Behaviour control infrastructure

Define how your model should behave, train against that intent, and catch regressions before they reach users.

npm i -g @tuned-tensor/cli
tt init
tt runs start <spec-id>

Open source CLI tool (MIT) on GitHub

How it works

A control loop for keeping model behaviour stable as prompts, models, and adapters change.

01

Define intended behaviour

Write the behaviours, constraints, examples, and failure modes your model must satisfy.

02

Train against the spec

Compile intent into training data, run fine-tunes, and keep every change tied to a behaviour version.

03

Measure drift and regressions

Evaluate before/after results, inspect failures, and learn from real model mistakes.

Want to learn more?

Explore the documentation to see how behaviour specs, runs, and evaluations work under the hood.