Feeding custom data set through OPEN AI API

Question

We are trying to feed some information to gpt model and get answer from model only based on the information passed.

1. How to do this using python?
2. How to pass dataset for training.
3. How to pass multiple datasets or a folder or a package of files.
4. How to get response only based on this specific data
5. Example use cases

Answer 1

@Prasad Bala Raghavendra, Technology Here are the answers to your questions below

How to do this using python?

You can get code snippets directly from the Azure OpenAI Studio UI by clicking on the View code button in the Chat View as shown below

How to pass dataset for training.

The Azure OpenAI on your Data uses cognitive search to index all your data and make it available for semantic search. You will have to first ingest data into Azure Cognitive Search, which could then be used to fetch data during requests.

How to pass multiple datasets or a folder or a package of files.

Like above, you would do this to Azure Cognitive Search first, which can later be used to retrieve relevant information during requests.

How to get response only based on this specific data

There is an option to do this as part of Azure OpenAI on your Data that you can enable.

Example use cases

The main use case for such a setup is to provide a natural language interface on existing knowledge bases. Along with semantic search and OpenAI models, you get a powerful interface for your users that not only allows them to quickly get the information they need but also in a way that is natural to the user. Instead of simply finding the right document, they get the exact information that they are looking for.