Not able to access fine tuned model using API

API
1

Hi, after I successfully fine tuned model, I am not able to access it via API call.
gptclient.chat.completions.create(model=modelid, messages=message).
The error message is as the follow.
The model ft:gpt-3.5-turbo-0125:personal::9vebcwxS does not exist or you do not have access to it.
I only have default project and not able to nail down the problem.

2

Welcome to the community!

As the fine-tuned model name says Personal, you have a fine-tuned model created with a Personal account. If you only have a Personal account, this is normal.

Possible points to consider are:

  • The fine-tuned model name is incorrect.
  • You are not using the API key created in your Personal account.

Please make sure that you are using the exact API key that you created during the fine-tuning process, and verify that the name of the tuned model is correct using the following method.

If you are using Python:

from openai import OpenAI
client = OpenAI()

client.fine_tuning.jobs.list()

If you are using Node.js:

import OpenAI from "openai";

const openai = new OpenAI();

async function main() {
  const list = await openai.fineTuning.jobs.list();

  for await (const fineTune of list) {
    console.log(fineTune);
  }
}

main();

This will get the exact name of the fine-tuning model.

1 Like
3

The funny part is, all my previously fine tuned models, I am able to access via API. Only the models which I fine tuned today are giving me this error. I was using API key created in my personal account all along, but today after facing the above problem, I created a new key at the project level (default project) and used that key to fine tune the next model. But, the problem persists. So, not able to figure out exact cause yet.

1 Like
4

I see, although the official status page shows All Systems Operational, but maybe there is some kind of system failure.

Unfortunately, OpenAI staff is not here, so there is no one able to solve the problem directly, but it might be helpful to report the problem via the help widget.

I am sorry that this is all I can say, but I hope that your problem will be solved as soon as possible.

5

I have the same problem. I’ve also noticed that after the fine-tune is ready there’s a check of model’s answers if they comply with the OpenAI policies:
“Evaluating model against our usage policies before enabling.” - it takes years.
Anyway, after the check is done and I try to access the model via API it is not available. FT models fine tuned today are not available for further fine tuning and even in the playground.

1 Like
6

How do we report a problem via the help widget?

7

I encountered the same problem, when trying to call a fine tuned model I get an error “The model ft:davinci-002:personal::***** does not exist or you do not have access to it.”
When calling a model fin tuned a few days ago, everything works, so the problem is definitely not in the correctness of the access token being transmitted
The model name was also checked many times…

8

Hi!
model=“gpt-3.5-turbo-1106” or “gpt-4o-mini-2024-07-18” works for me.

9

Seems like a bug. Finetuning 4o-mini takes X but models have a unusual tendency of getting stuck in Evaluating model against our usage policies before enabling. state taking almost 2X longer. More finetuned input tokens, multiplies this lag.