How to Fine Tune my data?

Hi all!
I am very new to GPT 3 . Can someone explain to me on how to fine tune GPT-3 with my very own dataset?

2 Likes

Hi @Sour_1

@m-a.schenk has shared the OpenAI guide (the most definitive ) to fine-tune your GPT-3 engine. Just here to add the OpenAI api reference for fine-tune

Also since you mentioned you’re new to GPT-3, perhaps the best place is to start with Docs, and explore the full range of functionality OpenAI’s GPT-3 has to offer. You may also want to read about prompt design and experiment in Playground

Cheers!

2 Likes

Thank you! I have been experimenting in the playground . Also read the documents you mentioned . I am started to get how to work with GPT-3.

Thank you @m-a.schenk! I have read the guide you had mentioned. I now have a fair idea about GPT-3. The problem I am facing is the token limitation in training the engine on how to respond to specific questions that might be asked to it.

1 Like

@m-a.schenk I am trying to create a response generator, which will respond to questions asked to it with some unique answers which I am currently trying to train the model on.

1 Like

I am trying to fine-tune my own model, not an existing one. I am not planning to use the answers endpoint. Yes I want to create embeddings of the answers.
I am hitting the token limit while filling the prompt with few shot learnings approach.

1 Like

Thanks a lot @m-a.schenk ! I do need help with what I am doing. Should I connect to you here via personal messages or any other platform?
Thanks in advance.