Hello,
I have an online store on the Opencart platform.
I am interested in making a chatbot through the Openai API.
But, I would like to fine-tune them according to the information on the website in order to be able to offer recommendations to customers or pertinent answers, including details about orders, etc.
Do you have any idea how I can start?
Thank you!
Sincerely,
Stelian
Fine-tuning is typically for behavioral changes, and let me ask you: if you stop carrying a product, or a property of the product changes what will you do? Fine-tune the fine-tune?
What happens if the question requires some logic to process? “What are your most recent 30 products?”.
What you are looking for is RAG. You need a database that can support embeddings (vectors). You want to keep knowledge separated, deterministic, easy to modify, and provide you an opportunity to perform background logic on the query.
Yes, RAG is the right approach and those links are great if you would like to implement it yourself. If you are looking for a solution that is already working well and does not require coding please take a look at cetient. I am a co-founder and would be glad to help you get set up. Thanks!
Hello @stelian99,
Im the creator of OpenAssistantGPT and I have many customer creating their chatbot using my app. They build their chatbot in OpenAI and once it is done they import it in our app and then add it to their app.
Let me know if this help!