8 papers
MI-MIDI: Mechanistic Interpretability of Text-to-MIDI Generation Models via Probing, Lenses and Steering
Jakub Poćwiardowski, Mateusz Modrzejewski
Mechanistic interpretability of music generation has concentrated on audio models, leaving symbolic models largely unexplored. We analyze two public text-to-MIDI systems of contras…
The Perceived Fragility of Explanations in Audio Models: Manipulation of Attribution with Unchanged Predictions
Piotr KitÅowski, Dominik WiÄ cek, Mateusz Modrzejewski
This paper investigates the fragility of post-hoc explanation methods in audio deepfake detection. While previous work on explanation manipulation focused on images using standard…
TADA! Tuning Audio Diffusion Models through Activation Steering
Åukasz Staniszewski, Katarzyna Zaleska, Mateusz Modrzejewski +1
Audio diffusion models can synthesize high-fidelity music from text, yet achieving fine-grained control over specific musical attributes remains challenging, as their internal mech…
Deep Dreams Are Made of This: Visualizing Monosemantic Features in Diffusion Models
Adam Szokalski, Mateusz Modrzejewski
This paper proposes latent visualization by optimization (LVO), a mechanistic interpretability technique that extends feature visualization by optimization - originally developed f…
ConceptCaps: a Distilled Concept Dataset for Interpretability in Music Models
Bruno Sienkiewicz, Åukasz Neumann, Mateusz Modrzejewski
Concept-based interpretability methods like TCAV require clean, well-separated positive and negative examples for each concept. Existing music datasets lack this structure: tags ar…
M6(GPT)3: Generating Multitrack Modifiable Multi-Minute MIDI Music from Text using Genetic algorithms, Probabilistic methods and GPT Models in any Progression and Time Signature
Jakub PoÄwiardowski, Mateusz Modrzejewski, Marek S. Tatara
This work introduces the M6(GPT)3 composer system, capable of generating complete, multi-minute musical compositions with complex structures in any time signature, in the MIDI doma…