Hello guys,
I want to update the training data given to the fine-tuned model which I created using davinci model.
Regards
Prakeet Singh
Hello guys,
I want to update the training data given to the fine-tuned model which I created using davinci model.
Regards
Prakeet Singh
You would need to fine-tune again, I believe, using that model’s name.
Ok, thank you, sir. That costs money. Actually, I was testing the model so needed to find tune again n again
That depends on your plan. Are you on the free plan? If so, fine-tuned models are not available after the session has closed (according to the docs, but they still show up in my list); however the “enterprise plans”, fine-tuned models are available to be used after a session is closed.
Sounds like you are on the free $18 worth of credits plan, @Prakeet ?
Did you list your models?
ruby-openai
gem)irb(main):006:1* Ruby::OpenAI.configure do |config|
irb(main):007:1* config.access_token = ENV.fetch('OPENAI_API_KEY')
irb(main):008:0> end
irb(main):009:0> client = OpenAI::Client.new
irb(main):010:0> client.models.list
HTH
We are using enterprise plan. I think we can re fine tune already fine tuned model right? (Like updating the data). Just was looking for the cli code in order to do that.
Yes, you can list all your models and re-fine-tune those models if you wish (especially if you are on the Enterprise plan as you mention). Let’s say the name of your fine-tuned model you want to use is:
davinci:ft-personal-2023-01-19-10-18-00
Then you can further fine-tune with something like this example:
response = client.finetunes.create(
parameters: {
training_file: file_id,
model: "davinci:ft-personal-2023-01-19-10-18-00"
})
fine_tune_id = JSON.parse(response.body)["id"]
Of course, the actually code is based on the programming language and libs you are using. The code snippet above is in Ruby using the ruby-openai
gem, and is only a snippet as an example.
HTH
Ok, thank you for your answer.
Is there any CLI code for updating the fine-tuned model, like the code for creating fine tuned model is " openai api fine_tunes.create -t <TRAIN_FILE_ID_OR_PATH> -m <BASE_MODEL> "
I just wanted to know to CLI code for updating the created model.
The code I shared with you previously is updating a fine-tuned model.
Do you recognize the naming convention of a fine-tuned model?
Alright, I’ll check with it. Thank you for your help!
Hello, I tested this way, but not working, I will generate a new model name with new time, only the new name model can work with newest fine-tune training data. This is not good for fine-tune a specified model. Is there the other way can be used just update/append training data for a specified model(do not generate a new model)? I found there is a interface “openai.FineTune.Update()”, but I don’t know what is it use for and how to use it, is there anyone can tell me?
Thanks.