Pyserini: A Python Toolkit for Reproducible Information Retrieval Research with Sparse and Dense Representations
Open accessThe DOI matches its Crossref record; 39 of the 41 references it cites were found in a registry or an index.
Abstract
Pyserini is a Python toolkit for reproducible information retrieval research with sparse and dense representations. It aims to provide effective, reproducible, and easy-to-use first-stage retrieval in a multi-stage ranking architecture. Our toolkit is self-contained as a standard Python package and comes with queries, relevance judgments, pre-built indexes, and evaluation scripts for many commonly used IR test collections. We aim to support, out of the box, the entire research lifecycle of efforts aimed at improving ranking with modern neural approaches. In particular, Pyserini supports sparse retrieval (e.g., BM25 scoring using bag-of-words representations), dense retrieval (e.g., nearest-neighbor search on transformer-encoded representations), as well as hybrid retrieval that integrates both approaches. This paper provides an overview of toolkit features and presents empirical results that illustrate its effectiveness on two popular ranking tasks. Around this toolkit, our group has built a culture of reproducibility through shared norms and tools that enable rigorous automated testing.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Interestingly, the classic BM25 model [18] remains competitive, surpassing some semantic models trained on search behavior datasets.”
From Generative Agents Navigating Digital Libraries (2025) · cites this paper as [18]
In the citing paper
“Sparse lexical methods, from classical BM25 [Lin et al., 2021] to lightweight full-text indexes such as SQLite FTS5, complement dense retrieval by matching exact terms.”
From MemX: A Local-First Long-Term Memory System for AI Assistants (2026) · cites this paper as Lin et al., 2021
Citations
Cites
PyTorch: An Imperative Style, High-Performance Deep Learning Library
Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al. (2019) · arXiv · doi:10.48550/arxiv.1912.01703 · Cited by 31,473 · cited 1×
“This is particularly the case for work on deep learning today, since the major toolkits (PyTorch [29] and Tensorflow [1]) have both adopted Python as their front-end This work is licensed under a Creative Commons Attribution International 4.”
BackgroundTensorFlow: A system for large-scale machine learning
Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., et al. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1605.08695 · Cited by 11,822
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929 · cited 1×
“also provides document expansion baselines using our doc2query method [28]; the latest model uses T5 [33] as described in Nogueira and Lin [26].”
MethodTransformers: State-of-the-Art Natural Language Processing
Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2020.emnlp-demos.6 · Cited by 8,741
Billion-scale similarity search with GPUs
Johnson, J., Douze, M., & Jegou, H. (2021) · IEEE Transactions on Big Data · doi:10.1109/tbdata.2019.2921572 · Cited by 4,228
Billion-scale similarity search with GPUs
Johnson, J., Douze, M., & Jégou, H. (2017) · arXiv · doi:10.48550/arxiv.1702.08734 · Cited by 4,140 · cited 3×
“Row (2a) refers to brute-force scans over the document vectors in Faiss [12], which provides exact nearest-neighbor search.”
BackgroundMethodDense passage retrieval for open-domain question answering
Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., Chen, D., & Yih, W.-T. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.550 · Cited by 3,575
MS MARCO: A Human Generated MAchine Reading COmprehension Dataset
Bajaj, P., Campos, D., Craswell, N., Deng, L., Gao, J., Liu, X., Majumder, R., McNamara, A., Mitra, B., Nguyen, T., et al. (2016) · arXiv · doi:10.48550/arxiv.1611.09268 · Cited by 2,473
Malkov, Y. A., & Yashunin, D. A. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2889473 · Cited by 1,904 · cited 3×
“To support dense and hybrid retrieval, Pyserini integrates Facebook’s Faiss library for efficient similarity search over dense vectors [12], which in turns integrates the HNSW library [23] to support low-latency querying.”
MethodColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT
Khattab, O., & Zaharia, M. (2020) · Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval · doi:10.1145/3397271.3401075 · Cited by 1,301 · cited 8×
“Row (3a) shows the results of combining TCT-ColBERT with BM25 bag-of-words search over the original texts, while row (3b) shows results that combine document expansion using doc2query with the T5 model.”
