# Forward Index Compression for Learned Sparse Retrieval

Source: [Forward Index Compression for Learned Sparse Retrieval](<https://www.sota2.com/research/paper/6985758ac97a0b3e3b5d5c72>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-07T05:00:04.109Z

Authors: Sebastian Bruch, Martino Fontana, Franco Maria Nardini, Cosimo Rulli, Rossano Venturini

Published: 2026-02-05T08:35:17.000Z

## Abstract

Text retrieval using learned sparse representations of queries and documents has, over the years, evolved into a highly effective approach to search. It is thanks to recent advances in approximate nearest neighbor search-with the emergence of highly efficient algorithms such as the inverted index-based Seismic and the graph-based Hnsw-that retrieval with sparse representations became viable in practice. In this work, we scrutinize the efficiency of sparse retrieval algorithms and focus particularly on the size of a data structure that is common to all algorithmic flavors and that constitutes a substantial fraction of the overall index size: the forward index. In particular, we seek compression techniques to reduce the storage footprint of the forward index without compromising search quality or inner product computation latency. In our examination with various integer compression techniques, we report that StreamVByte achieves the best trade-off between memory footprint, retrieval accuracy, and latency. We then improve StreamVByte by introducing DotVByte, a new algorithm tailored to inner product computation. Experiments on MsMarco show that our improvements lead to significant space savings while maintaining retrieval efficiency.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Sparse Retrieval Efficiency on MSMARCO v1 (test)](<https://www.sota2.com/research/sota/sparse-retrieval-efficiency-on-msmarco-v1-test.md>): Sparse Retrieval Efficiency; MSMARCO v1 (test); Latency @ 90% Acc (µs): 156.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.05445>)

[PDF](<https://arxiv.org/pdf/2602.05445.pdf>)