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
"""MoE specs and routing math (shared by resident and streaming paths)."""
from edge0.moe.routing import group_select_from_logits, select_from_logits
from edge0.moe.spec import (
MoESpec,
QuantSpec,
RouterKind,
WeightLayout,
)
__all__ = [
"MoESpec",
"QuantSpec",
"RouterKind",
"WeightLayout",
"select_from_logits",
"group_select_from_logits",
]