Latent Space Disentanglement via Activation Steering for Interpretable Attribute Control in Symbolic Music Generation
Transformer-based architectures have significantly advanced the generation of complex symbolic sequences, yet a significant gap remains in achieving fine-grained, interpretable control over discrete signal attributes. This paper investigates the mechanistic interpretability of the Multitrack Music Transformer (MMT) and proposes a framework for deterministic attribute modulation without retraining to bridge this gap via inference-time activation steering. Utilizing the Difference-in-Means (DiffMean) methodology, we isolate latent directions for signal attributes, specifically Pitch and Duration, within the residual stream. We validate the Linear Representation Hypothesis in this domain, achieving high correlation between steering magnitude and attribute shift. To address the inherent feature entanglement in multi-attribute steering, we introduce a Dual Steering framework utilizing Gram-Schmidt Orthogonalization. Experimental results demonstrate that this geometric decoupling reduces conceptual interference and signal degradation compared to naive vector addition, enabling independent deterministic control even against strong autoregressive conditioning.
Code (0)
등록된 구현이 없습니다.
Tasks
Music GenerationSimilar Papers 제목 키워드 기반
SAE-SSV: Supervised Steering in Sparse Representation Spaces for Reliable Control of Language Models
Large language models (LLMs) have demonstrated impressive capabilities in natural language understanding and generation, but controlling their behavior reliably remains challenging, especially in open-ended generation se…
Natural Language UnderstandingPost-Hoc Concept Disentanglement: From Correlated to Isolated Concept Representations
Concept Activation Vectors (CAVs) are widely used to model human-understandable concepts as directions within the latent space of neural networks. They are trained by identifying directions from the activations of concep…
DisentanglementDo LLMs Experience an Internal Polylogue? Investigating Reasoning through the Lens of Personas
Recent work shows that large language models (LLMs) encode behavioural traits ("personas") as linear directions in activation space, often called "persona vectors". Prior work has used such directions as static handles f…
Beyond Linear Activation Steering: Invertible Latent Transformations for Controlling LLM Behavior
Activation steering provides a lightweight inference-time mechanism for controlling large language models (LLMs) by modifying their internal activation vectors toward desired behaviors. Most existing methods compute a fi…
Activation Steering of Video Generation Models via Reduced-Order Linear Optimal Control
Text-to-video (T2V) models trained on large-scale web data can generate undesired content, motivating interventions that reduce harmful outputs without sacrificing visual quality. Activation steering offers an attractive…
Video Generation