That sounds like a very bad idea, chatgtp should not be a part of the infrastructure.
chatgpt api integrate in laravel project
I want to get the response from my custom database using openai package? how it is possible? how can i use the package that will train ai model according to my requirements?
I'm not sure what exactly you're trying to do. Sounds like an XY-Problem. What is your end goal?
@kiwi0134 actually i want to train gprt-3.5 turbo model on my custom data. like chat gpt respond on the general data but not have specific data e.g chatgpt does not know about jimmy location or salary but i have documents that have infromation of specific users. these documents are thousands so i have integrate the openai key. now i want to train the model. my problem is how can i train the model on my customdata? what techniques i will used? i am working in laravel so i want to know is laravel have any package for it?
@Rehmatweb First I have to clarify that I'm the exact opposite of an expert about this topic.
I did some basic research and found out, that you probably cannot upload documents to OpenAI to train a custom LLM. But you can develop your own service which makes use of OpenAI in combination with custom data.
Do these help you?
- https://medium.com/@sohaibshaheen/train-chatgpt-with-custom-data-and-create-your-own-chat-bot-using-macos-fb78c2f9646d
- https://medium.com/@sushwanth.n/how-to-train-chatgpt-with-your-custom-data-and-create-your-own-chatbot-6d525fc7f20f
They both look promising to me. Instead of using the UI the tutorials uses, you could build a simple internal endpoint for your service which accepts a query and returns the data.
I also found the following topic here on Laracasts:
But be careful to not mix data of different users, unless wanted.
@kiwi0134 thanks for guidence this article medium.com is related to python
I want to achieve the same goal but in laravel. like GPT-index and such type of packages/libraries are not found in laravel.
@Rehmatweb Yes, thats correct. There seems to be no Laravel package to do the same. Despite the idea to develop a small microservice in Python and then use that from Laravel, for example via an internal HTTP request to that microservice.
how can we tune a gpt-3.5-turbo model through customdata in laravel
Please or to participate in this conversation.