Unverified Commit 622f35b1 authored by Ömer Veysel Çağatan's avatar Ömer Veysel Çağatan Committed by GitHub
Browse files

fixed vae scaling (#5213)

parent 39baf0b4
......@@ -114,6 +114,8 @@ class LDMPipeline(DiffusionPipeline):
# compute the previous noisy sample x_t -> x_t-1
latents = self.scheduler.step(noise_prediction, t, latents, **extra_kwargs).prev_sample
# adjust latents with inverse of vae scale
latents = latents / self.vqvae.config.scaling_factor
# decode the image latents with the VAE
image = self.vqvae.decode(latents).sample
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment