StableLlama
StableLlama
AuraFlow is a new completely open source text2img model: https://huggingface.co/fal/AuraFlow It would be great when we could use kohya-ss to also train LoRAs and finetunes of AuraFlow.
There is a new paper, "u-μP: The Unit-Scaled Maximal Update Parametrization" at https://arxiv.org/abs/2407.17465 that promises selection of optimal hyperparameters, like the learning rate, even when using very small datatypes like...
### Feature Idea A feature for A1111 that I like was that on every new day a new folder with the current date is created in the output directory and...
As seen at https://huggingface.co/spaces/fancyfeast/joy-caption-pre-alpha it is a model with superb captioning skills. It output was nearly creating the test image in Flux again. So it would be great when it...
Doing some interactive work it would be great when I could drag an image (either from the image list or the currently shown image) to an external program that is...
Error message: ``` Loading microsoft/Florence-2-large-ft... Traceback (most recent call last): File "/home/abc/StableDiffusion/taggui/taggui/auto_captioning/captioning_thread.py", line 145, in run raise exception File "/home/abc/StableDiffusion/taggui/taggui/auto_captioning/captioning_thread.py", line 141, in run self.run_captioning() File "/home/abc/StableDiffusion/taggui/taggui/auto_captioning/captioning_thread.py", line 88, in...
### Is there an existing issue for this? - [X] I have searched the existing issues and checked the recent builds/commits ### What would your feature do ? Today https://blackforestlabs.ai/...
As written at https://www.reddit.com/r/StableDiffusion/comments/1f523bd/good_flux_loras_can_be_less_than_45mb_128_dim/ it can be sufficient just to train `single_transformer_blocks.7.proj_out` to get a very good and very small character LoRA. It would be great when I could do...
Working with models (esp. Flux) that use natural language captioning the current tag based view isn't easily readable, so I'd like to have the possibility to read the full caption...
It would be great to be able to have the image grid transposed. Currently having a batch = 4 and a few image / latent combine will always make a...