Welcome to edge0 Discussions!

[shuk777]

👋 Welcome!

edge0 is an open-source streaming MoE inference framework — SSD expert offload, Recover-LoRA, and prerouter routing prediction — for running large sparse MoEs on consumer hardware. One recipe, across Python, macOS, iOS, Android, and Windows.

We’re using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you’re wondering about — install, edge0 demo / edge0 serve, model download, a platform runtime, or why decode is slow on your machine.
• Share ideas — a model tier, a backend, a measurement, or a note on streaming, the prerouter, or Recover-LoRA.
• Show what you ran — platform, chip, memory, tier (edge0-35b or edge0-8b), and the numbers you got.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we build together 💪.

Bugs, performance regressions, feature requests, and model requests stay on Issues. Usage questions and show-and-tell belong here.

• Bug report
• Performance report
• Feature request
• Model request
• Security, or anything you cannot share publicly: samuel@edge0.ai

When you ask, include the platform, chip, memory, and — for the Python runtime — the mlx version. Start from the README, the platform guides (python · macos · ios · android · windows), and the technical report.

To get started, comment below with an introduction of yourself and tell us what you run edge0 on, and what you’re trying to do with it.