feat(llmobs): report experiment status and error to backend#16713
Conversation
Codeowners resolved as |
12a98d1 to
f7a7857
Compare
30618d4 to
3b45312
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 609a6a3074
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
gsvigruha
left a comment
There was a problem hiding this comment.
LGTM pending Chris's two comments
emmettbutler
left a comment
There was a problem hiding this comment.
Approved the release note
041a9b3 to
b931e44
Compare
…status When an experiment completes with errors (raise_errors=False), the status update now includes a descriptive error message (e.g. "3/10 tasks failed, 2 evaluator error(s)") so the backend can display why the experiment failed.
9b23d66 to
af4bdea
Compare
af4bdea to
8060c90
Compare
|
/merge |
|
View all feedbacks in Devflow UI.
The expected merge time in
|
Description
Add experiment_update() to LLMObsExperimentsClient that PATCHes experiment status/error via the backend API.
Update Experiment.run() to report status transitions:
Testing
Risks
Additional Notes