Error while running moderation eval refusals_v3 for snapshot ft:gpt-4o-mini-2024-07-18 Error while running eval for category harassment/threatening for snapshot ft:gpt-4o-mini-2024-07-18. Please try again

API
,
1

I have a problem with fine-tuning GPT-4o-mini as below:

Error while running moderation eval refusals_v3 for snapshot ft:gpt-4o-mini-2024-07-18:xxx: Error while running eval for category harassment/threatening for snapshot ft:gpt-4o-mini-2024-07-18:xxx. Please try again.

Retrying moderation eval refusals_v3 (attempt 3/3) due to an internal error.

Retrying moderation eval refusals_v3 (attempt 2/3) due to an internal error.

I believe that my training dataset is fine, since it is slightly modification of my previous working training dataset.

I re-run fine-tuning few times today but all failed. Could anyone advise how can I fix this problem?

2

I’m seeing the same problem with gpt-4.1-mini. This is a very benign data extraction prompt and I cannot imagine any reasonable way it would fail such a test.

Error while running moderation eval refusals_v3 for snapshot ft:gpt-4.1-mini-2025-04-14:xxxxxxxxxxxxxxxxxx Error while running eval for category harassment/threatening for snapshot ft:gpt-4.1-mini-2025-04-14:xxxxxxxxxxxxxxxxxx. Please try again.

3

Welcome to the community, and thanks for reporting!

I’ve passed this along. I know there were other fine-tuning problems today, but I sent this as a possible separate issue…