Intro

[00:00 - 00:02] All right, guys. Welcome.

[00:03 - 00:05] Welcome to the Live Workshop. I'm your host, Zao Yang.

[00:06 - 00:16] My name is pronounced wow, except with a Z. And so I'm the owner of New Line and a little bit background of myself.

[00:17 - 00:23] So I'm a software engineer by training. I created a farm bill, which scaled to 200 million users.

[00:24 - 00:31] And then I created a layer one in crypto, which has a $3 billion market cap. I am currently the owner of New Line.

[00:32 - 00:51] And over the years, I've been self-taught in machine learning, deep learning. And last year, when all the chat GPD kind of a phenomenon really got started, I spent a lot of time going into generative AI and the fundamentals.

[00:52 - 00:59] Alvin, do you want to give an overview of your background? Yeah, sounds good.

[01:00 - 01:05] I got my PhD two years ago from UC Berkeley. My work there was in pretty much making neural networks run really fast.

[01:06 - 01:19] That meant anything from understanding your hardware to improving safety or algorithms. At that time, computer vision was the probably closest to deployment that part of deep learning.

[01:20 - 01:25] And so a lot of my work was focused on making deep learning models run really fast. In the past two years, Sussmano, chat GPD took off.

[01:26 - 01:38] And so when I arrived at Apple, where I was a senior research scientist for two years, the focus shifted to LMS. And then as of next week, I'm going to be starting as a senior research scientist at OpenAI.

[01:39 - 01:42] So we'll see what they have in store for me. But a lot of my work will be the same thing.

[01:43 - 01:55] It's to make neural networks much smaller and much, much faster. Yeah, we constructed this workshop to solve a number of pains.

[01:56 - 02:07] And we sent an email out asking you for a survey of your biggest concern about concerns. And part of the problem is we're in a new transition in artificial intelligence.

[02:08 - 02:15] In software, you go through these transitions like every 10 years or so. So in the '90s, it was the internet.

[02:16 - 02:18] In the 2000s, it was Web 2.0. It was mobile.

[02:19 - 02:26] And now we're at the early stages of artificial intelligence. We're basically in year one-ish of AI.

[02:27 - 02:35] And the biggest problems that people have is they feel like they're unable to understand the language. They're unable to understand the lingo.

[02:36 - 02:56] They're unable to apply and judge what's good and what's not. They're unable to use and apply AI, be able to understand the internals of AI, and be able to build an AI foundational model yourself, and be able to adapt AI, which is retrieval augmented generation and fine tuning and these things.

[02:57 - 03:06] Of course, all of this is too much to cover in one workshop. And so we decided to focus this on the fundamentals of everything.

[03:07 - 03:16] So the fundamental technology is transformers. But rather than if you look at transformers online, you'll see a rather academic paper.

[03:17 - 03:25] It'll go into attention. And so the way we constructed this workshop is to have a mix of diagrams and code.

[03:26 - 03:40] So you can go from diagrams to understand the conceptual about high level, as well as go into the code. And so we'll be asking you to open a Jupyter Notebook in Google collab if you want to follow along with the code.

[03:41 - 03:51] The free version is perfectly fine, and you'll be able to execute about the code as we go along. And yeah, and then as for Q&A, Alvin will have his chat open.

[03:52 - 03:59] Please basically send Q&A and clarifying about questions as we go. There are no stupid questions.

[04:00 - 04:07] This is designed to basically be a zero to hero live workshop. And yeah, and please ask questions.

[04:08 - 04:15] And what he'll do is he'll stop once in a while at certain checkpoints. And then he'll basically answer about the questions in bulk.

[04:16 - 04:29] And yeah, did you guys-- before we get started, did you guys have any kind of questions, comments, concerns? Don't be shy.

[04:30 - 04:36] As I explained, I think you replied to my email. I'm brand new in AI.

[04:37 - 04:57] I haven't even taken a course in AI or deep learning or machine learning or deep learning. And I'm a full stack web developer just wanting to get into it because I know that in order to keep up with the competition, I have to say I have to get into AI and machine learning.

[04:58 - 05:12] I've got-- the only experience I really have with it is copilot installed in my visual code. I'm an Angular developer.net back end web API developer.

[05:13 - 05:24] I was just looking at my personal version of a visual code, visual studio code. And I do have Python, Jupyter, notebook loaded.

[05:25 - 05:27] I believe. Yeah, Jupyter.

[05:28 - 05:30] But I've never used it. Yeah.

[05:31 - 05:45] I can have Alvin elaborate a little bit more. But while constructing the materials for the course, we kept on going back to the persona, which is the persona that we're targeting toward is a full stack software engineer that wants to know about AI.

[05:46 - 06:00] And so the way we designed about this course is, well, this live workshop is we're going to go through the material. There's a Q&A. And then afterwards, there's a Discord channel where you can ask questions and follow up.

[06:01 - 06:08] And then we're going to upload this as a course on the New Line platform. And of course, all of you will have access to the recordings.

[06:09 - 06:24] And then the intent is to basically go through all the material. But if you haven't-- the intent is to go through all the material with the understanding that you're coming from a software background, not a machine learning background.

[06:25 - 06:36] But some of these material will take time to sink in. So you may have to basically rewatch the recording and then re-ask questions in the Discord channel about post-facto.

[06:37 - 06:40] So that's the way we designed it. So it's not like a one-off that you're done.

[06:41 - 06:47] And then you don't have any kind of support and things in the future. Yeah, that's basically just to reassure you.

[06:48 - 06:49] OK. Thank you.

[06:50 - 06:55] All right. Anyone else have any questions?

[06:56 - 07:01] If no one else have any questions, we can help. And why don't you get started?

[07:02 - 07:04] [VIDEO PLAYBACK]