Fine-tune models down for all users

[voidkandy]

This was an older post that was merged accidently. It can be seen here.

The post following this are recent (07/31/2023) and were correctly merged.

[artur.zhdan.g]

I am getting
error {
error: {
message: ‘internal error’,
type: ‘invalid_request_error’,
param: null,
code: null
}
}

on all retries.
it is a fine-tuned model

[sps]

Hi @artur.zhdan.g

Sharing your code making the api call will help us to find what went wrong.

[champjr]

Any api endpoints of fine-tuned models seem to be down, right now. We also are experiencing this problem, our product is down since 1 hour.

[terekita]

Seeing widespread reports of this, also experiencing the same thing

[terekita]

It appears that fine-tuned models are failing for anyone who tries to access them. I am seeing this widely reported today for any org that uses fine-tunes. This is a representative fail:

APIError: internal error {
“error”: {
“message”: “internal error”,
“type”: “invalid_request_error”,
“param”: null,
“code”: null
}
}
500 {‘error’: {‘message’: ‘internal error’, ‘type’: ‘invalid_request_error’, ‘param’: None, ‘code’: None}} {‘Date’: ‘Mon, 31 Jul 2023 18:05:09 GMT’, ‘Content-Type’: ‘application/json; charset=utf-8’, ‘Content-Length’: ‘147’, ‘Connection’: ‘keep-alive’, ‘vary’: ‘Origin’, ‘x-request-id’: ‘155ff9ca6037a44949f7dfef3db40d0a’, ‘strict-transport-security’: ‘max-age=15724800; includeSubDomains’, ‘CF-Cache-Status’: ‘DYNAMIC’, ‘Server’: ‘cloudflare’, ‘CF-RAY’: ‘7ef7c4938b0c2a45-ORD’, ‘alt-svc’: ‘h3=“:443”; ma=86400’}

[artur.zhdan.g]

I have the same issue as you; hope it will be resolved soon.

[artur.zhdan.g]

It, unfortunately, happens on any fine-tuned model I try and on any parameters (I have successfully used the models earlier today)

[aashutosh]

Same here! Should we just move away from fine-tuned ones, now?

[andre.diogo]

Same problem here, is there a workaround? My service is down for 2 hours

[andre.diogo]

Same problem :(, is there a workaround?

[srhf]

Also noticing this since about 5:10 PM UTC today. In our case, some (small) subset of finetuned models still work via completions endpoint, but most do not.

[AuXpAsoCQrGbF]

Exact same issue is occuring for me on multiple fine-tuned models. These had been working without issue for many months until about 4 hrs ago.

[curt.kennedy]

The outage of fine-tunes has apparently recovered:

It was showing an internal error earlier:

{'error': {'message': 'internal error', 'type': 'invalid_request_error', 'param': None, 'code': None}}

The outage lasted ~3 hrs.

[champjr]

Our fine-tuned models seem to be back and have been up and running smoothly for a few minutes now.

[terekita]

Same here! That was scary. Would still love to know the best way to let OpenAI know when stuff like this happens.

[curt.kennedy]

I agree, it wasn’t reported on the status page, and the outage was at least 3 hours!