If eligible (or elibible), you’d have a request access form available in the fine-tuning web interface (a button below “playground”).
1 Like
I believe the asker is wondering what makes a user eligible. Does anyone know what makes a user eligible?
1 Like
[quote=“tings, post:3, topic:496292, full:true”]Does anyone know what makes a user eligible?
[/quote]
From OpenAI support staff member:
The ability to request access for GPT-4 fine-tuning is only accessible to orgs that have actively fine-tuned other models. The reason is that for most orgs, 3.5 fine-tuning will be sufficient…
Also you’d have to determine your own eligibility in being able to spend hundreds or thousands on fine tuning in order to improve GPT-4.
3 Likes
I can confirm that the statement is true and was posted yesterday.
1 Like
For my use case I would like to fine tune using a multi-modal model, specifically image data in and text data out. I have the budget to do this, but I’m not eligible. Is there a way to petition to become eligible? 3.5 models are non-starters for me.
1 Like
Do you have any updates on it?
1 Like