# Two-Step SPLADE: Simple, Efficient and Effective Approximation of SPLADE

arXiv: https://arxiv.org/abs/2404.13357 · HTML: https://mlsanity.com/paper/2404.13357/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Apr 2024.

## Built on by (most cited first, 6 of 6)

- [DiSCo: LLM Knowledge Distillation for Efficient Sparse Retrieval in Conversational Search](https://mlsanity.com/paper/2410.14609/) (2024)
- [Few-shot Prompting for Pairwise Ranking: An Effective Non-Parametric Retrieval Model](https://mlsanity.com/paper/2409.17745/) (2024)
- [The Role of Vocabularies in Learning Sparse Representations for Ranking](https://mlsanity.com/paper/2509.16621/) (2025)
- [On the Challenges and Opportunities of Learned Sparse Retrieval for Code](https://mlsanity.com/paper/2603.22008/) (2026)
- [The Pre-Training Study of Expanded-SPLADE Models on Web Document Titles](https://mlsanity.com/paper/2605.01407/) (2026)
- [Inference-Free Multimodal Learned Sparse Retrieval for Production-Scale Visual Document Search](https://mlsanity.com/paper/2605.30917/) (2026)

## Abstract (arXiv)

Learned sparse models such as SPLADE have successfully shown how to incorporate the benefits of state-of-the-art neural information retrieval models into the classical inverted index data structure. Despite their improvements in effectiveness, learned sparse models are not as efficient as classical sparse model such as BM25. The problem has been investigated and addressed by recently developed strategies, such as guided traversal query processing and static pruning, with different degrees of success on in-domain and out-of-domain datasets. In this work, we propose a new query processing strategy for SPLADE based on a two-step cascade. The first step uses a pruned and reweighted version of the SPLADE sparse vectors, and the second step uses the original SPLADE vectors to re-score a sample of documents retrieved in the first stage. Our extensive experiments, performed on 30 different in-domain and out-of-domain datasets, show that our proposed strategy is able to improve mean and tail response times over the original single-stage SPLADE processing by up to 30× and 40×, respectively, for in-domain datasets, and by 12x to 25x, for mean response on out-of-domain datasets, while not incurring in statistical significant difference in 60% of datasets.