Vi Mrc Large

🚀 Vietnamese and English Question Answering Model

This model is designed for extractive question - answering tasks in both Vietnamese and English, leveraging fine - tuned XLM - RoBERTa on multiple datasets.

🚀 Quick Start

📦 Installation

The installation details are not provided in the original README. If you want to use this model, you can refer to the official documentation of the dependencies like transformers library.

💻 Usage Examples

🔍 Basic Usage (Hugging Face pipeline style - NOT using sum features strategy)

from transformers import pipeline
# model_checkpoint = "nguyenvulebinh/vi-mrc-large"
model_checkpoint = "nguyenvulebinh/vi-mrc-base"
nlp = pipeline('question-answering', model=model_checkpoint,
                   tokenizer=model_checkpoint)
QA_input = {
  'question': "Bình là chuyên gia về gì ?",
  'context': "Bình Nguyễn là một người đam mê với lĩnh vực xử lý ngôn ngữ tự nhiên . Anh nhận chứng chỉ Google Developer Expert năm 2020"
}
res = nlp(QA_input)
print('pipeline: {}'.format(res))
#{'score': 0.5782045125961304, 'start': 45, 'end': 68, 'answer': 'xử lý ngôn ngữ tự nhiên'}

🔎 Advanced Usage (More accurate infer process - Using sum features strategy)

from infer import tokenize_function, data_collator, extract_answer
from model.mrc_model import MRCQuestionAnswering
from transformers import AutoTokenizer

model_checkpoint = "nguyenvulebinh/vi-mrc-large"
#model_checkpoint = "nguyenvulebinh/vi-mrc-base"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
model = MRCQuestionAnswering.from_pretrained(model_checkpoint)

QA_input = {
  'question': "Bình được công nhận với danh hiệu gì ?",
  'context': "Bình Nguyễn là một người đam mê với lĩnh vực xử lý ngôn ngữ tự nhiên . Anh nhận chứng chỉ Google Developer Expert năm 2020"
}

inputs = [tokenize_function(*QA_input)]
inputs_ids = data_collator(inputs)
outputs = model(**inputs_ids)
answer = extract_answer(inputs, outputs, tokenizer)

print(answer)
# answer: Google Developer Expert. Score start: 0.9926977753639221, Score end: 0.9909810423851013

✨ Features

• Language Model: XLM - RoBERTa
• Fine - tune: MRCQuestionAnswering
• Supported Languages: Vietnamese, English
• Downstream - task: Extractive QA
• Combined Datasets:
  • Squad 2.0
  • mailong25
  • VLSP MRC 2021
  • MultiLingual Question Answering

This model is mainly for Vietnamese QA, but it also works well for English. The evaluation on the VLSP MRC 2021 test set shows that this experiment achieved TOP 1 on the leaderboard.

Public leaderboard	Private leaderboard

📚 Documentation

Model Details

MRCQuestionAnswering uses XLM - RoBERTa as a pre - trained language model. By default, XLM - RoBERTa splits words into sub - words. However, in this implementation, sub - words representations (after being encoded by the BERT layer) are recombined into word representations using the sum strategy.

📄 License

This model is licensed under the cc - by - nc - 4.0 license.

About

Built by Binh Nguyen For more details, visit the project repository.