Skip to content

fix(dreambooth): correct batch handling in _prepare_image_ids for flux2 img2img #13821

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
Jefsky wants to merge 1 commit into
base: main
Choose a base branch
from
+9 −6
Open

fix(dreambooth): correct batch handling in _prepare_image_ids for flux2 img2img #13821

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
15 changes: 9 additions & 6 deletions examples/dreambooth/train_dreambooth_lora_flux2_img2img.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1717,12 +1717,15 @@ def get_sigmas(timesteps, n_dim=4, dtype=torch.float32):
cond_model_input = (cond_model_input - latents_bn_mean) / latents_bn_std

model_input_ids = Flux2Pipeline._prepare_latent_ids(model_input).to(device=model_input.device)
cond_model_input_list = [cond_model_input[i].unsqueeze(0) for i in range(cond_model_input.shape[0])]
cond_model_input_ids = Flux2Pipeline._prepare_image_ids(cond_model_input_list).to(
device=cond_model_input.device
)
cond_model_input_ids = cond_model_input_ids.view(
cond_model_input.shape[0], -1, model_input_ids.shape[-1]
# Generate image IDs for a single sample, then expand across batch.
# Using _prepare_image_ids on individual batch elements assigns different
# temporal IDs per sample (T=10, T=20, ...) which is incorrect — batch
# elements are independent samples and should share the same temporal id.
cond_model_input_ids = Flux2Pipeline._prepare_image_ids(
[cond_model_input[0:1]]
).to(device=cond_model_input.device)
cond_model_input_ids = cond_model_input_ids.expand(
cond_model_input.shape[0], -1, -1
)

# Sample noise that we'll add to the latents
Expand Down
Loading