Baseline

Evaluation harness

1.0Purpose

What it does

Baseline is the measurement plane. It holds a fixed test corpus and a scoring method, so changing a model, a prompt or a threshold produces a number comparable to the last number.

It sits under the pipeline rather than inside it: it measures the other modules and never touches a production record.

2.0Input

What it accepts

The evaluation corpus: A frozen, versioned corpus under the same change control as the code. Changing the corpus is itself a recorded change.
Candidate configurations: Any model, prompt, rule set or threshold change proposed for release.

3.0Output

What it emits

Scores: The same scoring method applied to every candidate, so two runs are comparable by construction.

4.0Config

The configuration surface

Configuration belongs to the client. It encodes what your specialists already know, and it is versioned like code.

Corpus composition: What the corpus contains and what each case tests.
Scoring method: How outcomes are scored, versioned with the corpus.
Re-qualification thresholds: The score movement that triggers re-qualification, agreed in the statement of work.

5.0Evidence

What it produces as evidence

Evaluation records: A versioned record of every run: the corpus version, the candidate configuration, the scores and the comparison against the previous release.

6.0Acceptance

What it is tested against

Baseline is tested against acceptance criteria agreed in the statement of work: the corpus is versioned and frozen, every release carries an evaluation record, and no change ships without a score. Baseline is also how the other modules prove their own criteria are still met after a change.