Depth-Structured Music Recurrence: Budgeted Recurrent Attention for Full-Piece Symbolic Music Modeling

Researchers introduce Depth-Structured Music Recurrence (DSMR), a new AI training method for symbolic music generation that processes complete compositions efficiently. The technique uses stateful recurrent attention with distributed memory across layers, achieving similar performance to full-memory models while using 59% less GPU memory and 36% higher throughput.