The job failed due to an internal error, re-enqueued for retry

When I try to fine-tune a ChatGPT 3.5 model i keep getting an error indicating that there is an internal error. The message in the title appears three times, and then the fine tuning job eventually fails. what would be the problem?

Iā€™m running into this as well. No verbose information at all to help show what went wrong.

I am also facing the same issue - @52g @manouchehri did you guys find any solution?

We reduced the amount of max tokens per-example and the issue went away. Not sure why it was needed, since all of our examples were under the limit before.

1 Like