edgextract Rust → WebAssembly

Turn markdown into a knowledge graph with an ontology you write and a cutoff you own. It runs in this tab. Your text goes only to the decision model you name, by default Tev1 on WebGPU in this tab.

loading engine…

1 Your text

or drop a .md / .txt file on the box

2 Your ontology

3 The decision model

Every name and every link is a closed question. Same contract for both backends: the model answers; your cutoff decides. If it cannot answer, the graph stays empty.

Load Tev1 before Extract. Weights stay on this device; Load also warms the GPU.

Model location

Default: Hub id raphaelmansuy/tev1-0.8b-onnx-webgpu (Together Tev1 ONNX, ~1 GB, browser cache). Short ids load from models/<id>/ after make demo-webgpu-model (local demo only). Ollama tev1 remains the Metal/CUDA ceiling.

4 Your cutoff

Hover a link to see the sentence behind it. Drag a name to move it.

Decision host

Can't reach Ollama

This tab asked the model and got no answer, so the graph stays empty — nothing was invented.

To run Ollama:

ollama serve
ollama pull tev1