AI Foundationspredict · compress · act

concepts → System

System

GNoME

frontieras of 2026-093 connections

millions of candidate crystals screened by a graph network — candidates, not materials

Where it sits

The prism has six jobs across and eight layers down. Its primary cell is infer × L1. Hatched cells cannot exist — a GPU does not learn, an institution does not infer.

Runs at: datacenter. Evidence: benchmark.

What must come first — and what it unlocks

Left to right is reading order, derived from the prerequisite_of edges. Nothing here is hand-ordered: the diagram is the graph.

nothing comes first — this is a starting pointGNoMEnothing depends on it yet — a leaf in the reading order

Where to read it

The chapter that introduces it, and any chapter that uses it again.

64Simulating the Worldact XIV · AI for Science

Where it comes from

paperScaling Deep Learning for Materials DiscoveryMerchant et al., Google DeepMind · 2023

Every connection

All 3 edges touching this node, grouped by relation family — the sections above are highlights from this list. Colours match the relation families inthe atlas.

Structure · 1
is an instance ofMaterials discoveryTask
Flow · 2
usesGraph neural networkArchitecture
usesSurrogate modelMechanism

This page is a projection of one node in src/data/concepts.ts. It has no prose file of its own — 521 declared edges produce all 340 of these pages. Edit an edge and both endpoints change.