Kaggle is better graphic card?
Kaggle is better graphic card?
Fyep
Fit is 15 gb
Fshared vram will be ultra slow
Pi found that the problem of twins is often
Pin realist vision
Pbecause the better resolution is 512X768 but i have used 1024x768
Pi used 5.1 but i have 2.0 too
Pwith 8bit adam and xformers my training took 2:55hs now with lion and without xformers it will take almost 7 hours with 20 images in the dataset i will try kaggle
Pbut I will also test how it performs on 8bit adam locally at 512x768 and then using a 4x upscaler to generate the images.
Tdid u install Magic in comfyui? sr, last night, i go to sleep? 1AM in my country
PYes, I put it to run but it doesn't show any progress. It was in the attention message, can you tell if it takes a long time for the result to come?
PI had to look for an example openpose video to use with my image, but the result was black, I think it takes a long time to process, but it also didn't give any error messages.
PI need a comfyUI tutorial with Magic Animate to see if I didn't do anything wrong.
Tok.
PA short story for my niece and nephews:
Tools used:
Tools used:
- Stable diffusion
- GPT-4
- give my niece a nice present
- create a coherent story
- create coherent characters
FEven Faster SDXL Dreambooth Training Full BF16 Enabled
FStill further testing but results so far amazing
P
PI'm crying with 8gb vram haha
Ssame heh
Sit worked :DDDD, thank you so much for the help man, u savd my school project
S
Thmm, about 100Gb lol :v
V@Pelolisu I usually get exact dozens of images from hundreds of or even thousands of created images on the basis of trained lora, Furkan said dreambooth checkpoint is better than lora. I have not tried yet. If a object is too detailed and styled, lora results are not good, time wasting, many images are not to the standards you require.
F1 picture it is merged with horse though :d
Fhuge. did you do anything special?
Pthis one? The legs are under the wings 
it's supposed to be like that
it's supposed to be like that
POr this one, which is also intended.
The protagonist turns into a unicorn and then back into a human.
The protagonist turns into a unicorn and then back into a human.
Si think that's the one i still have installed right now, was hoping there was a working sdlx automatic1111 dreambooth at this point
SHey jet Thank u guys so much
Fthe guy left developing
Ssad news
Fye he is working for another ai company
Fdoing similar stuff i think
PI think the time of 2:55 hours for my training is acceptable with an 8GB RAM GPU. However, the result of the training still didn't satisfy me, but as I'm still learning stable diffusion it may be that I haven't yet set the prompt parameters correctly. I will also try to run it on Kaggle as Furkan suggested to compare the results. I think I can use a Lion optmizer there.
PI will also do some experiments with LoRA
X@Dr. Furkan Gözükara or anyone eles in here is there a course on how to do the deepfakes video ?
LI'm trying to get a baseline for SDXL Lora time frames. I see a bunch of different times about train a lora in 30 minutes etc but my use case is different. Most of my SDXL Lora trainings are 12-18 hours. Now I know this has to do in part im using 768X1344. Adjusting Network Dimm seems to reduce and learning rate seems to have an impact on the time but my concern quality will be impacted if I adjust these. My Anime models are coming out great despite the time. My person models are Meh despite following the guides and using the regularization images but I believe I missed something. I'm on an RTX 4090. Any insight?
FYep im using pagedadamw8bit, similar results to adafactor, still testing, got about 50 more trainings to test, will try prodigy next
Fcan you share your json file, also are you using the latest nvidia drivers with Sysmem fallback policy
LThat’s a good idea to check the drivers. I’ll grab the json when I have a moment
Fi also have a 4090 my trainings are usually 35 mins to 2 hours depending
LWhat settings are you using? I know that’s a large request. I’m all over the place even with the trainings and I’m still new.
Fdepends i use different settings, but usually adafactor and the extra arguments with the gradient checkpointing and sometimes xformers not always.
LDo you think adafactor is better than AdamW (not the 8 bit)
FFor me for faster training with lora i always use adamw8bit, dreambooth still testing but best is adafactor
Fand for sdxl i use the noise ofset 0.0357
Falso using a older version of kohya uses less vram i think 21.8.4
Falso if you want me to train your data set on my system i can. so we canm compare training times
