Music published on ArtaQuest, every track rendered by a generative model the notebook trains in its own cells. The trained weights are published beside the audio in the standard safetensors container, carrying the sha256 of the track they render, so anyone can fetch them from the work's permanent address and render the piece again.
Each piece is a public Kaggle notebook that has been run: its author picks which of the run's output files to publish, a reproducibility checklist reads Kaggle's public API — public notebook, public inputs, a finished run that produced these exact files, and whether it ran with the internet switched off — and the author confirms publication from their own inbox. Every published work carries a permanent DOI short link.