TomazTsql | Tomaz doing BI and DEV with SQL Server and R ...
In this Microsoft Azure AI series:
Documentation and Resources for Azure AI Foundry
Documentation and material for Azure AI Foundry are plentiful and growing on a daily basis, since the topic on AI and GenAI is evermore so popular. The general and best way to start is with Microsoft own product introduction website: https://azure.microsoft.com/en-us/products/ai-studio and later you can scale with different material from many different websites.
Detailed product documentation is on Microsoft learn: https://learn.microsoft.com/en-us/azure/ai-studio/
Video series:
Learning paths:
Books:
Code and samples:
Many introduction articles are available as well as many listings for the use cases. All of the code samples will be available on my Github.
In this Microsoft Azure AI series:
Evaluation in Azure AI Foundry
With evaluation you performing iterative, systematic evaluations with the right evaluators and measure and address potential response quality, safety, or security concerns throughout the AI development lifecycle, from initial model selection through post-production monitoring. With the Evaluation in Azure AI Foundry, you can evaluate the GenAI Ops Lifecycle production. In addition, it also gives you the ability to assess the frequency and severity of content risks or undesirable behavior in AI responses.
With each model, you can consider and evaluate:
- Evaluation of the models
- Evaluate different metrics
And you can start with evaluation when the deployment is ready, all relevant metrics selected, prompts added and data integration (input and output) defined. The results of the evaluation can be analyzed with the Python SDK as well.
Tomorrow we will look into documentation for Azure AI Foundry. All of the code samples will be available on my Github.
In this Microsoft Azure AI series:
Tracing with Azure AI Inference SDK
Tracing is a powerful tool that offers developers an in-depth understanding of the execution process of their generative AI applications. Though still in preview (in the time of writing this post), It provides a detailed view of the execution flow of the application and the essential information for debugging or optimisations.
Tracing with the Azure AI Inference SDK offers enhanced visibility and simplified troubleshooting for LLM-based applications, effectively supporting development, iteration, and production monitoring. Tracing follows the OpenTelemetry semantic conventions, capturing and visualizing the internal execution details of any AI application, enhancing the overall development experience.
Key advances of using tracing are:
- Creating a new app or connecting to an existing one
- Using the Kusto language for analyzing the logs for attributes, content, spans, evaluation events, and others
- Collecting and analyzing log traces with Python SDK
Tomorrow we will look into Evaluation in Azure AI Foundry. All of the code samples will be available on my Github.
In this Microsoft Azure AI series:
Prompt Flow in Azure AI Foundry
Prompt Flow is particularly beneficial for organizations leveraging AI to streamline operations, enhance customer experiences, and innovate in digital transformation projects.
With Python, you can start using prompt flow by installing the package and create a prompty file. For additional information on Prompt flow, visit: https://microsoft.github.io/promptflow/index.html
The main building blocks of Prompt flow are:
- Creating standard flow, chat flow, or evaluation flow
- Web Classification flow example
You can start building flow with the three predefined types or clone the prepared flows and deploy it based on your needs.
Tomorrow we will look into building prompt flow in VS Code using Python. All of the code samples will be available on my Github.
In this Microsoft Azure AI series:
Model Deployment in Azure AI Foundry
Models from the model catalog can be deployed using programming languages or using the Foundry studio. Model deployment has two types: Deploy from the base model or deploy from the fine-tuned model.
Each model has all the necessary detail information, as well all the metrics to check number of requests, prompt token counts, number of requests and additional information on model usage.
Models from the model catalog can also be deployed as a serverless API with pay-as-you-go billing. Deployment with Python SDK is straightforward.
Tomorrow we will look into the prompt flow in Azure AI Foundry. All of the code samples will be available on my Github.
