|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/generative-ai-the-future-of-ai-powered-creativity-pro-tm/
课程评论:没有评论
课程名称:生成性人工智能:AI驱动创意的未来 课程概述:解锁生成性人工智能的力量,这一21世纪最具变革性的技术。从文本和图像生成到多模态学习及增强检索生成(RAG),本课程提供了一个逐步探索生成性人工智能核心组成部分、现实世界应用和伦理考量的旅程。学习如何使用大型语言模型(LLMs),如GPT,变换器架构和领域特定的微调,重新塑造各行业的内容创作、自动化和人工智能创新。无论您是开发者、数据科学家、人工智能爱好者,还是数字创意人员,本课程将为您提供实践技能和战略洞察,帮助您构建前沿人工智能解决方案,包括使用LangChain和OpenAI构建聊天机器人。 课程内容摘要: 1. **生成性人工智能简介**:定义及其类型(文本、图像、音频、代码),应用于各行业(营销、医疗、教育)。 2. **生成性人工智能的工作原理**:从人工智能到机器学习,再到深度学习的概述,神经网络的关键概念及训练、推理、部署的区别。 3. **使用生成性人工智能进行文本生成**:了解大型语言模型(如GPT、BERT、T5)在文本摘要、翻译和讲故事中的应用。 4. **当前文本生成AI的挑战与局限性**:包括偏见、幻觉、提示限制、计算成本与令牌限制。 5. **增强检索生成(RAG)在AI中的重要性**:介绍RAG及其重要性,如何增强大型语言模型的知识。 6. **RAG与大型语言模型的工作机制**:RAG的架构、使用向量数据库构建管道,并提供实践示例(如问答系统)。 7. **使用生成性人工智能进行图像生成**:工具如DALL·E、Midjourney、Stable Diffusion,生成逼真、抽象和品牌视觉。 8. **使用生成性人工智能创作音乐与艺术**:AI的创意表达,从背景音乐到生成艺术,行业中的工具与平台。 9. **利用迭代生成AI增强艺术家工作流程**:与AI共同创作,风格迁移、迭代优化与创意加速。 10. **生成性人工智能中的变换器架构**:自注意力、编码器-解码器设计,以及变换器的发展历程(BERT→GPT→PaLM)。 11. **文本生成AI中的长距离依赖建模**:处理长文档的挑战及解决方案(注意力掩码、分层模型)。 12. **针对生成性人工智能的预训练模型微调技巧**:转移学习基础、低秩适应(LoRA)、PEFT和提示微调。 13. **针对生成性人工智能的领域特定微调技术**:医疗、法律、金融的定制化,以及数据集准备和微调管道。 14. **多模态生成模型的解释**:结合文本、图像、音频和视频的生成模型(如CLIP、Flamingo和Gemini)。 15. **多模态增强检索生成(RAG)**:使用RAG桥接不同模态,进行视觉问答和文档理解。 16. **生成性人工智能中的伦理考量**:关于虚假信息、深度伪造、同意等问题,以及监管框架(如欧盟人工智能法案、负责任的AI)。 17. **在AI模型开发中执行问责与责任**:审计AI系统、偏差检测和缓解策略,模型设计中的透明性。 18. **使用LangChain + OpenAI构建聊天机器人**:LangChain简介、OpenAI API集成、基于RAG的问答聊天机器人端到端项目。 这个课程为希望在生成性人工智能领域深入了解和实践的学习者提供了全面的学习体验。
Unlock the power of Generative AI, the most transformative technology of the 21st century. From text and image generation to multimodal learning and Retrieval-Augmented Generation (RAG), this course offers a step-by-step journey through the core components, real-world applications, and ethical considerations of Generative AI.Learn how Large Language Models (LLMs) like GPT, transformer architectures, and domain-specific fine-tuningare reshaping content creation, automation, and AI innovation across industries.Whether you're a developer, data scientist, AI enthusiast, or a digital creative, this course equips you with hands-on skills and strategic insights to build cutting-edge AI solutions - including chatbots using LangChain + OpenAI.1. Introduction to Generative AIWhat is Generative AI?Types: Text, Image, Audio, CodeUse cases across industries (marketing, healthcare, education)2. How Generative AI Works End-to-End AI → ML → DLOverview of AI, Machine Learning, Deep LearningKey concepts in Neural NetworksTraining vs Inference vs Deployment3. Text Generation Using Generative AIUnderstanding LLMs (GPT, BERT, T5)Applications in text summarization, translation, storytelling4. Challenges and Limitations of Current Text Generation AIBias, hallucination, prompt limitationsComputational cost, token limits5. Retrieval-Augmented Generation (RAG) in AI - Enhancing Model KnowledgeWhat is RAG and why it mattersAugmenting LLMs with external knowledge6. How RAG Works with LLMs - Mastering Retrieval-Augmented GenerationArchitecture of RAGBuilding pipelines with vector databasesPractical examples (e.g., Question Answering Systems)7. Introduction to Image Generation Using Generative AITools: DALL·E, Midjourney, Stable DiffusionGenerating realistic, abstract, and branded visuals8. Music and Art Creation Using Generative AIAI for creative expression: from background scores to generative artTools and platforms used in the industry9. Enhancing Artists' Workflow with Iterative Gen AICo-creation with AIStyle transfers, iterative refinement, ideation acceleration10. Transformer Architecture in Generative AISelf-attention, encoder-decoder designEvolution of Transformers: BERT → GPT → PaLM11. Modeling Long-Range Dependencies in Text Generation AIChallenges with long documentsSolutions: attention masks, hierarchical models12. Fine-Tuning Pre-Trained Models for Generative AITransfer learning basicsLow-Rank Adaptation (LoRA), PEFT, and prompt tuning13. Techniques for Domain-Specific Fine-Tuning in Generative AIMedical, legal, finance - customizing GenAIDataset preparation, fine-tuning pipelines14. Multimodal Generative Models ExplainedCombining text, image, audio, and videoCLIP, Flamingo, and Gemini models15. Multimodal Retrieval-Augmented Generation (RAG)Bridging modalities using RAGVisual QA, document understanding with RAG + LLMs16. Ethical Considerations in Generative AIMisinformation, deepfakes, consentRegulatory frameworks: EU AI Act, Responsible AI17. Enforcing Accountability & Responsibility in AI Model DevelopmentAuditing AI systemsBias detection, mitigation strategies, transparency in model design18. Building a Chatbot with LangChain + OpenAIIntroduction to LangChainIntegrating OpenAI APIsEnd-to-end project: RAG-based Q & A Chatbot