Azure OpenAI for your data brings you a fresh service that links Azure OpenAI's robust language models with your personal information. This lets you utilize Azure OpenAI to create text, convert languages, compose various types of imaginative content, and respond to your queries with insightful answers, all drawing from your own data.
In this article, you'll grasp the concept of Azure OpenAI on your data, its fundamental workings, getting started directions, and the model deployment procedure.
Table of Contents:
Data Formats and File Types
3. Get Started
5. Conclusion
What is Azure OpenAI on your Data?
Azure OpenAI on your data allows you to use OpenAI's powerful language models to chat with your customers and analyze your data. This is done without the need to train or fine-tune the models, which saves you time and money.
The model can access and reference specific sources to support its responses, so the answers are not only based on its pre-trained knowledge but also on the latest information available. This helps to ensure that the model is providing accurate and up-to-date information.
One of the key features of Azure OpenAI on your data is its ability to retrieve and utilize data in a way that enhances the model's output. The model will retrieve relevant data from your data source and append it to the user's prompt. This helps the model to provide more accurate and informative responses.
Here are some of the key benefits of Azure OpenAI on your data:
Accurate and up-to-date information: The model can access and reference specific sources to support its responses, so the answers are not only based on its pretrained knowledge but also on the latest information available. This helps to ensure that the model is providing accurate and up-to-date information.
Tailored content: The model can be tailored to your specific needs and data. This means that you can ensure that the model is providing the information that is most relevant to your customers.
Enhanced chat experience: Azure OpenAI on your data can be used to create a more engaging and informative chat experience for your customers. This can help to improve customer satisfaction and loyalty.
Improved decision-making: Azure OpenAI on your data can be used to gain insights from your data. This can help you to make better business decisions, identify trends and patterns, and optimize your operations.
Data Formats and File Types
Azure OpenAI on your data supports the following filetypes:
.txt
.md
.html
Microsoft Word files
Microsoft PowerPoint files
PDF
Azure OpenAI on your data has an upload limit.
The data upload limit for Azure OpenAI on your data is 16 MB per upload. This limit applies to all file types, including text, images, and audio.
If you need to upload more than 16 MB of data, you can split it into multiple 16 MB files and upload them separately. You can also use a third-party tool to compress your data before uploading it.
Here are some tips for optimizing your data upload:
Use a compression tool to compress your data before uploading it. This can significantly reduce the file size without affecting the quality of the data.
Split your data into smaller files. This will allow you to upload more data without exceeding the 16 MB limit.
Upload your data during off-peak hours. This will help you avoid any performance issues that may be caused by high traffic.
How it works
In this section, we will learn how the data is processed. But before that, you should know what kind of data is processed. Azure OpenAI processes three different types of data:
Prompts and generated content: When you use Azure OpenAI, you submit a prompt, which is a short piece of text that describes what you want the service to generate. The service then generates content, such as text, images, or embeddings, based on your prompt.
Augmented data included with prompts: When you use the "on your data" feature, Azure OpenAI retrieves relevant data from a configured data store and augments your prompt with this data. This helps the service to generate content that is more relevant to your data.
Training and validation data: You can also provide your own training data to Azure OpenAI. This data consists of prompt-completion pairs, where each pair is a prompt and the corresponding content that you want the service to generate. Azure OpenAI uses this data to fine-tune its models, which can improve the quality of the content that it generates.
Consider the below diagram that illustrates how your data is processed.
In the above image, data flows for inference and training in Azure OpenAI. The following are the key components of the data flow:
Prompts: These are the short pieces of text that are submitted by the user to Azure OpenAI.
Generated content: This is the text, images, or embeddings that are generated by Azure OpenAI based on the prompts.
Base models: These are the pre-trained models that are used by Azure OpenAI to generate content.
Finetuned models: These are the models that have been fine-tuned using customer data.
Files storage: This is where the customer data is stored.
The data flow works as follows:
The user submits a prompt to Azure OpenAI.
Azure OpenAI retrieves the relevant base model and finetuned models.
Azure OpenAI generates content based on the prompts and the models.
The generated content is returned to the user.
The data flows for inference and training are different in the following ways:
In inference, the customer data is not used to fine-tune the models.
In training, the customer data is used to fine-tune the models.
Get started
Azure OpenAI offers a no-code approach to exploring its capabilities through the Chat Playground. This environment allows you to experiment, iterate, and generate completions based on your prompts. It also provides an easy-to-use interface for exploring the capabilities of Azure OpenAI Service models and offers Python and curl code samples for integration into your own applications.
To access the Chat Playground, you will need the following:
An Azure subscription
Access granted to Azure OpenAI (currently available to approved enterprise customers and partners)
An Azure OpenAI resource with a Chat model (GPT-3 or GPT-4)
The Assigned Cognitive Services Contributor role for the Azure OpenAI resource
Here are the steps to access the Chat Playground:
STEP 1: Go to Azure OpenAI studio and Select "Chat Playground".
STEP 2: Now, under the "Assistant setup", select "+ Add a data source".
STEP 3: In the Data source tab, select "Upload files" under "Select or add data source".
Azure OpenAI needs both a storage resource and a search resource to access and index your data.
To access the storage account, you need to turn on Cross-origin resource sharing (CORS). Click Turn on CORS to turn it on.
CORS enables a web application running under one domain to access resources in another domain.
STEP 4: Select the Azure Cognitive Search resource and enter the Index name. After acknowledging that the connection to the Azure Cognitive Search account will incur usage to your account, click Next.
STEP 5: On the "Upload files" tab, select "Browse for a file" and select the files you want to upload". Select "Upload Files" and click Next.
STEP 6: On the Data Management tab, select whether you want to enable Semantic search or Vector search on your index. Semantic search and vector search require additional pricing.
STEP 7: Review all the details you entered, and select "Save and close".
Once you have completed these steps, you will be able to access the Chat Playground and start exploring the capabilities of Azure OpenAI.
How to Deploy the Model
After you connect Azure OpenAI to your data, you can deploy it using the following options:
Power Virtual Agents: You can deploy your model to Power Virtual Agents directly from Azure OpenAI Studio. This will allow you to create a chatbot that will respond using your data and connect it to the Power Platform.
Web app: You can also use the available web app to interact with your model using a graphical user interface. You can deploy this app using either Azure OpenAI Studio or a manual deployment.
You can also customize the web app's frontend and backend logic. For example, you could change the icon that appears in the center of the app by updating the /frontend/src/assets/Azure.svg file and then re-deploying the app using the Azure CLI.
Conclusion
Throughout the course of this article, you've acquired insights into Azure OpenAI integrated with your data—its operational mechanics and practical application for data generation.