mT5 Model for Chinese Question-Answering

This model is a fine-tuned version of google/mt5-base on DRCD. It is served as a Traditional Chinese QA model.

Training Dataset

Delta Reading Comprehension Dataset (DRCD)
The dataset's copyright belongs to 台達研究院(Delta Research Center)

Model Information

Developed by: Chia-Wen Lu and ckip lab
Model type: mT5 (multilingual pre-trained text-to-text transformer)
Language (NLP): Traditional Chinsese (zh)
Finetuned from model: google/mt5-base

Usage

Here is the quick example to run inference.

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained('chiawen0104/mt5-drcd-qa')
model = AutoModelForSeq2SeqLM.from_pretrained('chiawen0104/mt5-drcd-qa')
input_text = "question: 新教的主教制度源自於哪一教?  context: 主教制源自天主教的主教制度"
input = tokenizer.encode(input_text, return_tensors="pt")
output = model.generate(input)
output_text = tokenizer.decode(output[0], skip_special_tokens=True)

print(output_text)

天主教