We are excited to introduce a new larger and better quality Machine Translation dataset, MTet, which stands for Multi-domain Translation for English and VieTnamese. In our new release, we extend our previous dataset (v1.0) by adding more high-quality English-Vietnamese sentence pairs on various domains. In addition, we also show our new larger Transformer models can achieve state-of-the-art results on multiple test sets.
Get data and model at Google Cloud Storage
Visit our 📄 paper or 📝 blog post for more details.
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
model_name = "VietAI/envit5-translation"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
model.cuda()
inputs = [
"vi: VietAI là tổ chức phi lợi nhuận với sứ mệnh ươm mầm tài năng về trí tuệ nhân tạo và xây dựng một cộng đồng các chuyên gia trong lĩnh vực trí tuệ nhân tạo đẳng cấp quốc tế tại Việt Nam.",
"vi: Theo báo cáo mới nhất của Linkedin về danh sách việc làm triển vọng với mức lương hấp dẫn năm 2020, các chức danh công việc liên quan đến AI như Chuyên gia AI (Artificial Intelligence Specialist), Kỹ sư ML (Machine Learning Engineer) đều xếp thứ hạng cao.",
"en: Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.",
"en: We're on a journey to advance and democratize artificial intelligence through open source and open science."
]
outputs = model.generate(tokenizer(inputs, return_tensors="pt", padding=True).input_ids.to('cuda'), max_length=512)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
# ['en: VietAI is a non-profit organization with the mission of nurturing artificial intelligence talents and building an international - class community of artificial intelligence experts in Vietnam.',
# 'en: According to the latest LinkedIn report on the 2020 list of attractive and promising jobs, AI - related job titles such as AI Specialist, ML Engineer and ML Engineer all rank high.',
# 'vi: Nhóm chúng tôi khao khát tạo ra những khám phá có ảnh hưởng đến mọi người, và cốt lõi trong cách tiếp cận của chúng tôi là chia sẻ nghiên cứu và công cụ để thúc đẩy sự tiến bộ trong lĩnh vực này.',
# 'vi: Chúng ta đang trên hành trình tiến bộ và dân chủ hoá trí tuệ nhân tạo thông qua mã nguồn mở và khoa học mở.']
@misc{https://doi.org/10.48550/arxiv.2210.05610,
doi = {10.48550/ARXIV.2210.05610},
url = {https://arxiv.org/abs/2210.05610},
author = {Ngo, Chinh and Trinh, Trieu H. and Phan, Long and Tran, Hieu and Dang, Tai and Nguyen, Hieu and Nguyen, Minh and Luong, Minh-Thang},
keywords = {Computation and Language (cs.CL), Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {MTet: Multi-domain Translation for English and Vietnamese},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
This code is build on top of vietai/dab:
To prepare for training, generate tfrecords from raw text files:
python t2t_datagen.py \
--data_dir=$path_to_folder_contains_vocab_file \
--tmp_dir=$path_to_folder_that_contains_training_data \
--problem=$problem
To train a Transformer model on the generated tfrecords
python t2t_trainer.py \
--data_dir=$path_to_folder_contains_vocab_file_and_tf_records \
--problem=$problem \
--hparams_set=$hparams_set \
--model=transformer \
--output_dir=$path_to_folder_to_save_checkpoints
To run inference on the trained model:
python t2t_decoder.py \
--data_dir=$path_to_folde_contains_vocab_file_and_tf_records \
--problem=$problem \
--hparams_set=$hparams_set \
--model=transformer \
--output_dir=$path_to_folder_contains_checkpoints \
--checkpoint_path=$path_to_checkpoint
In this colab, we demonstrated how to run these three phases in the context of hosting data/model on Google Cloud Storage.
Our data contains roughly 4.2 million pairs of texts, ranging across multiple different domains such as medical publications, religious texts, engineering articles, literature, news, and poems. A more detail breakdown of our data is shown in the table below.
| v1 | v2 (MTet) | |
|---|---|---|
| Fictional Books | 333,189 | 473,306 |
| Legal Document | 1,150,266 | 1,134,813 |
| Medical Publication | 5,861 | 13,410 |
| Movies Subtitles | 250,000 | 721,174 |
| Software | 79,912 | 79,132 |
| TED Talk | 352,652 | 303,131 |
| Wikipedia | 645,326 | 1,094,248 |
| News | 18,449 | 18,389 |
| Religious texts | 124,389 | 48,927 |
| Educational content | 397,008 | 213,284 |
| No tag | 5,517 | 63,896 |
| Total | 3,362,569 | 4,163,710 |
Data sources is described in more details here.
We would like to thank Google for the support of Cloud credits and TPU quota!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent