Any guesses as to when gpt-4 will be available for fine-tuning? As of today, October 31, 2023, the OpenAI docs still say “We are working on enabling fine-tuning for GPT-4 and expect this feature to be available later this year.”. So, sometime in the next two months, I guess. Is there a more specific target date yet?
Look out for the Dev Day keynote on Nov 6. Maybe it will be mentioned then.
Per the keynote today and the docs, gpt-4-0613 (experimental) can now be used as a base model for fine-tuning for selected developers. How does someone get on the list?
FYI - I just saw this in the DevDay recap note:
" As quality and safety for GPT-4 fine-tuning improves, developers actively using GPT-3.5 fine-tuning will be presented with an option to apply to the GPT-4 program within their fine-tuning console."
I have not yet seen the gpt-4 option in my fine-tuning console, despite having tuned gpt-3.5. I’m not sure what “actively” means. Perhaps I need to be creating more fine-tuned models or use them more frequently or having been doing so for longer (my first gpt-3.5 fine tune was a little over a month ago). Or perhaps this is still in the category of “quality and safety” not being ready yet. I wish I had a bit more transparency into this.