Diren Kumaratilleke
IIIStructure

Crystara.

Finding emergent structure from latent embeddings — the architecture discovered at runtime rather than specified in advance.

1

What I Inverted

If this holds, capability moves along an axis that is not parameter count — which matters wherever compute binds and data does not.

Architectures are specified in advance and then scaled. Fix self-attention, hold depth and width, and vary parameters, tokens and steps. The reasons are good: a fixed architecture is analysable, reproducible and schedulable. Nobody reopens it because it has not needed reopening.

I made the architecture an output. A predictor fails in structured ways — differently on clusters, on cycles, and on the boundaries between them — and one fixed head must span all of it with a single inductive bias.

Three systems run in a loop. A ViT backbone with an EMA target publishes the energy surface. An explorer walks the regions where the predictor has no strong opinion, under Langevin dynamics preconditioned by the Fisher information metric, so exploration is isotropic in distribution space rather than in coordinate space.

z_{t+1} = z_t − η · G(z_t)⁻¹ ∇E(z_t) + √(2η / β) · ε_t

A crystallizer runs Vietoris–Rips persistent homology over those trajectories and turns stable features into typed modules — H₀ into attractors, H₁ into cycles, H₂ into boundaries.

Persistence is the right selection criterion because of the stability theorem: the bottleneck distance between two persistence diagrams is bounded by the sup-norm distance between the functions generating them. Perturb the trajectories slightly and the diagram moves slightly, with a proven bound rather than an empirical hope. Langevin trajectories are stochastic by construction, so that guarantee is exactly what a selection rule needs. Clustering at a chosen k gives none.

2

What It Shows

Self-supervised link prediction, AUC, on three real heterogeneous entity graphs against baseline JEPA and three supervised GNNs.

GraphCrystaraJEPAΔSupervised
CSET semiconductor (519)82.7%46.1%+36.6GAT 70.3 · GCN 63.9 · SAGE 33.8
GDELT news (380)69.1%47.0%+22.1GAT 92.1 · GCN 85.2 · SAGE 59.9
SEC EDGAR (9,725)66.4%46.4%+20.0GAT OOM · GCN ~7% cls

On the semiconductor graph the crystallizer produced sixteen modules mapping one-to-one onto real supply-chain structures, validated against CSET and industry trade data. No labels, no prompting, no supply-chain prior.

ModuleTypeStructure recovered
CMP pipelineH₀Chemical-mechanical polishing suppliers
ASML ecosystemH₀Netherlands lithography dependency cluster
Singapore ATPH₀Assembly–test–packaging corridor
China packagingH₀Back-end packaging cluster
Specialty chemicalsH₀Precursor and etchant suppliers
Design-to-fabH₁Cyclic design → foundry → validation loop
EUV ↔ etch/cleanH₁Litho–etch feedback coupling
AI ASICs ↔ HitachiH₁Non-obvious equipment dependency
Lithography ↔ CMPH₂Process-stage handoff boundary

Nine are shown; the remaining seven, with persistence scores and the full validation mapping, are in the repository. The H₁ and H₂ modules are the point — cyclic dependencies and stage boundaries are precisely what a single head averages away, and they are recovered here as typed objects rather than diffuse geometry.

On the controlled Two Rooms environment the same mechanism shows in the embedding: k-NN improves at every k (+23.0% at k=1, +13.8% at k=5, +42.3% at k=20) while the linear probe is unchanged, and cross-seed variance halves from 15.73% to 7.60%.

3

What It Doesn't Show

Three graphs, all heterogeneous entity networks. That is one modality, not a survey.

Not demonstrated on images, text, or time series. The Two Rooms result is a diagnostic on a 64×64 gridworld, not a vision benchmark. Any claim of generalisation to those modalities is unsupported by anything here.

Not a general claim about post-transformer training. This is one instance of growing an architecture rather than scaling one, evaluated on one modality at small scale, with no compute-matched scaled baseline for comparison.

It loses on GDELT. GAT beats it by 23 points. Event graphs appear to carry less persistent topological structure than physical supply chains — a real limit on where the method applies, not a tuning gap.

The SEC EDGAR result is about scale, not accuracy. 66.4% is not a strong absolute number; it is the only model that completes at 9,725 entities.

4

Sources