sd-scripts

sd-scripts copied to clipboard

FLUX training for 8GB VRAM?

12

I've tried the options for 12G, 16G, and 20G VRAM options here: https://github.com/kohya-ss/sd-scripts/tree/sd3?tab=readme-ov-file#flux1-lora-training and confirm they all work. But is it possible to do 8GB VRAM? Is there a specific...

cocktailpeanut

Specify blocks to Train w/ Finetuning?

1

Since it appears that Flux LoRA training can still be effective when only training specific layers, I am wondering if this functionality can be expanded to Finetuning, since this is...

setothegreat

FLUX LoRA Training - Text Encoder active training indicated when it should be disabled with --network_train_unet_only

2

The log for my training shows: INFO create LoRA network. base dim (rank): 128, alpha: 128 lora_flux.py:594 INFO neuron dropout: p=0.25, rank dropout: p=None, module dropout: p=None lora_flux.py:595 INFO split...

Enyakk

Thanks!

mostlyhuman

a67793581

Been mentioned that is one of Flux's greatest downfalls, but I wonder where it comes from (and how to fix it)? In my training I train one style I demand...

DarkAlchy

Flux Schnell lora size is too small (1.42 MB) and Does not generate images of the character that it was trained on.

3

I trained a custom character model using flux schnell, that training was success however the resulting lora is too small (1.42mb). And When I use that lora on the minimal...

Hassan-jr

Still actual: https://github.com/kohya-ss/sd-scripts/issues/280

lynrayy

Multi GPU train of flux Have Some Bugs

23

Multi GPU train flux, Is the script of flux train not supported now?

sx0404

essentially the title. i've tried the following: ``` transformer = diffusers.Flux2DTransformer.from_single_file(path_to_kohya_safetensors, torch.dtype=torch.bfloat16) pipeline = diffusers.FluxPipeline.from_pretrained(..., transformer=transformer, torch.dtype=torch.bfloat16) ``` however all generated images have weird artifacts, regardless of checkpoint number i've...

neuron-party