What I share, and what I don't.

The idea

A photograph, not a phrasebook

Most language models learn something like a huge phrasebook of what usually comes next. I'm building one that works more like a photograph of meaning.

Geometric representation means meaning captured as a continuous structure — the way brightness varies across a photograph, or loudness across a sound — rather than as statistics about word order. It's a different foundation, not a tweak to existing models.

The problem

Why a different foundation

Today's language models are expensive to train, need huge amounts of data, are hard to inspect, and will give confident answers when they don't actually know.

Researchers, developers, and software and firmware teams need tools they can understand, check and trust. That's who I'm building this for.

Meant to

Goals, not results

• Aims to learn structure efficiently — including from rare examples.
• Aims to be inspectable — so you can see why it gave an answer.
• Aims to say when it doesn't have enough information — rather than guess.

Status

Where it's at

Research prototype. Results will be published when properly tested.

I'll pre-register hypotheses and test on held-out data against baselines.

Patent pending (Canada).