central-utah-lug

Transcription from 2025-08-29T03:21:47Z

⚠️ <@342671004667215873>: I haven't received any audio from anyone in this recording, try switching to a different voice region if this problem persists.

Transcription from 2025-08-29T03:26:38Z

Transcription from 2025-08-29T03:31:49Z

Transcription from 2025-08-29T03:55:07Z

Transcription from 2025-08-29T04:28:45Z

Transcription from 2025-08-29T04:44:16Z

Transcription from 2025-08-29T04:54:41Z

Transcription from 2025-08-29T05:28:29Z

on and slap, slap, slap, slap.
 Thank you.

It's incredible if you can actually show me some of the actual changes.
 Yeah.

I'm supposed to be the first to show up at some of these games.
 You close these right now.
 Absolutely.

So, yeah, I can't tell you what I'm saying. I'm so glad.

No, you think you're wrong?
 You're the worst out of all, you're the worst in the world.
 You're the worst in the world.

I'm sorry.
 I just got kicked out of here.

I can't believe that.
 That's really going to be like, possibly, not a graphing.

What's up with your other people? It's time to talk to people. Yeah.

And it was so...
 Thank you, thank you, thank you, thank you, thank you, thank you, thank you, thank you, thank you.

Yeah, that's right, I didn't tell you.
 It was like, you lived in the back,
 you had to go to the attic,
 and you had to get to the attic,
 and then it was coming all the way,
 and I was feeling like I was in the attic, I think.

I'm going to get that in there like this.
 Thank you.
 I'm so glad to be here.

But if you want to know, please give us a chance, it sounds like I'm going to tell you as well, I'll set it up, I'm going to tell you as well.

You know anything you feel check your way?
 I think we'll have to.

You know, that's fine, it's a nice stuff, but we don't need to come make sure we say hi.
 You have to see who's nervous, which is not necessary, it looks like nothing's adjacent because it's going fine.

So they haven't, they're still sort of...

It's not something that's... that's what I was talking about stuff that's interesting, it's not something that's crazy about you there.
 So, it's actually one of the legados, part of the club, but it's not something that's...
 It's an international name, so I think that's not it.

I'm talking to the person who's in the sun, so I'm doing it. So for the day, so for the evening, so it's going to be. I think it's in my saffron of my safety chat, so I'm having some migrates. But like.

I see a lot of people think that's the kind of person to think.

and it's a good question because it's proven that a lot of people don't like to do these things as well.

That's the last thing. It's actually over the level, I can keep the general level as a middle level. So, it's our improvement.

Yeah, it's just hard to work, but it's a little light, so like, like, it's a little light here.

Oh, good night, honey.
 Okay, fine, I'm going to go.

Everybody got everything out.
 Thank you.

It's that... it's something that I'm making.
 It's something I'm making.

It sounds like the one that made me feel like it was fine, but the one that made me feel like the swan of me.

Just asking that, it's a completely new science club, and it's a scientific lab club, but that's the new computer lab.

even if I send it out to the room, it's going to.

Yeah, anyone, so yeah, we need to find a solution to these equations to do that. I'm not going to write that.

The professor finds that he looks at what's in the way.

a different seal saying no, no, no, no, no, no, no, no, no, no, no, no, no.

Micropon is very row body, I don't know if that is.

Yeah. I'm sorry. I need to be a staffer by the way.
 Go for a big deal. He's a Spencer. He has some trouble.

but it's on that top-microphone.
 So I'm here, I'm enjoying some of my songs,
 because I'm, uh, enjoying stuff.

Okay, this is off the phone, is this better?

But yeah, mostly just like a way to meet other people in the program.
 I think it's mostly software and juniors right now, the seniors, the computer science club
 kind of like, it's like an ebb and flow, because like depending on which students on the
 program, so there'll be like a group that's really into it and then they'll graduate,
 and then there won't be a club, and then a group will form and then they'll graduate,
 so it's a, right now it's in kind of the growing phase of things, so a good time to get in
 on it.

