High-Efficiency Diffusion Models for On-Device Image Generation and Editing with Hung Bui - #753

So we're realizing this open weight models, 7,000,000,000 parameter model. We're still getting complaint from the community in Vietnam that, oh, this model is too big. Right? We can't fit in in on on on our GPU. Right? And we said, okay. Fine. You know? Let let let let us go, you know, one more step. Try to, like, reduce the the number of parameters. Try to have the size of the model to, less than 4,000,000,000 parameter. And with the couple of improvement over the way we train it, we noticed that this model, less than 4,000,000,000 parameters, actually perform, you know, even better than the 7,000,000,000 parameter model. Alright, everyone. Welcome to another episode of the TwiML AI podcast. I am your host, Sam Charrington. Today, I'm joined by Hung Bui. Hung recently joined Qualcomm as VP of technology through the recent acquisition of VinAI Research, which ranked in the world's top 25 industrial AI labs based on research output in top conferences like ICML and NeurIPS. Before we get going, be sure to take a moment to hit that subscribe button wherever you're listening to today's show. Hung, welcome to the podcast. Thank you so much, Sam, and it's my great pleasure to be here. So we've got a bunch of really interesting topics to dig into, including your research into topics like diffusion models, image generation and editing, and more, and, of course, how to make all of that efficient on mobile devices. To get us started, I'd love to have you share a little bit about your background, which includes time at places like Google DeepMind and Adobe Research, among others, and tell us a little bit about how you got into AI. Actually, let's see. I I did my PhD almost thirty years ago. And a PhD is actually on this topic of multi agent system. And back then, it was also a very interesting AI topic. And the the the reason I got into AI is is is just by curiosity. During my undergrad, classes in Australia, I I learned about things like Turing test, and, I was, you know, very curious to myself that, you know, how we'll be able to book on a machine to pass the Turing test. And back then, I I I have to be honest that I don't think I'm gonna, live to see the machine passing the Turing test, which, you know, is kind of I take it for granted now. Tell us a little bit more about some of the work that you've done in your career. So, I think my first job, first real job after academia, was at a place called the AI Center at, SRI International. SRI is Stanford Research Institute? That that's right. Yeah. It it is is formally known as Stanford Research Institute. And back then, you know, we're talking about twenty years ago, I got a chance to work on some project called Calo. And back then, we tried …