A local model reasons step by step, and the reasoning becomes a knowledge graph that is used, not just drawn. Everything runs on your machine.
pip install mpe-lkg
mpe-lkgOpen http://localhost:5100. It talks to a local model through
Ollama and will tell you what is missing and the one command that
fixes it. The model matters more than anything else — measured, qwen3:4b-instruct-2507
answers 82.5% of the arithmetic battery against llama3.2:3b's 40%:
ollama pull qwen3:4b-instruct-2507-q4_K_MHarder questions: mode=explore answers each sub-question with its own run; mode=settle
finishes only when two independent runs agree. Any run can be taken as RDF:
curl -sX POST localhost:5100/jobs -H 'content-type: application/json' \
-d '{"query":"How many seconds are there in 23 weeks?","mode":"settle"}'
curl -s localhost:5100/jobs/<id>/rdf| How it works, and what the graph is for | the load-bearing measurements, the modules, the strongest-path search |
| Modes, jobs and RDF | reason / explore / settle, the vote, the probes, the vocabulary |
| Models and configuration | choosing models, every environment variable, troubleshooting |
| Embeddings from inside a model | reading a chosen layer instead of an embedding endpoint |
| Development | tests, the battery, and the gate that checks this documentation against measured data |
The mpedb License 1.0 — the same licence as mpedb and MPEqs, byte for byte.
Free of charge for every person and every organization, with one exception: a group whose revenue or valuation exceeds five billion dollars owes a one-time fee of seven US cents per device. Not an OSI-approved licence.
Published to PyPI as mpe-lkg — Morten Punnerud-Engelstad Local
Knowledge Graph.