Text-to-Image
Diffusers
stable-diffusion

Standard Configuration

#36
by StarGazer-Media - opened

Can you provide the best configuration for this model:
Scheduler:
num_inference_step:
Guidance_scale:
Image Resolution:

ByteDance org

Theoretically the model is trained for the following target settings:

  • scheduler: Euler, with trailing timestep (sgm_uniform)
  • num_inference_step: use the same inference step as the checkpoint.
  • guidance_scale: 1
  • resolution: 1024x1024

Sign up or log in to comment