mbart50-large-eng-yor-mt Open-source Model - Freely Achieve Efficient Machine Translation between English and Yoruba

Home

Mbart50 Large Eng Yor Mt

Developed by Davlan

An English-Yoruba machine translation model fine-tuned on mbart-large-50, trained on JW300 and Menyo-20k datasets

Machine Translation

Transformers

#English-Yoruba Translation #Multilingual Machine Translation #BLEU13.39 High Accuracy

Downloads 19

Release Time : 3/2/2022

Model Overview

This model is specifically designed for translating English text into Yoruba, establishing strong baseline performance for this language pair

Model Features

Strong Baseline Performance

Achieves a BLEU score of 13.39 on the Menyo-20k test set, significantly outperforming the mt5-base model's 9.82 score

Professional Corpus Training

Trained on a combination of JW300 professional religious texts and Menyo-20k modern Yoruba datasets

Language Code Adaptation

Innovatively uses Swahili (sw_KE) code to represent Yoruba, solving the original model's lack of Yoruba support

Model Capabilities

English to Yoruba text translation

Cross-language text conversion

Use Cases

Language Services

Religious Text Translation

Translating religious literature such as the Bible into Yoruba

Trained on JW300 professional religious corpus

Modern Content Localization

Translating modern English content into Yoruba

Menyo-20k dataset contains modern Yoruba expressions

🚀 mbart50-large-eng-yor-mt

This is a machine translation model that effectively translates texts from English to Yorùbá, establishing a strong baseline for this language pair.

🚀 Quick Start

This README provides an in - depth introduction to the mbart50-large-eng-yor-mt model, including its description, training data, training procedure, evaluation results, and citation information.

✨ Features

Language Support: Supports translation from English to Yorùbá.
Fine - Tuned Model: Based on the fine - tuned facebook/mbart-large-50 model.
Strong Baseline: Establishes a reliable baseline for English - Yorùbá translation.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

No code examples are provided in the original document, so this section is skipped.

📚 Documentation

Model Description

mbart50-large-eng-yor-mt is a machine translation model for translating from English to Yorùbá. It is based on a fine - tuned facebook/mbart-large-50 model and sets a strong baseline for automatic English - Yorùbá text translation.

Specifically, this model is a fine - tuned mbart-large-50 model on the JW300 Yorùbá corpus and Menyo-20k. Since the pre - trained model doesn't initially support Yorùbá, Swahili (sw_KE) was used as the language during training. Therefore, you need to use sw_KE as the language code when evaluating the model.

Limitations and Bias

This model is limited by its training dataset and may not generalize well for all use cases in different domains.

Training Data

This model was fine - tuned on the JW300 corpus and Menyo-20k dataset.

Training Procedure

This model was trained on an NVIDIA V100 GPU.

Eval Results on Test set (BLEU score)

Fine - tuning mbarr50-large achieves 13.39 BLEU on Menyo-20k test set, while mt5-base achieves 9.82.

Information Table

Property	Details
Model Type	Machine translation model from English to Yorùbá, based on fine - tuned `facebook/mbart-large-50`
Training Data	JW300 corpus and Menyo-20k dataset
Training Device	NVIDIA V100 GPU
Eval Result (BLEU)	13.39 on Menyo-20k test set

BibTeX entry and citation info

By David Adelani

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