Skip to content

fix: Fix image_seq_len in Lumina2Pipeline (channel count → patch tokens) #14514

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
Ricardo-M-L wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: Fix image_seq_len in Lumina2Pipeline (channel count → patch tokens) #14514

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
7 changes: 6 additions & 1 deletion src/diffusers/pipelines/lumina2/pipeline_lumina2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -640,7 +640,12 @@ def __call__(

# 5. Prepare timesteps
sigmas = np.linspace(1.0, 1 / num_inference_steps, num_inference_steps) if sigmas is None else sigmas
image_seq_len = latents.shape[1]
# `image_seq_len` is the number of patch tokens the transformer will see, not the
# number of latent channels. Lumina2's latents are unpacked `(B, C, H, W)`, so
# `latents.shape[1]` gives the channel count (typically 16) instead of the post-patch
# sequence length. The transformer patchifies with `config.patch_size` (default 2).
patch_size = self.transformer.config.patch_size
image_seq_len = (latents.shape[-2] // patch_size) * (latents.shape[-1] // patch_size)
mu = calculate_shift(
image_seq_len,
self.scheduler.config.get("base_image_seq_len", 256),
Expand Down
Loading