Generative AI for Cloud Solutions Architect Modern AI LLMs in Secure, Scalable, and Ethical Cloud Environments
Generative artificial intelligence technologies and services, including ChatGPT, are transforming our work, life, and communication landscapes. To thrive in this new era, harnessing the full potential of these technologies is crucial. Generative AI for Cloud Solutions is a comprehensive guide to und...
| Autor principal: | |
|---|---|
| Otros Autores: | , |
| Formato: | Libro electrónico |
| Idioma: | Inglés |
| Publicado: |
Birmingham :
Packt Publishing, Limited
2024.
|
| Edición: | 1st ed |
| Materias: | |
| Ver en Biblioteca Universitat Ramon Llull: | https://discovery.url.edu/permalink/34CSUC_URL/1im36ta/alma991009815725606719 |
Tabla de Contenidos:
- Cover
- Title page
- Copyright and credits
- Dedication
- Foreword
- Contributors
- Table of Contents
- Preface
- Part 1: Integrating Cloud Power with Language Breakthroughs
- Chapter 1: Cloud Computing Meets Generative AI: Bridging Infinite Impossibilities
- Evolution of conversation AI
- What is conversational AI?
- Evolution of conversational AI
- Introduction to generative AI
- The rise of generative AI in 2022-23
- Foundation models
- LLMs
- Core attributes of LLMs
- Relationship between generative AI, foundation models, and LLMs
- Deep dive - open source vs closed source/proprietary models
- Trending models, tasks, and business applications
- Text
- Image
- Audio
- Video
- Cloud computing for scalability, cost optimization, and security
- From vision to value - navigating the journey to production
- Summary
- References
- Chapter 2: NLP Evolution and Transformers: Exploring NLPs and LLMs
- NLP evolution and the rise of transformers
- The main drawbacks of RNNs and CNNs
- NLP and the strengths of generative AI in LLMs
- How do transformers work?
- Benefits of transformers
- Conversation prompts and completions - under the covers
- Prompt and completion flow simplified
- LLMs landscape, progression, and expansion
- Exploring the landscape of transformer architectures
- AutoGen
- Summary
- References
- Part 2: Techniques for Tailoring LLMs
- Chapter 3: Fine-Tuning - Building Domain-Specific LLM Applications
- What is fine-tuning and why does it matter?
- Fine-tuning applications
- Examining pre-training and fine-tuning processes
- Pre-training process
- Fine-tuning process
- Techniques for fine-tuning models
- Full fine-tuning
- PEFT
- RLHF - aligning models with human values
- How to evaluate fine-tuned model performance
- Evaluation metrics
- Benchmarks.
- Real-life examples of fine-tuning success
- InstructGPT
- Summary
- References
- Chapter 4: RAGs to Riches: Elevating AI with External Data
- A deep dive into vector DB essentials
- Vectors and vector embeddings
- Vector search strategies
- When to Use HNSW vs. FAISS
- Recommendation System for Articles
- Vector stores
- What is a vector database?
- Vector DB limitations
- Vector libraries
- Vector DBs vs. traditional databases - Understanding the key differences
- Vector DB sample scenario - Music recommendation system using a vector database
- Common vector DB applications
- The role of vector DBs in retrieval-augmented generation (RAG)
- First, the big question - Why?
- So, what is RAG, and how does it help LLMs?
- The critical role of vector DBs
- Business applications of RAG
- Chunking strategies
- What is chunking?
- But why is it needed?
- Popular chunking strategies
- Chunking considerations
- Evaluation of RAG using Azure Prompt Flow
- Case study - Global chat application deployment by a multinational organization
- Summary
- References
- Chapter 5: Effective Prompt Engineering Techniques: Unlocking Wisdom Through AI
- The essentials of prompt engineering
- ChatGPT prompts and completions
- Tokens
- What is prompt engineering?
