Hi I want to build an ai app for production to deploy for users to utilize it, i plan on commercializing this product as well at some point. What are best practices? currently using openai api and streamlit but wonder if thats the best thing to do moving forward. If i should go to fine tuning own custom model and then making inference from that. I will be training on specific use case. Any advice appreciated.
#Production Question
spare barn
tardy cedar
Hey @spare barn , fine-tuning is not necessary in most cases, usually RAG or retrieval augmented generation is enough. With RAG you retrieve relevant documents for a query using a vector database and then add those documents to your prompt.
To help you out more specifically I'll need to know more about your app. You can DM me if you dont wanna share details here.