- Using Kohya SS GUI as trainer script.
- Rented 8 GPU machine from RunPod and already doing 8 different trainings right now.
- If you have any input / feedback / suggestion let me know please.



Stable Diffusion 1.5 training experiments with Kohya SS GUI to find the best hyperparameters, running 8 trainings at once on an 8 GPU RunPod machine.
Started empirically experimenting to find the best training hyper parameters for Stable Diffusion 1.5 based models, shared as a progress note with screenshots and a request for feedback.




Kohya GUI configs for SD 1.5 based models DreamBooth / full fine-tuning, the very best of over 70 trainings, in three tiers from 10.2 GB to 22 GB VRAM.
Qwen Image 2512 is the new king for training: the video compares trained Base, 2512, Edit 2509 and Edit 2511 models and shows a big leap in realism.
Kohya and OneTrainer compared for FLUX Dev fine-tuning: OneTrainer with EMA and a copied Kohya config gave weaker results than the best Kohya config.
X's For You recommendation algorithm got a bigger update than it looks: commit e414c17 changed 187 files with 18,263 additions across the feed stack.
BF16, GGUF Q8, FP8 Scaled and NVFP4 precisions compared for speed and quality in ComfyUI with CUDA 13; NVFP4 ran 87% faster than GGUF Q8 on Z Image Turbo.
NVFP4 models run over 100% faster in ComfyUI and SwarmUI with CUDA 13; this full tutorial compares NVFP4 quality on FLUX 2, Z Image Turbo and FLUX 1 in grids.