Thanks for your interest. A note on what this repository is and isn't.
ann-research is the working lab notebook for an ongoing research program
on novel artificial neural network topologies. It is public so that the
discipline of writing artifacts that make sense to a stranger keeps the
researchers honest. It contains observations, open questions, hypotheses,
decisions, and dead ends — the trajectory of belief, not only the
conclusions.
This is not a project soliciting contributions. The vault is curated by
two people (@reh3376 plus an AI research assistant) with deep shared
context. PRs from outside that pair will rarely be merged, not because the
content isn't valuable, but because the vault's value depends on every
artifact being something the curators understand and stand behind.
- Issues — pointing out errors, asking questions about claims in the vault, suggesting research directions, or flagging missing context.
- Discussions — for broader engagement with the research themes.
- External work that builds on this — fork the repo, cite it, do your own research; we'd love to hear about it.
- PRs that add new artifacts (observations, hypotheses, etc.) without prior coordination. The vault is a coherent body of thought; new artifacts need to fit the existing trajectory.
- PRs that "tidy up" or restructure the vault. The mess is the point.
Open an issue describing what you're thinking. If a contribution makes
sense, we'll coordinate on a PR. The maintainer's review and approval
(@reh3376) is required for any merge to main; see BRANCH_PROTECTION.md.
- Do not include API keys, internal company information, customer data, or any non-public material in issues, discussions, or PRs.
- The repository is MIT licensed; contributions are under the same license.
Open an issue.