japanese
tgt_name	Portuguese
train_date	2020 - 06 - 17
src_alpha2	ja
tgt_alpha2	pt
prefer_old	False
long_pair	jpn - por
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