Yeah, you can also join clubs like unless a club has membership fees, you can kind of just join it through like the portal and just like see announcements and stuff and like.

It's like no commitment there, some clubs like the commitment of a fee because they're like,
 we want people that are involved, but the computer science club and the AI club are like.

talking about things. Love to have you, but no stress at all.

We are going to see all of the first places out of Australia.
 For the other, we are going to have a whole lot of work.

like social elements of going to college
 and that I want to meet people.

Yeah, that's what you're saying, I haven't met somebody who's done slick before because my understanding is as far as like.

And Dixie would turn it to Utah Tech and snow call it draw kind of like the smaller.

institutions. Maybe a little bit of Southern Utah University as well, but those are like the
 smaller institutions that prioritize costs and things like that.
 Yes.

So how far did you only do one semester there?
 Yeah, I just did one semester and it works.

pretty miserable, oh, okay, in what ways?

like no one wanted, I have one in person in one online class.

And in my papers in class, nobody wanted to talk to each other and there were like four people or so like it seems like they would want to talk to each other but no, no, no, we showed up to class, we just like sat there listening to the teacher for like.

like a couple hours or whatever long the class was out.
 And then everyone would just go home.
 And like, can you even when we have like little projects
 we're supposed to like help each other and work together
 and stuff, there's no one off?

And that, that was pretty bad.
 And then also I just...

I really don't like online classes, but...

the worst part is that I have to take public transit to get there and I just hate public transit.
 All right, it's just going to be out soon.

The other four students in your class were they like older or they all like

Transcription from 2025-09-05T00:04:02Z

Transcription from 2025-09-05T00:41:08Z

I don't know if I was actually doing all that, so sorry to Garrett, it was Janky.

but yeah, well saying this, I do found out it funny.

they handled all the internet stuff, actually implement hairpinning where it'll automatically,
 if I send something to the public, IP address from within the actual private network, it'll automatically
 re-estimate it, and I'll go back, yes, okay, which, the whole reason why I know what that's called is because I read that
 that killing, article, and that sort of one.

which is such a good article, but yeah, so I was not expecting.

it to be so good when I first read it, it's going to be like, this is like some of the best
 network of literature ever. I was expecting we have to be right.

I don't understand, but I read it and I'm like, oh, that makes sense.
 Yeah, it's, they're very good at presenting these concepts in a way that like...

I know everybody can understand, but everybody who stood up for Web Survey.

I give you that at some point and you think you're, you're like, we're entering the understanding.
 They're very good at building up the concepts.
 And after that, I'm like, oh!

That's the value out of jail skill, I don't want it.

I probably could figure it out, but like, oh, okay, yeah, I'm just sort of reading that,
 maybe be like, um, yeah, we're done, uh, oh, I do want to have that, yeah.

I know, but it seems like it would be a cool project to work on.
 Yeah, like the best of my things.
 Well, and...

A lot of the concepts are becoming more important in my opinion.
 So, like, especially...

Apparently Gemini has been doing this for a while and OpenAI is trying to replicate it, but I'll end off with the OpenAI side of it.

So there's this opening, I've real-time thing.

And I think Gemnight has been doing this for a while.

But the idea is, you connect over a web RTC connection to the OpenAI servers.
 WebRTC is like the Discord voice chat program, so it's not peer-to-peer.
 Oh

It's meant for peer-to-peer voting to seek communication.
 Yeah.
 It's like in Discord, you make a WebRTC connection with every network of the call and everybody
 kind of sends their audio through, it's not like a hub-and-spoke.

And with what so so there's the protocol of our TC and web our TC has these what the what the tail seal terms are to have these like

And I was part of the ClubRTC, is they have servers in between, so it's like if you don't have a direct connection, you stand up, one of these in every stage and everybody can at least stay local to their state for, like, moments and stuff.
 Yeah.
 And the protocol, like if you could figure your various connections and servers, things like that, kind of negotiates the fastest path, not mad at me.

and audio in like, there's like, you know, 16-bit audio buffers,
 you sent that over to OpenAI.
 So the handshake that you have would like your local server
 that you create, like your Lexi Sharp server, like.

generates a one time session token,
 and then you give that to your browser.

and the browser uses that to authenticate
 with the OpenAI servers and create this like.

voice chat call with that, and you can talk to the LLF, and so the audio stream gets sent
 over to OpenAI, they transcode it, and then give responses back. It has some nice, interrupted
 things where, like, if you pause, it starts sending responses, but you can interrupt the responses
 by talking, which causes them.

