# Faster Learned Sparse Retrieval with Block-Max Pruning

arXiv: https://arxiv.org/abs/2405.01117 · HTML: https://mlsanity.com/paper/2405.01117/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 May 2024.

## Built on by (most cited first, 10 of 10)

- [Efficient Constant-Space Multi-Vector Retrieval](https://mlsanity.com/paper/2504.01818/) (2025)
- [An Alternative to FLOPS Regularization to Effectively Productionize SPLADE-Doc](https://mlsanity.com/paper/2505.15070/) (2025)
- [SINDI: An Efficient Index for Sparse Vector Approximate Maximum Inner Product Search](https://mlsanity.com/paper/2509.08395/) (2025)
- [Efficient Sketching and Nearest Neighbor Search Algorithms for Sparse Vector Sets](https://mlsanity.com/paper/2509.24815/) (2025)
- [Efficiency and Effectiveness of SPLADE Models on Billion-Scale Web Document Title](https://mlsanity.com/paper/2511.22263/) (2025)
- [Exact Adaptive Hybrid Retrieval Without Fixed Top-L Cutoffs](https://mlsanity.com/paper/2608.07152/) (2026)
- [All-in-one Graph-based Indexing for Hybrid Search on GPUs](https://mlsanity.com/paper/2511.00855/) (2025)
- [From Tokens to Concepts: Leveraging SAE for SPLADE](https://mlsanity.com/paper/2604.21511/) (2026)
- [Hierarchical BM25: Lexical Search at Billion-Document Scale](https://mlsanity.com/paper/2608.00229/) (2026)
- [Static Pruning Across Sparse Retrieval Regimes: What Transfers, What Breaks, and What Still Helps](https://mlsanity.com/paper/2608.16309/) (2026)

## Abstract (arXiv)

Learned sparse retrieval systems aim to combine the effectiveness of contextualized language models with the scalability of conventional data structures such as inverted indexes. Nevertheless, the indexes generated by these systems exhibit significant deviations from the ones that use traditional retrieval models, leading to a discrepancy in the performance of existing query optimizations that were specifically developed for traditional structures. These disparities arise from structural variations in query and document statistics, including sub-word tokenization, leading to longer queries, smaller vocabularies, and different score distributions within posting lists. This paper introduces Block-Max Pruning (BMP), an innovative dynamic pruning strategy tailored for indexes arising in learned sparse retrieval environments. BMP employs a block filtering mechanism to divide the document space into small, consecutive document ranges, which are then aggregated and sorted on the fly, and fully processed only as necessary, guided by a defined safe early termination criterion or based on approximate retrieval requirements. Through rigorous experimentation, we show that BMP substantially outperforms existing dynamic pruning strategies, offering unparalleled efficiency in safe retrieval contexts and improved tradeoffs between precision and efficiency in approximate retrieval tasks.