# Localized Adaptation Reveals Distinct Learning Signatures in Transformers

arXiv: https://arxiv.org/abs/2607.25663 · HTML: https://mlsanity.com/paper/2607.25663/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 23 references that are in the map.
- First posted on arXiv 28 Jul 2026.

## Builds on (most cited first, 13 of 13)

- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [BERT Rediscovers the Classical NLP Pipeline](https://mlsanity.com/paper/1905.05950/) (2019)
- [Mass-Editing Memory in a Transformer](https://mlsanity.com/paper/2210.07229/) (2022)
- [Editing Factual Knowledge in Language Models](https://mlsanity.com/paper/2104.08164/) (2021)
- [Fast Model Editing at Scale](https://mlsanity.com/paper/2110.11309/) (2021)
- [Memory-Based Model Editing at Scale](https://mlsanity.com/paper/2206.06520/) (2022)
- [AI and the Everything in the Whole Wide World Benchmark](https://mlsanity.com/paper/2111.15366/) (2021)
- [Measurement and Fairness](https://mlsanity.com/paper/1912.05511/) (2019)
- [AdapterDrop: On the Efficiency of Adapters in Transformers](https://mlsanity.com/paper/2010.11918/) (2020)
- [Model Editing at Scale leads to Gradual and Catastrophic Forgetting](https://mlsanity.com/paper/2401.07453/) (2024)
- [The Butterfly Effect of Model Editing: Few Edits Can Trigger Large Language Models Collapse](https://mlsanity.com/paper/2402.09656/) (2024)
- [Should We Really Edit Language Models? On the Evaluation of Edited Language Models](https://mlsanity.com/paper/2410.18785/) (2024)

## Built on by (most cited first, 1 of 1)

- [Compiling Learning Problems into Adaptation Programs for Language Models](https://mlsanity.com/paper/2609.37371/) (2026)

## Abstract (arXiv)

Transformer adaptation is typically distributed across model depth, even when the intended change is narrow. We investigate how adaptation site shapes what a model learns, how well that learning generalizes, and how selectively it is applied. We introduce a controlled benchmark spanning five objectives (lexical binding, factual association, behavioral policy learning, causal mapping, and procedural reasoning) and define each objective's "adaptation geometry" as its profile of acquisition, transfer, and boundedness under full-stack and early-, middle-, or late-layer LoRA. The objectives exhibit distinct geometries. Lexical binding favors early-layer adaptation for acquisition and boundedness but requires broader updates for transfer; factual association favors later layers among localized adapters; behavioral learning separates late-layer action acquisition from middle-layer policy gating; and causal and procedural transfer benefit most from middle- or full-stack adaptation. These patterns largely persist under parameter-matched controls, and most corresponding directional contrasts replicate across five model families. These findings establish adaptation site as a key design variable for controlling what models learn, generalize, and leave unchanged.