MethodInfluential
Research works printed but not linked: 6
- Sebastian Hofstatter, Sophia Althammer, Michael Schröder, Mete Sertkan, and Allan Hanbury. 2020. Improving Efficient Neural Ranking Models with Cross-Architecture Knowledge Distillation. arXiv:2010.02666 (2020). · arXiv
- Sebastian Hofstätter, Sophia Althammer, Michael Schröder, Mete Sertkan, and Allan Hanbury. 2021. Improving Efficient Neural Ranking Models with Cross-Architecture Knowledge Distillation. arXiv:2010.02666 (2021). · arXiv
- Sean MacAvaney, Andrew Yates, Sergey Feldman, Doug Downey, Arman Cohan, and Nazli Goharian. 2021. Simplified Data Wrangling with textttir_datasets. arXiv:2103.02280 (2021). · arXiv
- Ronak Pradeep, Rodrigo Nogueira, and Jimmy Lin. 2021 c. The Expando-Mono-Duo Design Pattern for Text Ranking with Pretrained Sequence-to-Sequence Models. arXiv:2101.05667 (2021). · arXiv
- Chenyan Xiong, Zhenghao Liu, Si Sun, Zhuyun Dai, Kaitao Zhang, Shi Yu, Zhiyuan Liu, Hoifung Poon, Jianfeng Gao, and Paul Bennett. 2020 a. CMT in TREC-COVID Round 2: Mitigating the Generalization Gaps from Web to Special Domain Search. arXiv:2011.01580 (2020). · arXiv
- Yongze Yu, Jussi Karlgren, Hamed Bonab, Ann Clifton, Md Iftekhar Tanveer, and Rosie Jones. 2020. Spotify at the TREC 2020 Podcasts Track: Segment Retrieval. In Proceedings of the Twenty-Ninth Text REtrieval Conference (TREC 2020).
Cited by
Items 801–809 of 809
CoRA-Bench: Evaluating Evidence Preservation by Context Selectors Under Token BudgetsPreprint
Pardawala, H., Sukhani, S., Chava, S., & Shah, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2611.10238
Dritsas, E., & Τρίγκα, Μ. (2026) · Neural Computing and Applications · doi:10.1007/s00521-026-12561-7
RMIT-ADM+S at TREC RAG 2026: Research Agents, Retriever Ablations, and Open-Weight ReplicationPreprint
Zendel, O., Ran, K., Ma, C., Vo, N. K., & Wang, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2611.16425
Comparison and Dynamic Fusion of RAG One-Stage Retrieval Methods for Heterogeneous Queries
Chen, Y., & Feng, T. (2026) · ITM Web of Conferences · doi:10.1051/itmconf/20269301005
Generative Agents Navigating Digital Libraries
Zerhoudi, S., & Granitzer, M. (2026) · arXiv · Cited by 1
“Interestingly, the classic BM25 model [18] remains competitive, surpassing some semantic models trained on search behavior datasets.”
RankVicuna: Zero-Shot Listwise Document Reranking with Open-Source Large Language Models
Pradeep, R., Sharifymoghaddam, S., & Lin, J. (2023) · arXiv · doi:10.48550/arxiv.2309.15088
Self-citationJiang, P., Lin, J., Cao, L., Tian, R., Kang, S., Wang, Z., Sun, J., & Han, J. (2025) · arXiv · doi:10.48550/arxiv.2503.00223
““wikipedia-100w” refers to the English Wikipedia’s dump from 2018-12-20 in non-overlapping 100-word splits (Ma et al., 2021; Lin et al., 2021).”
InfluentialSelf-citationDecaf: Joint decoding of answers and logical forms for question answering over knowledge bases
Yu, D., Zhang, S., Ng, P., Zhu, H., Li, A. H., Wang, J., Hu, Y., Wang, W., Wang, Z., & Xiang, B. (2022) · arXiv · doi:10.48550/arxiv.2210.00063
“For the retrieval module of D EC AF, we use BM25 implemented by Pyserini (Lin et al., 2021) We evaluate D EC AF on four benchmark datasets: WebQSP (Yih et al., 2016), ComplexWebQues-tions (CWQ) (Talmor & Berant, 2018), FreebaseQA (Jiang et al., 2019), and GrailQA (Gu et al., 2021).”
MethodLevonian, Z., Li, C., Zhu, W., Gade, A., Henkel, O., Postle, M.-E., & Xing, W. (2023) · arXiv · doi:10.48550/arxiv.2310.03184
“(4)We chose dense retrieval both for its popularity in RAG implementations and its dominance on a related retrieval task (not reported here) compared to a strong sparse-retrieval baseline (BM25 Pyserini [23, 50]).”
Metadata
- Authors
- Lin, J., Ma, X., Lin, S.-C., Yang, J.-H., Pradeep, R., & Nogueira, R.
- Affiliations
- University of Waterloo (CA)
- Venue
- Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval
- Publisher
- ACM
- Published
- 2021-07-11
- Pages
- 2356-2362
- OpenAlex
- W3180230246
- Cited by
- 360
- References
- 41
- Citations by year
- 2026: 67 · 2025: 72 · 2024: 61 · 2023: 97 · 2022: 44 · 2021: 18
- FWCI
- 45.53
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) (publishedVersion, cc-by)
- Funders
- Natural Sciences and Engineering Research Council of Canada
- Waterloo-Huawei Joint Innovation Laboratory
- Canada First Research Excellence Fund
- Topics
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Information Systems › Recommender Systems and Techniques
- Keywords
- reproducibility, sparse retrieval, dense retrieval, hybrid retrieval, BM25, transformer representations, nearest neighbor search, neural ranking, relevance assessment
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.