AI GitHub
Pyserini for reproducible retrieval baselines
Run lexical and dense retrieval experiments with research-oriented tooling.
Key Takeaways
- Pyserini provides Python interfaces to search engines and documented collections or indexes.
- Choose a documented prebuilt index or reproduce one from a known corpus.
- Pick one documented benchmark and reproduce a lexical retrieval run.
- Scores depend on the corpus version and relevance judgments.
How it works
Pyserini provides Python interfaces to search engines and documented collections or indexes. It is useful for establishing a conventional baseline before adding a model-based ranker.
What you need
Choose a documented prebuilt index or reproduce one from a known corpus. Verify its document IDs and query set before comparing your run with a reported benchmark.
What to try first
Pick one documented benchmark and reproduce a lexical retrieval run. Save queries, run outputs and evaluation settings before changing the retriever.
Evaluate a query with an exact rare term and one with a paraphrase. This shows where a lexical baseline should remain part of the comparison.
Where to look for failures
Scores depend on the corpus version and relevance judgments. Verify Java and other backend requirements for the chosen search mode.
Related Franklin pages
Keep a lexical baseline while testing learned or late-interaction retrieval. Use the same labels and input segmentation, and compare index size and search time alongside relevance.
- Work with ColBERT-style representations and retrieval in a Python library. PyLate for training and using late-interaction retrievers
- Index and retrieve text using token-level late interaction through a Python interface. RAGatouille for ColBERT retrieval experiments
For the full path from source documents to generated answers, start with Franklin’s glossary. Retrieval-augmented generation
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