Ragnarök is a battleground for the best retrieval-augmented generation (RAG) models!

Create a new conda environment and install the dependencies:

conda create -n ragnarok python=3.10 -y
conda activate ragnarok
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 # For CUDA 11.8
pip install -r requirements.txt

pip install pyragnarok

We have a wide range of models supported by Ragnarök. To run the command-r-plus model on the rag24.researchy-dev topics using the top-20 bm25 results from the MS MARCO v2.1 segment collection, you can run the following command:

python src/ragnarok/scripts/run_ragnarok.py  --model_path=command-r-plus  --topk=20 \
  --dataset=rag24.researchy-dev  --retrieval_method=bm25 --prompt_mode=cohere  \
  --context_size=8192 --max_output_tokens=1024 

Or to run the gpt-4o model (ChatQA inspired format) on the rag24.raggy-dev topics with multi-stage retrieval + reranking ()bm25 followed by rank_zephyr_rho) and augmented-generation on the top-5 MS MARCO v2.1 segments, you can run the following command:

python src/ragnarok/scripts/run_ragnarok.py  --model_path=gpt-4o  --topk=100,5 \
    --dataset=rag24.raggy-dev  --retrieval_method=bm25,rank_zephyr_rho --prompt_mode=chatqa  \
    --context_size=8192 --max_output_tokens=1024  --use_azure_openai

If you would like to contribute to the project, please refer to the contribution guidelines.

Most LLMs supported by VLLM/FastChat should additionally be supported by Ragnarök too, albeit we do not test all of them. If you would like to see a specific model added, please open an issue or a pull request. The following is a table of generation models which we regularly use with Ragnarök:

If you use Ragnarök, please cite the following:

[2406.16828] Ragnarök: A Reusable RAG Framework and Baselines for TREC 2024 Retrieval-Augmented Generation Track

@ARTICLE{pradeep2024ragnarok,
  title   = {{Ragnarök}: A Reusable RAG Framework and Baselines for TREC 2024 Retrieval-Augmented Generation Track},
  author  = {Ronak Pradeep and Nandan Thakur and Sahel Sharifymoghaddam and Eric Zhang and Ryan Nguyen and Daniel Campos and Nick Craswell and Jimmy Lin},
  year    = {2024},
  journal = {arXiv:2406.16828},
}

This research is supported in part by the Natural Sciences and Engineering Research Council (NSERC) of Canada.