Hello, I am developer and wanted to train chatgpt with my data. I have read multiple articles about it but none of them seem to have a clear vision. Can anyone please guide me with the apis to use to train and fine tune my model. Thanks in advance.
You cannot fine-tune ChatGPT. ChatGPT is a application, not a model
You can fine tune a number of base models using the OpenAI API, and if you read the API doc and associated pages, you will have all you need to enjoy fine-tuning:
HTH
Yes, sorry meant fine tuning the model. Thanks for your reply. Just a quick question. whenever I am retrieving fine tune using openai.retrieveFineTune()
api I am getting the following response
validation_files: ,
result_files: ,
created_at: 1676548576,
updated_at: 1676548576,
status: ‘pending’,
fine_tuned_model: null,
Can you let me know why I am not getting the name of my trained model?
Because your new fine-tuned model is “pending” so the cake is not baked yet.
When the cake is baked, the system will assign the model a shiny, new name
How much time does it take to generally assign a name to model?
That depends on the size of the JSONL file and the number of n_epochs
you selected (and maybe other params I have not tested completely).
These days, it can take anywhere from 30 minutes to 4 hours, or more.
I ran 32 n_epochs
on a single-line JSONL a few days ago and it took four hours.
HTH