Sign In

Interview with Tyler: Creator of and the MEGA model

Tyler Is the Founder of and the creator of the very popular and hugely complex MEGA model. We had the pleasure of interviewing him on his work, thoughts on the future and ton of other subjects you can listen to here.

Q: What other Start ups have you worked at?

A: "I think the biggest one most people would have heard of is Groupon. I'm I was a pretty early engineer and then tactical leader over there. So started started there. I owned a software consultancy previous to that and we had about 50 engineers that worked worked for us. Groupon eventually bought us out. We became a lot of Groupon engineering team. And then I had various different, you know, technical and leadership roles there. I left in, I think, 2017 and have been back to, you know, early stage startup stuff since then."

Q: How Did you get started with SeekART?

A: "I have two co-founders founders, both folks I've worked with on multiple different projects throughout my career started out just me. So they're pretty recent. It was me in late August, you know, August, September, October, kind of just getting into the space and having fun playing with stuff and then rapidly deciding, Wow, I want to build something, right? I want to build something in a space. And I did that myself for a few months, built everything for seekart from scratch with my own two hands. Launch that in late middle October, completely rebuilt it, completely redesigned it in early November. Because I kind of suck at the frontend stuff and the things that stuff I put together initially just wasn't that great? And then mid-November, I brought on my first partner and just brought on my second co-founder today, actually.

Q: What got you into AI Where did it all start for you?

A:"It started with it started with, you know, mid journey initially.I was on the free plan and I went to their what I think is a $30 plan for their unlimited plans. And I still have it. I like to keep up with what's going on with M.J., of course. And it was just it was too slow, even even on the paid plan. The big one of the major things that got me to building my own stuff is mid journey was just too slow, right? And then I started running stable diffusion locally and I built out a bunch of scripts for myself to, you know, basically just run all the time. Right? So I'd set up a, like a lot of folks, like their scripts in order for this now right where you can Pramod tries a bunch of prompts and then you know lock away for 10 hours and come back to a bunch of art. Yeah. And I started doing I started doing that, but it didn't feel like very iterative, right? You know, like I wanted to make better stuff and, you know, even even with that kind of solution, right, there was just too much time between, like me, you know, cutting up the stuff for the for the particular problem, saying I wanted to generate and then coming back and looking at the art and then trying to decide what to do next. Mm hmm. Which, which is initially what got me excited about building something"

Q: What kind of things did you try to focus on while building your base Model?

A: "So it's an absolute just soup of stuff. I would say there's probably between. 10,000 and 100,000 concepts in here. If you think about the intersection of words and props. So there's only 40,000 images, but each caption is at least 20 words, if not more. So conceptually, there's a much larger space there."

Q: Where do you see things going in the next 6 months?

A: "Oh, my God, I have no idea. I'd say that the space is moving so fast and it's so unpredictable. And there's new mind blowing stuff that's getting released on a monthly, if not weekly basis that honestly, I have no idea. I'm sure stability will release, a new hopefully better model, better base model for whatever's next. I would love if they released their clip guidance stuff. So that's a big deal for output quality. But it seems like something they want to keep for themselves right now. Hey, we did it. But after that, it's tough, right? You never know what. M.J. is going to do, right? You don't know what new research paper is going to come out. You don't know what new, you know, startups going to enter the space and try to do something cool. So I think we're still in for a full year of surprises here."

Q: Where do you see all this leading as a society?

A: "Well, I mean, honestly, in some ways everything will change in other ways, nothing will change. What I mean by that is like, we've continued to go through big transformative changes in technologies. And you can go all the way back to the beginning. You can go all the way back to the turn of the 20th century. And find these articles about the the invention of the phonograph and the exact same headline, which I find really interesting, but it's like, you know: 'phonograph! who will ever go here live music again?' It's the same fear response that sells papers."

Are you an artist, model creator, or otherwise involved in this technology as it develops? Do you have some insights you like to share? Reach out to us at [email protected] so we can get in touch!