bert_question_answering

Question Answering Using Bert

input

QUESTION and CONTEXT.

output

Position of contextual text corresponding to the answer.

Usage

Set the QUESTION and CONTEXT as an argument.

$ python3 bert_question_answering.py -q "What is ailia SDK ?" -c "ailia SDK is a highly performant single inference engine for multiple platforms and hardware"
...
Answer :  [{'score': 0.5031098127365112, 'start': 13, 'end': 91, 'answer': 'a highly performant single inference engine for multiple platforms and hardware'}]

Reference

transformers

onnx-transformers

Framework

PyTorch 1.6.0

Model Format

ONNX opset = 11

Netron

• roberta-base-squad2.onnx.prototxt