Fine tuning gpt-4o-mini error: images are not supported for provided model

Question

This is my first attempt at fine tuning and I'm trying to tune GPT-4o-mini on some images, following a demo I saw from the Ignite conference.

However, I got this error:

Training file contains image input modalities and images are not supported for provided model. Please try again with a different model or different data.

Base model: gpt-4o-mini-2024-07-18

I essentially selected all the defaults since this is my first time.

Any ideas?

Answer 1

@Chris Harris Welcome to Microsoft Q&A Forum, Thank you for posting your query here!

.

The gpt-4o-mini-2024-07-18 model does not support image inputs for fine-tuning. This model is designed for text-to-text fine-tuning only. See here.

.

To resolve this, you can Use a Different Model: If you need to fine-tune with image data, consider using a model that supports image inputs, such as the gpt-4o-2024-08-06 model. More info here and here.

.

Hope this helps.

Answer 2

If you want to do fine-tuning with images, you have to follow a specified way to perform the "vision fine-tuning".

You could find more detail in the following link, Good luck.

https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/fine-tuning?tabs=azure-openai%2Cturbo%2Cpython-new&pivots=programming-language-studio#vision-fine-tuning