app.py

import gradio as gr
from transformers import pipeline
from transformers import pipeline,AutoTokenizer
from optimum.intel import OVModelForQuestionAnswering
model_checkpoint = "letrunglinh/qa_pnc"
global model_convert,tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
model_convert = OVModelForQuestionAnswering.from_pretrained(model_checkpoint,export=True, compile=False)
model_convert.half()
model_convert.compile()

def question_answer(context, question):
  
  model = pipeline('question-answering', model=model_convert,
  tokenizer=tokenizer)


  to_predict = [
    {
        "question": question,
        "context": context,
    }
    ]
  answers = model(to_predict)
  return answers['answer'], answers['score']

gr.Interface(fn=question_answer, inputs=["text", "text"], outputs=["textbox","textbox"], theme = "grass", share = True).launch()