# Harmonic Representations in Symbolic Music Transformers are Localized and Controllable

arXiv: https://arxiv.org/abs/2610.15957 · HTML: https://mlsanity.com/paper/2610.15957/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 43 of its 57 references that are in the map.
- First posted on arXiv 11 Oct 2026.

## Builds on (most cited first, 30 of 43)

- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [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)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Jukebox: A Generative Model for Music](https://mlsanity.com/paper/2005.00341/) (2020)
- [Simple and Controllable Music Generation](https://mlsanity.com/paper/2306.05284/) (2023)
- [MusicLM: Generating Music From Text](https://mlsanity.com/paper/2301.11325/) (2023)
- [Causal Abstractions of Neural Networks](https://mlsanity.com/paper/2106.02997/) (2021)
- [Towards Best Practices of Activation Patching in Language Models: Metrics and Methods](https://mlsanity.com/paper/2309.16042/) (2023)
- [AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining](https://mlsanity.com/paper/2308.05734/) (2023)
- [Music Transformer](https://mlsanity.com/paper/1809.04281/) (2018)
- [Stable Audio Open](https://mlsanity.com/paper/2407.14358/) (2024)
- [MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training](https://mlsanity.com/paper/2306.00107/) (2023)
- [Compound Word Transformer: Learning to Compose Full-Song Music over Dynamic Directed Hypergraphs](https://mlsanity.com/paper/2101.02402/) (2021)
- [Inducing Causal Structure for Interpretable Neural Networks](https://mlsanity.com/paper/2112.00826/) (2021)
- [DeepBach: a Steerable Model for Bach Chorales Generation](https://mlsanity.com/paper/1612.01010/) (2016)
- [ChatMusician: Understanding and Generating Music Intrinsically with LLM](https://mlsanity.com/paper/2402.16153/) (2024)
- [Codified audio language modeling learns useful representations for music information retrieval](https://mlsanity.com/paper/2107.05677/) (2021)
- [MuseCoco: Generating Symbolic Music from Text](https://mlsanity.com/paper/2306.00110/) (2023)
- [Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation](https://mlsanity.com/paper/2210.10349/) (2022)
- [A Foundation Model for Music Informatics](https://mlsanity.com/paper/2311.03318/) (2023)
- [Counterpoint by Convolution](https://mlsanity.com/paper/1903.07227/) (2019)
- [MuPT: A Generative Symbolic Music Pretrained Transformer](https://mlsanity.com/paper/2404.06393/) (2024)
- [NotaGen: Advancing Musicality in Symbolic Music Generation with Large Language Model Training Paradigms](https://mlsanity.com/paper/2502.18008/) (2025)
- [ACE-Step 1.5: Pushing the Boundaries of Open-Source Music Generation](https://mlsanity.com/paper/2602.00744/) (2026)
- [Whole-Song Hierarchical Generation of Symbolic Music Using Cascaded Diffusion Models](https://mlsanity.com/paper/2405.09901/) (2024)
- [Text2midi: Generating Symbolic Music from Captions](https://mlsanity.com/paper/2412.16526/) (2024)
- [Compose & Embellish: Well-Structured Piano Performance Generation via A Two-Stage Approach](https://mlsanity.com/paper/2209.08212/) (2022)

## Abstract (arXiv)

Transformer models trained on symbolic music have become remarkably capable. However, it is yet unclear how they represent music structure, and whether understanding their internal representations can be used to unlock methods for model control. We examine this through a case study of harmonic analysis: when a music transformer navigates a cadence, what does it internally "know" about its harmonic context, and can that knowledge be used to affect the generation? Using Aria, a LLaMA-based symbolic music transformer evaluated on Bach chorales, we find that harmonic decisions are causally localized progressively from middle layers onward. Linear probes show that mode becomes decodable in the same layer range where patching identifies growing causal effect, while key identity emerges later. We then confirm these representations are actionable for music generation control: steering via latent directions corresponding to harmonic changes reliably produces a deceptive or minor resolution, while preserving structural coherence and generation quality. This demonstrates that, at inference time, the model's internal harmonic knowledge can be directly manipulated to control the harmonic output.