Preference and ranking data partners.
Language models and agents are often tuned with judgements rather than rewards: which of two answers is better, how a set of responses ranks, whether an output meets a rubric. The quality of those judgements sets a ceiling on what the tuned model can learn.
If you run annotation or ranking programs, we would like to connect them to the fine-tuning runs Optimus sets up, so teams can request judgements in the domain they need and see how each batch affected the model. We work out task formats, quality checks, and terms with you.
How the preference and ranking data partnership works
Steps
- 01
Share access
You give us access to the product, its technical documentation, and a contact on your engineering team.
- 02
We integrate
We build the integration and write the documentation Optimus follows, so it knows when to use your product and how.
- 03
Test together
We run it in a real training setup and check the results with you before anything reaches a customer.
- 04
Go live
Optimus can select your product in customer runs, on commercial terms we agree with you.
For researchers
If you study data quality, imitation learning, or learning from human feedback, we would like to work with you. Run studies with Optimus on your own datasets, compare how different data changes a trained model, and publish what you find. We are open to writing results up together.
Academic labs, independent researchers, and open-source teams can apply for up to $30,000 in Iacon credits through our research grants.
Talk to us
If a training run consumes what you sell, we should talk. Book a time directly, or write to dev@iaconautonomics.com and tell us which layer you supply.
Book a call