Mirrors
/
diffusers
mirror of https://github.com/ShivamShrirao/diffusers.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix MPS scheduler indexing when using `mps` (#450) 

* Fix LMS scheduler indexing in `add_noise` #358.

* Fix DDIM and DDPM indexing with mps device.

* Verify format is PyTorch before using `.to()`
This commit is contained in:
Pedro Cuenca 2022-09-14 14:33:37 +02:00 committed by GitHub
parent 7c4b38baca
commit 1a69c6ff0e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 9 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/diffusers/schedulers/scheduling_ddim.py
View File

@ -250,6 +250,8 @@ class DDIMScheduler(SchedulerMixin, ConfigMixin):
noise: Union[torch.FloatTensor, np.ndarray], noise: Union[torch.FloatTensor, np.ndarray],
timesteps: Union[torch.IntTensor, np.ndarray], timesteps: Union[torch.IntTensor, np.ndarray],
) -> Union[torch.FloatTensor, np.ndarray]: ) -> Union[torch.FloatTensor, np.ndarray]:
if self.tensor_format == "pt":
timesteps = timesteps.to(self.alphas_cumprod.device)
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5 sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples) sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)
sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5 sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5

2
src/diffusers/schedulers/scheduling_ddpm.py
View File

@ -251,6 +251,8 @@ class DDPMScheduler(SchedulerMixin, ConfigMixin):
noise: Union[torch.FloatTensor, np.ndarray], noise: Union[torch.FloatTensor, np.ndarray],
timesteps: Union[torch.IntTensor, np.ndarray], timesteps: Union[torch.IntTensor, np.ndarray],
) -> Union[torch.FloatTensor, np.ndarray]: ) -> Union[torch.FloatTensor, np.ndarray]:
if self.tensor_format == "pt":
timesteps = timesteps.to(self.alphas_cumprod.device)
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5 sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples) sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)
sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5 sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5

4
src/diffusers/schedulers/scheduling_lms_discrete.py
View File

@ -120,7 +120,7 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
frac = np.mod(self.timesteps, 1.0) frac = np.mod(self.timesteps, 1.0)
sigmas = np.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5) sigmas = np.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5)
sigmas = (1 - frac) * sigmas[low_idx] + frac * sigmas[high_idx] sigmas = (1 - frac) * sigmas[low_idx] + frac * sigmas[high_idx]
self.sigmas = np.concatenate([sigmas, [0.0]]) self.sigmas = np.concatenate([sigmas, [0.0]]).astype(np.float32)
self.derivatives = [] self.derivatives = []
@ -183,6 +183,8 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
noise: Union[torch.FloatTensor, np.ndarray], noise: Union[torch.FloatTensor, np.ndarray],
timesteps: Union[torch.IntTensor, np.ndarray], timesteps: Union[torch.IntTensor, np.ndarray],
) -> Union[torch.FloatTensor, np.ndarray]: ) -> Union[torch.FloatTensor, np.ndarray]:
if self.tensor_format == "pt":
timesteps = timesteps.to(self.sigmas.device)
sigmas = self.match_shape(self.sigmas[timesteps], noise) sigmas = self.match_shape(self.sigmas[timesteps], noise)
noisy_samples = original_samples + noise * sigmas noisy_samples = original_samples + noise * sigmas

2
src/diffusers/schedulers/scheduling_pndm.py
View File

@ -367,7 +367,7 @@ class PNDMScheduler(SchedulerMixin, ConfigMixin):
noise: Union[torch.FloatTensor, np.ndarray], noise: Union[torch.FloatTensor, np.ndarray],
timesteps: Union[torch.IntTensor, np.ndarray], timesteps: Union[torch.IntTensor, np.ndarray],
) -> torch.Tensor: ) -> torch.Tensor:
# mps requires indices to be in the same device, so we use cpu as is the default with cuda if self.tensor_format == "pt":
timesteps = timesteps.to(self.alphas_cumprod.device) timesteps = timesteps.to(self.alphas_cumprod.device)
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5 sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples) sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)