Docs

The builder's reference

Everything you need to fit, evaluate and deploy your private model. Start with the quickstart if this is your first visit, or jump straight to the guide you need.

01Guides

Start here

Quickstart

Create an account, tailor a sample on the free tier, and send your first call in under ten minutes.

Quickstart

Your first fine-tune

Prepare a dataset, write a standard, run kethra fit, read the eval, and deploy a private endpoint.

Full guide

Deployment

How private endpoints work, versioning, rollback, and exporting your weights at any time.

Deployment

Evals

How a held-out eval suite is built from your standard, and how to read the delta between versions.

Evals

Security

Your data trains your model only. Where weights live, how access is controlled, and your audit log.

Security

Data

Data handling, retention, deletion rights, and how to bring your own storage for the weights.

Data

02Concepts

A short glossary

fit

The fine-tuning run that adapts a base model to your corpus and voice standard. Initiated with kethra fit.

eval

A scored test suite built from your examples. Every fit produces an eval delta showing the improvement over the base.

weights

The trained parameters of your model. On paid tiers the weights belong to you and can be exported at any time in a standard format.

deployment

The act of putting a fitted and evaluated model behind a private, versioned, OpenAI-compatible endpoint.

distillation

A technique used optionally during fitting to compress knowledge from a larger teacher model into a smaller, faster one.

Tailor your first sample.

Start free and tailor a sample in a minute, or book a fitting and we will build you a private model of your own.

Start freeBook a fitting