FLUX LoRA Single Block Training Research: Network Alpha & Rank Tests
FLUX LoRA single block / layer training tested in Kohya GUI: one block needed over 10 times the usual learning rate and still gave sub-par results.

Highlights
Research results also show the impact of a LoRA Network Alpha change when lowering LoRA Network Rank (Dimension). All runs used 15 images, 150 epochs and an RTX A6000 48 GB GPU, with full-resolution figures on Hugging Face.
Video tutorial
No Powerful GPU? Train a FLUX LoRA in the Cloud in Under 1 Hour (2024 Kohya Guide) · chapters and summary
More videos: Train FLUX on Your Own Photos With Just 8GB VRAM: Kohya GUI LoRA Tutorial (2024, Windows)
Related

Qwen Image 2512 Training Comparison: The Best Model to Train
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.

FLUX 2 vs FLUX SRPO & New Kohya SS GUI Premium FLUX Training App
FLUX 2 Pro compared with a trained FLUX SRPO model, plus the new Kohya trainer V35 that trains FLUX on GPUs with as little as 6 GB VRAM.

FLUX LoRA vs Fine Tuning / DreamBooth Comparison: Batch Size 1 vs 7
FLUX LoRA and Fine Tuning / DreamBooth compared with 15 and 256 training images at batch size 1 and 7, for both realism and stylization.