Paper-Conference

Evaluating Music Context Preservation: A Multi-facet Framework for Music Editing Systems
MuseCPBench evaluates whether music editing systems preserve the musical properties that an edit should leave unchanged. It compares editing methods across four groups of musical characteristics and identifies recurring weaknesses in preservation.
PDF
MixWeaver: Stem-aware Automatic DJ Transition Generation via Structured Discrete Optimization
MixWeaver generates DJ transitions by optimizing how individual stems enter and leave a mix. Its discrete search balances musical alignment, overlap, smoothness, and energy, making transition decisions interpretable.
Conference
SymSkill: Skill Graph for Symbolic Music Understanding
SymSkill organizes procedural music knowledge into a graph of skills to support language-model reasoning about symbolic music.
Conference
Toward Latent Language Model Skills Steering and Optimization: An Empirical Study
Skills, as a useful abstraction for the procedural capabilities of large language models (LLMs), capture how models perform structured, …
PDF
Reasoning Pathologies in Large Language Models: A Diagnostic Perspective
A diagnostic study of reasoning failures in large language models.
OpenReview
The Geometry of Agent Skills: Non-Commutative Composition in Representation Space
A study of the geometry of agent skills and how their composition depends on order in representation space.
OpenReview
Composer Vector: Style-steering Symbolic Music Generation in a Latent Space
Symbolic music generation has made significant progress, yet achieving fine-grained and flexible control over composer style remains …
PDF Code Demo
When Benchmarks Age: Temporal Misalignment through Large Language Model Factuality Evaluation
The rapid evolution of large language models (LLMs) and the real world has outpaced the static nature of widely used evaluation …
PDF Code DOI