Repository navigation
feat: make benchmark scores decision-safe and repeatable - #3
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
MediaBuyerBench can now compare models on a narrow, auditable claim: whether they make source-grounded paid-media decisions safely under a recorded harness. Headline results fail closed when safety evidence, repeated-run coverage, or judge-cache identity is incomplete.
What changed
Validation
python3 -m unittest discover -s tests- 81 tests passed.mediabuyerbenchandscripts.score --require-safety-pass.Code review completed: run
20260907-225515-4f421eaa. Four local reviewer lenses completed and an independent validator confirmed all six findings. The optional Claude cross-model review could not authenticate, so no cross-model review result was used.Limitations
Post-Deploy Monitoring & Validation
No additional operational monitoring required. This changes a local benchmark CLI and artifacts, not a production service.
summarize-panelsucceeds, and all three judge families return complete judgments.New concepts
Fail-closed benchmark scoring
The headline metric is emitted only when the evidence contract is complete. A missing safety result, incomplete cohort, mismatched model/runtime, or stale cached judgment becomes an error instead of an optimistic default. Use this pattern for release gates and leaderboards; avoid it for exploratory notebooks where partial output is explicitly useful.
Wilson confidence intervals
Wilson intervals express uncertainty in pass and error rates without the unstable edge behavior of a simple normal interval near 0% or 100%. Here they keep a small benchmark from presenting a point estimate as precise. They do not remove dependence between repeated runs or compensate for too few distinct cases.