کتاب هوش مصنوعی مولد در مایکروسافت آژور

مبانی فنی مدل‌ها و برنامه‌های کاربردی هوش مصنوعی مولد

Evolution of the Adoption Patterns • Types of Generative AI Models • Technical Concepts Behind Generative AI • Pretraining • Post-Training • Inference • Mapping to Microsoft Azure • Conclusion

از نمونه اولیه تا تولید: هوش مصنوعی در آژور

Overview of the Advanced AI Ecosystem on Azure • Deep Dive on Microsoft Foundry: Models, Workflows, and Safety • Key Capabilities of Microsoft Foundry • Model Diversity and Multiprovider Strategies • Azure AI Model Catalog: Many Models, One Platform • Model Providers • Deployment Patterns on Azure • Building Agents in Azure • Microsoft Copilot and Copilot Studio • Integration Patterns with Copilot Studio • Putting It Together: Copilot Integration Scenarios • Vector Databases • Azure AI Search • Azure Cosmos DB with Vector Search • Selection Criteria and Best Practices • Typical Use Cases for Vector DBs in Generative AI Solutions • Additional Data Platforms • Azure Databricks • Snowflake Cortex on Azure • NVIDIA NIM on Azure • Web Application Frameworks for AI • Azure App Service for AI Web Apps • Streamlit: Rapid UI for Data and AI • GitHub Copilot Models and Integrations • Models Behind GitHub Copilot • Copilot Integrations • GitHub Copilot Integrations and Ecosystem • Conclusion

انتخاب مدل و بهترین شیوه‌های مهندسی پرامپت

Considerations for Model Selection • Choosing Your Model on the Microsoft Foundry • Selection via Model Catalog • Model Deployment Types • Azure OpenAI Models • Additional Microsoft Foundry Capabilities • Prompt Engineering Best Practices • Laying the Foundations for Effective Prompts • Prompt Engineering Techniques • Key Takeaways • Prompt Engineering in Microsoft Foundry • Conclusion

تولید افزوده بازیابی

Fundamentals of RAG • The RAG Process • RAG in Azure • RAG with Azure AI Search • Data Ingestion and Indexing • Retrieval and Generation Workflow • Iterative Optimization of Your RAG Pipelines • New Foundry and Foundry IQ • Advanced RAG Scenarios • Multimodal RAG • GraphRAG • NL2SQL • Agentic RAG • RAFT • Looking Ahead with RAG • Conclusion

تنظیم دقیق مدل‌های هوش مصنوعی مولد در آژور

When to Fine-Tune Your Generative AI Models • Prompt Engineering Versus RAG Versus Fine-Tuning • Fine-Tuning Taxonomy • SFT • PEFT • Fine-Tuning in Microsoft Foundry • Fine-Tuning Azure OpenAI Models • Step-by-Step Fine-Tuning Workflow for Azure OpenAI Models • Step-by-Step Workflow for Fine-Tuning Other Catalog Models in Microsoft Foundry • Fine-Tuning Best Practices and Pitfalls • Conclusion

سیستم‌های عامل

When to Use Agentic Systems • Agentic Protocols • MCP in Detail • Agentic Design Patterns • Sequential Pattern • Workflow Pattern • Concurrent Pattern • Group-Chat Pattern • Self-Reflection/Cross-Reflection Pattern • Handoff Pattern/Human in the Loop • Supervisor Pattern • Magentic Orchestration Pattern • Summary of Patterns • Building Agentic Systems in Azure • How the Microsoft Foundry Agent Service Works • Supercharging Agents: Tool Catalog and Foundry IQ • Foundry Workflows: Orchestrating Multiagent Processes • Microsoft Agent Framework: From Semantic Kernel and AutoGen to Present • Hosted Agents: Deploying Agent Framework Apps to Azure • Conclusion

GenAIOps و LLMOps در آژور

From MLOps to GenAIOps: Why It Matters • Generative AI in Azure: The Operating Loop • Use Case: Multiagent Financial Compliance Assistant • Define and Explore • Quantitative Analysis and Monitoring: Core Model Performance Metrics • Model Selection and Evaluation • Tools and Frameworks • Build and Customize • Designing the Application Architecture • Build and Customize: Example • Experimentation and Prompt Engineering • Evaluation and Validation • Code Templates and Accelerators • Fine-Tuning Guidance • Observe and Optimize • CI/CD • Deployment and Release Management • Monitoring in Production • Automation of Model Lifecycle Management • Tools for Tracking, Versioning, and Reproducibility • Protect and Govern • Safety and Guardrails by Design • Security and Access Control • Governance and Accountability • Linking Protect and Govern with the GenAIOps Lifecycle • Conclusion

چارچوب حاکمیت هوش مصنوعی مولد

Thought Process for Your AI Governance • Levels of AI Governance • Typical AI Governance Assets • Internal Context at Organizational Level • Other Considerations • Key Building Blocks • Azure AI Content Safety • Microsoft Purview • Microsoft Defender • Microsoft Entra • Microsoft Responsible AI (RAI) Toolbox • PyRIT (Python Risk Identification Tool) • HAX (Human-AI eXperience) Toolkit • AI Bill of Materials (AI BOM) • HiddenLayer • How-To: Implementing Generative AI Governance on Azure • Preliminary, General First Steps • Human + AI Approach • Securing Your AI Systems Against Human Misuse • AI Models, Data, and Infrastructure • Last but Not Least: Are You the Responsible AI Champion? • Conclusion

مصاحبه با کارشناسان

Innovating and Exploring with Microsoft Azure, with Marco Casalaina • The Art of Designing AI Architectures, with James Serra • Building the Future of AI Platforms, with Eric Boyd • The Evolution and Future of Frontier AI, with Yina Arenas