# Latent Terms: Dense Retrievers Contain Trivially Extractable BM25-ready Zipfian Vocabularies

arXiv: https://arxiv.org/abs/2605.29384 · HTML: https://mlsanity.com/paper/2605.29384/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 41 references that are in the map.
- First posted on arXiv 28 May 2026.

## Builds on (most cited first, 28 of 28)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [MS MARCO: A Human Generated MAchine Reading COmprehension Dataset](https://mlsanity.com/paper/1611.09268/) (2016)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Models](https://mlsanity.com/paper/2104.08663/) (2021)
- [Unsupervised Dense Information Retrieval with Contrastive Learning](https://mlsanity.com/paper/2112.09118/) (2021)
- [ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT](https://mlsanity.com/paper/2004.12832/) (2020)
- [The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale](https://mlsanity.com/paper/2406.17557/) (2024)
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://mlsanity.com/paper/2007.00808/) (2020)
- [Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoders](https://mlsanity.com/paper/2407.14435/) (2024)
- [SPLADE: Sparse Lexical and Expansion Model for First Stage Ranking](https://mlsanity.com/paper/2107.05720/) (2021)
- [Pretrained Transformers for Text Ranking: BERT and Beyond](https://mlsanity.com/paper/2010.06467/) (2020)
- [Nomic Embed: Training a Reproducible Long Context Text Embedder](https://mlsanity.com/paper/2402.01613/) (2024)
- [Document Expansion by Query Prediction](https://mlsanity.com/paper/1904.08375/) (2019)
- [COIL: Revisit Exact Lexical Match in Information Retrieval with Contextualized Inverted List](https://mlsanity.com/paper/2104.07186/) (2021)
- [On the Theoretical Limitations of Embedding-Based Retrieval](https://mlsanity.com/paper/2508.21038/) (2025)
- [Sparse Autoencoders Do Not Find Canonical Units of Analysis](https://mlsanity.com/paper/2502.04878/) (2025)
- [SPLADE-v3: New baselines for SPLADE](https://mlsanity.com/paper/2403.06789/) (2024)
- [Learning Passage Impacts for Inverted Indexes](https://mlsanity.com/paper/2104.12016/) (2021)
- [Context-Aware Sentence/Passage Term Importance Estimation For First Stage Retrieval](https://mlsanity.com/paper/1910.10687/) (2019)
- [An Efficiency Study for SPLADE Models](https://mlsanity.com/paper/2207.03834/) (2022)
- [JaColBERTv2.5: Optimising Multi-Vector Retrievers to Create State-of-the-Art Japanese Retrievers with Constrained Resources](https://mlsanity.com/paper/2407.20750/) (2024)
- [Rethinking Agentic Search with Pi-Serini: Is Lexical Retrieval Sufficient?](https://mlsanity.com/paper/2605.10848/) (2026)
- [Zipf's law in 50 languages: its structural pattern, linguistic interpretation, and cognitive motivation](https://mlsanity.com/paper/1807.01855/) (2018)
- [Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense Retrieval](https://mlsanity.com/paper/2506.00041/) (2025)
- [Learning Retrieval Models with Sparse Autoencoders](https://mlsanity.com/paper/2603.13277/) (2026)
- [An Alternative to FLOPS Regularization to Effectively Productionize SPLADE-Doc](https://mlsanity.com/paper/2505.15070/) (2025)
- [Visual Words Meet BM25: Sparse Auto-Encoder Visual Word Scoring for Image Retrieval](https://mlsanity.com/paper/2603.05781/) (2026)

## Abstract (arXiv)

We propose Latent Terms, a method revealing that models trained for dense retrieval, whether single- or multi-vector, learn representations that can trivially be decomposed into retrieval-ready sparse features. When trained on frozen retrievers, Sparse Autoencoders without any retrieval-specific adjustments extract a latent vocabulary with approximately Zipfian collection statistics, directly suitable for classical sparse retrieval scoring via BM25. This approach enables sparse retrieval while requiring no learned expansion objective or sparse retrieval supervision whatsoever, and can be readily applied to any dense retriever. Latent Terms is able to match or outperform single-vector scoring methods from its own base model as well as comparable SPLADE variants. In addition, it substantially outperforms its base model on LIMIT, a task specifically designed to highlight the failures of single-vector retrieval. Overall, our results highlight that neural retrievers contain more expressive and indexable structure than their default scoring functions expose, but that other methods can nonetheless be leveraged.