Because things like this, I think you're going to become
 more important, just like really low-light C-L-L-L-L usage
 to make it feel more natural.

these like term servers or like jump servers and closer to real-time protocols
 kind of mesh networking things. I think they're going to become more important in

these domains, because I think this kind of stuff is going to end up being baked into a lot of different apps.
 Yeah. Well, I, I hope that some point we get to like, uh, technology to the point where, uh,
 what you can run all your own stuff locally.

I still don't have that idea of, like, yeah, let me go send up, like, all my...

things that I'm working on to help in AI to be their own.
 Yeah.

which is totally fair. I think that's one of the, I mean,

It's a bubble, but I don't know if it's going to pop like other bubbles and pop right like I think the pop will end up if there's a world where the pop could end up being a slow thing to just see useful things stay and use less things drop out.

but I think these smaller models are going to become more important
 because running the big models is expensive
 and nobody knows how to make money doing it, yeah?
 Yeah, that's true.

But it might be that no one is making money off of a pension.
 Um, and I think.

adding open source model for our clients to be the future.
 Thank you.

I don't really know how to describe it, is this...

Loads, stores, AI model that you can run locally, like you can.

get the raw data out since no one knows how LMS work anyway.
 Uh.

If there's not really any way to obscure how you're a specific element.

Yeah, you'd have to like create DRF tooling through the whole stack, which is not really
 how the ecosystem, but you'd have to re-implement the ecosystem.

which like can be done. That's basically what deep seek did.

And there's probably some, so there is debate over like

kind of if a model is free to download it has the licensing restrictions but
 doesn't have methodology at white papers or like any sets of reproducibility does
 that count? Source or just a weight?

But even if it's just an open-weight model, a permissive use open-weight model.

So maybe like there's not a white paper, it might cut it into your self, but it's something you get downloaded run

even if they're never like the focus, I think they're going to be a focus.

But even if those never become like the big focus,
 I think incidentally, they will eventually become good enough to use.

I haven't done anything to help in source models, but simply, if you are.

at least the ones that are small enough to, like, run without a data center.

Yeah, right because like deep sea as long as you have 400 games

the way I was running it is, if you don't get to pick it, it's grand so I've only run
 like a lot of them around that side.

And originally my Olauma configuration file was set up such they had they had been tried to use more RAM that like you do had it was just crash.
 not just crash, so.

It would craft my entire like desktop environment, so I would just get hooded back to the TTY.

get it to just offload onto the CPU and how you think the GPU is going to come from.

kind of stuck because it was slow but at least I could probably

one of my favorites to do. So the general model.

So, of course, the general model you can run makes smaller ones. This one is

I could get a GPU that does this without, like, too much fame.
 Okay.

But the general models have vision support.
 That's why there's like a couple embedding models,
 but really, like this is the LLM that I'm recommending
 suit to use, now it's 60 gigs, right?
 But the new, the GPT OSS, there's a 16 gig,
 or a 14 gig version.

And you know, give or take the gig sort of like exactly.

They're pretty good at two of them, as far as, like, stuff in those big classes of less than
 100 gigs, less than 60 gigs, I think those are, like, the best moments there.

Because like, yeah, you like like a deep six the best.

Yeah, TVK 2 is really good, but those are like multi-hundred gigs of stuff like he's doctor.

I do like these vision models where I can say like, uh, so, so yeah.

I just took a screenshot of my class roster, so oral.

or not my roster, a list of links across like.

Transcription from 2025-09-05T00:41:23Z

STT service timed out (SSRC 1407, session ID 500ea3d9-9a21-42e7-abc8-65991524dac9)