Search papers, labs, and topics across Lattice.
University of Padova Padova
4
0
6
Eight-bit quantization allows a $1.2$-billion-parameter model to run on an $8$\,GB Raspberry Pi without sacrificing audio quality or speed.
Executable LilyPond generation is achievable in zero-shot settings, but structural understanding remains a significant hurdle for LLMs.
Expertly curated data beats sheer data quantity: a small, musicologically-informed LilyPond dataset outperforms a 150x larger MIDI corpus for music understanding tasks.
Synthetically labeled music data can capture real human perceptions of flavor, unlocking new possibilities for large-scale cross-modal AI research.
