# One Adapter Pair per Model: A Universal Activation Interface for Language Models

arXiv: https://arxiv.org/abs/2608.09521 · HTML: https://mlsanity.com/paper/2608.09521/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability.
- Builds on 14 of its 22 references that are in the map.
- First posted on arXiv 10 Aug 2026.

## Placed in

- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

## Builds on (most cited first, 14 of 14)

- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Revisiting Model Stitching to Compare Neural Representations](https://mlsanity.com/paper/2106.07682/) (2021)
- [Relative representations enable zero-shot latent space communication](https://mlsanity.com/paper/2209.15430/) (2022)
- [Optimal whitening and decorrelation](https://mlsanity.com/paper/1512.00809/) (2015)
- [Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment](https://mlsanity.com/paper/2502.03714/) (2025)
- [Transferring Linear Features Across Language Models With Model Stitching](https://mlsanity.com/paper/2506.06609/) (2025)
- [SPARC: Concept-Aligned Sparse Autoencoders for Cross-Model and Cross-Modal Interpretability](https://mlsanity.com/paper/2507.06265/) (2025)
- [Ventriloquist LLMs: Linear Alignment of Late-Stage Representations](https://mlsanity.com/paper/2603.18908/) (2026)
- [A Negative Result on Cross-Model Activation Transfer in a Pythia Multi-Hop Setting](https://mlsanity.com/paper/2606.03280/) (2026)
- [Atlas-Alignment: Making Interpretability Transferable Across Language Models](https://mlsanity.com/paper/2510.27413/) (2025)
- [Universal Activation Verbalizer: A Unified Framework for Cross-Model Activation Explanation](https://mlsanity.com/paper/2605.25903/) (2026)
- [Cross-Family Universality of Behavioral Axes via Anchor-Projected Representations](https://mlsanity.com/paper/2605.09875/) (2026)

## Built on by (most cited first, 2 of 2)

- [Portable Semantics, Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells](https://mlsanity.com/paper/2609.11365/) (2026)
- [SharedSAE: One Feature Dictionary Across Language Models](https://mlsanity.com/paper/2609.04344/) (2026)

## Abstract (arXiv)

Activation-based tools are usually tied to one model's native hidden space, requiring probes, sparse autoencoders, and natural-language interpreters to be rebuilt or rediscovered for each new language model. We present a Universal Activation Bus, a framework that provides a common activation interface across compatible language models. Using a small set of source models, we learn a shared dense space together with one lightweight linear encoder--decoder adapter pair per model. After source training, the interface is frozen; a new model joins by fitting only its adapter pair on unlabeled matched text. The resulting interface allows activation-based tools to be shared across connected models, including common probes and SAE features as well as access to an NLA originally trained for a different model. Across five models, semantically related texts form consistent neighborhoods in the shared space, and an onboarded model reuses these tools effectively without retraining them. We further show that an intermediate activation from one model can be used by another model's frozen upper layers to produce predictions. These results establish a stable, model-wise activation contract for reusable tools across compatible language models.