Run the genetic algorithm on a free GitHub Actions runner (CPU). Results are committed to an
experiment branch exp/<name> and appear in the branch selector above when the run finishes.
Each branch is a lineage: with "continue" on, a new run starts from that branch's previous archive.
- Start via GitHub issue ↗ Opens a prefilled issue; press Create. Only the repo owner and collaborators can start runs. The issue gets a comment with a listen link when the run finishes, then closes.
- Fallback: GitHub's "Run workflow" form ↗
Press Run workflow there and enter the same values:
Produce explores what a branch already holds, without evolving it: one large burst of variations (thousands of candidates, far more and wilder than an evolution step) is scored against the knobs below and a light quality floor, and a diverse set is returned as a .mid + .mp3 zip. Nothing is committed; no branch changes.
Knobs percentile of real pop: 50 = typical, 90 = more than 90% of real pop. Unticked knobs stay free.
- Produce via GitHub issue ↗ Opens a prefilled issue; press Create. The issue gets a download link and a table of the pieces, then closes.
- Fallback: GitHub's "Run workflow" form ↗
Enter:
Pick a batch, listen to A and B, choose the one you like more. Labels are hidden until you answer. Your answers stay in this browser until you submit them as a GitHub issue.
A
B
Batch done, thank you! Now send your ratings below.
0 unsent ratings in this browser.
- Open GitHub with my ratings ↗ A new tab opens with the issue already filled in. Press the green Create / Submit new issue button there.
- Clears the local copy once GitHub shows your issue.
No GitHub account? and send it to Renaldas.