Skip to content

timestep scheduling with np.linspace #8623

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

Open
wants to merge 6 commits into
base: dev
Choose a base branch
from
Open

timestep scheduling with np.linspace #8623

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
b96ec40
timestep scheduling with np.linspace
ytl0623 Nov 5, 2025
3a2ffc5
remove step_ratio variable
ytl0623 Nov 5, 2025
aa3cce6
steps_offset causes out-of-bounds timestep indices
ytl0623 Nov 5, 2025
616f5f2
Refactor timesteps calculation in ddim.py
ytl0623 Nov 5, 2025
f7ca165
Fix timesteps calculation in ddpm.py
ytl0623 Nov 5, 2025
ef7e83c
With the linspace approach, max_timestep = (num_train_timesteps - 1 -…
ytl0623 Nov 6, 2025
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
19 changes: 8 additions & 11 deletions monai/networks/schedulers/ddim.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,17 +117,14 @@ def set_timesteps(self, num_inference_steps: int, device: str | torch.device | N
)

self.num_inference_steps = num_inference_steps
step_ratio = self.num_train_timesteps // self.num_inference_steps
if self.steps_offset >= step_ratio:
raise ValueError(
f"`steps_offset`: {self.steps_offset} cannot be greater than or equal to "
f"`num_train_timesteps // num_inference_steps : {step_ratio}` as this will cause timesteps to exceed"
f" the max train timestep."
)

# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
timesteps = (np.arange(0, num_inference_steps) * step_ratio).round()[::-1].copy().astype(np.int64)
if self.steps_offset < 0 or self.steps_offset >= self.num_train_timesteps:
raise ValueError(f"`steps_offset`: {self.steps_offset} must be in range [0, {self.num_train_timesteps}).")

timesteps = (
np.linspace((self.num_train_timesteps - 1) - self.steps_offset, 0, num_inference_steps)
.round()
.astype(np.int64)
)
self.timesteps = torch.from_numpy(timesteps).to(device)
self.timesteps += self.steps_offset

Expand Down
5 changes: 1 addition & 4 deletions monai/networks/schedulers/ddpm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,10 +122,7 @@ def set_timesteps(self, num_inference_steps: int, device: str | torch.device | N
)

self.num_inference_steps = num_inference_steps
step_ratio = self.num_train_timesteps // self.num_inference_steps
# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
timesteps = (np.arange(0, num_inference_steps) * step_ratio).round()[::-1].astype(np.int64)
timesteps = np.linspace(self.num_train_timesteps - 1, 0, self.num_inference_steps).round().astype(np.int64)
self.timesteps = torch.from_numpy(timesteps).to(device)

def _get_mean(self, timestep: int, x_0: torch.Tensor, x_t: torch.Tensor) -> torch.Tensor:
Expand Down
Loading