with the inspect emedding training program, do we have to check magnitude for overtraining or just s
with the inspect emedding training program, do we have to check magnitude for overtraining or just strength?
B2.1 modeli çalıştırınca siyah box çıktısı alıyor sadece nasıl çözebilirim?
Fyou dont have to. best way is always comparing checkpoints
IOK, I reinstalled everything on a new machine, and I get the same erorr. But I'm not using filewords.
Fyour classification directory is invalid
Fput / beginning @I A
IYes I did..white space?
Fby the way you written them with incorrect order. above one is training below one is classification
Fyou noticed that right?
INO! Let me check...thanks
IOh yeah, it's working now. Your the man! Thanks
II will update you later.
Fyou are welcome
IMeanwhile I used some extra settings, based on your conclusions, like offset noise to 0.07, image genertion to DEISMultistep, and testing to deterministic.
Ztraining lower rate for longer seems to raise the magnitude which makes the embedding not play well with other prompts, so it seems like 0.005 is about as good as it's gonna get. Wonder if i make the backgrounds transparent if that will allow the ai to focus on the model better
ISo far so good, but gradio froze, so I cannot track the progress. I assume if I close the window, the training will stop? I can see it running on the terminal, tough.
Ffollow the terminal. yes unfortunately gradio freezes
IThank you again. When the training is over, I will do some tests and comparisons, and share my conclusions, based on your video tutorial.
II'm not sure if you covered this, but it would be great to see some tutorials on Deforum, Animations, and all the video stuff on A1111 or Runpod.
Fi plan to do an animation tutorial
ITraining finished. To generate images it asked me to add these --no-half --disable-nan-check so I did
IUnfortunately all images generate are black squares 
INo error messages though, just black images.
Scan anyone tell the differences between
Slora, locon, loha
Sand which is better so far
Sif anyone had experience with training i mean
FHI!
FI just did this tutorial> https://www.youtube.com/watch?v=O01BrQwOd-Q and have this error > AssertionError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
YouTubeSECourses - Software Engineering Courses
Our Discord : https://discord.gg/HbqgGaZVmr. This video I am showing how to downgrade CUDA and xformers version for proper training and I am showing how to do LoRA training with 8GB GPU. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on 
https://www.patreon.com/SECourse...
https://www.patreon.com/SECourse...FI don´t know nothing about code, please help
IBy the way Dr. Furkan, I noticed some changes on A1111 side too. Doing a fresh new install, everything works fine, but not Dreambooth. Dreambooth tab is not sowing in A1111 after the installtion. Just keeping you updated with my experience.
EWhat is the best training method to train model certain object that is the same shape and similar form but it differs in decorations and filling. I have tried Dreambooth and Textual Inversion. With Dreambooth I used LORA for training and main problem is that when Im generating new things I can't specify how the object should look like because it is generated randomly and my prompts in most cases doesn't work, or work poorly. With textual inversion i feel like generated objects are poorer quality than loras. I have more control with the prompts but objects are generated in different perspecitve and in most cases not as a whole but zommed in.
EI had simmilar problem with dreambooth not showing. I installed it by cloning extension from github and updating certain requirements. Also I installed python 3.10.6. With the older/newer it didn't work for me.
Fare you using sd 1.5?
Fi am gonna make a new video today short one to show how to install dreambooth manually
Fhi welcome
Fwhat gpu do you have?
Fi still couldnt find to experiment with objects unfortunately. currently about to finish training styles
Zwhat should my batch size and gradient accumulation steps be for about 10gb of vram? using about 20 images
Zi've been keeping it 1 to 1 but i've heard that increasing this will improve quality of images
Z
Factually it is also said that opposite
Fso it is a debated thin
Zah ok
Fyou can test different values until you get out of memory error
Zi'll try one with 1 and 1 and one with batch size 5 gradient 2
Fsure
Zit never made sense until someone on reddit made the analogy of it's trying to make a statue out of a boulder. You want to start with a sledgehammer, and then move onto a chisel
Zi'm using a transparent channel on my subject tho so i lowered the learning rate significantly and checked "Use PNG Alpha channel as loss weight"
Fi think you need extra processing for this. it was like that initially "Use PNG Alpha channel as loss weight"
Fi mean pre processing of images