Is fine-tuning the solution for bringing some context to specifics topics


I can’t find if fine-tuning is a solution for my needs :

I wan’t to bring to openai some informations, mainly press articles and local datas, in order to make text generation/summarization about specifics topics that openai doesn’t even know.

My datasets are not very large, typically 5 to 10 articles for an amount of 8-10.000 words on which i want Open ai proceed summarization.

It’s too big for classic prompts contect in playgroud, so how can i bring this data to openai ? Is fine tuning the solution ?

Thank you very much for your help !

I just found this openai blog article about Book summarization :

I also search about Embeddings but I don’t see information about text generation from embeddings.

Please help me find out what could suite my needs, i’m absolutely lost ! :wink:

Hello Myself,

Just to tell you that you move to this very interresting topic : Train (fine-tune) a model with text from books or articles - #40 by Pierre-Jean

You should explore the use of embeddings for this task.