t5-en-vi-base Open Source English-Vietnamese Translation Model - Achieve Accurate Text-to-Text Conversion for Free

T5 En Vi Base

Developed by NlpHUST

A text-to-text transfer Transformer pre-trained model optimized for English-Vietnamese translation, based on the T5 architecture

Machine Translation #English-Vietnamese machine translation #T5 pre-trained model #High BLEU score

Downloads 51

Release Time : 3/2/2022

Model Overview

This model is a bidirectional English-Vietnamese translation model based on the Transformer architecture, utilizing the T5 framework for pre-training and fine-tuning, specifically designed for high-quality machine translation tasks between English and Vietnamese.

Model Features

High-performance translation

Achieves a BLEU score of 32.38 (case-sensitive) on the IWSLT'15 test set, outperforming traditional neural machine translation models

Pre-training + fine-tuning architecture

Supports two-stage training with unsupervised pre-training and supervised fine-tuning to enhance model generalization

End-to-end text conversion

Adopts T5's unified text-to-text framework to simplify translation task processing

Model Capabilities

English to Vietnamese translation

Vietnamese to English translation

Long-text translation

Domain-adaptive translation

Use Cases

Education

Educational material translation

Automatically translates English teaching materials into Vietnamese versions

Maintains accuracy of professional terminology and contextual coherence

International communication

Cross-language communication

Real-time translation for business meetings or social conversations

Sample output demonstrates proper handling of culturally specific expressions

🚀 T5-EN-VI-BASE: Pretraining Text-To-Text Transfer Transformer for English Vietnamese Translation

This project focuses on pretraining a Text-To-Text Transfer Transformer for English - Vietnamese translation, aiming to provide high - quality translation solutions.

📦 Dataset

The IWSLT'15 English - Vietnamese data is sourced from the Stanford NLP group.

For all experiments, the corpus was divided into training, development, and test sets:

Data set	Sentences	Download
Training	133,317	via GitHub or located in `data/train-en-vi.tgz`
Development	1,553	via GitHub or located in `data/dev-2012-en-vi.tgz`
Test	1,268	via GitHub or located in `data/test-2013-en-vi.tgz`

📊 Results

The following are the results on the test set:

Model	BLEU (Beam Search)
Luong & Manning (2015)	23.30
Sequence - to - sequence model with attention	26.10
Neural Phrase - based Machine Translation Huang et. al. (2017)	27.69
Neural Phrase - based Machine Translation + LM Huang et. al. (2017)	28.07
t5 - en - vi - small (pretraining, without training data)	28.46 (cased) / 29.23 (uncased)
t5 - en - vi - small (fineturning with training data)	32.38 (cased) / 33.19 (uncased)
t5 - en - vi - base (pretraining, without training data)	29.66 (cased) / 30.37 (uncased)

💻 Usage Examples

Basic Usage

import torch

from transformers import T5ForConditionalGeneration, T5Tokenizer
import torch
if torch.cuda.is_available():       
    device = torch.device("cuda")

    print('There are %d GPU(s) available.' % torch.cuda.device_count())

    print('We will use the GPU:', torch.cuda.get_device_name(0))
else:
    print('No GPU available, using the CPU instead.')
    device = torch.device("cpu")

model = T5ForConditionalGeneration.from_pretrained("NlpHUST/t5-en-vi-small")
tokenizer = T5Tokenizer.from_pretrained("NlpHUST/t5-en-vi-small")
model.to(device)

src = "In school , we spent a lot of time studying the history of Kim Il - Sung , but we never learned much about the outside world , except that America , South Korea , Japan are the enemies ."
tokenized_text = tokenizer.encode(src, return_tensors="pt").to(device)
model.eval()
summary_ids = model.generate(
                    tokenized_text,
                    max_length=128, 
                    num_beams=5,
                    repetition_penalty=2.5, 
                    length_penalty=1.0, 
                    early_stopping=True
                )
output = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
print(output)

Output

Ở trường, chúng tôi dành nhiều thời gian để nghiên cứu về lịch sử Kim Il - Sung, nhưng chúng tôi chưa bao giờ học được nhiều về thế giới bên ngoài, ngoại trừ Mỹ, Hàn Quốc, Nhật Bản là kẻ thù.

📞 Contact Information

For personal communication related to this project, please contact Nha Nguyen Van (nha282@gmail.com).

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