Skip to content

[Dreambooth] Add a function to log interim generation #1711

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

Closed
wants to merge 4 commits into from
Closed

[Dreambooth] Add a function to log interim generation #1711

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
cb2acab
add func logging interim generation
discus0434 Dec 15, 2022
84cdc3e
formatted with black
discus0434 Dec 15, 2022
c09698a
improve generator seeding
discus0434 Dec 15, 2022
4258eb8
align code style (maybe)
discus0434 Dec 15, 2022
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
24 changes: 24 additions & 0 deletions examples/dreambooth/train_dreambooth.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,12 @@ def parse_args(input_args=None):
" training using `--resume_from_checkpoint`."
),
)
parser.add_argument(
"--generating_progress_steps",
type=int,
default=500,
help=("Save an image generated from the model every X steps."),
)
parser.add_argument(
"--resume_from_checkpoint",
type=str,
Expand Down Expand Up @@ -416,6 +422,11 @@ def main(args):

if args.seed is not None:
set_seed(args.seed)
g = torch.Generator(device=accelerator.device.type)
g.manual_seed(args.seed)
else:
g = torch.Generator(device=accelerator.device.type)
g.manual_seed(42)

if args.with_prior_preservation:
class_images_dir = Path(args.class_data_dir)
Expand Down Expand Up @@ -739,6 +750,19 @@ def main(args):
accelerator.save_state(save_path)
logger.info(f"Saved state to {save_path}")

if global_step % args.generating_progress_steps == 0:
if accelerator.is_main_process:
# generate and save the image
pipeline = DiffusionPipeline.from_pretrained(
args.pretrained_model_name_or_path,
unet=accelerator.unwrap_model(unet),
text_encoder=accelerator.unwrap_model(text_encoder),
torch_dtype=weight_dtype,
).to(accelerator.device.type)

image = pipeline(args.instance_prompt, generator=g).images[0]
image.save(f"{args.output_dir}/logs/image_{global_step}.png")

logs = {"loss": loss.detach().item(), "lr": lr_scheduler.get_last_lr()[0]}
progress_bar.set_postfix(**logs)
accelerator.log(logs, step=global_step)
Expand Down