Open
Description
I want to use this workflow:
- run an eval
- look at outputs in the dashboard; maybe filter for outputs that were evaluated as "bad"
- add a bunch of traces to labeling queue
If there were checkboxes in this part of the UI where I've added red starts, and a "bulk add to labeling queue" button it would support this workflow.
Thank you!
Alternatively - add a recipe in your docs to use your SDK to:
- fetch the most recent output for a given eval (specified by the eval
groupName) - fetch the example traces (the ones in the screenshot above)
- filter them in javascript or python
- bulk-upload some of them to a labeling queue specified by name
Metadata
Metadata
Assignees
Labels
No labels