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
| Name | Last commit message | Last commit date |
|---|---|---|
| .. | ||
| mem_budget.cc 1.9 KB | ||
| mem_budget.h 675 B | ||
| prefetch.cc 43.8 KB | ||
| prefetch.h 3.1 KB | ||