# SAGE-Music: Low-Latency Symbolic Music Generation via Attribute-Specialized Key-Value Head Sharing

arXiv: https://arxiv.org/abs/2510.00395 · HTML: https://mlsanity.com/paper/2510.00395/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 38 references that are in the map.
- First posted on arXiv 1 Oct 2025.

## Builds on (most cited first, 28 of 28)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [Neural Machine Translation of Rare Words with Subword Units](https://mlsanity.com/paper/1508.07909/) (2015)
- [Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention](https://mlsanity.com/paper/2006.16236/) (2020)
- [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://mlsanity.com/paper/1901.02860/) (2019)
- [Generating Long Sequences with Sparse Transformers](https://mlsanity.com/paper/1904.10509/) (2019)
- [Linformer: Self-Attention with Linear Complexity](https://mlsanity.com/paper/2006.04768/) (2020)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [Fast Transformer Decoding: One Write-Head is All You Need](https://mlsanity.com/paper/1911.02150/) (2019)
- [Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset](https://mlsanity.com/paper/1810.12247/) (2018)
- [Music Transformer](https://mlsanity.com/paper/1809.04281/) (2018)
- [Compound Word Transformer: Learning to Compose Full-Song Music over Dynamic Directed Hypergraphs](https://mlsanity.com/paper/2101.02402/) (2021)
- [MusicBERT: Symbolic Music Understanding with Large-Scale Pre-Training](https://mlsanity.com/paper/2106.05630/) (2021)
- [POP909: A Pop-song Dataset for Music Arrangement Generation](https://mlsanity.com/paper/2008.07142/) (2020)
- [PopMAG: Pop Music Accompaniment Generation](https://mlsanity.com/paper/2008.07703/) (2020)
- [Poolingformer: Long Document Modeling with Pooling Attention](https://mlsanity.com/paper/2105.04371/) (2021)
- [LakhNES: Improving multi-instrumental music generation with cross-domain pre-training](https://mlsanity.com/paper/1907.04868/) (2019)
- [MuseCoco: Generating Symbolic Music from Text](https://mlsanity.com/paper/2306.00110/) (2023)
- [Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation](https://mlsanity.com/paper/2210.10349/) (2022)
- [MuseMorphose: Full-Song and Fine-Grained Piano Music Style Transfer with One Transformer VAE](https://mlsanity.com/paper/2105.04090/) (2021)
- [Multitrack Music Transformer](https://mlsanity.com/paper/2207.06983/) (2022)
- [Byte Pair Encoding for Symbolic Music](https://mlsanity.com/paper/2301.11975/) (2023)
- [ReaLJam: Real-Time Human-AI Music Jamming with Reinforcement Learning-Tuned Transformers](https://mlsanity.com/paper/2502.21267/) (2025)
- [Structure-informed Positional Encoding for Music Generation](https://mlsanity.com/paper/2402.13301/) (2024)
- [Equipping Pretrained Unconditional Music Transformers with Instrument and Genre Controls](https://mlsanity.com/paper/2311.12257/) (2023)
- [LZMidi: Compression-Based Symbolic Music Generation](https://mlsanity.com/paper/2503.17654/) (2025)

## Abstract (arXiv)

Low-latency symbolic music generation is essential for real-time improvisation and human-AI co-creation. Existing transformer-based models, however, face a trade-off between inference speed and musical quality. Traditional acceleration techniques such as embedding pooling significantly degrade quality, while recently proposed Byte Pair Encoding (BPE) methods - though effective on single-track piano data - suffer large performance drops in multi-track settings, as revealed by our analysis. We propose Attribute-Specialized Key-Value Head Sharing (AS-KVHS), adapted to music's structured symbolic representation, achieving about 30% inference speedup with only a negligible (about 0.4%) quality drop in objective evaluations and slight improvements in subjective listening tests. Our main contributions are (1) the first systematic study of BPE's generalizability in multi-track symbolic music, and (2) the introduction of AS-KVHS for low-latency symbolic music generation. Beyond these, we also release SAGE-Music, an open-source benchmark that matches or surpasses state-of-the-art models in generation quality.