Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
90 lines (68 loc) · 3.81 KB

CONTRIBUTING.md

File metadata and controls

90 lines (68 loc) · 3.81 KB
Raw

Contributing to edge0

Thanks for your interest in contributing! edge0 is an open-source streaming MoE inference framework, and contributions of all kinds are welcome — bug reports, performance measurements, documentation fixes, new model support, and new platform/backend work.

Ways to contribute

  • Report a bug / performance issue / request a feature or model — open an issue using one of the issue forms. The forms ask for the platform, backend, chip and memory — these fields resolve the large majority of reports, so please fill them in.
  • Ask a question or share results — please use GitHub Discussions rather than an issue.
  • Send a pull request — see the workflow below.

Repository layout

Path What
edge0-infer/ The program you run: the native inference engine (C++, GGML), its CLI and local OpenAI- and Anthropic-compatible server, the model asset tools, and the chat app examples for desktop, Android and iOS
edge0-strata/ The engine for edge0-145b and edge0-120b (NVIDIA GPU, Linux or Windows), a fork of Strata that edge0-infer starts for these models
docs/ Model design docs: MoE routing, the prerouter, and the model tiers
paper/ The technical report

edge0-infer/ is developed in a separate repository and synced here. Open issues and pull requests for it in this repository as usual: an accepted change is applied there first and arrives here with the next sync.

Development setup

Build the engine and get the model assets as described in Getting started. It covers macOS, iOS, Android, Linux and Windows. For edge0-145b and edge0-120b, also build edge0-strata/ as edge0-strata/EDGE0.md describes.

Running the tests

The tests are CTest tests with a level label from L0 (static checks) to L4 (acceptance on a device). From edge0-infer/:

cmake --preset cpu && cmake --build --preset cpu
ctest --preset cpu -L "L0|L1"     # no model assets needed

Testing describes every level, what each needs (the tiny test fixtures, the real assets, a device) and the full regression commands.

Adding a new model

A model has two parts in edge0-infer/: its asset tools in edge0-infer/tools/assets/<model>/ (see Adding a model) and its computation in edge0-infer/src/models/<model>/ (see Architecture). Please include a short note on the architecture (layers / experts / routing width / attention type) in your PR.

Pull request workflow

  1. Fork the repo and create a feature branch off main (e.g. fix/expert-slot-oob, docs/clarify-memory, feat/tool-calling).
  2. Make your change; add or update tests for new behavior.
  3. Run the tests for the levels your change touches (at least L0 and L1) locally.
  4. Open a PR against Edge0-AI/Edge0:main and fill in the PR template. Link the issue it closes (Fixes #NN).
  5. Keep PRs focused — one logical change per PR is easier to review and merge.

Commit / PR style

This repo uses conventional-commit-style prefixes, e.g. fix(experts): ..., feat(server): ..., docs(readme): ..., perf(decode): ..., chore: .... Please follow the same convention.

Reporting security issues

Do not open a public issue for security problems. Email samuel@edge0.ai instead.

License

By contributing, you agree that your contributions will be licensed under the project's Apache-2.0 license. Contributions to edge0-strata/ are licensed under that directory's MIT license.