The most wanted feature that you can see during Keynote at Build 2023. SO you can add custom data to gpt-3.5-turbo and than use the chat gpt. Also you see how to configure chat gpt.
See step by step – how I build the public ChatGPT than add a Custom Data to answer questions about company products. Finally deploy the chat.
Some important info about configuration that we can adjust:
Azure OpenAI on your data enables enterprise users to utilize OpenAI’s powerful conversational AI models, such as ChatGPT and GPT-4, on their own data while complying with their organizational policies. Azure OpenAI service is combined with Azure Cognitive Search to index and retrieve data that is private and external to the ChatGPT large language model. The retrieval step in Azure Cognitive Search finds the most relevant pieces of information and presents the top ranked results to the language model. And because the knowledge lives outside of the ChatGPT model, you’re in control — it is not used to train the model.
Your data source is used to help ground the model with specific data. Azure OpenAI on your data provides many options of data storage to select from, including an existing Azure Cognitive Search index, Azure Storage container, or upload local files. For this lab, we have already created the blob storage container, uploaded the PDF documents containing Contoso’s employee and medical policies and created the Azure Cognitive Search resource in advance. When you try out this feature outside of the lab, make sure that all this is in place before you move forward.
System message is included at the beginning of the prompt and gives the model instructions about how it should behave and any context it should reference when generating a response. You can describe the assistant’s personality, tell it what it should and shouldn’t answer, and tell it how to format responses.
Few-shot examples show the chat what responses you want. The model will try to mimic any responses you add here so make sure they match the rules you laid out in the system message.
Parameters can be adjusted to control the behavior of the model and improve the quality of the generated output for a scenario. For example, temperature is the sampling temperature the model uses, between 0 and 1. Higher values means the model will take more risks. Try 0.9 for more creative applications, and 0 for ones that need a well-defined answer
Deploy to a web app is a new feature within the ChatGPT playground that enables you to create a ChatGPT powered web interface with all your specified parameter values, data connectors and other settings, with the clicks of a few buttons.
You can try-out the application without leaving the Studio and share it with other members of your organization to try out.
The web app is a lightweight Python Flask application that runs in a Linux environment hosted on Azure App Service.
The video taken from LAB i did during Microsoft Build 2023.