# Aligning Sentence Embeddings to Human Concepts via Sparse Autoencoders

arXiv: https://arxiv.org/abs/2607.00023 · HTML: https://mlsanity.com/paper/2607.00023/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Generation and context integration, one of 15 directions in Retrieval-augmented generation.
- Builds on 10 of its 16 references that are in the map.
- First posted on arXiv 19 Jun 2026.

## Placed in

- Retrieval-augmented generation / [Generation and context integration](https://mlsanity.com/rag/generation-and-context-integration/)
- Mechanistic interpretability / [SAE features and their applications](https://mlsanity.com/mechinterp/sae-features-and-their-applications/)

## Builds on (most cited first, 10 of 10)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Toy Models of Superposition](https://mlsanity.com/paper/2209.10652/) (2022)
- [Scaling and evaluating sparse autoencoders](https://mlsanity.com/paper/2406.04093/) (2024)
- [Sparse, Dense, and Attentional Representations for Text Retrieval](https://mlsanity.com/paper/2005.00181/) (2020)
- [Linear Algebraic Structure of Word Senses, with Applications to Polysemy](https://mlsanity.com/paper/1601.03764/) (2016)
- [BatchTopK Sparse Autoencoders](https://mlsanity.com/paper/2412.06410/) (2024)
- [Interpreting CLIP with Hierarchical Sparse Autoencoders](https://mlsanity.com/paper/2502.20578/) (2025)

## Abstract (arXiv)

Dense sentence embeddings are fundamental to modern Retrieval-Augmented Generation (RAG) systems but suffer from a lack of interpretability due to feature superposition. This opacity hinders the alignment of retrieval processes with human intent, as the entangled representations are difficult to analyze or control. In this work, we propose a method to disentangle the dense representations of sentence transformers (e.g., E5) into human-interpretable concepts using Top-k Sparse Autoencoders (SAEs). We demonstrate that these disentangled features align with specific semantic, syntactic, and pragmatic categories. Furthermore, we introduce an activation steering mechanism that allows for precise intervention in the retrieval process. By clamping specific latent features, we show that it is possible to re-rank search results to better align with user constraints without retraining the backbone model. Our findings suggest that SAE-based decomposition offers a viable path toward transparent and steerable neural information retrieval.