Text-to-Image
Diffusers
Safetensors
StableDiffusionXLPipeline
stable-diffusion
sdxl
flash
sdxl-flash
lightning
turbo
lcm
hyper
fast
fast-sdxl
sd-community
Instructions to use sd-community/sdxl-flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use sd-community/sdxl-flash with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("sd-community/sdxl-flash", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
Loss of diversity
#3
by jychoi - opened
Even with the CFG scale set to 1 to disable guidance, it seems the model still generates similar high-quality images, suggesting that diversity has been significantly sacrificed. Could you share how this model was trained?
Even with the CFG scale set to 1 to disable guidance, it seems the model still generates similar high-quality images, suggesting that diversity has been significantly sacrificed. Could you share how this model was trained?
you can use COMFY's automatic CFG node, set high CFG to restore any accelerated model to multiple artistic styles (without affecting the acceleration effect).



