|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/ethical-secure-ai-guardrailsai-nvidia-nemo-guardrails/
课程评论：没有评论
课程名称：AI Guardrails - 安全与负责任的生成式AI [实践项目] 课程概述： 在人工智能这一快速发展的领域，我们推出了“AI Guardrails：安全生成式AI应用”的密集课程，旨在帮助学习者深入理解引导AI技术安全发展的伦理框架。课程将探讨如何抵御恶意人类交互对大型语言模型（LLM）的威胁，并介绍多种实现AI护栏的技术，包括平台（如AWS Bedrock）、模型（如提示注入、主题管理、幻觉检测）和框架（如GuardrailsAI、Nemo、Haystack）。虽然我们正在开发网络护栏内容，但该课程不包括此主题，并且不会涉及如何微调模型以实现AI护栏的概念。 学习目标： - **AI伦理基础**：概述AI开发中重要的伦理考量，包括公平性、隐私与责任。 - **安全性**：学习基于模型的方法，保护人类对LLM的访问安全。 - **识别与实施AI护栏**：通过简明的讲座和互动场景，学习如何建立和执行护栏，以防止AI被滥用，并确保与人类价值观一致。 - **实际应用**：审视真实案例，强调忽视AI护栏的后果及其风险缓解措施。 - **实用工具**：了解评估AI系统、识别潜在风险并确保AI在伦理边界内运行的工具、框架和方法论。 - **用户输入护栏**：使用来自Llama 3.1系列的开源模型（如Prompt-Guard和Llama Guard 3）检测提示注入和内容管理。 - **LLM响应护栏**：使用开源微调模型，如phi3-hallucination-judge和hallucination-evaluation-model，专注于幻觉检测和答案相关性。 - **基于提示的护栏**：掌握如LLM-As-A-Judge、上下文相关性等技术。 - **扫描LLM漏洞**：使用开源工具Garak进行实践操作，查找LLM的漏洞。 - **AI代理**：借助开源框架CrewAI，学习生成式AI多代理的渗透测试，扫描网络漏洞。 - **多模态营养AI代理**：结合Haystack、FastRag、HuggingFace与多模态Phi-3.5-vision-instruct，执行多代理用例。 - **AWS Bedrock平台的护栏**：学习如何配置、部署和运行AWS Bedrock上的护栏。 - **Haystack框架**：了解Haystack管道的介绍及使用。 - **评估者**：学习通过指标驱动的评估来评估RAG管道。 课程亮点： - **聚焦课程**：深入AI伦理及护栏的精髓，便于立即应用。 - **实践学习**：参与模拟真实世界挑战的互动练习，设计以契合紧凑的课程格式。 - **专家指导**：从行业领袖和伦理学家的经验中获益，学习可操作的伦理AI治理策略。 适合人群： 本课程适合AI开发者、数据科学家、商业领袖及渴望快速增强伦理AI实践理解的爱好者。无论您是希望将伦理考量应用于当前项目，还是希望拓宽对AI安全措施的知识，本课程都将为您提供负责的AI开发所需的深入见解。 加入我们： 抓住机会，通过将伦理考量和安全措施融入AI技术的核心，塑造AI的未来。立即注册“AI Guardrails：确保伦理与安全的AI部署”，为负责任和安全的AI部署迈出重要一步。
Navigate the intersection of innovation and ethics in the dynamic field of Artificial Intelligence with our intensive course, "AI Guardrails: Secure GenAI Applications" This course is meticulously crafted to provide learners with a condensed, yet profound understanding of the ethical frameworks necessary to guide AI technologies safely. This course will explore different ways to achieve Guardrails against malicious human interaction with LLM. In the course we will explore various techniques - platforms(AWS Bedrock), models(prompt injection, topical moderation, hallucination) and frameworks (GuardrailsAI, Nemo, Haystack) to achieve GenAI Guardrails. We are still working on Cyber Guardrails and it is not included. The course will not cover DS concepts like fine tuning models to achieve AI Guardrails. What You'll Learn:Foundations of AI Ethics: An overview of the ethical considerations critical to AI development, including fairness, privacy, and accountability.Security: Learn to apply security using model based approach for human access to LLMIdentifying and Implementing AI Guardrails: Learn through concise lectures and interactive scenarios how to establish and enforce guardrails that prevent AI misuse and ensure its alignment with human valuesReal-World Applications: Examine case studies that underscore the consequences of neglecting AI guardrails and the steps taken to mitigate such risksPractical Tools: Gain insights into the tools, frameworks and methodologies for assessing AI systems, identifying potential risks, and ensuring that AI operates within ethical boundariesUser Input Guardrail: Use Open Source Models from Llama 3.1 family (like Prompt-Guard and Llama Guard 3) to detect Prompt Injection and Content moderationLLM Response Guardrails: Use Open Source fine tuned models like phi3-hallucination-judge and hallucination-evaluation-model focused on Hallucination detection and Answer Relevancy Prompt based Guardrail: Techniques like LLM-As-A-Judge, Context RelevancyScan LLM Vulnerabilities: Explore and do Hands On with Open Source tool Garak to find vulnerabilities on LLMAI Agents: Penetration Testing with GenAI Multi-Agent - Learn about AI Agents and do a Hand On to scan Web Vulnerabilities for Cyber Security Penetration Testing using Open Source framework, CrewAI.Multimodal Nutritional AI Agent - We will use Open Source components like Haystack, FastRag, HuggingFace with Multimodel modal Phi-3.5-vision-instruct to run multi Agentic use case. We will also cover multi agentic Tools with Multi-Hop and ReAct Prompt.Guardrails on AWS Bedrock Platform: Learn how to configure, deploy and run Guardrails using AWS BedrockHaystack Framework: Introduction to Haystack pipelineEvaluators: Learn to Evaluate RAG pipelines using metric driven evaluationCourse Highlights:Focused Curriculum: Dive into the essentials of AI ethics and guardrails, tailored for immediate application.Hands-On Learning: Participate in engaging exercises that simulate real-world challenges, designed to fit within the course's compact format.Expert Guidance: Benefit from the distilled wisdom of industry leaders and ethicists, sharing actionable strategies for ethical AI governance.Who Should Enroll:This course is ideal for AI developers, data scientists, business leaders, and enthusiasts eager to enhance their understanding of ethical AI practices quickly. Whether you aim to apply ethical considerations to current projects or seek to broaden your knowledge of AI safety measures, this course will equip you with the insights needed for responsible AI development.Join Us:Embrace the opportunity to shape the future of AI by embedding ethical considerations and safety measures into the fabric of AI technologies. Enroll in "AI Guardrails: Ensuring Ethical and Safe AI Deployments" and take a significant step towards responsible and safe AI deployment.
