Azure OpenAI Service provides REST API access to OpenAI's powerful language models, including the GPT-3, Codex, and Embeddings model families. Also, the new GPT-4 and ChatGPT (gpt-35-turbo) model series have been officially launched. These models can be easily adapted to your specific tasks, including but not limited to content generation, summarization, semantic search, and natural language to code translation. Users can access the service through the REST API, Python SDK, or our web-based interface in Azure OpenAI Studio.
Feature Overview
Features
Azure OpenAI
## Available models
New GPT-4 Series GPT-3 Basic Series New Charter (gpt-35-turbo)Code Series Embedded Series Learn more on our models page.
Nudge
Ada
Babbage Curie Cushman Davinci Nudge is currently not available for new customers.
Price
Available here
Virtual Network Support and Private Link Support
Yes
Managed Identity
Yes, through Azure Active Directory
User Interface Experience
Azure
Portal for account and resource management, Azure OpenAI Service Studio for model exploration and fine-tuning
Regional Availability
Eastern United States South Central United States
Western Europe
France Central
Content filtering
Use automated systems to filter prompts according to our content policies and complete for evaluation. High severity content will be filtered.
Responsible Artificial Intelligence
At Microsoft, we are committed to advancing AI with human-centered principles. Generative models, such as those available in Azure OpenAI, have significant potential benefits, but without careful design and thoughtful mitigation, such models can generate incorrect or even harmful content. Microsoft has made significant investments to help prevent abuse and unintentional harm, including requiring applicants to demonstrate well-defined use cases, incorporating Microsoft's principles for responsible use of AI, building content filters to support customers, and providing onboarded customers with a responsible Responsible AI Implementation Guide.
How to access Azure OpenAI?
How to access Azure OpenAI?
Currently, access is restricted because we understand the height of market demand, upcoming product improvements, and Microsoft's commitment to responsible AI. We are currently working with customers who have partnerships with Microsoft, lower-risk use cases, and customers who are working on incorporating mitigations.
More specific information is included in the application form. Thank you for your patience as we work to ensure Azure OpenAI is more widely accessible and held accountable.
Apply here for access:
Apply now
Comparing Azure OpenAI and OpenAI
Azure OpenAI Service gives customers advanced language AI with OpenAI GPT-4, GPT-3, Codex, and DALL-E models with the security and enterprise promise of Azure. Azure OpenAI co-develops the APIs with OpenAI, ensuring compatibility and a smooth transition from one to the other.
With Azure OpenAI, customers get the security capabilities of Microsoft Azure while running the same models as OpenAI. Azure OpenAI offers private networking, regional availability, and responsible AI content filtering.
Key Concepts
Tips and Completions
The completion endpoint is the core component of the API service. This API provides access to the text input and text output interfaces of the model. The user simply provides input hints containing English text commands, and the model generates text completions.
Here is a simple tip and completed example:
Tip:""" count to 5 in a for loop """
Complete: for i in range(1, 6): print(i)
Token
Azure OpenAI processes text by breaking it into tokens. Markers can be words or just blocks of characters. For example, the word "hamburger" is broken down into the tokens "ham", "bur" and "ger", while short and common words like "pear" are single tokens. Many tokens start with a space, such as "hello" and "bye".
The total number of tokens processed in a given request depends on the length of the input, output, and request parameters. The number of tokens being processed also affects the model's response latency and throughput.
Resources
Azure OpenAI is a new product on Azure. You can start using Azure OpenAI like any other Azure product by creating a resource or service instance in your Azure subscription. You can read more about Azure resource management design.
Deployment
After you create an Azure OpenAI resource, you must deploy the model before you can start making API calls and generating text. This can be done using the deployment API. These APIs allow you to specify the model to use.
Contextual Learning
The model used by Azure OpenAI uses natural language instructions and examples provided during the build call to determine the requested task and required skills. When using this method, the first part of the prompt includes natural language instructions and/or examples of the specific task required. The model then completes its task by predicting the most likely next piece of text. This technique is called "in-context" learning. Predictions will be provided based on the context contained in the hints, these models will not be retrained in this step.
There are three main methods of contextual learning: few-shot, single-shot and zero-shot. These methods vary depending on the amount of task-specific data provided to the model:
Few Shots: In this case, the user includes several examples in the call prompt that demonstrate the expected response format and content. The following example shows a several shot prompt where we provide multiple examples (the model will generate the last answer):
Convert the questions to a command:Q: Ask Constance if we need some bread.A: send-msg `find constance` Do we need some bread?Q: Send a message to Greg to figure out if things are ready for Wednesday.A: send-msg `find greg` Is everything ready for Wednesday?Q: Ask Ilya if we're still having our meeting this evening.A: send-msg `find ilya` Are we still having a meeting this evening?Q: Contact the ski store and figure out if I can get my skis fixed before I leave on Thursday.A: send-msg `find ski store` Would it be possible to get my skis fixed before I leave on Thursday?Q: Thank Nicolas for lunch.A: send-msg `find nicolas` Thank you for lunch!Q: Tell Constance that I won't be home before 19:30 tonight — unmovable meeting.A: send-msg `find constance` I won't be home before 19:30 tonight. I have a meeting I can't move.Q: Tell John that I need to book an appointment at 10:30.A:
The specific number of examples is determined by the number of examples that can be accommodated by the maximum input length of a single prompt, which is usually between 0 and 100. The maximum input length may vary depending on the specific model you are using. Few-shot learning can significantly reduce the amount of task-specific data required for accurate predictions. This method is generally less accurate than fine-tuned models.
Single Shot: This case is the same as the few shot method, except that only one example is provided.
Zero Shot: In this case, no examples will be provided to the model, only task requests.
Models
The service provides users with access to several different models. Each model offers different features and price points.
The GPT-4 model is the latest available. Due to high demand, this model range is currently only available on request. To request access, existing Azure OpenAI customers can do so by filling out this form
The GPT-3 base models are known as Leonardo da Vinci, Curie, Babbage and Ada, in order of decreasing capability and Arranged in increasing order of speed.
The Codex family of models are descendants of GPT-3 trained on natural language and code, designed to support natural language to code conversion. Learn more about each model on our model concepts page.
Related Articles
How to Highlight Duplicates in Excel
How to Highlight Duplicates in Excel The following is about how to highlight duplicates in Excel Steps to highlight duplicates in Excel: Select the data that you suspect may be duplicates. Click Home > Conditional Formatting > Highlight Cell Rules > Duplicate Values. By default, when you open the Duplicate Values dialog box, "Fill...
How to use Nearby Sharing to share text from Windows PC to Android
How to use "Nearby Sharing" Whether you are sharing files, photos, videos, or now sharing text and URLs, this versatile tool has simplified our digital lives and made sharing a breeze. Listed below are the tools for sharing text from PC Nearby Sharing steps required to get to Android:...
Advanced Formula Environment is becoming Excel Labs, a Microsoft Garage Project!
The experiment is The key to unlocking innovation in any field. It's how we learn and evolve, and it's what drives us to keep creating more features that help you multitask. So today, we're excited to announce that we're doing just that by launching Microsoft Garage Project Excel Labs continues to invest in...
ChromeOS Flex USB can now be set up on Linux systems
Google launches Flex USB for desktops and laptops ChromeOS Flex. One very annoying decision Google made at the time was that the installer USB had to be set up using a Chromebook recovery utility that was incompatible with Linux distributions like Ubuntu and Linux Mint. Unless you have another one lying around...
The above is the detailed content of What is Azure OpenAI Service?. For more information, please follow other related articles on the PHP Chinese website!
