Skip to content

[2737]: Add DPMSolverMultistepScheduler to CLIP guided community pipeline #2779

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Mar 23, 2023
Merged

[2737]: Add DPMSolverMultistepScheduler to CLIP guided community pipeline #2779

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
94fb856
[2737]: Add DPMSolverMultistepScheduler to CLIP guided community pipe…
Mar 22, 2023
5362102
Merge branch 'main' into njindal/2737
patrickvonplaten Mar 23, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 4 additions & 8 deletions examples/community/clip_guided_stable_diffusion.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,7 @@
AutoencoderKL,
DDIMScheduler,
DiffusionPipeline,
DPMSolverMultistepScheduler,
LMSDiscreteScheduler,
PNDMScheduler,
UNet2DConditionModel,
Expand Down Expand Up @@ -63,7 +64,7 @@ def __init__(
clip_model: CLIPModel,
tokenizer: CLIPTokenizer,
unet: UNet2DConditionModel,
scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler],
scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler, DPMSolverMultistepScheduler],
feature_extractor: CLIPImageProcessor,
):
super().__init__()
Expand Down Expand Up @@ -125,17 +126,12 @@ def cond_fn(
):
latents = latents.detach().requires_grad_()

if isinstance(self.scheduler, LMSDiscreteScheduler):
sigma = self.scheduler.sigmas[index]
# the model input needs to be scaled to match the continuous ODE formulation in K-LMS
latent_model_input = latents / ((sigma**2 + 1) ** 0.5)
else:
latent_model_input = latents
Comment on lines -128 to -133
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you mind elaborating on this change?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This looks correct! This should not be done in the pipeline :-)

latent_model_input = self.scheduler.scale_model_input(latents, timestep)

# predict the noise residual
noise_pred = self.unet(latent_model_input, timestep, encoder_hidden_states=text_embeddings).sample

if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler)):
if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler)):
alpha_prod_t = self.scheduler.alphas_cumprod[timestep]
beta_prod_t = 1 - alpha_prod_t
# compute predicted original sample from predicted noise also called
Expand Down
12 changes: 4 additions & 8 deletions examples/community/clip_guided_stable_diffusion_img2img.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
AutoencoderKL,
DDIMScheduler,
DiffusionPipeline,
DPMSolverMultistepScheduler,
LMSDiscreteScheduler,
PNDMScheduler,
UNet2DConditionModel,
Expand Down Expand Up @@ -140,7 +141,7 @@ def __init__(
clip_model: CLIPModel,
tokenizer: CLIPTokenizer,
unet: UNet2DConditionModel,
scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler],
scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler, DPMSolverMultistepScheduler],
feature_extractor: CLIPFeatureExtractor,
):
super().__init__()
Expand Down Expand Up @@ -263,17 +264,12 @@ def cond_fn(
):
latents = latents.detach().requires_grad_()

if isinstance(self.scheduler, LMSDiscreteScheduler):
sigma = self.scheduler.sigmas[index]
# the model input needs to be scaled to match the continuous ODE formulation in K-LMS
latent_model_input = latents / ((sigma**2 + 1) ** 0.5)
else:
latent_model_input = latents
latent_model_input = self.scheduler.scale_model_input(latents, timestep)

# predict the noise residual
noise_pred = self.unet(latent_model_input, timestep, encoder_hidden_states=text_embeddings).sample

if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler)):
if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler)):
alpha_prod_t = self.scheduler.alphas_cumprod[timestep]
beta_prod_t = 1 - alpha_prod_t
# compute predicted original sample from predicted noise also called
Expand Down