Run 35B MoE models in ~2.5 GB of RAM. Weights stream from SSD; a trained router predicts loads a step ahead. Mac, iPhone, Android, Windows supported
Python
README: move demo video above intro paragraph
linyubupade1d1490055f
diff --git a/README.md b/README.mdindex 6f09534..2c449a3 100644--- a/README.md+++ b/README.md@@ -15,6 +15,18 @@ English | [中文](README_zh.md)</div>+<div align="center">++<video+ src="https://huggingface.co/Edge0/Edge0-35B-A3B-preview/resolve/main/20260910-105854.mp4"+ controls+ playsinline+ preload="metadata"+ width="80%">+</video>++</div>+**edge0** is an open-source streaming MoE inference framework. Itgeneralizes the production-proven recipe — **SSD expert offload +Recover-LoRA + prerouter routing prediction** — into an extensible@@ -67,18 +79,6 @@ are co-located with each checkpoint and load automatically, soand that model's adapters; upgrading adapters swaps adapterfiles only — the base stays read-only and is never merged.-<div align="center">--<video- src="https://huggingface.co/Edge0/Edge0-35B-A3B-preview/resolve/main/20260910-105854.mp4"- controls- playsinline- preload="metadata"- width="80%">-</video>--</div>-## Core mechanisms- **SSD expert offload**: expert weights are streamed from storage on