Fine tuning easily

Hi. Has anyone coded a simple way to upload documents to fine tune GPT3? I suppose, what I’m asking is, is there a simple way to fine tune GPT3 for people with zero coding skills?

There’s this info

but I suppose what I want is an API where I can upload documents and all the steps outlined therein are taken care of.

Is anyone reading this prepared to code this or know of it already existing?

Thx in advance :grinning:

Your best bet would be to just use somebody’s code. It shouldnt be too hard to figure out how to use it, even with 0 experience.

1 Like

Dave Shapiro regularly posts on these forums and has a few videos on the topic.

Perhaps one of his tutorials will help.

2 Likes

Thx for all your help :green_heart:

once you figure it out, its very easy. figuring it out is the problem.

see here:

simple spreadhseet for fine tuning rules. everyone makes this stuff way too complicated.

extract the last column into a text file and name it fine-tuning-to-upload.jsonl

then from ubuntu linux command line:

openai tools fine_tunes.prepare_data -f fine-tuning-to-upload.jsonl

openai api fine_tunes.create --model “ada” --suffix “ada-new-rules” -t “fine-tuning-to-upload.jsonl”

it will return somehting like this when done:

ada:ft-your-company-name:ada-new-rules-2022-09-30-18-19-58

then do use this model to do your completions:

https://api.openai.com/v1/engines/ada:ft-your-company-name:ada-new-rules-2022-09-30-18-19-58/completions

super simple. but finding this out is painful.

3 Likes

Hi, in my AI Content Labs platform, we have the option to fine tune without code. You can see the documentation of the process in this link.

Regards