IDEA-CCNL

/

Taiyi-Stable-Diffusion-XL-3.5B

like 59

Follow

Fengshenbang-LM 241

可以的。我们主要的测试都是在Fooocus上做的，因为提供了很多预设的质量词，风格词和negative prompt

有diffusers加载加载代码参考吗

这个和SD-XL加载一样就行https://huggingface.co./stabilityai/stable-diffusion-xl-base-1.0：

from diffusers import StableDiffusionXLPipeline

model_id = "IDEA-CCNL/Taiyi-Stable-Diffusion-XL-3.5B"
pipeline = StableDiffusionXLPipeline.from_pretrained(model_id)

prompt = "A majestic lion jumping from a big stone at night"
image = pipeline(prompt).images[0]

嗯多谢，好像在体验网址上测试效果一般呢，有什么方法提升效果吗

多尝试不同的风格，我们看到有些网友的测试效果挺不错的：

Prompt: tango dancers embracing

Negative Prompt:

Fooocus V2 Expansion:

Styles: ['Artstyle Impressionist'], Performance: Quality

Resolution: (896, 1088), Sharpness: 17.7

Guidance Scale: 8.63, ADM Guidance: (1.5, 0.8, 0.3)

Base Model: taiyi_diffusion_xl.safetensors, Refiner Model: None

Refiner Switch: 0.5, Sampler: euler_ancestral

Scheduler: karras, Seed: 2907879249181077976

嗯，主要想测试下中文效果

2024-02-01_12-07-06_6017.png

Prompt: 探戈舞者抱在一起

Negative Prompt:

Fooocus V2 Expansion:

Styles: ['Watercolor 2'], Performance: Speed

Resolution: (1216, 832), Sharpness: 0

Guidance Scale: 7, ADM Guidance: (1.5, 0.8, 0.3)

Base Model: taiyi_diffusion_xl.safetensors, Refiner Model: None

Refiner Switch: 0.5, Sampler: euler_ancestral

Scheduler: karras, Seed: 4240181031045946292

hi,我发现我用这套代码无法推出正常的图片，是因为有哪些库不对应么，我用diffusers=0.24.0或者更高都不行
from diffusers import StableDiffusionXLPipeline

model_id = "IDEA-CCNL/Taiyi-Stable-Diffusion-XL-3.5B"
pipeline = StableDiffusionXLPipeline.from_pretrained(model_id)

prompt = "A majestic lion jumping from a big stone at night"
image = pipeline(prompt).images[0]

我们发布训练代码了：

https://zhuanlan.zhihu.com/p/689784367
https://github.com/IDEA-CCNL/Taiyi-Diffusion-XL