Text-to-Image
Diffusers
Safetensors
StableDiffusionPipeline
stable-diffusion
Inference Endpoints

How do I use a seed with the prompt?

#20
by LaferriereJC - opened

I've tried
Seed:
Seed =
and
using the seed number #

none appear to work.

Please try
import torch
seed = 512 #any
generator = torch.Generator("cuda").manual_seed(seed)
out = pipe(prompt, generator=generator).images[0]

Sign up or log in to comment