ml-pokedex · small models that dream in RDF

ml-pokedex<br>a collection of tiny RDF experiments

Four small experiments with language models trained on RDF<br>instead of English, over a single Pokémon knowledge graph. Each runs entirely in your<br>browser. Each draws what the model says, not what the page knows.

My first go at models that read and write the Semantic Web, published as a<br>curiosity. The honest verdict is on the about page: a neural net<br>loses to gzip as a database, but the rules it learns are the interesting part.

recall · 80 MB model<br>The Memoriser live

Pick real Pokémon. The model recalls their stored facts (typing, abilities, egg group)<br>and draws the constellation, sprites and all. It memorised the whole graph.

pick subjects → draw what it remembers

imagination · 11 MB model<br>The Minter live

Name a Pokémon that never existed. The model can't look it up, so it invents one: a<br>coherent typing, weaknesses, moves, a habitat, all made up on the spot.

a fresh IRI → a whole new creature

reasoning · same 11 MB model<br>The Reasoner live

Choose a typing. The model derives its weaknesses from a type chart it was never handed,<br>and the page grades it live against the real chart. It aces the typings it saw and<br>stumbles on the ones held back.

a typing → derived weaknesses, graded

the story · captured samples<br>The Arc

Scrub through training, epoch by epoch, and watch RDF structure appear in layers: first<br>syntax, then vocabulary, then schema, and reasoning last of all. The RDF tinyshakespeare.

garbage bytes → grammar → meaning

One rule across all four: the page holds names, never<br>relations. It knows that pikachu and electric exist. It<br>does not know Pikachu is Electric. Every fact you see came out of a model. Delete the model<br>files and these pages could still list and lay out subjects, knowing nothing about them.