Instructions to use F16/krea2-turbo-sda with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use F16/krea2-turbo-sda with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("krea/Krea-2-Turbo", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("F16/krea2-turbo-sda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
F佬,多LORA人物崩肢体严重
#6
by junguang - opened
F佬,LORA一多挂,尤其是几个给K2洗脑的LORA 加上各种NSFW的lORA,导致肢体就崩的比较厉害,各种多残 肢,非要挂10个LORA,那么强度只能开到0.3了(但是多样性几乎消失)
我现在想了法子用低显存LORA (用于绕过输出NSFW检查)+ 角色LORA勉强可以用
junguang changed discussion title from 666!我这就去把聪明种子节点删了~ to F佬,请问这个支持K2 SPEED节点吗?
junguang changed discussion title from F佬,请问这个支持K2 SPEED节点吗? to 666,成功了!
DiversityBoost的方法不适合krea2 turbo,所以我又把SDA翻出来更新了一下,后面还会不会继续改进DiversityBoost不好说,但起码暂时没有计划
说起来这些都只是”打补丁“式临时用用的东西,随着蒸馏方法的改进,早晚不再需要这些。AI时代,今天不学,明天可能就不需要学 :-)
junguang changed discussion status to closed
junguang changed discussion title from 666,成功了! to F佬,多LORA堆崩肢体
junguang changed discussion status to open
junguang changed discussion title from F佬,多LORA堆崩肢体 to F佬,多LORA人物崩肢体严重