# Sense Representations Are Inducible Interfaces

arXiv: https://arxiv.org/abs/2605.28669 · HTML: https://mlsanity.com/paper/2605.28669/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 31 references that are in the map.
- First posted on arXiv 27 May 2026.

## Builds on (most cited first, 18 of 18)

- [Language Models as Knowledge Bases?](https://mlsanity.com/paper/1909.01066/) (2019)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Plug and Play Language Models: A Simple Approach to Controlled Text Generation](https://mlsanity.com/paper/1912.02164/) (2019)
- [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019)
- [The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translation](https://mlsanity.com/paper/2106.03193/) (2021)
- [FUDGE: Controlled Text Generation With Future Discriminators](https://mlsanity.com/paper/2104.05218/) (2021)
- [XL-Sum: Large-Scale Multilingual Abstractive Summarization for 44 Languages](https://mlsanity.com/paper/2106.13822/) (2021)
- [XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning](https://mlsanity.com/paper/2005.00333/) (2020)
- [The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants](https://mlsanity.com/paper/2308.16884/) (2023)
- [AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders](https://mlsanity.com/paper/2501.17148/) (2025)
- [Improving Steering Vectors by Targeting Sparse Autoencoder Features](https://mlsanity.com/paper/2411.02193/) (2024)
- [SAEs Are Good for Steering -- If You Select the Right Features](https://mlsanity.com/paper/2505.20063/) (2025)
- [Moving Down the Long Tail of Word Sense Disambiguation with Gloss-Informed Biencoders](https://mlsanity.com/paper/2005.02590/) (2020)
- [Codebook Features: Sparse and Discrete Interpretability for Neural Networks](https://mlsanity.com/paper/2310.17230/) (2023)
- [Steering off Course: Reliability Challenges in Steering Language Models](https://mlsanity.com/paper/2504.04635/) (2025)
- [LMMS Reloaded: Transformer-based Sense Embeddings for Disambiguation and Beyond](https://mlsanity.com/paper/2105.12449/) (2021)
- [Multi-Sense Embeddings for Language Models and Knowledge Distillation](https://mlsanity.com/paper/2504.06036/) (2025)

## Abstract (arXiv)

Sense representations (explicit, per-token meaning decompositions) are useful for disambiguation, steering, and cross-lingual alignment, but existing approaches require models to be pretrained with sense structure baked in. We introduce ACROS, which induces an explicit sense pathway into a frozen pretrained decoder LM through a gated residual addition. On SmolLM2-360M, ACROS preserves base LM quality while supporting three uses of the same induced variables: zero-shot word-sense disambiguation (64.95 F1 on Raganato ALL, competitive with the WordNet first-sense heuristic), low-KL lexical steering across 5,161 CoInCo cases where a simple non-oracle proxy recovers about 90% of positive shifts, and SENSIA cross-lingual adaptation to four languages (mean R@1 0.988, target FLORES PPL 7.94). ACROS makes sense representations an inducible interface for ordinary pretrained LMs.