- 1) Get a basic framework set up
- Dataset: https://pastebin.com/n3AbdZUc
- Using 1 model: https://huggingface.co/deepset/roberta-base-squad2 (Most downloads among other QA models)
- 2) Set up basic framework with multiple models
- Same dataset
- Models: https://huggingface.co/models?pipeline_tag=question-answering
- 3) Test framework with practical datasets
- Dataset: HTX website + speeches (by Pang Wei) and News section (ownself scrape)
- 4) Install and test alternate open-source systems
- DeepPavlov (https://github.com/deepmipt/DeepPavlov) [Not Tested]
- Haystack (https://github.com/deepset-ai/haystack) [Tested]
- Demo: https://colab.research.google.com/drive/1usrI9eyKYbE-pEwMVstulL35RjHDbFXY?usp=sharing (No custom pipeline like multuple retrievers)
IMPORTANT: This Repo would not be updated after 6th August 2021