sd-scripts

sd-scripts copied to clipboard

very bad results on saved checkpoits

2

Hello, im training on JuggernautXL and while im training the results are insanely good in the samples.. But once the checkpoint is saved, it looks very very very bad. Like...

Deathawaits4

Failed to start DreamBooth training: `INFO prepare tokenizer`

5

``` /usr/local/lib/python3.10/dist-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead. _torch_pytree._register_pytree_node( /usr/local/lib/python3.10/dist-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead. _torch_pytree._register_pytree_node( The following values were not passed to `accelerate launch`...

nickkolok

Soft min SNR gamma

3

In "[Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformers](https://arxiv.org/abs/2401.11605)" they came up with Soft min SNR gamma which smooths out the transition area. ![Screenshot 2024-01-23 at 23-13-30 Scalable High-Resolution...

rockerBOO

failed run kohya for training lora model

8

hello guys, i have an issue running the kohya, please someone give me a solution for this, thank you this is the error message:    06:34:57-522654 INFO Start...

afifulinuha

Broken training that appeared suddenly

4

I get this issue after training (I'm using kohya, but manually calling the training script: ```python Traceback (most recent call last): File "C:\Users\waxel\kohya\kohya_ss\sd-scripts\sdxl_train_network.py", line 184, in trainer.train(args) File "C:\Users\waxel\kohya\kohya_ss\sd-scripts\train_network.py", line...

axel578

How to change HuggingFace directory

10

I'm sorry to use an issue for this. My C: drive is very full and the hugginface cache is using about 30Gb so I want to move it to one...

mbastias

Training sample parameters

4

Hi i am currently training loras with sdxl 1.0. The sample images during training are flawless, hovewer if i rty to use the resulting lora i become disorted results, for...

shockman100

Style and concept separation training https://github.com/yardenfren1996/B-LoRA

sdbds

Feature request: different batch size per GPU

1

Assume you have two GPUs: 1x 3090 24GB and 1x 3060 12GB. When using DDP or DeepSpeed, the 3090 will not be able to use more than 12GB and will...

Trojaner

### Reproduce Using v1-5-pruned.safetensors in https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main Add ``` import copy copy.deepcopy(unet) ``` after this line https://github.com/kohya-ss/sd-scripts/blob/f9317052edb4ab3b3c531ac3b28825ee78b4a966/gen_img.py#L1556 ### Error stack trace ``` File "/home/tsogzark/miniconda3/envs/kohyass/lib/python3.10/copy.py", line 172, in deepcopy y = _reconstruct(x,...

tsogzark