Hi everyone,
I’m running into the same issue described in the “OpenAI Evals bug: Unknown Parameter” thread, which appears to have been closed without a fix.
When I run my eval I have a error : Unknown parameter: ‘data_source.sampling_params.text.verbosity’.
I’ve tried reproducing the setup as cleanly as possible based on the suggestions in the thread, but I’m still getting the same error.
For context, I followed this setup :
-
Created a new eval from scratch in the OpenAI Platform dashboard
-
Uploaded a CSV file as test data
-
Enabled “Generate new responses”
-
Only used text columns (no images)
-
Did not select “Import as responses”
-
Used a simple developer/user prompt
-
Added a semantic similarity grader
However, the issue still persists, and I haven’t been able to identify where verbosity is being injected from or how to disable it.
Has anyone found a workaround or fix for this?
Thanks a lot in advance ![]()