Skip to content

Fix multistage timestep boundary error causing training crashes #519

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
relaxis wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix multistage timestep boundary error causing training crashes #519

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 9 additions & 1 deletion jobs/process/BaseSDTrainProcess.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1265,7 +1265,15 @@ def process_general_training_batch(self, batch: 'DataLoaderBatchDTO'):
with self.timer('convert_timestep_indices_to_timesteps'):
# convert the timestep_indices to a timestep
timesteps = self.sd.noise_scheduler.timesteps[timestep_indices.long()]


# Clamp timesteps to multistage boundary range to prevent floating point edge cases
if self.sd.is_multistage:
boundaries = [1] + self.sd.multistage_boundaries
boundary_max, boundary_min = boundaries[self.current_boundary_index], boundaries[self.current_boundary_index + 1]
# Ensure timesteps stay strictly within (boundary_min * 1000, boundary_max * 1000) range
# Subtract small epsilon from max to avoid edge case where timestep equals boundary
timesteps = torch.clamp(timesteps, min=boundary_min * 1000, max=(boundary_max * 1000) - 0.01)

with self.timer('prepare_noise'):
# get noise
noise = self.get_noise(latents, batch_size, dtype=dtype, batch=batch, timestep=timesteps)
Expand Down