Find evidence before answering
Wikipedia question answering combines BM25 and QLD retrieval, BGE reranking, and few-shot generation.
Try the method
Interactive explainerHow it works.
Retrieve candidate passages
Fuse lexical BM25 and query-likelihood retrieval across entity and question variants.
Continue in the source.
Open the source notebook in Jupyter, Colab, or the environment described in the README. Data and model downloads may be required.
git clone https://github.com/eforus-overseer/RAG-QA-Wikipedia.gitProject artifacts.
notebooks/rag-analysis-notebook.ipynb ↗notebooks/rag-q-a-v5-8.ipynb ↗notebooks/rag_pipeline_databricks.py ↗
Source links point to the original public repository. Credit belongs to the project authors and the dependencies credited there.