Yadia / About
About Yadia
Yadia is a research lab building multimodal AI that reasons causally. We post-train models with reinforcement learning to work out what causes what, and test by intervention that their reasoning drives their answers.
Yadia is a research lab operated by Myelin Limited. We build multimodal AI that reasons causally: models that work out what causes what, rather than what usually goes together, and whose stated reasoning can be shown, by intervention, to drive their answers. We get there through post-training, mainly reinforcement learning, starting with medical imaging such as CT. Products will follow from the research; for now the lab publishes methods and results.
How we work
- Evidence before claims. Achieved improvements are stated only with supporting results; their evidential status is shown alongside them.
- Controls and pre-registration. Every trained model is compared with a matched control, and pass criteria and held-out tests are fixed before training starts.
- Observation is not explanation. A written reasoning trace is an observation. A faithful causal account requires separate evidence.
Research programmes
Multimodal reasoning post-training
Can reinforcement learning teach multimodal models to reason about cause and effect, and how do we know the gain is real?
Current research
Causal faithfulness
Does the reasoning a model states actually cause its answer? We test it by intervention.
Current research
Factual representations
What do models represent internally, and which of it does their reasoning actually use?
Current research
Contact
For research, compute partnership and press enquiries, email hello@yadia.tech.