Hello everyone. I am Dr. Furkan Gözükara. PhD Computer Engineer. SECourses is a dedicated YouTube channel for the following topics : Tech, AI, News, Science, Robotics, Singularity, ComfyUI, SwarmUI, ML, Artificial Intelligence, Humanoid Robots, Wan 2.2, FLUX, Krea, Qwen Image, VLMs, Stable Diffusion
And I also have a question about training full dreambooth checkpoint @Furkan Gözükara SECourses. Since you said training person with dreambooth should separate into 2 part of training. Part 1 to take the epoch which is flexible to style. Part 2 is train new epoch with the base is flexible style model from part 1. What is the different with continue training compare with separte with 2 parts like this. since we used constant learning rate. For example: Part 1 - I choose epoch 30 to use as based model to retrain. Part 2 - I train 100 more epoch with based model in Part 1. But that just similar to train 130 epoch from Part 1 right ? Because we use constant learning rate.
So as a general rule, the more training images you have of 1 character, the less epochs you would need to get good result. Am I getting the logic right? So 300 pics you might get good result with 10 or 20 epochs, but with 15 pics, you will need more epochs, like 100-200.
@Furkan Gözükara SECourses are you in contact with anyone who is doing large finetunes like RealVis/Juggernaut, and is attempting to do the same for Flux? I might have something that helps.
@Furkan Gözükara SECourses ss_network_args": "{"loraplus_lr_ratio": "16", "loraplus_text_encoder_lr_ratio": "4", "dropout": null}" do you know what is this? if know how to do this?
@Furkan Gözükara SECourses Doc, When preparing the images, is the resizing step necessary? I cropped using your tool and I'm not training on faces. Most images to train on are larger than 1024p and in various aspect ratios. in that case, should I change the max resolution setting in Kohya?
In this tutorial, I am going to show you how to install OneTrainer from scratch on your computer and do a Stable Diffusion SDXL (Full Fine-Tuning 10.3 GB VRAM) and SD 1.5 (Full Fine-Tuning 7GB VRAM) based models training on your computer and also do the same training on a very cheap cloud machine from MassedCompute if you don't have such compute...
If you want to train FLUX with maximum possible quality, this is the tutorial looking for. In this comprehensive tutorial, you will learn how to install Kohya GUI and use it to fully Fine-Tune / DreamBooth FLUX model. After that how to use SwarmUI to compare generated checkpoints / models and find the very best one to generate most amazing image...