Software Engineering Courses (SECourses)•10mo ago

its a bit confusing how such big models can be run with consumer gpus, but I'm hopeful

Ddaroth its a bit confusing how such big models can be run with consumer gpus, but I'm h...

Timson•2/26/25, 10:14 AM

with extreme quantization and quality loss

Timson•2/26/25, 10:15 AM

Consumer GPU need to get better, not the other way around

Timson•2/26/25, 10:16 AM

I was hopeful 5090 release would drop GPU compute prices but they continue to rise. L40S time is ever more expensive.

Timson•2/26/25, 10:16 AM

maybe because libraries does not fully support 5090 as Dr. Furkan reporting.
Seems vary strange how long it takes

TrumpYouAll•2/26/25, 10:24 AM

@Dr. Furkan Gözükara Hello, I watched this video https://www.youtube.com/watch?v=jHlGzaDLkto
which is about how to use 5090 for Swarm UI
However, I got this error .

I got several questions:

I got this error (can you help me solve it?) Is it 5090-specific?
I forgot (I asked you before but I since my computer was fried so I don't have the record) where to put the trained Flux Dev models with Dreambooth ; I remember somewhere I need to put something like 'diffusion_models' (?). Can you give me a a specific answer where to create this folder (or if my folder name is corrrct to put my Dreambooth-trained models)
I forgot how you type the yoloface command in the input textbox (again sorry, I lost the record when my PC is fried)

Thanks in advance ! I always love ur videos!!!! U are the king in AI image !

@Dr. Furkan Gözükara

YouTubeSECourses

RTX 5090 Tested Against FLUX DEV, SD 3.5 Large, SD 3.5 Medium, SDXL...

In this video I have intensively compared RTX 5090 speed on FLUX DEV, FLUX Fill, SD 3.5 Large, SD 3.5 Medium, Stable Diffusion XL (SDXL) and Stable Diffusion 1.5 (SD 1.5) models. For each benchmark, I have compared RTX 5090 against RTX 3090 TI so we see the speed improvement. Moreover, I have tested FP8 vs 16-bit precision for FLUX and SD 3.5 La...

AAiInfluence @Dr. Furkan Gözükara what gpu should I rent for the large 14b model on massed ?

Furkan Gözükara SECourses•2/26/25, 10:33 AM

80gb best

Furkan Gözükara SECourses•2/26/25, 10:33 AM

but 48gb will work too

Ddaroth its a bit confusing how such big models can be run with consumer gpus, but I'm h...

Furkan Gözükara SECourses•2/26/25, 10:33 AM

it works

Furkan Gözükara SECourses•2/26/25, 10:34 AM

especially lower resolution

Furkan Gözükara SECourses•2/26/25, 10:34 AM

i am gonna update our app today

Furkan Gözükara SECourses•2/26/25, 10:34 AM

for lower resolutions

Furkan Gözükara SECourses•2/26/25, 10:54 AM

anyone tested float8_e4m3fn with wan 2.1?

TTrumpYouAll @Dr. Furkan Gözükara Hello, I watched this video https://www.youtube.com/watch?v...

Furkan Gözükara SECourses•2/26/25, 10:54 AM

hi for 5090

Furkan Gözükara SECourses•2/26/25, 10:54 AM

did you use my installer?

Furkan Gözükara SECourses•2/26/25, 10:54 AM

flux dev put into diffusion_models folder

Furkan Gözükara SECourses•2/26/25, 10:55 AM

yolo format : <segment:yolo-face_yolov9c.pt-1,0.7,0.5>photograph of ohwx man

FFurkan Gözükara SECourses flux dev put into diffusion_models folder

TrumpYouAll•2/26/25, 10:59 AM

@Dr. Furkan Gözükara thanks for your prompt reply, yes, I did use your installer from Patreon
"flux dev put into diffusion_models folder", may I know if my Dreambooth-trained models are put inside this folder, called 'diffusion_models' ? Is this folder inside 'Models' folder ?

Thanks in advance !