primeqa

The Prime Repository for State-of-the-Art Multilingual Question Answering Research and Development.

PrimeQA is a public open source repository that enables researchers and developers to train state-of-the-art models for question answering (QA). By using PrimeQA, a researcher can replicate the experiments outlined in a paper published in the latest NLP conference while also enjoying the capability to download pre-trained models (from an online repository) and run them on their own custom data. PrimeQA is built on top of the Transformers toolkit and uses datasets and models that are directly downloadable.

The models within PrimeQA supports End-to-end Question Answering. PrimeQA answers questions via

• Information Retrieval: Retrieving documents and passages using both traditional (e.g. BM25) and neural (e.g. ColBERT) models
• Multilingual Machine Reading Comprehension: Extract and/ or generate answers given the source document or passage.
• Multilingual Question Generation: Supports generation of questions for effective domain adaptation over tables and multilingual text.
• Retrieval Augmented Generation: Generate answers using the GPT-3/ChatGPT pretrained models, conditioned on retrieved passages.

Some examples of models (applicable on benchmark datasets) supported are :

• Traditional IR with BM25 Pyserini
• Neural IR with ColBERT, DPR (collaboration with Stanford NLP IR led by Chris Potts & Matei Zaharia). Replicating the experiments that Dr. Decr (Li et. al, 2022) performed to reach the top of the XOR TyDI leaderboard.
• Machine Reading Comprehension with XLM-R: to replicate the experiments to get to the top of the TyDI leaderboard similar to the performance of the IBM GAAMA system. Coming soon: code to replicate GAAMA's performance on Natural Questions.

🏅 Top of the Leaderboard

PrimeQA is at the top of several leaderboards: XOR-TyDi, TyDiQA-main, OTT-QA and HybridQA.

XOR-TyDi

TyDiQA-main

OTT-QA

HybridQA

✔️ Getting Started

Installation

Installation doc

# cd to project root

# If you want to run on GPU make sure to install torch appropriately

# E.g. for torch 1.11 + CUDA 11.3:
pip install 'torch~=1.11.0' --extra-index-url https://download.pytorch.org/whl/cu113

# Install as editable (-e) or non-editable using pip, with extras (e.g. tests) as desired
# Example installation commands:

# Minimal install (non-editable)
pip install .

# GPU support
pip install .[gpu]

# Full install (editable)
pip install -e .[all]

Please note that dependencies (specified in setup.py) are pinned to provide a stable experience. When installing from source these can be modified, however this is not officially supported.

Note: in many environments, conda-forge based faiss libraries perform substantially better than the default ones installed with pip. To install faiss libraries from conda-forge, use the following steps:

• Create and activate a conda environment
• Install faiss libraries, using a command

conda install -c conda-forge faiss=1.7.0 faiss-gpu=1.7.0

• In setup.py, remove the faiss-related lines:

"faiss-cpu~=1.7.2": ["install", "gpu"],
"faiss-gpu~=1.7.2": ["gpu"],

• Continue with the pip install commands as desctibed above.

JAVA requirements

Java 11 is required for BM25 retrieval. Install java as follows:

conda install -c conda-forge openjdk=11

💬 Blog Posts

There're several blog posts by members of the open source community on how they've been using PrimeQA for their needs. Read some of them:

1. PrimeQA and GPT 3
2. Enterprise search with PrimeQA
3. A search engine for Trivia geeks

🧪 Unit Tests

Testing doc

To run the unit tests you first need to install PrimeQA. Make sure to install with the [tests] or [all] extras from pip.

From there you can run the tests via pytest, for example:

pytest --cov PrimeQA --cov-config .coveragerc tests/

For more information, see:

• Our tox.ini
• The pytest and tox documentation

🔭 Learn more

❤️ PrimeQA collaborators include

	Stanford NLP		University of Illinois
	University of Stuttgart		University of Notre Dame
	Ohio State University		Carnegie Mellon University
	University of Massachusetts		IBM Research