Skip to content

Cast ControlNet image columns before preprocessing #14565

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
medisean wants to merge 1 commit into
base: main
Choose a base branch
from
+7 −0
Open

Cast ControlNet image columns before preprocessing #14565

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: 7 additions & 0 deletions examples/controlnet/train_controlnet.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@
from accelerate import Accelerator
from accelerate.logging import get_logger
from accelerate.utils import ProjectConfiguration, set_seed
from datasets import Image as DatasetImage
from datasets import load_dataset
from huggingface_hub import create_repo, upload_folder
from packaging import version
Expand Down Expand Up @@ -657,6 +658,12 @@ def make_train_dataset(args, tokenizer, accelerator):
f"`--conditioning_image_column` value '{args.conditioning_image_column}' not found in dataset columns. Dataset columns are: {', '.join(column_names)}"
)

# Local copies of datasets such as fill50k may store image paths as
# strings, while the preprocessing below expects decoded PIL images.
for column in (image_column, conditioning_image_column):
if not isinstance(dataset.features[column], DatasetImage):
dataset = dataset.cast_column(column, DatasetImage())

def tokenize_captions(examples, is_train=True):
captions = []
for caption in examples[caption_column]:
Expand Down
Loading