Question-Answering

Context based Question Answering using Generative QA.

Tested

• Transformers from Hugging Face
  • multiple models -> they can be found at their website
  • supports multiple types of QA building
  • fast
  • efficient
  • Website: https://huggingface.co/
  • Code source: https://github.com/huggingface/transformers
• haystack by deepset
  • LLM(Large Language Models)
  • multiple ways to build a QA system
  • BERT models
    • you can finetune models
  • extensive documentation
  • efficient
  • heavy
  • multiple answers based on given texts
  • Website for haystack: https://haystack.deepset.ai/
  • Tutorials: https://haystack.deepset.ai/tutorials
  • Deepset website: https://www.deepset.ai/
  • Source: https://github.com/deepset-ai/haystack
• Deeppavlov
  • russian/en
  • 2 en only models
    • squad_bert
    • qa_squard2_bert
  • simple
  • efficient
  • based on transformers
  • Source code: https://github.com/deeppavlov/DeepPavlov

Run

$ pip install -r requirements.txt
$ git clone https://github.com/pinguimbotsathome/Question-Answering
$ cd Question-Answering
$ python qa.py "question"

Notes

Only outputs short answers.