> 歡迎訂閱我的 [AI Engineer 電子報](https://aihao.eo.page/6tcs9) 和瀏覽 [[Generative AI Engineer 知識庫]] - 課程網址: https://www.cloudskillsboost.google/journeys/118 - 中文翻譯: https://www.youtube.com/playlist?list=PLiuLMb-dLdWJPpybrCYNhi6D9Vd4vz16i * 課程心得 * 因為是 Google 出的影片,所以內容總是會參雜一些產品業配 * 5-11 理論的部分,明明是複雜的深度學習演算法,儘管圖表再精美,要在如此簡短的篇幅內講完,還是聽不懂(對非機器學習專業的工程師來說),只感受到不明覺厲.... XD * 看得懂的部分就是 Introduction 而已 以下只有節錄有收穫的部分 ## 1-2 Introduction to Large Language Models * AI > ML > Deep Learning * Gen AI 是 deep learning 的 subset * deep learning model 可分為 discriminative 和 generative 兩種 * GenAI 輸出 自然語言、影像、聲音。Not GenAI 輸出 數字、分類、機率等 ![[Pasted image 20230922131447.png]] ![[Pasted image 20230922132102.png]] ![[Pasted image 20230922132445.png]] ![[Pasted image 20230922133338.png]] ![[Pasted image 20230922133540.png]] ![[Pasted image 20230922133703.png]] ![[Pasted image 20230922133829.png]] ![[Pasted image 20230922133806.png]] ![[Pasted image 20230922133846.png]] ![[Pasted image 20230922133909.png]] ![[Pasted image 20230922133950.png]] ## 3 Introduction to Responsible AI (略) ## 4 Introduction to Image Generation 主要介紹一下 Diffusion model ## 5-6 Encoder-Decoder Architecture ![[Pasted image 20230922141342.png]] 用 RNN 教,不明覺厲 ## 7 Attention Mechanism 不明覺厲 ## 8-9 Transformer Models and BERT Model ![[Pasted image 20230922144904.png]] 不明覺厲 ## 10-11 Create Image Captioning Models 不明覺厲 ## 12 Introduction to Generative AI Studio ![[Pasted image 20230922145959.png]] > Q: Generative AI Studio 跟 MakerSuite 有啥不同??? > A: MakerSuite 的定位似乎是拿來做 PoC,如果需要 Scale, 企業支援, SLA 等,Google 建議改用 Vertex AI ![[Pasted image 20230922152958.png]] https://cloud.google.com/vertex-ai/docs/generative-ai/text/text-prompts