it's the best currently what do you mean they actually did remove ??
it's the best currently what do you mean they actually did remove ??
AI suppose their abusing monopoly until a competitor who comes beat them
GHi @Furkan Gözükara SECourses,
I've tried both
on 48 images with faces pre-captioned with joy captioning and replaced with
it takes too much time - estimated 45 days! is that make sense??
Shared GPU raised to 14.2 and GPU RAM is on 7.5 GB.
I've attached all screenshots
I've tried both
flux1-dev-fp8.safetensorsflux1-dev.safetensors8GB_GPU_Quality_Tier_4_FP8_6850MB_13.2_Second_IT.jsonon 48 images with faces pre-captioned with joy captioning and replaced with
ohwxit takes too much time - estimated 45 days! is that make sense??
Shared GPU raised to 14.2 and GPU RAM is on 7.5 GB.
I've attached all screenshots
Ganother screenshot with more lines from training console
Fi will share hopefully after finalizing
VHi, I also have a problem when training with kohya: I have a geforce rtx 3900 with 24 vram, I use the models from your youtube tutorial video (flux dev, fp16, etc.etc). Before training, I clear vram (restart, autostart apps off, etc). When I check, I have enough vram under 500 mb. When I start training, it goes up to 23 GB and ALSO the shared Vram goes up. Why? Is that normal? using the 24GB_GPU_Quality_Tier_3_23150MB_5.9_Second_IT
Fi dont plan to research this atm
Fclick load model
Fit will then use all 4
For generate like 16 
Fyour folder repeating is 40
Ftherefore you are doing 392k steps
Fmake folder name like 1_ohwx man
Fyour speed is good
F10 second / it
Ftotally normal
Fyou can increase block swap count
Fand compare speed
�Okay but can you tell me how I can run this in kohya? I am not able to understand.
Vokay. thanks. it trains like 6-7 hours
Vand works
Vwas just wondering
Fi dont know either didnt look at it 
Ftotally good
Funless you have rtx 5090 you have to use shared vram
�Please take a look whenever you get time.
Tdaily credits removed from all plans, only monthly remained, effective for new subs after 20 feb
Athis is bad wow
ABut as I said I think they abuse monopoly rn
TI don't think even current monetisation is profitable for them. video generation compute is crazy expensive
Ais it ? For 5seconds and from ab image frame?
Tyeah, 5 seconds at 30 fps is 150 frames each taking unknown number of steps for diffusion model
Twe are talking many minutes of a100 probably per generation
A3~
Aso 0.50$
TSo you get 660 monthly credits for 80$ yearly, 35 per 5 s generation, so you pay ~0.35$ per gen for their top model 1.6 in pro mode
Twithout daily credits
T- a ton of freeloaders with free plan
AI give it 2 months for a better open source model to come
AOmni human and Goku are on their way too so
Dfor the sort by similarity app, what kind of speeds should i be expecting? right now it takes about an hour to process 20 images
Fit is wrong
Fit means not using your gpu
Fit uses tensorflow so it is extremely delicate to the cuda and cudnn libraries
Fin your system
Fit sucks from this perspective actually
Dyeah its all on the cpu atm. not sure why
Di have all the prereqs installed. i think it might be defualting to a cached version of keras and protobuf that aren't compatible
Dthank you. might be time for a clean install again lol
Fit is related to cuda and cudnn 99%