Equivariant Music Transformer accepted to ISMIR 2026 with Best Paper Award nomination
umpedronosapato · x · 2026-10-27
- The authors announce that Equivariant Music Transformer has been accepted to ISMIR 2026 with a Best Paper Award nomination.
- Key finding: music transformers become less equivariant as they scale or train longer, suggesting they memorize each musically-shifted version of the same data separately instead of learning the underlying symmetry of musical structure.
More from Research
- A 'Hugging Face for steerable neurons' launches with 300+ cataloged entries — According_Air_3815 · 2026-10-27
- Constrained Decoding Isn't Inherently Harmful: Constraint Type Decides LLM Impact — Nokia · 2026-10-27
- DialFork Benchmark Uses 36,278 Dialogues to Catch Procedural Hallucinations — DialFork · 2026-10-27
- OffensiveCon Tokyo demo: finding iOS 0-days with Muse Spark — dyn___ · 2026-10-27
- Michigan's KIDBench finds LLM child safety varies sharply with age awareness, language, culture — radamihalcea · 2026-10-27
- EdgeRock architecture keeps KV cache at 72 MiB for 256K context, 99.1% smaller than Qwen3.5-4B — alifcoder · 2026-10-27