Repository navigation
feat(llma): support per-item offline evaluation models - #1130
Radu-Raicea wants to merge 1 commit into
Conversation
posthog-python Compliance ReportDate: 2026-10-26T19:11:35.263265+00:00 ✅ All Tests Passed!121/121 tests passed Capture_V1 Tests✅ 95/95 tests passed View Details
Capture_Ai Tests✅ 5/5 tests passed View Details
Feature_Flags Tests✅ 17/17 tests passed View Details
Feature_Flags_Local_Evaluation Tests✅ 4/4 tests passed View Details
|
There was a problem hiding this comment.
Not approved — this change needs a human reviewer.
Re-adding the stamphog label gives the same result unless the changed files change.
stamphog can't auto-approve this pull request. Two gates refused it. The deny-list gate matched public_api, because the change alters the EvaluationItem constructor signature and updates references/public_api_snapshot.txt. The tier gate classified it as T2-never (344 lines across 7 files, cross-cutting, feature), so it is not eligible for automated approval. The size gate itself passed, since the substantive change is only 8 lines in 1 file.
Please ask a human reviewer to look at it. The public API surface change is the part that needs their sign-off.
Gate mechanics and policy version
| Gate | Result | |
|---|---|---|
| prerequisites | ✓ | all clear |
| deny-list | ✗ | matches: public_api |
| size | ✓ | 8L, 1F substantive, 344L/7F incl. docs/generated/snapshots — within ceiling |
| tier | ✗ | classified as T2-never: T2-never (344L, 7F, cross-cutting, feat) |
| stamphog 2.6.1 | .stamphog/policy.yml @ unknown · reviewed head 5b918b6 |
💡 Motivation and Context
Add optional
model_versiontoEvaluationItemandEvaluationItem.from_dataset_itemso one experiment can compare executions from different models, following sdk-specs#149.Explicit models use the existing identifier validation and remain part of the item declaration through serialization, uploads, retries, and bulk recovery. Omitted/
Nonemodels remain unset for backend experiment fallback and legacy replay compatibility. Includes examples, the public API snapshot, and a minor release changeset. No divergence from the new spec requirements.💚 How did you test it?
📝 Checklist
If releasing new changes
sampo addto generate a changeset file.🤖 Agent context
Autonomy: Human-driven (agent-assisted).
Codex used Git/GitHub, uv, Sampo, and the repository's checks to implement the merged spec. Existing declaration serialization already carries the model through deduplication and recovery, so the runtime change is limited to the two construction paths. Validation uses mocked service responses; no live backend conformance is claimed.