- Elements of a good prompt design
- Prompt parameters
- ChatGPT roles
- Techniques for effective prompt engineering
- N-shot prompting
- Chain-of-thought (CoT) prompting
- Program-aided language (PAL) models
- Prompt engineering best practices
- Bonus tips and tricks
- Ethical guidelines for prompt engineering
- Summary
- References
- Part 3: Developing, Operationalizing, and Scaling Generative AI Applications
- Chapter 6: Developing and Operationalizing LLM-based Apps: Exploring Dev Frameworks and LLMOps
- Copilots and agents.
- Generative AI application development frameworks
- Semantic Kernel
- LangChain
- LlamaIndex
- Autonomous agents
- Agent collaboration frameworks
- AutoGen
- TaskWeaver
- AutoGPT
- LLMOps - Operationalizing LLM apps in production
- What is LLMOps?
- Why do we need LLMOps?
- LLM lifecycle management
- Essential components of LLMOps
- Benefits of LLMOps
- Comparing MLOps and LLMOps
- Platform - using Prompt Flow for LLMOps
- Putting it all together
- LLMOps - case study and best practices
- LLMOps field case study
- LLMOps best practices
- Summary
- References
- Chapter 7: Deploying ChatGPT in the Cloud: Architecture Design and Scaling Strategies
- Understanding limits
- Cloud scaling and design patterns
- What is scaling?
- Understanding TPM, RPM, and PTUs
- Scaling Design patterns
- Retries with exponential backoff - the scaling special sauce
- Rate Limiting Policy in Azure API Management
- Monitoring, logging, and HTTP return codes
- Monitoring and logging
- HTTP return codes
- Costs, training and support
- Costs
- Training
- Support
- Summary
- References
- Part 4: Building Safe and Secure AI - Security and Ethical Considerations
- Chapter 8: Security and Privacy Considerations for Gen AI - Building Safe and Secure LLMs
- Understanding and mitigating security risks in generative AI
- Emerging security threats - a look at attack vectors and future challenges
- Model denial of service (DoS)
- Jailbreaks and prompt injections
- Training data poisoning
- Insecure plugin (assistant) design
- Insecure output handling
- Applying security controls in your organization
- Content filtering
- Managed identities
- Key management system
- What is privacy?
- Privacy in the cloud
- Securing data in the generative AI era
- Red-teaming, auditing, and reporting
- Auditing
- Reporting
- Summary.
- References
- Chapter 9: Responsible Development of AI Solutions: Building with Integrity and Care
- Understanding responsible AI design
- What is responsible AI?
- Key principles of RAI
- Ethical and explainable
- Fairness and inclusiveness
- Reliability and safety
- Transparency
- Privacy and security
- Accountability
- Addressing LLM challenges with RAI principles
- Intellectual property issues (Transparency and Accountability)
- Hallucinations (Reliability and Safety)
- Toxicity (Fairness and Inclusiveness)
- Rising Deepfake concern
- What is Deepfake?
- Some real-world examples of Deepfake
- Detrimental effects on society
- How to spot a Deepfake
- Mitigation strategies
- Building applications using a responsible AI-first approach
- Ideating/exploration loop
- Building/augmenting loop
- Operationalizing/deployment loop
- Role of AI architects and leadership
- AI, the cloud, and the law - understanding compliance and regulations
- Compliance considerations
- Global and United States AI regulatory landscape
- Biden Executive Order on AI
- Startup ecosystem in RAI
- Summary
- References
- Part 5: Generative AI - What's Next?
- Chapter 10: The Future of Generative AI - Trends and Emerging Use Cases
- The era of multimodal interactions
- GPT-4 Turbo Vision and beyond - a closer look at this LMM
- Video prompts for video understanding
- Video generation models - a far-fetched dream?
- Can AI smell?
- Industry-specific generative AI apps
- The rise of small language models (SLMs)
- Integrating generative AI with intelligent edge devices
- More important emerging trends and 2024-2025 predictions
- From quantum computing to AGI - charting ChatGPT's future trajectory
- What is AGI?
- Quantum computing and AI
- The impact of AGI on society
- Conclusion
- References
- Index
- Other Books You May Enjoy.
- _Int_V1jQ29D8
- _Hlk161251332.
