This repository is for the course: L101 Machine Learning for Language Processing.

Project: An In-depth Analysis of the Readers in Open-domain Question Answering Pipeline Systems

---

This repository is largely based on SF-QA original repo.

SF-QA helps you to evaluate your open-domain QA without building the entire open-domain QA pipeline. It supports:

• Efficient Reader Comparison
• Reproducible Research
• Knowledge Source for Applications

Features

✨ Easy evaluation framework: Build especially for open-domain QA
✨ Pre-trained Wiki dataset: No need to train it yourself
✨ Scaleable: set your own configurations and evaluate on the open domain scale.
✨ Open source: Everyone can contribute

Relevant repositories:

• DrQA [codes] [paper]
• BERTserini [codes] [paper]
• DS-QA [paper] [codes]
• Paragraph Ranker [codes]
• Multi-passage BERT [codes] [paper]
• Answer re-ranker [codes]
• SPARTA [codes] [paper]
• FID [paper] [codes]

Notes:

• I will (try to) sort out the full version of codes after the release of the final grade.