All findings

COMMON. / FINDINGS / RESEARCH

Research.

Talking to other labs and reading their work. What they have already found out, and the places where our own results stopped agreeing with the theory we started from.

Research

Specialists can be blended on demand.

Can a specialist be built without training one? A blend of three public LoRA adapters beat the base model by 9 points on 200 questions none of them had seen. The first run of the…

Read 16 min
Research

The graph is the reasoning layer.

We said CGLA was zero-hallucination and that more was coming. Here are the numbers from two working systems, and what that claim looks like once it is measured.

Read 3 min
Research

What breaks at the seam.

The full results from three experiments on handoffs between small specialist models: what we measured, what the noise floor said, why one fix worked, why the actual MoE bet…

Read 12 min
Research

In early conversation with AIML.

We have started talking with researchers at the Australian Institute for Machine Learning at the University of Adelaide. Early conversations around causal reasoning and world…

Read 1 min