I created a fine tune job nearly 15 hours ago and it still isn’t enqueued (the message about the cost hasn’t yet appeared). The fine_tunes.follow
command just returns the Created fine-tune message and nothing else. Is this expected or are we experiencing some sort of outage of service degradations?
I ran a similar fine tune operation last month and it ran successfully.
Same here, been 23 hours since having uploaded a fine tune job. In OpenAI status (https://status.openai.com/), there’s no mention of outage, but this much delay is something quite worrying!! Hope these are not signs of what we have to expect in future with such models.
Same for me! Currently been waiting 20 hours for a fine-tune. A similar job that I run on Monday morning took around 7 hours to process. Have you had any updates on yours yet? I’ve read reports of them taking up to 24 hours but I guess it just depends on how long the queue is.
Nope, no update yet.
The openai api fine_tunes.list
command continues to show the job status as pending.
I have half a mind to cancel this one and start a new one but i am not sure if that “restarts the clock” so to speak. The plan is to wait another 24 hours.
Ah, that is frustrating. I’d be interested to hear from OpenAI (or anyone who knows) if this is expected behaviour for fine-tuning and how their queuing system works. Fingers crossed yours gets processed soon
I am having the same issue. I have been finetuning jobs for a couple months now without any issues (other than it taking several hours) but yesterday I tried further finetuning a finetuned job (my first time) and it has been stuck in pending status for the last 15 hours
Just for added support for everyone… same thing is happening here. Been pending since yesterday.
(no clue if it matters if people see my ft id)
I guess its same for all. It’s been more than 33 hours in pending state. Dataset i uploaded is yet to be even enqueued, forget running epochs. strangely there is no update on any sort of outage.
Same for me. I’ve launched 3 fine-tune jobs of different sizes over the last 24+ hours, starting as early as 7a ET yesterday (29 hours ago), and none of have queued yet. This is clearly affected by an outage of some kind, considering all of the concurrent reports, and probably will create a large backlog of finetunes to process.
It’s been 48 hours, and my fine-tuning job is still pending.
Looks like they’ve finally updated the status page, reporting a “Partial outage on training jobs on fine-tuned model creation”.
It’s seems to be working now. Got my data fine tuned finally after 46 Hrs. It was uploaded on 08-08 at 11:00 hrs
Yes, I can confirm my job is now processed as well
2023-08-08 16:39:27 - Created fine-tune
2023-08-10 11:52:20 - Fine-tune succeeded
Nearly 48 hours.
Hopefully it was a temporary glitch and things will go back to normal soon.
My fine-tuning job has been pending more than 48 hours…
Hi and welcome to the developer forum!
Hopefully your jobs will start to process soon, there was an issue which has now been resolved.