that means 1 epoch is actually 30 epochs of sd dreambooth extension
that means 1 epoch is actually 30 epochs of sd dreambooth extension
Fbecause it trains 30 times each image in 1 epoch of kohya lora script
F1 epoch simply means every training data is trained 1 time. the step count for this also depends on your batch size
Flets say i have 20 images
F1 epoch is 20 steps
Fif i make batch size 4
Fthen in 5 steps = 1 epoch
Fbecause each step it trains 4 images
A@Furkan Gözükara SECourses have you tried the Tile resample control net for upscaling images?
Fnot yet. i was waiting controlnet 1.1 to become stable
Ffor a new video
AOh okay. Apparently it works like magic by adding a lot of details. I am not able to figure out how to properly use it. Will wait for your video 
Fsince automatic1111 published a new release i think time has come
BAre there any best-practices on the settings/math for training Kohya_SS styles? ie) 100-300 images.
Fwell best way is saving frequent epoch and afterwards doing x/y/z comparison
Fi have to go the gym. will be afk for a while
Ris there a batch processing tool in kohya_ss to process multiple json config settings consecutively, to test variations of LoRA's? @Furkan Gözükara SECourses
RI thought there was one
PIs the new automatic111 stable and is dreambooth fully functional? What is the best method for upgrading it if so?
PI am hoping the upgrade will fix the LDSR issue I have.
Ithanks, makes sense now
RGetting 4it/s on Kohya_ss torch 1.13... isn't 2.0 faster? You said it wasn't working so well, is updating to 2.0 not recommended? @Furkan Gözükara SECourses thx
RUpdating should in principle speed up the training 4x no (got 16it/s on webui generating but not sure how it carries over into kohya training)? Can you qualify what you meant?
ROh I see. Slow thinking. You are using captions to add specific flexible details in the portraits. Sweet. Amazing how varied methods can be. Nothing is written on stone
Fi saw they fixed LDSR issue
Fhavent tested dreambooth yet
Ftorch 2 didnt do faster for me but you can test not sure why it wasnt
FGitHub
Best Stable Diffusion and AI Tutorials, Guides, News, Tips and Tricks - GitHub - FurkanGozukara/Stable-Diffusion: Best Stable Diffusion and AI Tutorials, Guides, News, Tips and Tricks
Awhats the problem??
Fit means space character error. replace that cell with below code
FFyou see your from has space problem
Alemme try again now
Agc??
Fthat means you skipped this step
Aand what about these cells
Fplease restart begining and execute every cell with order
Fif you skip a cell it wont work
Ai did that
A
Fin your code ther eis no gc
Abut i dont know why i see some new cells that i did not see on video
Fi will update link wait a sec
Aokay sir :4827spidermanok:
Fthanks for bringing into my attention here
FExplore and run machine learning code with Kaggle Notebooks | Using data from T5 prompts DeepFloyd IF
Ayeah its working perfect now
A
M@Furkan Gözükara SECourses can you help me solve this, ive been at it for 2 days i still havent been able to use kohya
M
MRuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 8.00 GiB total capacity; 7.20 GiB already allocated; 0 bytes free; 7.29 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.